Episode Transcript
Rhett: All right. On today's episode of Pipeline Things, we bring back a return guest and I might say an old problem. We discussed one of the longest named titles in IPC history, which I'm just going to rename to Hard Spots There and Back Again, which if you are a Lord of the Rings fan, you will get that reference.
For the rest of you, it's a great episode. Special shout out and thanks to the TDW data science team for their contributions to this project and TDW and Williams for giving us this. Excellent paper to discuss.
That's okay. You're tall enough to reach me. Welcome to today's edition of Pipeline Things. If you are watching on the YouTube channel, just forewarning, you might actually want to watch this one on YouTube because we're going to have some graphics, but the room's probably going to give away the guest. If you have a good memory, the minute you see the room, you should know the guest.
Chris: It's fun to be back here.
Rhett: It is, man. It's got a new name. It's got a new name.
Chris: It does.
Rhett: It's Odin's lair now.
Chris: Yeah.
Rhett: I can't remember what it was. Dragon's lair before. It was.
Chris: You see Thor walking around?
Rhett: Just Thor's walking around?
Chris: Yeah. He must have come in after. He's a little kid in like a Thor outfit. So I guess the guy that now owns it's Odin. I don't know.
Rhett: There's some serious, serious gaming going on outside, though. I know that tabletop stuff. Chris.
Chris: What's up?
Rhett: How many times a year do you think you fly?
Chris: Oh. No, I'm just guesstimating maybe no less than 30 times.
Rhett: No less than 30 times. And every time you get in the plane, you expect the plane to take off and land and do exactly what it's supposed to do.
Chris: Yeah, we have a new saying, right? It's not safe travels.
Rhett: You're not one of those people that collapse whenever the plane lands, right?
Chris: Oh, no. It's not safe travels. What have we heard lately? Eventless travel is what we want. Eventless travel is what the expectation is now.
Rhett: That's reasonable. And this ties directly to today's episode because I didn't realize this. Do you know, just out of curiosity, what training pilots are required to do in order to fly those 30 segments that you do?
Chris: I cannot say I know what all the training is. I do know that there's an hours component to it. And then once it's hours, then there's a qualification for each aircraft that you have to meet. But I cannot. describe what that qualification process is so it's a combination of hours and qualification for the aircraft
Rhett: it is so let me and if to our audience if you remember we actually talked about some of this on the Boeing episode because that was what contributed to 737 max issues because they didn't want to retrain people but get this I didn't know this so pilots are required to take simulator simulator training and proficiency checks every six months
Chris: oh wow
Rhett: they have to have 20 to 40 hours of annual online training They have to have a medical exam every 12 months or six months if they are over six years old.
Chris: If they're over six years old?
Rhett: 60 years old.
Chris: Holy smokes. I was like, what country are we talking about, bro? Is this like AI where now I was six-year-old? Can you use it? I mean, Mati's three, and he dominates the iOS service. So the iOS, Mati just crushes it. He changes the color on the screen and the brightness. So maybe flying a plane at six is reasonable.
Rhett: But we're getting there. AI, right? I don't get to do it. But no, it speaks to the fact that I think what's interesting is we get in a plane and nobody ever thinks about what the pilots have to go through.
Chris: We think about it.
Rhett: Nobody ever reflects on that.
Chris: You just trust it. You don't question it. That's the issue is you don't question it. Right? We trust the system. And I'll get a little bit personal here. So I've asked Amanda a couple times, like, hey, you know, why don't we take a PJ and go to Austin instead of driving or something like, why don't we look into this and see what it's like? Because we have a friend that may or may not have a PJ private jet. And she's like, never.
Rhett: Did that feel like name dropping? Did that feel like it was dropping? That felt like a little like boom. I'm just asking the audience out there.
Chris: No, no. We know them. We haven't necessarily been invited.
Rhett: I don't know anybody that has a PJ. I didn't even know that was an acronym.
Chris: And she's like, never. We are commercial all the way. And I was like, why? And she's like, because I trust that system better. Whereas when you're in a PJ, you're less sure that certain protocols or processes are being followed.
And I was like, that is interesting because, yeah, it's kind of similar, right? It's like this whole idea of I think a lot of times we get in these commercial airlines. It's we generally trust the process. We trust the system. And we don't audit it.
Rhett: The segue is clear as we bring on our guest. I want you guys to think about because a lot of you out there, I know a lot of our audience travels. You get on the plane and you don't think about everything that goes into making sure that that plane lands and takes off. You just expect it to do our job, right?
