Hi, I'm Nissa Darbonne, executive editor-at-large for Hart Energy. Thank you for joining us. Today we're visiting with Jon Ludwig. Jon is president and co-founder of Novi Labs. He was just among speakers this afternoon at Hart Energy's Executive Oil Conference & Expo here in Midland, Texas.

Jon, thank you very much for all that we learned from you this afternoon. Among the topics in particular was discussion of AI and machine learning (ML). Just areas that I've been hearing commentary on recently. It was actually, oddly, motors in drilling using AI and machine learning where operators and your drilling contractors are able to predict in advance of a motor failure, so they can trip out and obviously save themselves a lot of time for not having to fish.

Is this kind of application like a critical application of AI? There are so many areas.

Jon Ludwig, president and co-founder, Novi Labs: I mean, yeah, there are uses for AI across all the way from planning to operations. So, pump rates on wells that are operating during the drilling, during the frac. I would argue the most important thing is trying to drill the right well to begin with, so evaluating the financial prospect of each well or pad really that the decisions are being made at this stage. Drilling is a great application. Anything that's really expensive -- and generates a lot of data -- is a very good chance to optimize with AI and machine learning.

So, if you can prevent a massive mistake like "I drilled the wrong well to begin with" or "While I was drilling the well, I destroyed the drillbit and had I known there was a problem about to happen, I could have stopped it" as you said.

And then on the operations side, these wells operate 24/7, so there are lots of opportunities to look at the things that lead to failure or that lead to suboptimal production -- therefore suboptimal returns -- and try to address those things as they're happening.

ND: What are applications currently for AI and machine learning that are ready to roll out that may help an operator reduce, let's say, overhead?

JL: Yeah, so I mean the most expensive thing -- I'll speak specifically on onshore drilling, which is my expertise, -- you do is drill and complete the wells. That's the lion's share of the cost. I think the best application of AI and ML is trying to reduce that cost, and that has nothing to do with reducing people. It's more about increasing the odds of success and decreasing the odds of a cataclysmic failure. Those are probably the best applications.

I would say maybe a second category are what I would call "AI advisors." So these are essentially bot-driven, leveraging all of the data that's ever been created from any drilling activity and putting that at the fingertips of an operator. So whether they're a well operator or lease operator, operating a drilling rig or a frac crew, it can tell them things in real time or give them the ability to ask questions in real time and provide real device.

I mentioned in my talk earlier about this crew change, and that's a real thing. There have definitely been a lot less petroleum engineering graduates just because of the ups and downs of the industry, and they figured other forms of engineering might be better, but that's a real thing.

So if there are less people entering the workforce. How are you going to make the people that are in the workforce as productive as possible? Reduce the risk of failure as much as possible? And replace expertise that is being lost due to retirements and other things where folks are leaving the industry and there are a lot less people around with 30 years of experience or 40 years of experience?

So you have to replace that somehow. All of their experience is reflected in documents and data and all sorts of things, but it's very difficult to make that accessible in such [a way] that it makes a difference in real time. That's really where I think AI applications in particular can help a lot.

ND: How many years into the future do you think AI may be able to find, let's call it, "the next bright spot?"

JL: It is almost kind of hard to tell because you sort of don't know what you don't know.

But you can look at things like -- I was just looking at statistics the other day that my company published on LinkedIn about -- how many more feet per day that every single drilling rig drills. Some of that's due to learning. Some of that is due to the sophistication of AI- and ML-based applications that are telling people, "Hey, if this combination of things is happening, it leads to this result, which you may or may not like."

So I think the combination of all those things is probably going to lead to bright spots in the future. One in particular that we're really intrigued by is looking at upside exploration formations -- so like Barnett and Woodford or what if Wolfcamp A has been developed and I'm looking at an overfill or an underfill situation, what's the real loss from parent or child developments?

You've got to have forward-looking ways of trying to guess what that future might look like before you drop $10 million drilling the well. So that's probably a really strong way I think that AI and ML could contribute, especially in the U.S. at least, as these basins are maturing over time.

ND: That's really exciting. Thank you very much, Jon. And thank you for joining us. Stay tuned right here for more actionable energy intelligence.

Comments