Subscribe
Share
Share
Embed
Welcome to Big Digital Energy with Chuck Yates, Mark Meyer and Kirk Coburn. Weekly news in energy covering oil and gas and cleantech.
0:00 Hey, Clyde, welcome to an emergency episode of BDE. Isn't that pompous emergency emergency episode? Let's redo this. Hey, Clyde, welcome to breaking news on BDE. Oh, you've already released a
0:13 press release, right? That's not really breaking anything. Yeah, but it took a while for the press release to come out. Okay. It's definitely been in the works for a while. Then in the works
0:23 for a while. Okay. So lay out the press release to me. What are you two guys doing? Well, so NOV is going to be managing the real time data aggregation and visualization across multiple data sets,
0:36 whether that's coil tubing, wire line, pump down your, um, any offset sensor data that you have, they're going to capture that in real time. And it's going to be time synchronized across all the
0:48 data sets, then that streams directly into the well data labs platform where you can start doing analytics on your data So, as the the old oil and gas, finance, pro. The way I remember it used to
1:03 work, and anytime I got anywhere near operations, my engineering partners would scream, but you go do a frack job, you get all the data, and then you'd analyze it, and you'd improve on the next
1:17 well. Are you here telling me
1:27 we could do stuff on the fly in real time? That's right, so it's more impactful now. So like Eric said, we're aggregating everything on location, we're transmitting it in real time. People can
1:33 look at it and say, this is good, this is bad, we need to adjust, things like that, all pretty much instantly. Oh wow, that's the real promise of the internet, telecommunications, big compute,
1:51 all that coming together. You got it. Oh, that's very cool. Walk me through, give me kind of a dummy down situation of an example of this, like. All right, so our team goes out. We install
2:06 all of the hardware. We get everything synchronized coming into one box. So, you know, even if we take a step back there, we're on a frack job. Okay, got it. We're on a frack job. We have
2:18 lots of different providers out there, right? You got your frack company, your wire line company, your pump down company, maybe even some well-head sensors that are another company. Maybe you
2:30 have your offset well sensors that are another company. Our job at NOV is to bring all of that information together into one synchronized timestamp. So now, instead of having to get data from
2:43 different providers, probably in a CSV, probably after the stage or after a couple of stages, We're bringing it in into one. time stamp into one box, into one feed. So now everything is together,
2:58 you only have to worry about one feed instead of these multiple feeds, and you're able to visualize it in max completions. And then in addition to that, we've partnered with Well Data Labs to do
3:09 that advanced analytics in real time. So you're the, I'm the completions manager maybe, and I'm sitting there, I've got your front end pulled up, the analytics data's feeding in Yes, so an
3:25 example that I can provide where NOV really stepped up and helped us out is when we were doing some of our diagnostics projects or a company is wanting to bring in time series data to overlay with
3:38 their FRAC data, they ended up going out, the operator was using their own gauges, which were sometimes not calibrated or they had issues with transmitting the data or someone shut a valve and they
3:51 didn't know that. NOV was able to go out there within a 12-hour period and install a defect quality gauge on the wellhead so that they can start collecting all this time series data to really
4:04 understand what was happening out in the field. So they knew by using the well data lab workflows, what should they be seeing? And they weren't seeing that in real time, but NOV stepped in and
4:16 with our partnership, they were able to assist and get this data collected in real time I'm a music guy, one of my favorite stories is Butch Vig, the music producer, producer Nirvana's Never Mind,
4:29 their first album, and it blew up, right? It changed the world, smells like Teen Spirit, as iconic as long as we've had in the music business. And he got a little bit of criticism that, you
4:41 know, it was overproduced, it was too poppy, et cetera, and Butch's response was, you know what I did? I just had them tune their instruments
4:53 Yeah. So very cool. So we're thinking through kind of use cases on this. We're out on a frack. What would be a problem that potentially can get solved now that we couldn't solve before you guys
5:09 came together? One that immediately comes to mind is if the operator is using a friction reducer and they're trying to evaluate is this friction reducer doing what it's supposed to or if they're
5:21 going to trial do like an AB test they want to say I want to try pods versus a PLA diverter so they want they want to understand like which one's actually going to have a better impact based on some
5:33 of these measurements so they might have multiple wells censored and then they're just going to want to evaluate that in real time and compare it against some of their historical data that we can