I think a lot of times in our industry, we expect the same things of inline inspection technology, right?
Like I just want to put it in the pipe and it needs to do its job. I don't want to invest time in making sure that it's improved. Like how many times have we heard that old adage? I'm not paying you to do exploratory digs on your tool. But the reality is today's IPC paper.
And as I introduce the guests, we're going to be discussing IPC 2026-185228 with the longest title name ever. Hard spot assessment using high and low field magnetic flux leakage inline inspection with targeted non-destructive examination. It was a collaborative publication between Williams and TDW.
And I think it's a good example of why. You need to invest sometimes in the ILI technology.
Chris: Yeah, because the analogy of getting on a plane is the wrong analogy, right? Like that should not be the mindset.
Rhett: Exactly.
Chris: The mindset should definitely be much more of like at a restaurant since, you know, you love me at restaurants.
Rhett: Is this where you assault the waiter with questions?
Chris: Is it assault? Or is it ensuring that there's enough communication so that your expectations are met, right? You deliver an order. And then the waiter asks and says, did I get the order right? And then as the food is being brought, you're inspecting it. And then they ask you and say, is this what you wanted? That's really what you're looking for, right?
Rhett: Usually when you're like, no, I didn't order a hamburger patty, I ordered a steak. I think those were your exact words.
Chris: This guy's just putting all my business out there.
Rhett: So, all right. I want to go ahead and bring our guest on. Adrian, welcome back to the show as Adrian comes in behind us. He'll shake your hand. It's harder for him to shake my hand. Oh, there we go.
Chris: Watch the corner. We got you.
Rhett: So to our audience, our guest is Adrian Bellinger. We are bringing him back from TDW. And Adrian, really, really excited to have you on. And thanks for joining us. So it's been, as Chris plays with your mic in the audience, you get to watch Chris. Chris, you make sure his mic's in the right place.
Chris: All right.
Rhett: Adrian, it has been, we discovered, almost two years to the date since we've had you on the show last time.
Adrian: Wow. Yes.
Rhett: And tell us, where have you been the last two years? What's been going on?
Adrian: Oh, well, since the last time we talked, it was on the data science and getting good data. So as usual, I'm always trying to get the best data I can to use for our RLI tools.
But then hard spots came up and I had to go back to my past. And because I worked with the original hard spot tools with Tubiscope.
And so. I started working with Williams. There was a failure they had, and they gave us a piece of pipe to run our tool through it. And so we collected a lot of data. Am I getting ahead of myself with this?
Rhett: Yeah, no, it's fine, but I was going to sidebar. Wait, I want to sidebar with you on that. I got to ask you a career question. Do you feel like things come and go in the pipeline industry like that?
Adrian: Oh, yes. All the time. All the time.
Rhett: Didn't I deal with this 20 years ago? I'm feeling like deja vu.
Adrian: Yeah.
Rhett: So you worked with hard spots in the tubascope days, however many years ago that was.
Adrian: I started in 96 in the business. So, yeah.
Rhett: And then it comes back around and 26, 30 years later or 2025 and you're thinking like. Didn't I do this 30 years ago?
Adrian: Yeah, yeah, yeah, yeah.
Rhett: I've had that happen a few times in my career. Wrinkles, I feel like I've happened with that. Wrinkles? I'm like, oh my god, wrinkles.
Chris: So did you greet it like an old friend? Like, hey, how are you, buddy? Yeah, good to see you again.
Adrian: You know, that's why you say that, because in a way it is. Because I really like it, because it's always the problem with operators. It's like when I see nasty corrosion, I say, oh, that's beautiful. It's like a doctor looking at pathology.
Rhett: You have the most beautiful, terrifying corrosion.
Adrian: You want to analyze it, and you want to break it down. And with hard spots, it's all the metallurgical differences that involve there, all the magnetics. To be frank, the magnetics has not progressed since the Patel days, really.
Everybody keeps focusing on the hard spot. It's just the actual hardness and then the metallurgical makeup. But the way we look at hard spots is magnetically. Fundamentally has not changed.
Chris: We need to qualify that a bit. I want to pause there. While the magnetics haven't changed, we would say that the sensor configuration has changed on how we're addressing hard spots.
Adrian: We've changed it into the density of the sensors. You do the components. You get triaxial now instead of uniaxial. Back in the day, we were using coils. It was just integrated voltages.
That's gotten better, but it's still of the same physical phenomenon. We hadn't really penetrated deep into the physics behind it.
