If at first you don’t succeed with your Industrial Internet of Things (IIoT) initiative, try, try again. You could also get some helpful insight from those who have already tried, tried again.
If National Oilwell Varco (NOV) had given up after its first go-round (or its second), the oil drilling rig manufacturer would not have the kind of success it’s having today. It likely would not have survived the stark downturn in oil prices without its workforce facing the brunt of the cost reductions.
The end results of NOV’s IIoT efforts are convincing. Along with the rest of the oil and gas industry, NOV’s revenue faced a steep decline when oil prices began falling in 2014. And yet the rig company not only survived, but fluorished through the reduced costs afforded by connected assets, noted Ashe Menon, senior vice president of global manufacturing for NOV, Grant Prideco.
Most convincing: At a customer’s rig in West Texas, its first ever implementation of IIoT led to a reduction in downtime costs of more than 80 percent.
But those successes did not come overnight, and were the result of repeated attempts and reevaluations along the way. There is no easy button for IIoT, Menon said, outlining some key challenges and efforts for attendees of Industry 4.0 ThinkTank, a two-day conference held recently in Chicago.
The challenges for NOV’s customers are many, considering their location in what are typically extremely remote, extremely rugged oil drilling locations with a low level of connectivity. Given the remote locations, it’s very expensive when something breaks. The industry tends to be very technology-averse, particularly if any projects have failed in the past, Menon explained.
Add to that any trepidation about new technologies—not to mention skepticism about the hype behind IIoT—and the concept of connected machines predicting failure could be a further challenge.
“Forget about all the IoT nomenclature. It’s just machines telling us what to do,” Menon advised. “We had to make it simple to make sure we’d succeed. Let’s not waste our time monitoring things that don’t mean anything.”
To begin, NOV analyzed data gathered from a very critical piece of equipment—the top drive. “When it stops working, you’re not drilling,” Menon explained. “Imagine if you had a locomotive engine and stood it on its head; that’s what it is. When it stops working, we have big problems.”
NOV focused in on risk management, asset management and condition monitoring, analyzing failure data, criticality, probability, consequences, etc. “Without understanding that data, it wasn’t going to work,” Menon said.
They started with 10 years worth of data, and finally managed to whittle that down to three years worth of good data. “And then we found the one largest cause of downtime,” Menon said. “Other.” In other words, after all that effort, they knew pretty much nothing.
So they tried again, going back to the point of data entry. But the results were not much better the second time around, with explanations like “break down” or “worn out.” Menon urged, “If you don’t pay attention to what you’re doing on the front end, it will mean nothing on the back end.”
Ultimately, NOV borrowed a lesson from the finance industry, Menon said, describing a call he got from MasterCard while he was on vacation in Las Vegas. They were concerned about his spending habits because it was abnormal activity for him. He explained that it wasn’t normal activity for when he was at home, but it was normal for when he’s in Vegas. The next time Menon was in Vegas, MasterCard didn’t call. “Because now they know the pattern,” he said.
NOV used the same kind of supervised learning to overcome differences in location for its top drives, which are manufactured in California, but shipped all over the world. How a top drive operates in extreme heat, for example, could be very different from how it operates in cold temperatures. “We want to make sure the systems are smart enough to know: When I’m doing this in this place, I don’t need an alarm,” Menon said. “We knew how every single piece of equipment could break, and what was the criticality of failure. So the first time it sends a notification, we train it to say no, this is normal. So they stop getting false alarms.”
The system got more sophisticated over time. “The system could tell you where to look first,” Menon described. “Once it learned it, it wouldn’t tell us where to look; it would just tell us we’re going to have a seal failure.”
They were getting great results, like the savings mentioned earlier in West Texas.
“And then 2014 happened and oil prices cratered,” Menon said. “Oh man, we were so ready for this. But we decided hey, we’ll just use it ourselves.”
Cut to two years later, with oil prices still low, and the company facing a 70 percent drop in revenue. But the IIoT initiative was a success, despite the shaky start. The biggest challenges of getting a project started, Menon said, is the fear of failure. “People say, ‘We don’t understand it. It’s not going to work here,’” he said. “But the onus is on us. We have to keep on trying.”