This episode of the Pipeliners Podcast looks at why leak alarms are considered the most difficult alarm a pipeline controller can face. Host Russell Treat is joined by Angie Schrader, co-founder and COO of Flowstate, to unpack what drives alarm sensitivity and false alarms, and how operators can build stronger leak alarm response and restart procedures. The conversation also touches on API RP 1130 and PHMSA’s Rupture Mitigation Valve Rule.

Leak Alarm Show Notes, Links, and Insider Terms

• API RP 1130 – the American Petroleum Institute recommended practice that defines four criteria (sensitivity, reliability, robustness, and accuracy) for evaluating computational pipeline monitoring, or leak detection, systems.
• Computational Pipeline Monitoring (CPM) – the industry term, defined under API RP 1130, for algorithmic leak detection systems that use flow, pressure, and other operational data to detect hydraulic anomalies in a pipeline.
• Sensitivity – in leak detection, how small a leak a system can reliably detect, and how quickly it can catch a large one, discussed at length as one of the hardest variables to tune.
• PHMSA – the Pipeline and Hazardous Materials Safety Administration, the federal agency regulating pipeline safety, referenced in connection with rupture response timing.
• PHMSA Rupture Mitigation Valve Rule – the federal rule requiring operators to isolate a confirmed pipeline rupture as soon as practicable, and no more than 30 minutes after identification, cited as context for leak alarm response time.
• False alarm (“cry wolf”) problem – the tendency for frequent false alarms to erode a controller’s trust in and responsiveness to a leak detection system.
• Dynamic thresholds – adjustable alarm thresholds used to reduce false alarms during known, non-leak operating conditions.
• Line pack / packing and unpacking – the temporary buildup or release of fluid volume in a pipeline segment during operating transitions, a major source of false alarms and a key risk factor during restart.
• Mass balance calculation – the underlying method leak detection systems use, comparing volume in versus volume out, compensated for hydraulic effects, to identify a possible leak.
• Sub-segmentation – dividing a longer pipeline segment into smaller monitored sections to improve alarm location and confidence.
• Bridger Pipeline – the Wyoming-based crude oil pipeline operator that Flowstate originated from as an internal research and development project, and remains a close development partner.
• Flowstate – the leak detection software company, co-founded by Angie Schrader, that provides the Flowstate Leak Detection System discussed throughout the episode.
• NTSB report on the Enbridge Marshall, Michigan spill – the 2012 National Transportation Safety Board accident report on the 2010 Line 6B rupture, cited as a cautionary example of controllers misreading a line pack event during restart.
• Restart / safe-to-restart procedures – the process of confirming a pipeline is safe to bring back online after a shutdown, discussed as one of the riskiest and least standardized parts of leak alarm management.

Leak Alarm Full Episode Transcript

Russell Treat: Welcome to the Pipeliners Podcast, episode 455, sponsored by Flowstate, providers of pipeline leak detection software built to remove barriers and empower operators. Flowstate helps teams detect anomalies, reduce false alarms, and improve response with practical tools designed for diverse pipeline operations. Learn more at flowstatesolutions.ai.

The Pipeliners Podcast, where professionals, Bubba geeks, and industry insiders share their knowledge and experience about technology, projects, and pipeline operations. And now your host, Russell Treat.

Thanks for listening to the Pipeliners Podcast. I appreciate you taking the time, and to show that appreciation, we give away a customized Yeti tumbler to one listener every episode. This week, our winner is Jamie Stout with Enstar Natural Gas. Congratulations, Jamie, your Yeti is on its way. To learn how you can win this prize, stick around until the end of the episode.

This week we speak with Angie Schrader, with Flowstate, about leak alarms and how they’re the most complex alarm a controller can get.

Hey, Angie, welcome to the Pipeliners Podcast.

Angie Schrader: Yeah, thanks, Russell. It’s good to be here. I’m looking forward to the conversation.

Russell Treat: Yes, me as well. This is… you know, oftentimes on this podcast I talk about things I don’t know very much about. In this case, I know a fair amount about the subject, so fair warning, I guess.