Chris: I just wanted to clarify. The measurement principle in theory is about the same, but the sensors that we use... are tuned such that we could do a better job at it.
Adrian: So we can measure that magnetic field much more accurately, is the way you put it. But relating that magnetic field to the actual metallurgical composition, that's still...
Rhett: The underlying physics principle hasn't changed, but our ability to dig in and interrogate it has improved significantly because of the instruments we're using.
Chris: And honestly, our understanding of hard spots is a lot better now, too.
Adrian: Yeah, I would agree. And that's why we are coming up better algorithms for it.
Chris: I always like to bring this up. Verification digs 20 years ago were truly to say, what is it? We see the indication, but we don't know what it is. That was kind of like the progression towards POI.
Now we're much more seasoned and we're like, we know what we're looking for. We understand how hard spots are happening a little bit better. And now we can look at the ILI data and say, hmm, I think maybe this is why it is.
Adrian: And to build on that, it would be like, what is the hardest to go out there with basically a ball gauge? Now we got the UCI tools. But just what is the maximum hardness I measured? And call it that.
Rhett: We could go on a whole sidebar there for sure. No, because so many things I want to say. But this project starts, right? So this is collaborative between you and Williams.
Can you give us the genesis? So obviously it lands on your desk 30 years later. But what starts this guy, this project again with y'all? Can you set the stage a little bit for the story?
Adrian: Going back to, well, Williams had a leak. And they remember the section of pipe that leaked on it. And they gave it to all the ILI vendors. It wasn't just us, but we did pull through it.
So we already had a hardness algorithm we had built up. We had samples of A.O. Smith pipe that had been sleeved in the past. It actually had leaked under the sleeve because it was a type A repair and it wasn't done properly.
So they were just eventually removed. And we developed our initial algorithm on these original A.O. Smith hard spots.
Rhett: Now, this is pre-Williams. Pre-Williams approaching you all, you have a tool, you have a hard spot service, and you have an algorithm.
And Williams gives you a piece of pipe and says, run your algorithm. And a pull test, which has a limited number of features.
Adrian: And we're collecting all this high-resolution data like Chris has mentioned about. So, I mean, when you look at the actual view of the data, it's like every one of the hard spots was... popping up very clear, well-defined.
Rhett: In the pool test.
Adrian: In the pool test, yeah. And the one that had the leak, we basically nailed it by tuber nail. Right on.
The other smaller ones, we weren't as close. We were undercalling it, overcalling it, but we were within tolerance. So we thought, oh, the algorithm's working on that.
Rhett: On the pool test.
Adrian: On the pool test, yeah. We were very comfortable with our algorithm at that point. It's like a hard spot's a hard spot.
Rhett: You're thinking 30 years is done. I've solved this problem 30 years later. I'm still solving it, but we're good.
Adrian: We just need to increase the tolerance. That's what we're always getting the pressure for.
Rhett: And so no big learnings or anything like groundbreaking from that pull test from your perspective?
Adrian: Yeah, for us on our end, it was just confirmation, validation.
Rhett: And so they move from there, give the audience a story. Y'all move from there into you, and the paper says this, y'all were given 13 segments. So it's a run. And you ran the first segment. What happened on the first segment?
Adrian: Well, they chose some digs. They were feature-rich. And they went out there. And this is what made me want to write the paper. Williams does the gold standard on how you collect hard spot data. They removed it all, the coding. They did a spine tool over it.
And before they did the spine tool, they did this slow-tiered. uh blasting on it so you didn't hard hard work the surface they found some anomalies that we didn't call which is a good thing uh so we could go back in our data and find why we didn't call it but they collected and they collected high resolution hardness grids so that we can go back and compare that to our actual data for our tools but what was happening was the algorithm was consistently calling Sometimes it was within tolerance, but when it was out of tolerance, it was consistently high.
And no amount of linearization.
Rhett: You were calling it consistently high, or the feature was higher than you were calling it?
Adrian: Meaning that when we were out of tolerance, it was always on the high end. It wasn't just a large spread. And the way the errors were, we couldn't just correct it with a linear gain offset adjustment. Because that's typical. You have to do that from run to run. You have to do a linearization just to normalize it.
But we did notice, looking at those signatures, it's always that we found out that we developed a different class for these newer hard spots that were failing. And we called that the class B and the old one the class A.
Rhett: So did you find the new features by virtue of the fact, was it a, I'm going to say, because ILI goes through detection and then identification, were you all detecting the features and then just sizing them wrong? Or was it also becoming aware of the fact that there were new features? I'm just curious.
