David Greenfield, automation scientist
0:13
Welcome to the world of automation, get your questions answered, as we connect with industry experts to get the answers you need about industrial automation technologies. Find more answers by subscribing to Automation World at subscribeaw.com. I’m David Greenfield, Editor-in-Chief at Automation World. The topic we will cover in this episode is understanding equipment failures and how to respond to them. Joining me in answering this question is Stephen Lacy with Belden Corporation, a supplier of industrial automation cables, connectors, networking, and cybersecurity technologies. So thanks for joining me, Stephen.
Stephen Lacey, Belden
0:51
Thank you, David. It’s nice to be here.
David Greenfield, automation scientist
0:53
So, Stephen, you know, although of course different types of equipment will fail for different reasons. What do you see as the main mechanical, electrical or used causes of equipment failure in industrial processes.
Stephen Lacey, Belden
1:09
We’ve seen some heavy hitters out there. The lifespan of automation and electrical equipment is 15, 20, 25 years and it’s aging fast. We see high absolute temperatures either in a container or in the production environment itself. The daily seasonal temperature cycles, places mechanical stresses on the chips, and desiccates the condensate. Dust accumulates on casing filters and printed circuit boards. It is a bit conductive, which helps speed up failures. Then finally the vibration. For equipment that has stopped on machinery, whether it is slight or high vibration, it wears out the connector, contact points and solder joints over time of mechanical equipment, motors and machine devices. We see failure modes there bearing and joint challenges, overloading of motors, misalignment in the X, Y or Z planes of a driven load connected to two motors or servos. Materials out of specification, many times can cause equipment bottlenecks, and eventually operator errors, quite as a normal part of business practice and production environment management.
David Greenfield, automation scientist
2:39
So, given those types of failures that I just described, are there certain types of equipment that are more likely to fail because of these and other reasons?
Stephen Lacey, Belden
2:48
Yes, we are most familiar with the Pareto chart type of heavy hitters. Again, variable frequency drives, and power supplies, mostly 24 volts. But of course, there are other voltages, and non-industrial rated network switches are very common and can cause a lot of grief, as non-industrial, non-hardened computers are used with HMIs. Ethernet media converters definitely exist, especially the less expensive ones that have an external power supply that plugs directly into the wall, eg protocol converters, they seem to age and when they start to fail it’s kind of the present intermittent intermittent problems that make them more difficult to tell which one It gives you a headache. Then finally, AC and DC motors due to coil tolerance issues or brush failures.
David Greenfield, automation scientist
4:00
So with all of these common causes and types of failure that we’ve discussed here so far, you know, there’s still the fact that many manufacturers still rely on reactive maintenance rather than being more proactive. Based on your experience working with end users and industry, can you talk about typical cost factors associated with this most commonly used interactive approach?
Stephen Lacey, Belden
4:27
Obviously repairing or replacing failed components is the first thing but that is assuming they can be picked up and shipped quickly and we all know this can be very difficult nowadays there is maintenance labor to disassemble to repair and then reassemble the equipment. And depending on where the failed component is located, you know, within the machinery or plant area, that may require working at height, which having a partner has risks associated with it. There is labor in the field of makeup production that may be required during overtime hours. Unplanned maintenance can also contain an element of safety again depending on the risk involved in random breakdowns that put the facility in an abnormal operating condition for which the operations team may not be prepared. So, you definitely want to know, and you want to avoid that if you win when you can. Equipment failures can significantly lower KPIs in a plant or facility. Unreliable production equipment can create uncertainty about whether production targets can be achieved. So the goal here is to really make sure that the production environment is configured and robust enough. So achieving production goals can only depend on personnel performance, not on machine or equipment issues. So you definitely want to get rid of those. Finally, production delays are always a concern, as it may cause your customer to try a competing product.
David Greenfield, automation scientist
6:28
This is an area you know and want to get away from.
For all the obvious reasons. They sure are, yeah. You know, now, you know, we’re talking about reactive maintenance versus proactive maintenance. But, you know, across the industry right now, there’s a lot of interest in predictive maintenance to allow for higher levels of proactive maintenance. But is it possible without specialized predictive maintenance software to predict when equipment will fail?