Angie Schrader: Yeah, I’m looking forward to it too. I’ll admit that I’ve only been in the industry for six years now, but I’ve focused purely on leak detection and have really enjoyed learning a lot about it and working with the folks in the industry.

Russell Treat: Well, that’s a great tip. Tell us a little bit about your background and how you found yourself working at a company doing leak detection.

Angie Schrader: Sure, yeah, good question. So I’m a mechanical engineer by background, so not a complete imposter, but I had worked in missile defense for a while. Then I had an opportunity to come back to my home state of Wyoming, where I was working with my current partner on a previous startup doing simulation software for composite materials. That company got acquired by Autodesk, so we learned a lot in that process, not only about running a small company and building a software solution, but also how a world-class organization operates and delivers software.

I was still doing that when my partner was starting Flowstate. It actually started as a research and development project out of Bridger Pipeline. They were kicking around an R&D effort, and when they decided it was ready to grow some legs and become a product, they brought my partner, Jerad, and me in to take it to market and spin it out into a company. So it was a little bit of engineering background, but more project and product management, that brought me into Flowstate.

Our secret sauce is working closely with Bridger to learn about leak detection and develop the product. Having a partnership like that between a technology company and an operator is so valuable, so crucial. It’s hard to achieve for a lot of people.

Russell Treat: Well, look, I brought you on to talk about leak alarms and why they’re so freaking difficult. I think where I want to start is to ask this question to tee up our conversation: what do you think is the most challenging thing about leak alarms in general?

Angie Schrader: Oh, that’s such a great question. My partner Jerad and I joke that when we got into this six years ago, we thought, how hard can this be? What goes in, what comes out? That math seems pretty simple. But it’s incredibly difficult, and every pipeline segment is different. It’s its own beast and presents different challenges. You’re dealing with aged infrastructure, environmental conditions, different operating conditions, and different operators run things differently. You’re dealing with instrumentation error and weather. It’s amazing the variables that get thrown into this. Simple things, like changing densities of the fluid because I’m taking product from place A versus place B…

Russell Treat: Yes, and yes.

Angie Schrader: So you’ve got technology you can apply to this problem, but you’ve got to account for all those various things and make the technology work for that particular circumstance. And each operator is different too, in terms of their expectations of the system, how they want to use it, and their tolerance for risk versus their tolerance for false alarms. So we’re constantly working to strike that balance. Here’s what we can do in terms of sensitivity, but you are going to get false alarms, or here are some things we can do to try to mitigate those false alarms. It’s just a constant challenge.

Russell Treat: Yeah, so I think one of the biggest challenges in this domain of leak detection software is false alarms. And the reason that’s so… I mean, it’s human nature, as you well know, that if I get alarms and they’re always false, then I get to the point where I just ignore them.

Angie Schrader: Yep, the cry wolf problem.

Russell Treat: Most alarms coming from a leak detection system are false.

Angie Schrader: Yep. Ideally, you don’t get the leak alarm, right? And in truth, there are very few leaks, contrary to maybe what we’d gather from the news. There are very few leaks. So yeah, most alarms are going to be false alarms. Our challenge is, as I said, we work closely with the operator to try to dial that in. If we understand the operations and the instrumentation, we can use the technology we bring to the table to mitigate false alarms. We’ve got an advantage with the technology we offer, but we’ve got other tools too, like dynamic thresholds, or we can work with warnings versus alarms. There are different knobs we can turn to try to minimize false alarms for known operations.

But then you get those various things, like the controller decided to change the pump configuration, or we decided to run a pig and didn’t have a pig signal for it. These things come up. So our approach is to work closely with the controller so they understand the system well. We’ve dialed it in so they’re not getting false alarms during known operations, and when they do get an alarm, they have the tools at hand to quickly dive in and see, okay…

Russell Treat: Yep.

