The Pipeliners Podcast is excited to deliver a series of episodes with Giancarlo Milano of Atmos International. In this fourth episode of the series on leak detection, Russel Treat and Giancarlo discuss rupture detection.
In this episode, you will learn what type of leak classifies as a rupture, the importance of your systems and measurements to provide visibility and alarming if a leak reaches the level of a rupture, how to properly address a rupture in a complex system, and other key topics related to pipeline ruptures.
In the next episode on leak detection, Russel and Giancarlo will complete this series by discussing leak alarm response and training.
Rupture Detection: Show Notes, Links, and Insider Terms
- Giancarlo Milano is the Senior Simulation Support Engineer at Atmos International. Connect with Giancarlo on LinkedIn.
- As part of this series with Giancarlo, enter to win our free book giveaway contest for the “Introduction to Pipeline Leak Detection” book by Atmos founders Michael Twomey and Jun Zhang.
- Leak detection systems include external and internal methods.
- External methods are based on observing external factors within the pipeline to see if any product is released outside the line.
- Internal methods are based on measuring parameters of the hydraulics of the pipeline such as flow rate, pressure, density, or temperature. The information is placed in a computational algorithm to determines whether there is a leak.
- Rupture detection is the process of quickly and reliably identifying a rupture in a pipeline and shutting down the pipeline to prevent catastrophic events.
- A Rupture Detection System (RDS) uses algorithms and patterns to provide the pipeline operator with accurate information to shut down a pipeline in the event of a rupture.
- A pump trip is a pattern in the system that helps detect the possibility of a rupture in the pipeline.
- Dynamic Model Analysis (DMA) is a pattern that detects ruptures on all areas of a pipeline.
- The Inventory Method for rupture detection is used to detect pipelines that go slack – e.g. what is coming out of the pipeline is less than what is going in.
- The Low Friction Factor Method for rupture detection is used near pump stations where the product is easier to pump.
- Slack line flow occurs when the pressure in a pipeline falls below the vapor pressure of the liquid in the pipeline during abnormal operating conditions.
- The San Bruno or PG&E Incident in September 2010 refers to a ruptured pipeline operated by the Pacific Gas & Electric Company. The rupture created a crater near San Bruno, California, caused an explosion after natural gas was released and ignited, and resulted in fires causing loss to life and property. [Read the full NTSB Accident Report.]
- The Marshall Incident refers to the Enbridge Incorporated Hazardous Liquid Rupture and Release incident, which occurred on July 25, 2010, in Marshall, Michigan. [Read the full NTSB Accident Report.]
- The Macondo well is a large reservoir of natural resources in the Gulf of Mexico off the coast of Louisiana. The Deepwater Horizon drilling rig was set up for deepwater exploration before exploding in April 2010.
- API 1130 is a recommended practice published by the American Petroleum Institute and incorporated by reference into the U.S. pipeline regulations in 49 CFR 195.134 and 49 CFR 195.444 for how pipeline operators should design, operate, and maintain their computational pipeline monitoring (CPM) systems. While this standard was not discussed during the podcast, it is a critical document for any pipeline operator with CPM-based pipeline leak detection.
Rupture Detection: Full Episode Transcript
Russel Treat: Welcome to the “Pipeliners Podcast,” episode 27.
Announcer: The Pipeliners Podcast, where professionals, Bubba geeks, and industry insiders share their knowledge and experience about technology, projects, and pipeline operations. Now your host, Russel Treat.
Russel: Thank you for listening to the Pipeliners Podcast. We appreciate you taking your time, and to show our appreciation, we’re giving away a customized YETI tumbler to one listener each episode. This week, our winner is Michael Pruitt with Boardwalk Pipelines. Congratulations, Michael, your YETI is on its way.
To learn more about how you can win this signature prize pack, stick around to the end of the episode.
Giancarlo, welcome back. This is now the fourth episode in our leak detection series, and we’re going to talk about rupture detection. Welcome aboard.
Giancarlo Milano: Thank you, Russel. That’s right. There’s been a lot of talk about leak detection over the last few weeks. Great to be here once again.
Russel: We’re talking about rupture detection this week, so how is rupture detection different than leak detection?
Giancarlo: First of all, let’s start that a rupture is a leak. We just have to get that right off the bat. Now the difference between a rupture and a leak is that a rupture is going to be a very large, very significant leak that you have on your pipeline.
