Why you need machine learning if you’re spending millions on expensive equipment

Back in the 1930s and ‘40s in New York, an Irishman named James ‘Smelly’ Kelly roamed the sewers and subway system checking for leaks. Nowadays, New York’s subways come equipp ed with sensors that sample the air quality and monitor them for build ups.

However, less than a century ago that responsibility belonged entirely to Smelly Kelly.

According to Atlas Obscura, Kelly had an, “uncanny ability to locate leaks that no one else could find.”

Finding small leaks before they turned into major catastrophes, in one instance, Kelly was called in to investigate an odour that was unexplainable, but which was overpowering subway commuters waiting on their platform.

He arrived at the scene, took a few deep breaths through his legendary nose and declared that the source of the smell was elephants. People thought Kelly was crazy because elephants were as rare a sight in the subways then as they are now.

Nevertheless, he was insistent: the source of the smell was elephants.

The elephant in the room

The remarkable thing is that Kelly was absolutely correct. It turns out that the station in question had been built under what had been the New York Hippodrome. It had frequently been the location for circus performances, and layers of elephant dung had been buried at the site. A broken water main subsequently ran through the covered matter, which started to leak into the subway.

But what on earth does Smelly Kelly have to do with machine learning? Well, Smelly Kelly was an outlier. He was one-in-a-million. He could predict things before they happened, and single-handedly prevented any major leaks in the New York underground for three decades.

That skill was incomprehensible in the ‘30s and ‘40s but it’s within the grasp of most organisations these days, because predictive maintenance gives that same power.

Machine learning can forecast when machinery or equipment will need servicing before it breaks down. Predictive machinery is vital, because it lets you plan when to most efficiently make repairs and replacements.

Your goal isn’t zero downtime – as that’s near-impossible.

Instead, it’s zero unplanned downtime.

Driving costs down

Anyone involved in a business where the machinery is expensive understands the cost of maintenance. That cost can be calculated by comparing the maintenance budget with the replacement cost of the machinery.

Let’s say your maintenance budget is €200,000 per year and it costs €10 million to replace the machine. Then your maintenance cost is two percent. (According to the experts, the goal should be anywhere under three percent, but some companies are averaging less than two percent.)

Driving the maintenance cost down makes the machine more efficient, but also has the potential to drastically improve a company’s bottom line.

A study by the US Department of Energy claimed that a, “functional predictive maintenance program can reduce maintenance cost by 30 percent, reduce downtime by 45 percent, and eliminate breakdowns by as much as 75 percent.”

Further, the cost of a comprehensive predictive maintenance programme is relatively inexpensive, especially when compared with the increasingly high costs of industrial machinery.

Greg Fell is the CIO of Terex, which makes industrial cranes and heavy construction equipment. He says that, “the basic idea of predictive maintenance isn’t new. What’s changed is that it’s much less expensive to get data off a machine today than it was in the past. 20 years ago, an accelerometer cost thousands of dollars. Today, every smartphone has one built into it. The cost has fallen dramatically.”

Maintenance savings: a case study

Subway stations have drastically moved on since Smelly Kelly’s day and now come equipped with the latest high-tech ventilation systems to ensure clean air for commuters. These ventilation systems are made up of huge industrial fans that cost upwards of €500,000 apiece and there are 64 such fans in the Chongqing Subway in China.

The traditional method used in China and most subways around the world was a ‘run to failure’ model whereby the machinery was only repaired after it broke down. However, the Chongqing authorities turned to a predictive maintenance model based on machine learning.

The result: less breakdowns and a saving of €9,000 per fan per year, excluding labour costs.

Now that’s a result even Smelly Kelly wouldn’t turn his nose up at.

Do you want to use machine learning to save your expensive machinery?

If you want to harness the power of artificial intelligence in your business, Statwolf’s data science service can help, with advanced online data visualisation and analysis simply running in your web browser. 

We offer a range of custom services to suit your needs: advanced data analysis and modelling, custom algorithm creation and implementation with a particular focus on Predictive Maintenance. 

Our advanced data science consultancy can team up with you to interpret your data and make your business work more efficiently so get in touch today to unleash your business’ potential.

Want to make sense of your data? Download our comprehensive guide: The Pedictive Maintenance Cookbook.

New Call-to-action