Angie Schrader: …I know that was because I lost power on the pump, it dropped, I lost pressure, it gave an alarm, but I know what I’m going to be watching now to make sure that recovers and stabilizes. Some of our operators are more tolerant of alarms because they understand what’s going on and what causes those alarms. They don’t want to compromise on sensitivity to give up those alarms. It’s more like getting feedback from a well-tuned machine that you’re driving.

Russell Treat: Yeah, so let’s talk a little bit about sensitivity, because that’s a term we use in leak detection that means something very specific. We probably ought to unpack that for the listeners. So if you look at API RP 1130, they have four different criteria for evaluating the effectiveness of an alarm system, one of which is sensitivity. Sensitivity, simply stated, is how small a leak you can find.

Angie Schrader: Yep.

Russell Treat: And most of the metering we have in liquid pipelining, if it’s running really well, is plus or minus one percent, reliably. If I try to dial it in more than that, I’m getting underneath the level of error of the instrument. Now I have to do some things to make sure I’m getting rid of all the instrument bias and that kind of stuff. So what would you say… some operators want greater sensitivity, and some operators will dial that sensitivity down so an alarm becomes a more “real” alarm. Where is that border between highly sensitive and not very sensitive? Do you understand the question I’m asking?

Angie Schrader: Yeah, I do. It’s kind of a trap question, I admit. I think of so many different case studies we’ve had over the years tackling this problem. There are different things that cause struggles with sensitivity. First of all, sensitivity isn’t only how small a leak you can catch, but also how quickly you can catch the big leaks. How quickly are you ready to alarm without potentially false-alarming on what was just a transient in your hydraulic operations? We’re constantly trying to straddle both of those, and it really depends on the system.

We take a real bespoke approach to each segment and look at what the limiting factor on sensitivity is for that segment. Is it that there’s a lot of noise in the imbalance? And when I say imbalance, I mean simply the difference between what’s coming in and what’s going out, based on your flow meters. Is there a lot of noise in that, meaning there’s packing and unpacking, or injections coming on, and stuff like that? Then we’re having to work around that noise to bring sensitivity down, and figure out what’s going to drive false alarms in that scenario and how we can mitigate it.

How tolerant are we, versus, say, a transmission line, where maybe it’s a batched line and changes in density are driving the offset differences? That’s the limiting factor on sensitivity there. So it really depends on the situation. We’ll work with the operator to determine how far they want to go. Let’s dial it down as close as possible, as sensitive as we can, see how it’s operating, see what’s driving false alarms, and see how we can mitigate that.

Russell Treat: What do you think the most common driver of false alarms is?

Angie Schrader: In my experience… we’ve got our system deployed on a variety of systems, so you’ve got your transmission lines, but we actually service a lot of gathering lines too. Gathering lines are tremendously difficult.

Russell Treat: Yeah.

Angie Schrader: They have a lot of noise. It’s just a spider web network, a lot of noise. You’ve got limited instrumentation. Sometimes you’ve got great meters, sometimes not. That’s really a driver of false alarms there. In those scenarios, we’re really working to become familiar with the system so we know what scenarios may drive false alarms. Can we either mitigate that, or make it so the controller is very familiar with, oh yeah, I know that when we have to revert to truck dump on this injection, I need to keep an eye on it because I may have false alarms there, or I need to suspend something in that scenario.

Russell Treat: Yeah. So, okay, I’m going to kind of answer my own question, because everything you’re saying I know to be true, but I’m trying to put a little sharper edge on the answer. What I would say the biggest driver of false alarms is, is two things. One, operating transitions. I’m moving from one state to another, starting and stopping. I’m adding an injection, I’m taking an injection out, I’m adding a delivery, I’m taking a delivery offline. That’s number one. Number two is any other kind of operating change.

Angie Schrader: Absolutely.

Russell Treat: Changing the pump speed, changing the pipeline pressure, changing the density of the fluid. All of those things will drive false alarms, because fundamentally what you’re doing is measuring the volume going in, measuring the volume going out, and doing some kind of compensated mass balance calculation for things that are occurring hydraulically in the line as you’re moving the fluid, which makes the math not that simple.