Now, when you have a rupture, the damages that are being caused on the environment or the surroundings around the pipe are going to be or could be catastrophic. Whether it’s a gas pipeline or a liquid pipeline, if you have a rupture, you want to make sure that you know that you have a rupture right away. You can’t just wait until your leak detection system takes its time to alarm or someone is walking by and report it. You have to make sure that as soon as that rupture occurs in your pipeline, you get an immediate response.
The other thing is you got to make sure that the operator is trusting that response that you’re getting from your leak detection or rupture detection system. That’s a big difference when we talk rupture.
Russel: Probably for the listeners this might be helpful. Historically, the focus in leak detection has been on the ability to find a smaller leak and to locate that leak. That’s where the focus has been. We’ve been doing things to optimize systems to find smaller leaks. Where the focus on rupture detection is certain knowledge I’ve got a rupture, so that’s something different than leak detection.
Giancarlo: Absolutely. When you look at history, every project that it’s involved in, the leak detection, the consumer is always looking at the smallest leak size that they’re able to find in the fastest time possible. By doing that, in a way they are sacrificing the big leaks. When you’re looking for a big leak and you’re also looking for a small leak within the same system, you are sacrificing the reliability of that system.
If you’re looking for the small leak, you’re going to be more prompt to false alarms, and whichever technology you’re using, there’s going to be false alarms a month a year, depending of course of how well those systems are tune-optimized.
Unfortunately for rupture detection, that’s not good enough. When you have a rupture, you want to be able to trust that signal, that alarm, 100 percent. You want to be able to shut down the pipeline without question.
In some cases, the ideal situation is when your rupture detection system, it’s tied up to your control system. If you receive that rupture alarm, then your pumps are shut down, your valves are closing to isolate the section where that leak is contained — or for the whole pipeline. You’re able to minimize the catastrophic effect that that event will have in your pipeline.
Russel: To take it back to the API 1130 stuff we’ve been talking about about reliability being false alarms and sensitivity being the ability to identify and locate a small leak, the focus here is on reliability. We want something that’s so reliable that if we get that alarm we know we have a problem and we’re comfortable automatically shutting things down.
Giancarlo: That’s the most important part of rupture detection. There has to be no doubt from the operator’s perspective that something is happening in your pipeline, and they can go ahead and shut that down safely and without doubt.
Russel: With the focus there being without doubt. Safety’s always a focus, but in this particular case, it’s the without doubt that’s really material when you start talking about rupture detections.
In leak detection, there’s always a process I go through to determine whether or not the alarm is valid or not. In rupture detection, that process should not be required. It should just be valid.
Giancarlo: You want to make sure that if you receive that alarm, you shut down the pipeline no questions asked.
Russel: Let’s talk about what are some of the techniques you can apply for rupture detection.
Giancarlo: There’s two techniques that have been applied through history. One of them being the systems are able to detect a rupture within the SCADA and the other one where you have a more dedicated rupture detection algorithm that is looking at different signals or different patterns in order to react to a rupture.
When you’re looking at the rupture detection within the SCADA system, you’re looking at the flows and the pressures and identifying patterns that could be representative of a rupture, a large leak. When you’re doing that, then you’re looking at pressure drops, flow drops, or drastic flow increases that will be representative of that rupture happening in your pipeline.
In some instances, the process signals such as the flow and the pressure, they could also be combined for other patterns such as pump trips or maybe let’s say a drastic increase on the pump speed, or maybe a control valve closing drastically to make up for the set point that it’s trying to meet.
Using these variables that come into the SCADA, an operator can write a simple algorithm or script within the SCADA to be able to identify these instances and alarm for a rupture. That’s one of the ways of doing it.
Russel: If you think about a liquid pipeline, and I’ve got pumps and flow control valves, if I get a leak, depending on the control methodology I have in place whether I’m upstream and I’m controlling to a pressure point, and downstream I’m floating, or I’m controlling to a flow signal or something, the method of control is going to dictate the leak signature.
If I’m controlling on pressure, and I get a leak, my pump’s just going to pick up and pump harder.
Giancarlo: The RPM, you’re going to see that pressure drop. It’s no longer meeting that setpoint that you have set, so the RPM of the pumps will increase. Maybe if you have an intermediate pump station, you see that drastic large pressure drop passing through a station, what’s going to happen is the pump is going to trip all of a sudden.