Adrian: We were identifying hard spots correctly. It's just that in order to get the correct sizing, we needed to identify the sub classification of those features because even from the first pull test you could see that the signatures were different than hard spots we were used to
so we knew that these guys were showing up differently in the uh especially in the low field data, high field data it's a little bit ambiguous but the low field data they were definitely had a signature that you could identify them properly and then it left room for a new sizing model
Rhett: cool and so audience i want to y'all to hold tight definitely after the break we're going to get into looking at what some of these images look like so that adrian can walk us through actual descriptions around what he means when he's talking about the high field and the low field and it varying.
Um, but I think what I'm curious about is that you've done this now 30 years apart and we actually didn't show prep on that, which makes it entertaining for me.
Um, cause I'm curious cause 30 years ago, obviously what's happening in the field has also advanced significantly. And so I just want to pause from, cause I don't, I think it's worthwhile for the audience and us as. ILI people to reflect on that right it's not just that the tool technology is getting better how much difference did you see in the ditch 30 years later with the data that williams is
Adrian: huge huge amount um used to get so like uh the previous hard spots um back when i developed it like about 30 years ago it was with the gulf south um they would actually give a grid it was one inch by one inch grid and it was like handwritten numbers which was actually incredible at that time because normally you get a sheet uh it would just say hard spot might not even get a hardness value um or you just get that singular value
whereas what williams is doing is they're going and gridding it they make five separate measurement in every half inch grid take the average of that and so you have this high resolution multiple measurements you can see how much variation you get and then you you distill it down into a half inch grid with the average values of it for the hardness
that actually allows because you have a spatial map now um actually and circumferentially you can line that up with the mfl data and there you go there's there's all your
Rhett: and it's funny too because you also mentioned the spine tool which that wasn't around 30 years ago as well and so um It's cool to think about how much more data you're getting now and how that could actually be, I'm going to say, influencing and enhancing the learnings, which I think is neat.
And to our audience, we talked about some of that with Jing Wang before, and I'd say that that's been published a lot by Enbridge and TC out there, those procedures.
So separate topic, but the information for actually understanding that procedure now has really been documented. That's probably what, Chris, the last six years is really where we've learned that.
Chris: Since 19, right? So Danville was 19 is when kind of this curve went up and everybody started paying more attention to it.
I want to ask you, so 1996 to 2026, I heard, I remember you saying, you know, you worked on this early on. I want to talk, pick your brain about your position. Like what is your opinion on the importance of having more looks at the magnetic field through things like low field or ID discrimination versus just purely looking at a high field MFL?
Adrian: Oh, well, it's the property. The research was done by Nestle, Roth, and Crouch way back in the early 90s. It's a GRI paper out there.
They went in and did all the magnetic properties and the material properties of a whole bunch of pipeline steels. So if you judge from that report, the cohesivity, which is a property of the magnetic curve, is highly correlated with hardness.
Now, there are other factors that can affect hardness too, but to get the coercivity, you have to have those two magnetic measurements because you have to see how the top end of the curve, how much you can saturate it, and then you need the bottom end is how much it takes to get rid of the magnetization because that's actually a point of our low field. It's in the opposite direction.
So what happens is that it's actually taking the magnetic field. When the magnetizer passes it, you're in residual state. You're actually removing the residual and taking it down into the negative, into that other quadrant. And that will give you, that's what you can correlate to with the hardness.
Rhett: And can I ask, that's a calibrated low field.
Adrian: Yes, yeah, yeah.
Chris: It's not a trailing sensor.
Adrian: It's not a trailing sensor.
Chris: It's calibrated low field.
Adrian: If it were a trailing sensor, it would essentially be a residual.
Chris: Yeah, my one of my takes is, and this is just my observation from just this conversation, right? Is it's, you know, I hear you, I think of 1996, early days of MFL hard spots. And then I instantly think of Danville, right?
Where it was like, hey, we had done a hard spot assessment before. Then we look at the data again in 2019 and learn, oh my God, there's other 2019, 2020. Hey, we could look at this data differently now.
Rhett: Yeah.
Adrian: Right.
Chris: And then there's learnings. And then even still. it's staged that way almost on this project, right? To where it's like, hey, we did a pull test and we nailed it within like 2% of the Brunel reading. So you come in very confident, right?
Adrian: Yeah, yeah.
Chris And then we get into a feature rich line and you start doing digs and it's like, oh, there's still more to learn.
Adrian: Yes.