Angie Schrader: Right, yep.

Russell Treat: So that means the operator has to have a level of knowledge of the hydraulics and what things are going to impact the performance of the leak detection system.

Angie Schrader: Yes, which, going back to API 1130, is a robustness conversation.

Russell Treat: Yes. So how, in your experience, how easy is it to get the operator, the person actually on the console running the pipeline, to really understand this?

Angie Schrader: Oh, that’s a great question. We really enjoy… let me put it this way, we enjoy working with operators when we can get them to invest the time in learning about leak detection. And I don’t necessarily mean just learning about our particular software and the methods we use, but just learning about leak detection in general.

When we’re working with a customer or an operator to stand up leak detection, we like to start with thresholds as sensitive as we think are viable and work from there. When you have an opportunity to work with an operator from that standpoint, we’re like, okay, let’s stand up this system, let’s start with these thresholds, and let’s just see how it runs for a few days or a couple of weeks. Then we know what this looks like on startup, what this looks like when we change a batch, what this looks like when we change pump configurations, or something like that.

We provide a lot of supporting trending tools so they can see not only their own operational data, but the output data from our system, so they can see how it’s responding. We find the best success when those operators are in there, understanding not only how our solution is working, but just how their pipeline responds to what they’re doing, and what that means from a leak detection standpoint.

I’ve talked before, at API and elsewhere, about how, if you need to change rate by a thousand barrels an hour, the difference in the hydraulics on your line between a full one-thousand-barrel-per-hour step change versus stepping it up a hundred at a time… the hydraulic effects are significant, and the effects on your leak detection can be the same, in terms of whether or not you’re going to get a false alarm because your system wasn’t ready for that. Being able to work with an operator where they can learn to monitor that is key.

Russell Treat: You’re actually saying something that I think is often lost, which is simply… how I operate the pipeline will impact how the leak detection system performs.

Angie Schrader: Absolutely.

Russell Treat: Because if I turn it on and I just add a thousand barrels, that’s a very different hydraulic profile versus adding fifty at a time over a longer period. So am I going to turn on a thousand now, or am I going to turn on fifty at a time over the next two hours? Those are decisions you make. Maybe you have to operate it one way, but you understand what that’s going to do. From a leak detection standpoint, you know you’re going to get an alarm, but you understand what it’s going to look like and you’re okay with that.

Angie Schrader: So, having that understanding, when we get to that point with operators, it’s so fantastic, because then they start having ideas too. Like, oh, I know we’ve had trouble on startup, will it help to add back pressure? They start leaning into the solution. I’m learning that if I’m correlating how I operate to whether or not I’m going to get a false alarm, how I operate, how I have it instrumented, how I have it segmented… that’s when it’s really rewarding to work with an operator, and you feel like we’re kind of in this together, to find the best solution.

Russell Treat: That is a deep comment. That is a very, very deep comment, because in my experience… this is probably too broad, but I’ll say there are people who think, I buy a leak detection system, I put it in, and I’m good. But the reality is, no, you need to make sure you have a program for managing your measurement and instrumentation and understand how that program impacts leak detection. And you need to make sure your operators are educated in a way that they’re able to operate, and they understand the operating decisions they make and the impact that has on leak detection.

Angie Schrader: Absolutely. And I think that’s where you get to the worst-case scenario for all of us: leak detection vendors, operators, the public, is that there’s going to be a leak. And that first mindset you described, where, okay, I invest in my leak detection, I install it, and I walk away, I don’t look at it because I’m just looking at SCADA, I hope to never hear from leak detection… then you get a leak, you get a leak alarm, and you don’t know what to do because you never used this software. You don’t understand how it works, you’re not looking at the trends, you’re not integrating that into how you operate the pipeline. You don’t remember what’s normal, you don’t remember how to investigate or use this tool.

