A Human Centered Approach To Machine Monitoring

With recent advances in IIoT and sensor technology, many manufacturers have implemented machine monitoring programs to extract the most value from their assets.

Modern machine monitoring solutions can do way more than calculate OEE and display engine temperature in real-time. Using AI, smart sensors, and the cloud, machine monitoring can give manufacturers a variety of predictive, diagnostic, and prescriptive analytics. Together, these present a holistic, real-time picture of a machine's performance across an operation.

Well, almost holistic.

Many machine monitoring systems have a glaring blind spot: the human. Given that most machine problems are attributable to usage, not just mechanical factors, machine monitoring needs to account for how operators interact with assets. To be truly effective, machine monitoring should track not just what happens in a machine, but what happens around it.

In short, manufacturing needs a human-centered approach to machine monitoring.

In order to explain what a human-centered approach to machine monitoring looks like, it helps to use a metaphor. Let’s consider a simpler scenario of shared machine usage: two roommates sharing a bike.

Roommate A rides the bike on her commute to work, while roommate B rides the bike for exercise on weekends. A, the taller roommate, needs the seat to be high, so she adjusts it before every ride. She commutes at a leisurely pace on paved roads, and travels 10 total miles in a week.

When roommate B rides the bike, she lowers the seat to match her height. And because she wants vigorous exercise, she rides the bike hard and fast on gravel paths. In a single weekend, she might ride 50-75 miles. At the end of each weekend, she cleans the dirt from the frame and cassettes and puts air in the tires as a courtesy to roommate A.

About every 200 miles, the tires need to be replaced, the brakes tightened, and derailleurs adjusted. Along the way, both roommates perform preventative maintenance by tightening loose nuts, keeping the tires filled, and preventing debris buildup in the chain.

Thanks for bearing with me through the metaphor. Here’s the point: Suppose we had only the data relating to the bike’s performance and maintenance (miles rode, conditions, adjustments, tune-ups). Would we know which factors contributed most to the degradation and loss of efficiency in the bike? If our bike had a top-of-the-line machine monitoring setup, we could monitor the miles ridden, the “uptime” the roommates put on the bike, and the RPM of each as they rode.

Is this information enough to predict potential failures? Is it enough for the roommates to determine what actions to take to extend the life of the bike? For example, how do we know that the roommates were adequately cleaning the bike? Did they always put air in the tires? Is routine maintenance actually performed, and when it’s performed, is it done right? Does one of the roommates put extra pressure on the petal? Do they both store it properly?

What I mean to show is that there is a long list of potential machine usage factors that could contribute to poor machine efficiency and lower than optimal part lifetime.

And this is for a (fairly) basic machine. At this point you should have no problem imagining how this idea transfers to a complex industrial machine like a CNC.

If you’re only monitoring spindle speed, tool condition, motor and ambient temperature, vibration, and other machine parameters, are you really getting the whole picture? Or do you need a way of assuring that the tools were set properly, that changeovers were performed as prescribed, that machines were fully cleaned between runs, and that maintenance was carried out on schedule to specification?

When it comes down to it, monitoring machines means monitoring how humans use machines. At its best, machine monitoring doesn’t just provide a set of numbers about machine performance and health. It gives engineers the information they need to make decisions that bring value to the company. It cuts to root causes.

If you’re not accounting for the largest cluster of factors that account for lagging performance, are you really getting the most out of your machine monitoring?

Tulip’s frontline operations platform gives you full visibility across your operations. Curious how Tulip can help you institute a human-centered approach to machine monitoring? Get in touch for a demo.