Chris: And I think, I just think for me, that's a big message, right? Is it's, you know, a lot of times kind of with the flight, you know, we expect things to just work when it comes to ILI. But the reminder is, A lot of these ILI specifications are based on essential variables that the ILI vendor set at a time, at a specific time.
And so as an operator, this is why I relate it back to like our analogy of I want a steak, not a hamburger patty, is it's we need to be clear as operators, we not retinite, but rather we as an industry as to what our expectation is. And you need to communicate that to the ILI vendor to collaborate with them to say, what do you need to meet my expectation?
And it sounds really nice on this project for you guys, because it sounds like Williams in this case already said, I'm willing to do this in the ditch. I'm willing to do this many runs. It wasn't one. It was 13. Right. I'm willing to do this many digs and expose more than just the feature. And so that collaborativeness kind of allows you guys to not just say bonk on the head. Hey, you missed this stuff, but rather, hey, let's keep moving this technology forward. Right.
And I don't think this was project was for development. Right. This was really like. This is an assessment. However, the feeling comes across as it's collaborative, right? And that moves the technology forward for you guys, and it's a win-win for everybody.
Rhett: And I think it's, Chris, you know what I was thinking about in that, just for our audience to think about, even in the original Danville failure, which we discussed, remember the original vendor went back and suddenly found a whole bunch of hard spots based on their data.
I think what's interesting to think about is that learnings aren't static technology is not static but what we learn isn't static either right and so it's a good it really speaks sometimes i think to the importance of reassessing but also reassessing our technologies right that like sometimes it's not just the technology that's improved it's what that vendor knows that's improved you've given them additional information whether in the ditch or by virtue i think a failure finding us
it's making me wonder like in 10 years will we be like You remember how we analyzed cracks back in 2014? Like that's what I'm thinking in my mind. Like I wonder if that'll be us as old people.
Chris: This whole airplane, you and I do this all the time, right? Like the whole airplane analogy is so good, right? Like the airline industries, they're very forward about lessons learned and information sharing, right?
And if we look at like the, just the information sharing, and we're using this just subjectively of like 1996 and even 2019 as it relates to understanding hard spots, because that's what our topic is today.
I mean, you could use any of your AI platforms and say, find me papers or presentations on pipelines related to ILI and hard spots. And you'll find all kinds of stuff, right? And that's public. So operators can use it. Consultants can use it. Vendors can use it. And so this ability to share information, I mean, it's just, it's good. It's a new era, right? I think it's super helpful.
Rhett: So tell you what, we're going to take a real quick break audience and I want you to come back. And if you're not on the YouTube channel, you might want to, cause we're going to let Adrian dive into some of the signals and what they learned from that first run. So hang on. We'll be right back.
Rhett: All right. We're back from our break with our guest, Adrian Belliger talking about the IPC 2026 paper that has too long of a name to actually read again on the air.
Um, so. I'm going to recap for you guys. We set the stage for this collaborative project on hard spots between TDW and Williams.
Prior to the break, Adrian took us through them being approached by Williams, given some hard spots to run on a typical pool test, doing very well on that, and then getting into a real-world first segment and having some learnings.
What we're going to do, again, I want to encourage you, we're going to jump over. We're going to be sharing a lot of images from the paper as Adrian walks us through their learnings.
But Adrian, so you guys go through run one, and y'all learned some new things about hard spots, right?
Adrian: Yes, we did. Because that first pull test, we had three points, so you really can't do a statistical evaluation with three points. But after collecting numerous digs from the first run, we saw that there was an over-calling from the original algorithm developed on the A.O. Smith original hard spots.
But looking at the signatures, we were able to come up with a different classification. And so we classified the old ones as a class A. You want me to do the description now?
Rhett: Yeah, sure. So let's go. So class A, we're going to flip over to the screen right now. What you're seeing if you're watching the video is a class A hard spot. And what is a class A hard spot?
Adrian: So a class A hard spot is most probably produced during the milling process where the...The line would pause, and the cooling jets would just over-quench.
Rhett: In some way it gets quenched, but this is what we think of as conventional. This is what the industry for a long time has called a quote-unquote hard spot.
Adrian: Yeah, they always thought it was just quenching of the plate before the rolling process. Because it had a water, and it used to be even, they thought there was puddles of water on there.
Rhett: Yeah, we shot that down.
Adrian: Yeah, exactly. That's not enough to actually rapidly quench it to get a hard spot.
Rhett: So what are we seeing here?
Adrian: So in the high field, which is what most of your MFL corrosion tools are doing, you see a depressed signal compared to the surrounding signal from the main pipe.