I think if you’re using it all the time… even if it’s just, we try to provide a way for operators to get some use out of it other than just leak detection, like, okay, you guys brought on a new injection, let’s look at how that’s operating, let’s look at how it’s affecting the rest of the line, or let’s look at how much benefit you got out of that back-pressure valve, is it packing up quicker? The more you use that tool for other things, then in that worst-case scenario, when you get a leak alarm at eleven o’clock at night, you’re using something that’s familiar to you as part of your normal workflow, and you know how to dig in.

Russell Treat: No, that is very, very important. There are other words in my head, but they’re not coming out.

Angie Schrader: It’s a hard job.

Russell Treat: You know, I’ve heard you talk before about human factor science and cognitive overload. Very difficult job. I think all the regulations we have to work toward are all there because they were written to try to help the controller do his job, and do it safely and swiftly.

Angie Schrader: In fact, when we started Flowstate, Tad True, who’s the vice president of Bridger Pipeline… it was his company that we spun out of, his directive to us was simple: improve the life of my controllers. So that’s one of our operating mantras. When we’re implementing a feature or a new design, is this going to help them make that decision better?

Russell Treat: And that is a non-trivial ask. Because what I will say is a lot of engineers and software developers give notional assent to that, but they don’t really understand what it means to sit in that chair.

Angie Schrader: Right, and I think you’re right. When we’re working in isolation, sitting at our desks at a software company, you have all these ideas about, oh, it would be cool to add this, oh, I could do that. Especially now with artificial intelligence, and as things are moving so fast… and even controllers, operators, sometimes they’ll Monday-morning quarterback, oh, it would be interesting if you could provide this information in the drill-down card, or… but that’s their idea in the moment, it’s not necessarily the team’s idea.

Russell Treat: Well, and it’s their idea at the time. And sure, now that you have the benefit of hindsight, it’d be good to have that information on the UI. But do you want it every time?

Angie Schrader: And every time we add one of those, there have been cases where we’ve had these great ideas, and then a little further down the line you think, well, that was kind of a one-off scenario. Now we’ve started to contribute to cognitive overload. We’ve made the UI more complicated without really adding an equivalent level of value. So what we’re trying to do now is provide options for operators to utilize those features at their discretion. If you find it useful to get a warning before you get an alarm, you can do that. If you just want the alarm, you can do that. Making it more flexible so they can use it however it suits them.

Russell Treat: So I want to pivot a little bit. I want to talk about leak alarm response.

Angie Schrader: Sure.

Russell Treat: So a lot of what we’ve been talking about is what you do to manage your false alarms, so my total alarm count comes down. But the other side of that is, when I get an alarm, what do I do? I would say that, given what’s going on with PHMSA and their regulations, particularly the Rupture Mitigation Valve Rule and the expectations they laid out, if you get an indication that you may have a leak, you need to have the pipeline shut down within thirty minutes.

Angie Schrader: Mhm.

Russell Treat: That adds… there’s a lot that goes into that. What is an indication that I may have a leak? Is that a leak alarm, or is that something else? And then how long do I have to perform that analytical process before I have to start shutting down? Is that five minutes, or five hours?

Angie Schrader: Such a great question. And of course, we could probably do a whole other episode on the Rupture Rule.

Russell Treat: Oh, yeah.

Angie Schrader: Yeah, and I probably will, it’s about time.

Russell Treat: Yeah, but yes, absolutely. So I think what I’m driving at, Angie, is what are your thoughts about helping the operator, helping the controller, be more effective at leak alarm response?

Angie Schrader: Sure. Yeah, that’s a good question. So much of a leak detection program is dependent on the operator’s own procedures. We’ve kind of learned that over the years too. We can’t design a system that’s one-size-fits-all, because we’ve got small operators who are kind of a one-man shop, it’s one guy running SCADA but doing field response too, and then we’ve got the big guys who have a control center with ten consoles. So it really comes down to their own procedures.