Stephen Lacey, Belden
6:52
It is, yes, that, the conditional statement, but it’s getting better and better and easier all the time. We first start by identifying all production components and critical equipment in the automation, network sensors and electrical components world. If the factory facility does not have accurate, accurate drawings or drawings, schematic drawings are fine, perhaps with pictures. Once you know what you have, the next step is to start by installing inexpensive wireless sensors. These are great because they work beyond your existing bandwidth you know and OT networks which are crucial to production. The sensors available nowadays can measure location, current movement, distance, pressure, speed, you name it. So you’re there an inexpensive sensor to determine most, excuse me, most, most, most of any critical parameter. Then, the failed power supplies are replaced with smart monitored power supplies. There are some modules with IO jumper built into them. And some of the parameters included in this link are actually the remaining useful life of the supply, which plays into the predictive nature of exercise analytics. Network monitoring software installed to detect when a network switch may start to malfunction or fail, you will see spikes in network traffic, which are not consistent with anything else and these may indicate an imminent failure. You can install wireless engine vibration and temperature sensors on the engines using an analysis software package. I would just like to point out that the solution ideas described can either run on existing utility networks or they can be on a smaller separate network dedicated solely to equipment monitoring and analytics.
David Greenfield, automation scientist
9:10
So, given all of those choices that manufacturers face, you know when it comes to equipment failure prediction methods, where do you suggest manufacturers start?
Stephen Lacey, Belden
9:21
I’ll start with a quick estimate of the costs associated with downtime for a particular machine or plant area. This is really important because it helps determine your ROI or at least gives you an initial ROI and enables you to quickly get a proof of concept approved internally. If you haven’t started, you know that any data-driven industrial 4.0 or legacy software starts small on a single machine or process area which limits cost and risk. concept results. Then suggest better, faster, and cheaper ways to do the following. This approach also allows time for the utility staff to understand the value of the effort and take advantage of them in the approach.
David Greenfield, automation scientist
So back to the cost issue for all of this earlier, we were talking about reactive maintenance costs, but how expensive would it be to implement these predictive methods that you’re recommending?
Stephen Lacey, Belden
10:37
It’s true in Industry 4.0, and the Industrial Internet of Things by its very nature, is designed not to be expensive, to be able to be small, nimble, and scalable. The main thing here is to start a small process – one machine. You can start by adding a few new sensors to your existing equipment, adding the signals already available from your existing hardware, PLCs, PLCs, and/or DCS if you have one. As the journey continues, you may find additional value-added references from other existing factory systems, such as warehouse management system, integrated enterprise resource planning system, enterprise resource planning system, manufacturing execution systems, quality assurance systems, quality control, laboratory management systems, etc. • Assist in identifying correlations and identifying the root cause of the problem. Separately, small factory floor displays can be added so that all employees know from top to bottom the root causes of production loss. And note that everything described above is usually seller-neutral. So you have a lot of options in the market.
David Greenfield, automation scientist
12:05
To help put all of this into perspective. Stephen, can you share some details about the equipment failure issues that Belden helps its customers with?
Stephen Lacey, Belden
12:13
That’s why we install wireless vibration and temperature sensors on all of our engines and all of our manufacturing facilities. Then machine learning software is used to detect the defects. It analyzes these two variables and then alerts via email or text when they go out when they go beyond their normal operating range. This is one area where we start with one of the customers that we’re working with now, they have a problem with polypropylene plastic conveyor belts that fail a lot, belt failures stop all finished products from leaving the factory and have a significant six-figure cost per year. to process it. We have designed a custom set to monitor belt tension and belt speed, and a small control panel that combines these signals with other VFD signals present across an edge device and sends them to the cloud. We’re in the cloud, building a predictive model for every remaining useful life belt. The model will automatically adjust over time as the causes of belt breaks are identified and corrected. So we’re really excited about those two use cases. We look forward to doing more.
David Greenfield, automation scientist
13:39
Well, very interesting, you know, thank you again for joining me on this podcast, Stephen. And thanks, of course, to all of our listeners. And please keep watching this space for more Automation World installments and to get your questions answered. And remember, you can find us online at automation world.com, subscribe to our print magazine, and subscribe to aw.com to stay up-to-date with the latest industrial automation technology insights, trends, and news.