5:44 provide that analytics on. Oh very cool very cool so in terms of of thinking those through, give me another possible use case. So really just taking it back down to the ground level,
6:01 there are some companies out there that are still having a hard time getting data from the field. So that is really, so the strength of this partnership is NOV has the infrastructure, the reach,
6:16 the field staff to deploy a really tiny, like true tested hardware solution and have that reliable data feed out of the field. So Eric's team and well data labs, they do all of the really cool
6:33 fancy stuff, but I like to really bring it back to the point of we're still having data transfer issues in FRAC locations today. And that's really where NOV comes in. The synchronization is really
6:47 cool, it's really awesome, but if you can't get it out of the field, It's not going to do anybody any good. pretty much, right? So there's that aspect. And then there's also, okay, you get it
6:58 out of the field. How are you looking at it? And then what are you doing with it? So a lot of folks that we talk to still are doing kind of like a team share right now. So they're relying on
7:11 really good internet in the field, in the frack fan, and all they can do is see what's going on in real time. They're not able to go back a stage or two stages or two a different well and see,
7:21 okay, let me look and see what was going on there. They're not able to have that control of what they're looking at to analyze. Interesting, 'cause, you know, we've been out with our AI
7:34 enterprise software solution, talking to a lot of folks, and one company, the way they've sort of handled this is they literally have a WhatsApp channel for each well. And, you know, the head
7:48 operations we're having a meeting with And he's like, I got 1500 WhatsApp messages this morning. Yeah. He goes, There's no way for me to kind of track that. And how many of those do you read?
8:02 Probably? Yeah. One. Yes. Right. Yeah. Exactly, exactly to that point. So very cool. Do you want to show it to us? Yeah, well, just one quick thing. Like I was mentioning friction reducer,
8:14 when they want to evaluate a friction reducer with our stage variance tool, you can actually look at multiple wells at one time. So right here, I just have three demo wells pulled up, where we're
8:27 looking at average treating pressure, the friction reducer and the slurry volume. And then down here on the bottom is actually gonna be like the, where you can see what the ranges are from 10, 000,
8:38 all the way down to 9, 300 for your average treating pressure. So it kind of gives you a quick visual of what's happening. So if you only took two of these variables, friction reducer, volume
8:49 pump versus average treating pressure, you might not necessarily see something. This gives you an opportunity to compare multiple data sets historically against each other across multiple wells. If
9:02 you were trying to evaluate like a AB test, friction reducer, or you're trying to say night crew versus day crew, or you're trying to say service provider one versus service provider two, whatever
9:14 variable you want, you can have added into this, and you would essentially just click this little plus button, and then you would bring in, you can have up to three variables that you want to look
9:23 at and compare against. So whatever's on the drop down menu, whatever channels that you're bringing in or collecting real time through NOV, those are the channels that you can have to compare
9:33 against. So I'm making this point when I'm out talking, and I'm gonna throw it out there, and both y'all are pying on this. My feeling on the state of AI today It's really good at showing you
9:50 correlations, even correlations you don't know exist, but it's not there in terms of giving you causation. You're still the subject matter expert. You're the one that has to say, Does this
10:02 actually matter? Why does it matter? Does it not matter? But I have found things in oil and gas, and I think I'm pretty smart. Don't tell me otherwise. This is my show, I'm very shallow Let's
10:17 not bash the host. But I've even been just playing around with data sets and oil and gas. I have been suggested correlations. I'm like, holy cow, that might actually matter. Thoughts? Yeah, so
10:32 again, just from the NOV perspective, this is the direction that we are looking to go. We're not in the data realm, just to kind of collect data and visualize it, right? It's all leading towards
10:45 automation Um. AI really figuring out what the data means and looking at it from a different perspective to your point. So we have, you know, our customers are super experienced in fracks and they
11:01 know exactly what's going on, but maybe just taking a different look at it or having something pop up to say, hey, did you think about this would be really beneficial? I think AI is still kind of
11:13 scary in the, specifically in the completions and in the frack market. And we wanna make sure that, you know, our customers are ready for that and see the good that it can provide. When I think
11:27 about like what NOV is doing, they actually announced that they're doing like predictive analytics. It's, you need this data to be able to make decisions, but you also need quality data coming in.