In many instances, there has been cases where an operator sees the pump tripped. They restart that pump, and not to realize that what they’re doing at that moment is just feeding a rupture. That product is not going anywhere other than the environment.
By being able to identify these patterns, one can be able to have some sort of rupture detection within your SCADA system to take care of these instances.
Russel: Absolutely. Like everything else, it’s not as easy as you might think. One of the advantages of thinking about rupture detection is the absolute signal quality is not a big a deal because it’s about change.
Russel: I’m looking for changes that cannot be explained given the operating condition I’m in, flow control valves that are pinching down, pumps that are speeding up, pressure changes across a station, those kinds of things that can’t be explained. In particular, rupture detection can be very helpful when things are steady state because it’s easier to pick out the change. It’s more difficult in transients.
Giancarlo: We have transients, that’s when you’re looking at a more dedicated type of rupture detection system. You’re just basing the leak detection system or your rupture detection system on the SCADA signals alone might not be enough. You might need a more details algorithm that are able to pick up on different patterns for your pipelines.
When we’re talking about this dedicated rupture detectors, there’s few types. You have a type that is called dynamic model analysis (DMA). What that’s doing, it’s looking at the patterns of operation of flow and pressures at the individual stations as well as other equipment, not only flows and pressures, but also pump speed and the control valves.
Then there’s another method that’s called the inventory method. What that one is doing, it’s identifying a rupture when your inventory on your pipeline changes drastically.
Then, there is a third method, which is the low friction factor method. The way that it will work is if you have a rupture in your pipeline, then the friction within your pipeline is not going to be as high and more product is going to be able to flow a lot more easy into the pipeline. That will be able to be picked up by the rupture detection module and alert on those patterns and conditions.
Russel: It’s like thinking about that as flow versus back pressure. If my flow is increasing and my back pressure is dropping and I’m not filling a line, I probably got a problem.
Giancarlo: That is correct, yes. On the dedicated rupture detection, the way that these three algorithms work is that they’re all active at the same time.
Depending on the topography of the pipeline, depending on the fluid that you’re moving, or the way that you operate your pipeline, it could be that the dynamic model analysis method is not good enough. It might not be able to pick up rupture, but the inventory or the Low KL method will.
In a way, by having three different methods that are working independently but together at the end, will be able to cover you in the case that you have rupture.
Russel: The other thing to be aware of if I’m looking at these patterns, if I’m looking at them on an upstream pump and I’m pumping into a segment, that’s going to look very different than if I’m looking at a suction pressure of a downstream station.
When you look at these things, when you think about it, if I have a rupture closer to the upstream pump, it’s probably going to be easier to see than if it’s very close to the end of the segment. The flip side of that is I probably don’t have as much pressure towards the end of a segment.
Giancarlo: Correct. The changes are going to be fewer, and it’s going to be harder to identify.
Russel: That’s exactly right. That’s exactly right. I think one of the things that certainly I’ve seen in conversations I’ve had about rupture detection is there’s a lot of, “Well, how big’s a rupture?”
Giancarlo: The size of a rupture?
Russel: Yeah. What’s the size of a rupture?
Giancarlo: That is a big question. It’s a million dollar question. There is not a straight answer to that.
Russel: I assert to you that’s probably more like a $10 to $100 million question.
Giancarlo: There’s no simple answer to it. It’s going to depend on your pipeline and your operation. It’s going to depend what type of fluid you’re moving in your pipeline, highly volatile fluids that are operating at the vapor pressure, maybe a pipeline that is flowing based on a gravity at a slope, and you have the outlet, you seen a lot of product volume moved through that segment.
It’s obviously going to be dependent as well on the complexity of your pipeline and whether you’re working on a single transmission line, or a gathering, or a distribution system. How much can you afford to lose before you call it a catastrophe?
The bottom line is that the smallest leak is going to be bad enough. When it comes to identifying a rupture, it could be classified as anywhere from 20 percent of the nominal flow rate to 40 percent of the nominal flow rate, and obviously to 100 percent of the flow rate if your pipeline just, it’s broken and breaks in two pieces altogether and no product is going to pass that segment.
Russel: What you said earlier about when does it become catastrophic, that’s a material question for an operator to ask.
That question is not so much about the size of the leak as it is the consequence, and to some degree, the visibility of the leak. What I mean by that if you think about San Bruno, which happened in the San Francisco area, or you think about Marshall, which happened in the Illinois area, both of those were very large and had a lot of press.