But how we’ve kind of tackled that problem is by providing access to details about the alarm. My favorite supporting tool is trends. We make trends readily available in our software and our supporting analysis package, so you get an alarm, and it’s one click away to get to a trend of what’s going on. So you can see: was there an event, and if so, do I know what that was? I’ll say one of the biggest contributors to false alarms is data outage. So here you can see, oh, it’s in data outage. Not that there’s not maybe an issue, because I’m kind of blind in one eye, but I know what caused the alarm. So I have a minute to figure out, do I have any other indication that there might be a problem, or is this just a data outage? Because that usually suggests a big leak alarm, because you’ve completely lost output flow.

The other thing with the trends and the information we provide: do we think this is a highly… like a ten, twenty percent of your flow rate? Or do we think this is a small leak rate where we’re right at that noise threshold? Say you wanted one-percent sensitivity and were alarming at one percent, that changes your approach. Are you looking at a ten-percent leak? You probably need to go into a protocol where you have five minutes to make a judgment decision, or you’re shutting down. Or is it a one-percent leak, in which case this is going to take a minute? I’m going to look at trends, I’m going to look at pressures, I’m going to make some calls, I’m going to look at whether I’ve got some time to look at other things that may be causing an issue.

Russell Treat: I think there’s another thing you can do too, and I don’t know if you all do this, but if you think about a simple example, like a transmission pipe… if I define the pipe as a segment being from a meter to a meter, and then I have another segment that’s a meter to past the next meter, to the meter following that, right? So now I have three segments, all of which somewhat represent the same flow. So now, when I get an alarm, if I get an alarm in a short segment, I can look and see, well, am I getting the same alarm in the longer segment? And that’s a way to have more confidence that it’s a real alarm.

Angie Schrader: That’s a great scenario. We have done that with one of our operators, where we do like to sub-segment a larger segment, so you get that higher resolution and location. They’ve known, okay, we know that this substation has com issues, so we’ll do those two sub-segments, but we’ve got the overall system coverage as well.

Russell Treat: Yep, that’s a great scenario. And again, if you’re keeping an eye on your operations, your system health, alarm history, and so on, you can start to understand where you have those vulnerabilities and think about what your Plan Bs are in those scenarios.

Angie Schrader: Yeah.

Russell Treat: There’s a lot bound up in that conversation. We could probably do a whole episode just on leak alarm response. I think trends are absolutely critical, and not just trending pressures and flow in the pipe, but trending whatever the analytical output of the CPM is.

Angie Schrader: Yep.

Russell Treat: That’s really critical, because we as human beings are way better at seeing trends and determining what that means, recognizing a pattern, or seeing a pattern that looks like something we’ve never seen before. We’ll see that way before we’ll get there looking at numbers or alarms alone. That’s critical. I think the other thing is, I need a way to know if I’ve had an operating change occur, and I need to know that really, really quick. Did somebody turn a pump on? Did they turn a pump off? Did a delivery come on into the system, or go off?

Angie Schrader: And we look at trends over various scales. Nominally, we provide at minimum a one-hour trend. But for a small leak, one hour isn’t good enough, because our system’s probably been tracking that leak for over an hour before it alarms on a very small leak rate. So you’re going back hours, maybe days, to look at when this started having an issue. Did it start tracking all of a sudden, in which case that may be a leak, let’s start digging into that? Or have we started tracking more often for several days now, and that lines up with when the temperature changed from forty degrees outside to ninety degrees outside? Maybe we’ve got a meter issue. So, yeah, looking at long-term trends, we do provide a trend of our leak detection system output, so you can see, was this thing really never tracking before, and now all of a sudden it’s tracking all the time? That’s probably pointing toward an instrumentation issue, versus something that just happened, like a pinhole leak.

Russell Treat: Well, I think there are even things you could do on the instrumentation, where you start trending the hysteresis on the instrumentation, so you know it’s acting flaky.

Angie Schrader: We do. Change in meter offset is one of our biggest challenges, so we do track that over time so we can see, well, yeah, it does look like that was starting to drift a few days, weeks, or a month ago. Same thing on pressures and other densitometers and other things.