11:41 I guess, well, I was in the Navy for 10 years, shit and shit out. So if you don't have good data to begin with, well, then what's the point of collecting data? So we want to see good data
11:53 coming in so that you can compare it against your frack models. Is this going to be performing as you're anticipating? How else are you going to integrate these data sets into your workflows? Those
12:04 are the type of questions that we like to ask and understand when we're dealing with these operators. Another consideration is when like there was a merger that was announced today. So looking at
12:15 these mergers, you have two completely different data sets, the quality, how do you combine them? How do you access them? Those are the other big considerations when you think of how do you
12:24 utilize AI effectively? With well data labs, we actually just launched a chat WDL where you can go in there and ask questions of your data. So it's like a more interactive way to find out like,
12:38 what is my slurry volume pumped for these wells or. How do I compare two other operators in the area? So it's finding ways to make it more useful and impactful. 'Cause I think the under-appreciated
12:50 thing in my running joke is always, I'm old enough to remember when AI was called statistics.
12:56 But, you know, at this point in AI world, and it's getting scary better every day, but if we have a data set where I opine on a frack and you opine on a frack, and you opine on the frack, it
13:12 should not listen to what I say about the frack. It should be Chuck's an idiot, but if I'm in the data set, I get equal waiting. Exactly. You know, until we figure it out. So your point about
13:25 getting the data in really good shape, we're working with one company and I'll make this up. I think they have 39 terabytes of data and 26 of those terabytes or duplicates. Oh.
13:42 test data and put it on their hard drive, manipulate it, save it there, and so figuring out how to get just the true data in a shape where you can use it. It's going to be really important very
13:55 much. I just spoke with a customer a couple of weeks ago who has been collecting data for years, years, and so what do you do with it? And they're like, well, when there's an emergency or
14:12 something happened on location, we go back and look at it. So you are sitting on all of this data and have never cleaned it, looked at it, put it to good use. And I think part of that is because
14:24 it's really hard to do that, especially if you don't know what you're looking for. So that's where I think this AI transition is really going to make a difference in the world is a lot of people are
14:36 sitting on a lot of data. And then this can go in and kind of say, Hey, Again, have you ever looked about at it like this? Or we see this trend or we see every time this happens, this happens,
14:48 you know, is it correlated? Does that have to do with an operation? Does it have to do with a different chemical? What, you know, how can we make operations more efficient? My story to that
15:02 point is I'm sitting in a room with 15 people kind of demoing our software and being a software salesman as I'm coming to learn is kind of like being a lawyer. You don't ask a question unless you
15:15 know the answer. Oh, yeah. I'm an idiot. I throw caution to the wind. So a geoscientist was sitting there and she asked a really thoughtful question. She's like, Chuck, you're telling me that
15:25 I'm the subject matter expert still and that at the end of the day, I'm getting answers quicker and I'm getting data. But then you're telling me when I connect these disparate data sources and I
15:39 start looking at it. I'm going to find correlations that exist that I, the subject matter expert, didn't even appreciate. She goes, I can't wrap my head around that. Can you help me? And I had
15:50 a data set of call it six well files. And I'm like, demoing. I'm like, why don't we just try this? I go, how would you compare and contrast these six wells? And the story came out of, well,
16:05 number one, I'd look at casing design. I'd look at completion recipe I'd look at the formations, all the stuff that you go, okay, well, no shit, Sherlock. But then it did this, which I
16:17 thought was really interesting. It said, I would compare and contrast the EURs or the IP 30s by vendor that worked on the well. And I kind of went, whoa. I did a lot of early stage lease and
16:29 drills when I was at Cane. We may have done 100, 125 of them. And I think we got pretty good being able to sort through, you know, here 200 wells. here at eight that are really good. I, it was
16:42 the fine mesh sand or whatever in the completion recipe. Never thought that we should be going per vendor. And so anyway, she was, she was kind of funny. She goes, oh my gosh, okay, I get it.