This usually happens when you have some metal that's in close proximity, something that takes away the flux, takes away the magnetism. So what that is is because martensite, being really hard, it's hard to magnetize.
So it's still kind of a contradiction. It's still permeable because you need more field to magnetize it. So it's absorbing the field where the rest of the steel is saturated and it's spilling over into it.
At the low field, and the way the low field is constructed, it's a weaker magnetizer, but the field's in the opposite direction. So when the original magnetizer passes over it and then it's gone, there's a residual magnetization.
Some tools may put sensors out there to capture the residual magnetization, but we go the extra step of magnetizing it in the other direction. but at a lower rate.
So this actually gives you the differentiation of the hard spot from the rest of the pipeline steel. And so that does the LFM that has real texture.
You see there's bright spots, dark spots, because there's regions of Martin's side and other combinations of byproducts of the quenching process. But it still has that same kind of pool.
Rhett: The shape, like one looks like a shadow, and one looks like a perimeter around the shadow.
Adrian: And it does. It picks up the edge of the hard spot very well. It's more delineated on the edge.
Now, we have other data sets. We use them as confirmation that there's a hard spot. The deformation, which is def, is because the hard spot is formed at the plate level. When they roll the hard spot and weld it, it resists the rounding, and it actually shows up as a little flat spot in the high-resolution deformation.
The IDP sensors, these are sensors used to discriminate for corrosion for internal or external, but they also are sensitive to the change in permeability on the inner surface of the pipeline.
So these class A's are through wall hardnesses. They manifest on the inside, and you can measure them with the IDP sensor.
Rhett: All right, so traditional. This is what we, for the last 30 years, have been hunting. The first thing you guys encountered, which you labeled as a type B, and we'll flip to now. What am I seeing when I look at the type B?
Obviously, I'm just going to give the audience what I think, and then you tell me whether I'm right or wrong.
I see a less pronounced shadow, and I see maybe even a less pronounced periphery, but the periphery looks different. It looks brighter, almost like a real bright halo.
Adrian: Yes.
Rhett: All right. What is this?
Adrian: So these are the newer heart spots that are starting to fail in the aging pipeline, Bethlehem being like the first case. So we call these Class B, Bethlehem.
Rhett: B for Bethlehem.
Adrian: But the metallurgical samples that have been taken from these types of heart spots show they have more bainite in it. They tend to be more dish-like. So that's where that sharp perimeter you noted in the LFM is associated with.
It's believed, here's another industry consensus, is that these are probably remnants of, in the ladle, when you're doing a pour, probably there's some segregation and then they get into the next batch.
And you have these, just basically these little islands of different metallurgy. So it's a mixture of martensitic and bayonetic. They tend, at their highest hardness, they tend to be as hard as the class A's. but they can still be above that 327.
Rhett: Okay. So we have traditional hard spots as A, the Bethlehem hard spots as B, and then we get this type C. And the type C that we're going to see on the screen.
Adrian: Well, I would just like to touch base on the deformation sensor. They're still flattening, so it was still at the plate manufacturing part. And then the IDP, this particular one goes all the way through.
It does manifest on the inner surface, so we are getting. But they don't always do, whereas the Class A's pretty much, unless they're not very hard, they usually have a manifestation on the inside.
Rhett: So you don't always see the interior manifestation.
Adrian: So that's why I say those are secondary. You can still have a hard spot and not have signal there. Or even with the deaf, it could, and they might not be flat enough to measure.
Rhett: Got you. But distinguishing characteristics from a signal data between both of them.
Adrian: Yes.
Rhett: so then we have the type c hard spots and the type c's this looks like what i so to just describe to the audience it looks like a plot from the matrix on the top left where you got the black numbers or it looks like yes it looks like rain falling down a window in both the mfl and the lfm
Adrian; well c in this case was we started with c circumferential because we seem this they seem to be associated with other you see those straight lines in the picture there's a rolling pattern probably when the pipe was being rolled or when it was being, with expanders, when it was being rounded out.
We think it's cold working, causing these. Because if you notice, those straight marks, they disappear in the IDP and the deaf. We're not seeing anything there. So they're not manifesting on the inside at all.
But the irregularity was indicative of the old type A's. But these never showed up near as hard. They do have elevated hardness, so we do try to call them.
So we call C. C now is for kind of more of a catch-all. We do have some classifications where you see it in the MFL and the LFM. The distinguishing characteristics is it's always negative in MFL. always negative in LFM.
Rhett: For type C.
