Get Your Big Data Walking

I really need to make a point of talking with my neighbors more often. I learn interesting things. Like the time I found out the woman across the street is an accountant for a major machine builder in the Chicago area, and is good friends with the very engineer we needed to talk to for an upcoming feature. Or the time I found out that when I sing while wearing headphones and mowing the lawn, you actually can hear it from inside the next house…with the windows closed…while running the vacuum cleaner.

Today I found out that my next-door neighbor is directly involved in advising industrial clients about how to manage and take advantage of the reams of data coming from their sensors. Who knew? I always thought he was a financial analyst or something along those lines. He wanted to know if I write about said Big Data.

Do we ever! I told him it’s a consistently popular topic among our readers, who are trying to figure out how to deal with the terabytes of data coming off their automation devices and how to take advantage of that data to create, say, predictive rather than reactive maintenance systems. OK, so here’s his concern: His clients come to him thinking that they can magically put all that data in one end and get a predictive maintenance system out the other.

“But whether it’s a machine on an oil rig or a Caterpillar or a Deere or whatever, you need a lot more than just the data coming off the sensor,” he says, explaining that there are so many other factors besides hours of operation that go into predicting failure, like climate data, environmental data, and more. He says that industrial clients are essentially trying to go from crawling straight to running. And they need to learn how to walk first.

At this point in my story, I am obliged to refer you to the standing-in-your-side-yard rule of speaking off the record, so you’ll just have to trust me when I say that he’s in a position to know what he’s talking about—playing a key role with a major software and hardware supplier in our space.

He went on to explain how sensor data might indicate the likelihood of a failure, but without loads of other supporting data and the right tools to analyze that data, you won’t know where that failure is going to occur. He explained: “It’s like if I said to you, ‘Aaron, I can tell you that something in your house is going to fail tomorrow. But I can’t tell you what. So you’ll need to have a plumber on hand, and an electrician, and an HVAC guy, and whoever else you might need to fix whatever happens to go wrong.’”

I had to concede that we as trade journalists have a tendency to get caught up in the Gee Whiz side of technology. I used to specialize in the coverage of lithography for semiconductor manufacturing, where leading-edge technology was using 193 nm light to print 32 nm circuits, and where we liked to talk about which advances could get us to those 6 nm lines that pushed the limits of Moore’s Law. Although the debate was fun, what so many of our readers really needed to know was how to keep their 248 nm lithography tool up and running to get as many logic devices produced as possible.

Big Data—or the Internet of Things or Industrial Internet or what have you—certainly falls in this Gee Whiz category. We like to talk about all the neat things you could do with that data and all the ways that machines in a factory or systems within a network of oil rigs could communicate with each other to ease operations. Jim Pinto rounds up a lot of these concepts in a recent column about Automation Technology Futures. Another column from ARC Advisory Group insists that manufacturers need to stop dragging their heels and embrace such new information technologies.

But my neighbor’s advice? Learn how to walk first. He points to a good example from a forklift manufacturer. They get all the data from their forklifts to see how many hours they’re in operation. Rather than use that data for predictive maintenance, they’re doing something simpler: Seeing when it’s time to send their sales guys out to offer up more forklifts to a company whose current supply is nearly maxed out.

When it comes right down to it, my neighbor said, industrial manufacturers are actually spending a very small percentage of their IT budgets on Big Data. Nonetheless, it remains an intriguing idea.