16:54 Now I kind of see what you're saying. The wild thing is I was talking to an old school EMP guy and he goes, he goes, oh man, Chuck, that's always been my secret sauce. And I go, what do you
17:07 mean? How'd you know that? He goes, man, back in the financial crisis, oh, 809, I had 10 wells drilled. I had to hold my leases. There was literally one company that was fracking. So they
17:18 sent their best sales guy over and the sales guy was like pitching, oh, we had these special additives to make the fracks work better. And he goes, dude, you're the only game in town. Just go,
17:29 go frack my wells. And he said, you know, 18 months later, I'm doing my EURs and those wells really were 20, 25 better So he said, I paid more for frack. jobs because of that. So, you know,
17:43 it's a, it's an interesting world out there. So, uh, give me the, uh, let's break some news here on BDE. Yeah. Uh, how'd the, how'd this deal go down? Oh, man. Did, did, did, was it?
17:55 Did somebody slide into somebody's D. Ims?
17:59 Did we go grab a drink? What happened? Well, for, uh, for me, that's directly involved with my CEO. I think it was just a relationship that they had established and it just kind of really
18:09 worked out. It's like, one of those epiphany moments like, Hey, this really makes sense. Like, let's talk. And then we mentioned that it took a while for it to come out just because we had to
18:18 wait for quarterly financials to be released and we had to wait for dotted lines and lawyers. And so it took, that's why it took a little while. Yeah, I think so from NOV standpoint, we, we know
18:32 that we can't be everything for everybody. Um, we, what we do really good is the hardware. in the visualization and we look outside of ourselves to find people like Well Data Labs who really kind
18:45 of fill the gaps for what people are looking for. And I mean, Eric, you're pretty cool. So that's kind of, you know, the first thing is make sure that we can work well, you know, with others
18:58 and that it fills a need in the industry and it just kind of, it fit. It was like perfect puzzle pieces that came together Yeah, we had we had customers with immediate needs. So as soon as we
19:09 could get this off the ground, we did because it was like, okay, you guys can help with instrumentation. We help with the analytics, but they also need visualizations and having, having them
19:20 being able to collect all of the data out in the field was very impactful. Gotcha. And so do we have teams that come together and work on it and figure out integration? Did we have a client that
19:34 you worked on together to build the first?
19:38 build it and then want to go to clients. What was - Oh, I can - All of it? All of it.
19:44 I love this inside Scoob. So I think, you know, it really just goes to show how strong both of us are independently. So
19:53 when a customer did come to us together, needing the solution, we deployed our solution. You guys deployed your solution. And then some of the software geniuses in the back made it work So the
20:07 first one, I think, was a little challenging. But I think as we kind of deployed on more and more of these together, we really harden that back end via the API that we have set up. So now, we
20:22 can go on location. Somebody could call us today. We could be on location tomorrow with a fully set up solution. Gotcha, so I'm CEO of an oil and gas company. got to help my shareholders. I
20:38 would not be a good CEO of an oil and gas company. But so I'm CEO of an oil and gas company. I've listened to this podcast and go, wow, this is really cool. I think I need this. How does this
20:49 work? Are we calling NOV? We call them well data labs. How does that? Well, essentially we made it so that it's both parties. We can sell on behalf of each other. So we we've worked that out We
21:03 we've done training where I'm familiar with the NOV max completions and they keep me aware of all of the new updates. As they continue to grow their team, I'm working with their new hires to
21:15 actually teach them about the the analytics that we offer. So we can effectively reach the market together with the same, same pitch. And I think too, with the reach that both of us already have
21:27 in the market, you know, different reach if we have an existing customer that we have a relationship with. They can come to NOV. Well, Data Labs has an existing customer they have a relationship
21:38 with. They can go directly to them, but ultimately they're gonna get the same experience with this partnership. They're not having to deal with two vendors. They're just having to deal with one on
21:49 all of the backend side. Gotcha. So again, I'm the CEO and I hear every day from my shareholders about AI. I need to be using AI I'm
22:02 slightly skeptical,
22:06 walk me through, 'cause I'm a finance bro by DNA. So walk me through ballpark-ish, how much this costs and how are y'all thinking about the ROI for me as your client?