Russell Treat: Here, I always want to make sure I make this statement, because I think it’s really important for operators to understand. I want to set some rules of thumb that I think operators ought to consider as guidelines. One is, if I’m running a transmission line and I get a leak alarm, I should be able to diagnose and make a shutdown determination in ten to twenty minutes. I want your response to this, tell me if you agree or disagree.

Angie Schrader: So ten to twenty minutes, again, it’s difficult if it’s that small leak in the one-percent range or something.

Russell Treat: If I get a leak alarm, I should be able to make a shutdown decision in ten to twenty minutes.

Angie Schrader: Okay, I agree with that. Because if I’m the controller, and maybe I’ve decided this is a one-percent leak, I don’t see anything that’s happened, I’m going to spend some time investigating this… if I can’t make that decision in ten to twenty minutes, I need to adjust the sensitivity.

Russell Treat: I agree with that one hundred percent. That gets back to, if I need to see something at the one-percent level, I’m going to handle that as a warning, and that diagnostic is different.

Angie Schrader: Agreed. I absolutely agree with that. And that gets back to, do you understand your system and your leak detection solution.

Russell Treat: Okay, now, here’s the corollary. In a gathering system, if I get a leak alarm, I should be able to make a determination whether or not that’s a real leak within one to two hours, tops.

Angie Schrader: That one is definitely more challenging, but one to two hours, I think, so…

Russell Treat: I think less would be better.

Angie Schrader: Less would be better, but one to two hours is reasonable.

Russell Treat: So if I find myself in a situation where I don’t know, and I’m in transmission and it’s longer than twenty minutes, then I should shut down. And if I find myself in a situation where I’m in a gathering system, and I don’t know, I got a leak alarm but I don’t know if it’s real, and I’ve had that going on for one to two hours, then I should shut down.

Angie Schrader: I revise my answer, Russell, I absolutely agree with that. I think about our customers who have transmission lines, and the wheels were turning as you were elaborating. I think most of our operators do have that approach, that if you don’t understand what’s going on, you can’t make a decision within twenty minutes, you’re shutting down. I think all of our operators of transmission lines, that’s how they operate. So we know we need to support that workflow. We need to present them with the information they need to make that decision, or they’re shutting it down. Obviously, there are various operational and business impacts to shutting down, so that’s a big deal.

Russell Treat: It is a big deal, and there’s a lot bound up in the decision. There’s another declarative statement I want to make before we wrap. The point I’m trying to drive home here for operators is, I think it’s very important that you understand, if I’m in that situation of I don’t know what’s going on, I cannot live there very long. Any incident, particularly those with significant consequences, and significant fines and remediation costs, somebody lived in that “I don’t know what’s going on” for too long. This goes for any kind of alarming, but it’s more critical in leak alarming, because leak alarming is more squishy than other kinds of alarms. We always encourage operators to have a statement that says, if you’re a controller and you don’t understand the operating condition, you’re required to shut down.

Angie Schrader: Yeah.

Russell Treat: So if I get a leak alarm and I don’t know why I have it, I don’t understand the operating condition, I’m required to shut down. And then, how long can I be in that? Okay, so I want to make one more declarative statement. I think the riskiest part of leak alarm management is the restart after a shutdown following a leak alarm. So how do you deal with that, if at all, in your leak detection approach?

Angie Schrader: I should have known that one was coming, because I know it’s one of your favorite questions. You’ve probably heard me ask it before.

Russell Treat: I have, but I’ve never recorded you, so here we go.

Angie Schrader: Yeah, that’s tricky. Again, I think being familiar with the system and the information we’re providing matters. We provide trends on everything that’s happening, including pressure. So if you’re holding pressure… I think that’s one of the first things our folks look at. And again, were you able to determine what caused that alarm? Do you think it was your pump configuration, or a data outage, or unknown? And if you start up, what are you going to do? Are you going to be watching pressure? I think our job is to provide the information they need to aid in that decision.