When you look at those and you ask, if you’re a member of the public and you don’t know anything about pipelining, the question that’s going to come to your mind is how is it that you didn’t know that that was occurring.
The analogy being if we lose an airplane, we know immediately we lost an airplane. Everybody knows. If you lose a commercial airplane, everybody knows you lost a commercial airplane. The people in the control centers and all that know that before anybody else.
Question would be for a pipeline operator is how could you not know. RDS is really designed to address that issue of we did know and we were taking action.
Giancarlo: It’s very important.
Russel: If it was pipelining according to Giancarlo, what would be your approach to doing rupture detection? How would you approach it?
Giancarlo: I would definitely go with a dedicated type of rupture detection where I’m using different methods in order to determine that a rupture has happened in my pipeline and making sure that I have different methodologies in place that will make my system failsafe.
If one doesn’t cover it, then I’m going to be certain that another methodology that is running in the background as well is going to be able to pick that up.
When we’re talking about the rupture detection, the things that would be more important to me is that we’re able to detect ruptures, as you mentioned, at timely manner.
They have to be detected as fast as possible because you’re not talking about a few barrels of product. You’re talking about a lot of volume that’s coming out of your pipeline. The faster you react, the less of an effect on the environment you will have.
The other one, I want to make sure that I don’t have any false alarms, whether I have instrument failures, or communication, or the SCADA system failing as well. I want to make sure that if any of those happens, that rupture detection module is not going to be giving me a false alarm because of those.
The other thing is transient operations. When a pipeline is in transient conditions, that’s when you’re giving the pipeline the most stress. When that’s happening, that’s when the pipeline is going to be more subject to having a rupture due to those drastic conditions changing. I want to make sure that I cover those.
Another factor that we have spoken very little through the series is slack conditions or maybe operations when we’re draining or filling. During those situations, I also want to make sure that I’m covering for rupture.
Having a way that is able to tell me to account for a rupture during all of those conditions would be of high importance to me and I’m sure that to any operator up there.
Russel: The way that you should approach that is you start by saying, ‘Okay, what does a rupture look like in steady state conditions?’ Then you start tuning your algorithm. You say, ‘Okay, yes,’ but if we’re starting up, then I need to say, ‘Yeah, but if it looks like this, that’s not a rupture.’ If I’m shutting down, I need to say, ‘If it looks like this, that’s not a rupture,’ so that you end up tuning your model across those transients. The hard part about this is without real data on real ruptures, this can become very hard to do.
Giancarlo: When we’re tuning a leak detection system, we’re tuning to a false alarm free system that the operator will be able to trust just from looking at the small leak. When you’re looking for rupture, you’re raising that threshold, and you still want to maintain the false alarm but in a reliable way, 100 percent. Now, we cannot have any false alarms.
The best thing that you can do is tune the system with the operational data that you have and just raising the threshold to make sure that there’s no false alarms during those conditions. That includes the operations. That includes the communication failure, the SCADA failure. Anything that could go wrong needs to be taken into account in the data that you have collected.
Unfortunately, nobody wants to have a rupture just for the purpose of tuning the system. You have to do the best you can with the data that you’re able to collect.
Russel: That’s right. That’s a big challenge. That’s really a big challenge. Some of the techniques that work well for tuning the real-time transients and the statistical volume balances, they don’t work really for looking at a rupture. It’s just different. What you’re doing is just different.
Giancarlo: All of these systems are looking at leak size. They could be configured for high or large leak size, which would consider a rupture, let’s say anywhere in the 20, 30, 40 percent and up. You have to make sure that there’s no false alarms during those instances.
Basing the rupture detector just on the process itself, it’s not enough. You got to take into account other conditions to make sure that you’re covered all across the board.
Russel: You made the comment about slack line. If a pipeline can be operated without slack line, that’s preferred. That’s not always possible, but certainly it makes all these things that we’re talking about doing throughout this whole series, it’s easier if you never have slack.
Giancarlo: Absolutely. Not having the slack on the pipeline, that’s the ideal condition, especially when you’re get into high elevation or drastic elevation changes. That’s when the slack comes in.
It also depends on the type of fluid that you’re moving and the vapor pressure at which they run. If you have critical products pipeline such as ethane or methane versus the heavier crudes or gasolines, refined product, that’s going to behave slightly different.