22:22 Yeah, so the cost per customer are going to vary. So I can't just give a ballpark number. And I wouldn't ask him for a number Sure. Or just, well, how do we think about that? Is it seed? Well,
22:36 so for well data labs, we want to make it so that we were talking earlier about you use your geology example or your geophysicist example. We want to make sure that all of the teams can be
22:46 integrated and utilize it. So we're not trying to limit the amount of seats, where essentially how many wells are you going to be completing per year, so that we understand what is the workload on
22:57 our end that's required, and then for NOV, I would assume it's going to be, you know, how much field operations do they need to support. So for us, once we understand that, that's going to be
23:08 our workload, and then we'll just be able to turn on access from that. And do you need to upload any historical data? That's another consideration, but that's usually something that the operator
23:19 can handle on their end. It's essentially just a dropping drag that you can have access to the data, or if you have snowflake, you can essentially just link it to us, and then we can just transfer
23:30 the data. drop box, drop box, yep. Yeah, and from the NOV side too, our goal is for people to use it, right? So we're not gonna limit by seats or anything like that. We do typically charge by
23:43 pad. You know, you got a one well pad, you got an eight well pad, yeah, there might be a little bit different, you know, charging there. Again, because of humans and hardware that are
23:54 typically involved in that. But at the end of the day, our goal is to make this information accessible to the most amount of people so that they can make the decisions that they need. Right, and I
24:07 know we're early days, so the answer is probably to be determined. But are we setting a baseline for previous fracks and if the fracks using y'all solution are better? Is that how we're gonna
24:26 calculate ROI for me, the client? How are we thinking about that? that and I know the answer will be, well Chuck, we'll come back in two weeks and tell you how we're thinking about it now and
24:37 then two weeks later we'll come back again, but. Oh, that's a good point. Well, like what is their metric? How does that change, if it's a public company, how does that change based on what
24:46 the shareholders want? If it's impacting their dividend, well then they're gonna wanna find ways to keep that dividend but also reduce costs elsewhere. So that's how they can use the combination of
24:57 the platforms to identify with historical data, how can I improve and then validate that and show it in real time through the NOV monitoring. Yeah, I was out on location a few months ago and got a
25:10 real reality check because real time means different things to different people and I was talking to a company man in the data van and we were talking about real time and he said, Well, I can screen
25:25 out a well in three seconds. So anything less than three seconds, I'm not really interested in. You know, I really need it in real time. If I'm expected to make real time decisions, I need the
25:36 time to do it. So, you know, that is our goal, is to provide real time information to make those real time decisions and prevent screenouts and cleanouts, having to bring coil in in the middle of
25:49 a frack job. Things like that is really what we're looking to do. And without all of that information, all your wire line pump down, they're offset wells, things like that, coming in together,
25:58 you're just not able to do it, especially if you're trying to move people out of the van, out of location, or you have a high profile job, and people are watching it from a remote ops center,
26:09 something like that. You're not able to really get a handle on that kind of stuff that happens instantaneously without a real-time platform. And then again, with the advanced analytics that Well
26:22 Data Labs provides, it just kind of is the cherry on top to make even more data-driven decisions. So we could, we could reduction of screen outs, uh, that's a good metric to, uh, to run some
26:35 numbers on better performance. Interesting. Cause you know, the weird thing we're having on the AI side is it's this kind of, what I'll say push or pull or dance of, you know, what does the
26:53 client want to see in terms of an ROI or a payback, but it's also incumbent on us to semi define that, you know, to some degree. Hey, if we can do this part of it's just an expectation setting,
27:06 but the other part of it is you've got to be level setting with where the technology is today, you know, so. Absolutely. And I think we're still in that transition period of people don't know what
27:18 to do with their data or what they can do with their data. And it's our responsibility to kind of help guide them through this transition period. because if not, they're just gonna collect data for
27:29 the sake of collecting data, sit on a mound of it and only look at it when a problem occurs. So this is gonna be an interesting question to me because we've all been in the energy business long