Adrian: It's a different type of hardening. Martin said it's probably not being created. You're just getting cold working.
Rhett: Gotcha. As we get away, we can go back to the thing I want to ask because I think this is important for the audience.
Chris: Why is it important?
Rhett: You find hard spot first. You detect. You're identifying or classing into A, B, and C. And there's a fourth class D we won't chat about, but A, B, and C.
Adrian: We throw those in C, too.
Rhett: D's and C's. C's and D's together. C catches them all.
Why is the classification important?
Adrian: So that brings us back to the dig results we were getting on those class Bs where we were overgrading it. So by identifying properly whether it's class A or class B, then we can use the correct sizing algorithm.
Rhett: So you don't have a one-size-fits-all sizing algorithm?
Adrian: No, I do not.
Rhett: I think that's really important. So you actually have different sizing algorithms depending on how the hard spot feature is classed.
Adrian: Yes.
Chris: How important is knowing the pipe manufacturer in A's and B's?
Adrian: It is. We have found... collecting enough DIG feedback over several manufacturers that the A.O. Smith type Class A Bs are harder than the other manufacturers, which was Merberg and I think National Tube was the other one that we had DIG data on.
And then we had some others that just weren't identified. They're unknown manufacturers. Yeah, yeah.
Chris: Because the... I think the Bethlehem, you mentioned it a couple of times. I think that's relevant. And guys, don't don't fact check me too much. I'm just pointing a direction. I think in 23, TC had a failure in Bethlehem pipe, right? And it was hard spot.
And what I remember is it's the measured hard spot was around 200 for now, which is interesting. And that had hydrogen-induced cracking, hydrogen-induced cracking. And so it kind of jives a little bit with what you were saying, right?
So the big takeaway, not fact-finding everything, is just I think I heard you say that typically the Bethlehem ones were coming in a little bit less hard. And I think that's really important because when we looked at the data, the high field, if I remember correctly, it's just not as pronounced, right?
Because you said there's more elements of... They're more bayonetic than purely martensitic, which means they're less hard, which means they're a little bit easier to magnetize, which means, to me, the so what is it?
If you're only running a high field, you may not be getting what you need if you have Bethlehem Steel. And I'm not saying that's the case, guys. Don't take that as a hot take.
What I'm saying is I'm painting this posture back to ordering your food, right, is saying there is a component of... you need to know what kind of pipe you're inspecting and what type of hard spots you're inspecting for such that you can correlate that to will the ILI system be set up to find it?
Because, I mean, I think you guys are looking at four data sets here.
Adrian: Yes.
Chris: And it's much easier for you guys to say, hey, these are similar, but they're still different. And you can start segregating by either metallurgy or if you want to round up to, say, by manufacturer.
But if you don't have those data sets and you don't know what the pipe is, you don't know what you're not finding, right? Because if you're only going for stuff over 324 Burnell, you wouldn't have found TC's failure in 2023 in the Bethlehem Steel. So just a takeaway.
Adrian: No, it's a great plug for LFM.
Rhett: Yeah, it's a hot take on how I feel. So talk to me now. So you guys developed this after the first segment, and then you had 12 more segments after that. Can you talk to us a little bit about what was the performance like in the remaining 12 seconds?
Adrian: Well, after with the new classification, going back and regrading with the new classification, we were consistently doing better on digs. Some we're still a little bit out, but that's typical of a hard spot. That's why we need dig.
We can recalibrate those and get in the tolerance. So we can do that. You get the picture up here?
Rhett: We can put it back if you want to compare the type A's and type B's.
Adrian: Because, yeah, this was the one where...
Chris: This is the A and B. This is it, right, guys?
Adrian: Knowing how the class B's are working, there was an original algorithm development. There was a hard spot. This was in the A.O. Smith that always was kind of like stuck in my craw. because it just came out.
It's a much lower hardness than the other two that look very similar to it. And I could never explain what it was. I just had to wash away. It was like, well, metallurgical differences or whatnot.
But by developing the Class B and going back and reassessing it, if you compare the Class B one on the bottom row there and look at the LFM, you can see it's actually a composite of like three different... Class B-looking signatures.
And that is something that we have found with multiples digs, that you can have these stat Class Bs that if you're only looking in the high field, because if you look in the high field over there, it's just looking regular Class A, just call it a hard spot, grade it with the Class A.
But with the Class B, I could break that hard spot up and grade each one of those separately.
Rhett: What he's talking about there, as you can see, audience, if you look, it looks like three fingers that kind of run almost horizontally through the hard spot. So that's really interesting. And you guys haven't. All right.