What I was going to say earlier is, our hope is that in improving the life of the controller, which is our directive, when they get that alarm and have to make that shutdown decision, one of the first things they go to is our software. Why did I get that alarm? What made the system think it was a leak? What do my trends look like? Similarly, we can provide supporting information for that startup procedure.

Russell Treat: I haven’t thought about that, but there’s no reason you couldn’t use a CPM for modeling pressure behavior in a shut-in pipeline and get a second level of assurance that you’re good to restart. I don’t know if you’re doing any of that. Are you aware of anybody doing anything like that?

Angie Schrader: Doing what specifically?

Russell Treat: Let me frame this a little bit. So if I had a leak alarm, I shut the pipeline in, I would expect pressure to do what?

Angie Schrader: Well, it depends. At a basic level, for a short period of time, I would expect pressure to hold. For a longer period, depending on ground temperature and product temperature, I would expect that pressure to build, probably build, because the product’s going to cool down, and as it cools down, it’s going to get more dense, and that’s going to cause more pressure.

Russell Treat: I’m getting a little over my skis here, but what I’m getting at is, you could have a restart, a safe-to-restart notification, based on, I’m holding pressure.

Angie Schrader: Yep, and in the few scenarios where we’ve actually had that conversation with an operator, that’s been their approach: okay, we’re going to restart, but we’re going to really watch pressure, and if we’re not holding pressure, then we’re going to go walk the right of way, or whatever.

Russell Treat: I could create a procedure or a policy that says, if we do a shutdown based on a leak alarm, then we stay shut down for this duration, we verify that we’re holding pressure, and then we’re safe to restart. The other scenario that keeps me up at night is packing. When you have a startup and you’re having to pack the line, and you have that temporary difference between what’s coming and what’s going out… I’ve read the NTSB report on the Enbridge Marshall, Michigan spill, and those poor controllers thought they were just packing the lines, so they kept pumping and pumping and pumping. I think being able to look at, okay, what does a normal startup on this line look like? That packing should take five minutes, twenty-five minutes, forty-five minutes, whatever it is, and this time it seems like it’s taking longer… nope, let’s shut it down, this isn’t looking right, we’re not pressurizing like we normally do.

Angie Schrader: Yeah, and again, that’s something else a good leak detection system can help with, because it’s really not leak detection we’re trying to do, it’s leak prevention. Leak detection is one aspect of that, but another aspect is all the things you do on restart.

Russell Treat: Right. Yeah, well, I think that’s a good place to wind it up. So my takeaway from this conversation is, I need to quit thinking about leak detection and start thinking about leak prevention. What does that look like for a CPM?

Angie Schrader: Yeah, and I think, unfortunately, leak alarms mean you’ve had an issue probably, but if we can provide you tools to help you operate safer and have a leak prevention mindset, I think that’s an overall improvement.

Russell Treat: Yes, a leak prevention mindset, because when you think about leak detection, that gets you all focused on generating the alarm. If you think about leak prevention, it’s, well, how do I make sure I’m not getting alarms that are false, and when I get an alarm and need to restart, how do I make sure I’m safe to restart, and how do I build all that capability into a CPM?

Angie Schrader: Agreed. Boy, that’s such a great place to start. This has been a great conversation. I could talk about this for another couple of hours.

Russell Treat: Me too, me too.

Angie Schrader: It’s been really rewarding, and I’m passionate about the topic, so I appreciate having the chance to discuss it with you.

Russell Treat: I know you are as well. Yeah, thanks, Angie, for being our guest.

Angie Schrader: All right, take care.

Russell Treat: I hope you enjoyed this week’s episode of the Pipeliners Podcast and our conversation with Angie. Just a reminder, you should register to win our customized Pipeliners Podcast Yeti tumbler. Simply visit PipelinePodcastNetwork.com/win and enter yourself in the drawing. If you have ideas, questions, or topics you’d be interested in, or if you’d like to be a guest, please let me know on the Contact Us page at PipelinePodcastNetwork.com, or reach out to me on LinkedIn. Thanks for listening. I’ll talk to you next week.