27:44 enough where we quote unquote made a lot of money because we knew something the market didn't know, right? And then we pissed it all the way in spades, trying to figure out spacing on our own I got
27:57 a buddy, I got a buddy in Silicon Valley who periodically every three months calls me in just goes. So let me get this again. Y'all have two sections of land and you spend250 million and maybe you
28:12 get100 million of oil out of there and you did that without talking to your neighbor? Oh yeah, every day, that was an everyday occurrence. Do we see potentially customers allowing y'all to
28:27 anonymize data and sharing. Is that potentially part of the future here? I'm not 100 sure we're gonna get there 'cause it always seems like there's gonna be some secret sauce recipe. Like I don't
28:39 wanna tell them what I'm doing. I wanna keep this completely confidential because it impacts their stock price. So I think that
28:47 there are some issues there. But is there a way for us to take all the data that we have and offer historical frat curves? That's one of the tools that we have. So you can go back and say, How do
29:00 I compare to other operators in the area? Those are some things that we can do where it's like anonymized data. We're that you're just trying to say, How does this frat design compare to something
29:11 in this county? You can get as granular as that, but. And that's why I asked the question, 'cause I knew we were gonna have an answer of all of us kind of scratching our heads Yeah. I think where
29:22 we have seen it is everybody. but he wants that data, but doesn't want to share their data. So that's the fun part, right?
29:33 At what level are we able to really make this available? 'Cause the way that I look at it too is like, all of this data and AI and everything that we're doing is really to make the industry better
29:45 as well, right? And in order to make the industry better as a whole, yeah, we may have to start sharing some non-proprietary information so that we can grow and build on what everybody else is
29:56 doing. Yeah, well, it's really interesting 'cause if you look at Stack Overflow, where the software nerds hang out, Google and Microsoft engineers will share code with each other. Oh, really?
30:11 And having this problem, you know, and a Microsoft guy, you know, software engineer will go, Hey, I wrote some code, does this help you? And it's kind of crazy that there is that level of.
30:23 kind of what I'll call open sourcing in tech. And literally we won't tell each other if we figured out how to keep a well pump going two extra days. Well, and I think it's about through official
30:36 channels, right? 'Cause I think our industry is so small. And if you've been around long enough, you have friends that work for other companies. And I think there are phone calls that are
30:46 happening on the side of like, hey, this is a situation I've gotten myself into. Have you experienced it? What have you done, things like that? So I think it is happening just through unofficial
30:56 channels. Yeah, no, I think you're right. It definitely happens in the field. Oh yeah. You know, my old boss, you pick up the phone and you call me, this is going down, what do I do? So
31:06 yeah, but it'll be interesting to watch this in the AI world in terms of how much are we willing to share. So at the end of the day, the more we share, the better off we'll be as an industry. One
31:19 thing to note is that well data labs, We're approaching roughly two million, two million stages of frack data that we have, access to that we've trained our machine learning models on. So whether
31:30 you're trying to quickly have it set a flag in the data for your ISIP, so that you don't have multiple engineers picking specific points, it's consistent on how we trained our machine learning
31:41 models. It can also do that for target rate, your ball seat, what is your, like if you're trying to evaluate fracture-driven interaction When does it set that flag? How does it recognize that
31:53 differential in pressure using some of the patents that we have? So we can leverage that data set for training machine learning models. Like if someone wants to get improved efficiency to identify
32:05 when I stop fracking or stop pumping sand to when I start pumping sand, what is that downtime? So how can they narrow that window, increase efficiencies? Those are the things that we're looking at.
32:17 So if we have somebody with a unique way that say, how can we leverage this data or all of this historical data to improve efficiencies? That's what we're looking at. Yeah, no, I think, I mean,
32:28 I think literally, and I'll make this up, machine learning as simple as, okay, six hours before the event, let's look at every variable and just find our correlations. I mean, just as simple as
32:45 that is going to be a game changer All right. It just has to be. So, okay. So, how do folks get in touch with you? So with me at Well Data Labs, it's going to be Eric at Well Data Labs dot com.
33:02 Cool. Mine's a little bit more complicated because I have a challenging name. But mine is Kayla dot S C H E R E R at N O V dot com. You can also look me up on LinkedIn Cool, cool.