So the key takeaway from this, and I think it's really important, is that with 30 years of hindsight and new technology, better technology, new procedures in the ditch, you guys are able to refine your process and improve both the identification and now the sizing of features that some might have thought we'd put to bed. And I think that that is really cool. And you guys aren't done, right?
Adrian: No, no.
Rhett: So tell us a little bit. In the last couple of minutes, Adrian, where are you guys going forward now?
Adrian: Well, from the dig results we've gotten, we've, as Chris made me bring up earlier, is that we have seen a manufacturer difference between A.O. Smith type Bs and type. uh from the other manufacturer type b so that's going to be the topic and uh in ppim we're we're presenting a paper on that so we actually have a separate sizing based on manufacturer and that is important important why
Rhett: so now we have sizing per manufacturer
Adrian: yeah
Rhett: and per type
Adrian: yes per type yeah so that's why it is important operators provide us uh as accurate manufacturer detail because that just only refine our process uh sizing
Chris: which is great right i mean the idea is to move away from thing finders That way you can build appropriate responses for each of these features, right?
And that's not anything more than saying we shouldn't dig it. It's rather it allows us to take information and integrate it and say what makes sense.
Rhett: Yeah, you know, but there's a theme here, right? And so I'm going to let the audience know because I don't know the order that the IPC podcast will appear. But there's a consistent theme across all of our IPC podcasts, which is when you give the vendors more data, they can do better things with it, right?
So whether or not, again, for the audience on the Bubnik episode, you're going to hear him talk about if they can identify, they can size better, right? On the marathon episode, you're going to hear a very collaborative discussion between an operator and a vendor to improve an incident.
And I think what's really cool about this is, again, you see that theme of collaborating with, the vendor to produce improved results, improved sizing.
And I'm going to say learn on an issue that many people, I think 15 years ago would have thought we put to bed. If 2015, I think people would have thought we put hard spots to bed.
Chris: I don't know if it's to bed. I think it was, I'll go another way. Maybe this is a hot take. It's a stable flaw, right?
Most of these had already been through some kind of hydro testing, either in the mill or in situ, but we didn't understand the importance of hydrogen, right? So like CP systems and the influence of the environment, right? I would say, I think that shift was-
Rhett: Reverse and flow.
Chris: Yes. And that's what I'm getting to, right? I think it was more, we thought they were stable and now they're not. And now we need to better understand where there's changes. And so what do we always do? We turn to ILI and say, how much more data can you give me so we can do some kind of engineering assessment?
Adrian: Yeah, this is only one of the three legs on that stool there. So it's the HIC that fails, not the hard spot.
Rhett: I like that. That's like the pipe manufacturers where we have problems with welds and the pipe manufacturers are like, well, that's a welding problem.
Chris: Catering to our environment. Have y'all seen the Spider-Mans where they're like pointing at each other? It's kind of like the manufacturers pointing at, you know, the, the, the environment and operations and then operations is pointing at the manufacturer and the environment. It's so good.
Rhett: Well, Adrian, I want to say thanks for coming on and joining us again.
Adrian: It's always a blast.
Rhett: I look forward to it. Maybe it'll be another two years.
Adrian: You guys got Tom out of retirement to do this, huh?
Rhett: We did get Tom out of retirement to come do this. Yeah, so, you know, if I don't know the order again, I never know the order of the episodes. Yeah. But whether, if you haven't heard him yet, you'll need to go listen to him again. It's another, I mean, Tom's always, it's great.
Adrian: Oh, yeah.
Rhett: I aspire to be where Tom or Mike Rosenfeld are one day in my career. And maybe that's a high goal aspiration that I'll never achieve. But I feel like if I can achieve what those guys got in the pipeline industry, I'll be set.
Adrian: Man, aim for the stars.
Chris: You'll just end up out of orbit. Be careful what you ask for.
Rhett: So, again, thanks for joining us. Thanks to TDW and for Williams for putting this paper together at IPC. Audience, I'd encourage you to listen to it. The rest of our audience, hang on. We'll be back in two weeks as we continue our IPC discussion.
And this is Rhett and Chris signing off on Pipeline Things.
This episode of Pipeline Things was executively produced by Kara Turner. Special thanks to our guest, Adrian Bellinger from TDW, and to my life partner and co-host, Christopher DeLeon, and also to Odin's Place and Tomball for letting us film.
Finally, I'd also like to recognize TDW and Williams both for their contributions to the IPC paper and the continuing advancement of knowledge in the pipeline industry.