All of these things, they need to be taken into account when you are optimizing the eruption detection methods for your particular pipeline and operations. Very important. Again, reliability, no false alarms, 100 percent trust of the operator, that’s the most important thing.
Russel: My experience with rupture detection, and it’s not very broad, but my experience with rupture detection is most of the rupture detection approaches are developed by the pipeline operators themselves. I’ve not really seen a lot on the market that I would call a commercial rupture detection. Is that your experience as well, or what’s your experience with that?
Giancarlo: Yes. It’s not something that’s been available for a long time.
Within Atmos, we developed a rupture module because a few years back, anywhere from four, five years ago if not a little more maybe, we started seeing a lot of emphasis on ruptures. It goes back years of course. Around that time period, anywhere between five, six years ago, there was a big emphasis on rupture detection. At least, that’s when we picked it up.
We said that we have to do something about this that we can provide our customers a system where yes, we can do leak detection, but let’s take it one step further and develop this module so we’re able to 100 percent be able to tell them that they have a rupture in their system and react accordingly.
There’s not that many companies out there that provide a rupture module. The game, it’s evolving when it comes to leak detection. It certainly has been evolving over the last three to five years, and I think it’s going to continue to do so.
Russel: Certainly 2010 was an important year in pipeline safety because we had a number of large incidents that got national news attention. They all occurred relatively close together and at the same time that we had the Macondo Well in the Gulf, which has nothing to do with pipelining, but the public doesn’t necessarily make that distinction.
The thing about rupture detection — the people that I’m aware of that are applying it — their typical method of application is if I get one of those alarms, I automatically shut down. Now that may not mean shut down and sectionalize, but it means I quit operating that line and treat it as if I have a rupture until I know certainly I do not.
Giancarlo: It was certainly a wake-up call in the industry to say hey, we have to make sure that we cover this. We’re doing everything possible that we can to be aware of these situations and shut down as soon as possible.
Russel: If you look at San Bruno, most of the damage from San Bruno was well after they knew they had a leak. It took what again I would say the public would consider unreasonably long to turn the gas off. If I can have a system that says that’s a leak and it turns the gas off automatically, then that’s a much better place to be.
Giancarlo: The faster you can react, the better that you’re going to be. Maybe better is the wrong thing to say because at the end of the day that release is happening. It would definitely be better for the communities around it that are experiencing that to shut it down and minimize it so it won’t be as large.
Russel: Exactly. I think again this is a good place to wrap up this episode. We’re going to do one more episode in this series, and what we’re going to talk about, we’re going to change the tables a little bit.
We’re going to have Giancarlo interview me, and I’m going to attempt to talk about leak alarm response, in other words, what happens in the control center in the pipeline operation when you get one of these alarms. Getting them is one part, but responding to them is equally as important.
Giancarlo: Most definitely. I am very interested to listen what you have to say about that. From my point of view, it’s always been, ‘Okay, let’s make sure we give the alarms to the operators so they can rely on them and trust them.’ It’s another picture when you’re looking at it from a different side and see what do you do with it. How do you interpret that in your SCADA and your alarm management system? What kind of action you’re supposed to take? Looking forward to interviewing you on that, Russel.
Russel: Awesome. I’m looking forward to that as well.
For the listeners, Atmos is being very generous, and they are offering a book giveaway. They’re giving away the book, “Introduction to Pipeline Leak Detection.” The first five listeners that go to the website for Pipeliners Podcast for this episode and click through the link in the show notes will have the opportunity to just sign up and get a copy of that book.
Please do that, and please hurry. I’m going to tell you I highly recommend this book. I’ve read it. I think it’s an excellent overview of leak detection beyond the things we’ve been talking about around the software approach to detecting leaks. I would encourage you guys to go do that.
Giancarlo, thank you very much. We’ll be with you again next week.
Giancarlo: Absolutely. Thank you very much for having me, Russel, and thank you to the listeners out there.
Russel: Hope you enjoyed this week episode of the Pipeliners Podcast and our continuing conversation with Giancarlo Milano of Atmos International about leak detection. Just a reminder before you go, you should register to win our customized Pipeliners Podcast YETI tumbler. Simply visit pipelinerspodcast.com/win to enter yourself in the drawing.
Russel: If you have ideas, questions, or topics you’d be interested in, please let us know on the Contact Us page at pipelinerspodcast.com or reach out to me on LinkedIn. My name’s Russel Treat. Thanks for listening. I’ll talk to you next week.
Transcription by CastingWords