For the first time on the Pipeliners Podcast, host Russel Treat dives into the fundamentals of pipeline leak detection with guest Giancarlo Milano of Atmos International.
In this episode, you will learn about the fundamentals of leak detection including external and internal prominent approaches to leak detection. Then, Giancarlo walks listeners through various computerized approaches beginning with basic mass balance through to the more complex real-time transient model (RTTM) for leak detection, Statistical Volume Balance, and Negative Pressure Wave.
Also, you will learn about the latest leak detection technology that is being used in the field and the importance of adhering to the API 1175 standard. Download this informative episode today!
Pipeline Leak Detection Fundamentals: Show Notes, Links, and Insider Terms
- Giancarlo Milano is the Senior Simulation Support Engineer at Atmos International. Connect with Giancarlo on LinkedIn.
- Find the book “Introduction to Pipeline Leak Detection” by Atmos founders Michael Twomey and Jun Zhang on Amazon.com.
- 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.
- The Real-Time Transient Model (RTTM) simulates the behavior of a pipeline using computational algorithms. The model, which is driven by the field instrumentation, monitors discrepancy between the measured and calculated values potential caused by a leak. RTTM uses flow, pressure, temperature, and density among many other variables.
- Statistical Volume Balance is a method using the volume in and out of a pipeline, along with pressure changes to account for the pipeline inventory in real-time. This method is also capable of detecting smaller leaks while coping with transient conditions. A statistical approach using Sequential Probability Ratio Test (SPRT) evaluates the probability of a leak in the pipeline.
- Negative Pressure Wave is a method to detect the occurrence and location of leak incidents in a pipeline based purely on the pressure drop due to a leak as it travels up and down the pipeline. Three core technical challenges include data quality, dynamic slope, and false alarms causing changes to normal working conditions.
- API 1175 is a recommended practice published by the American Petroleum Institute addressing how pipeline operators should maintain their leak detection program. The goal of the standard is to have the best leak detection system possible by always looking for continuous improvements to the individual LDS components achieving operational buy-in with the culture, strategies, KPIs, and testing.
- 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.
Pipeline Leak Detection Fundamentals: Episode Transcript
Russel Treat: Welcome to the “Pipeliners Podcast,” Episode 14.
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: Thanks for taking the time to listen to the Pipeliners Podcast. We’re offering a free, customized YETI tumbler to one listener every episode. The winner this week is Elizabeth Oakley with Kinder Morgan. Congratulations, Elizabeth. If you’d like to know how you could win a free, customized YETI tumbler, then stick around and we’ll tell you how at the end of the episode.
This week we have with us Giancarlo Milano. Giancarlo is a leak detection engineer with Atmos International. He’s going to help us learn about the fundamentals of pipeline leak detection. Giancarlo, welcome to the Pipeliners Podcast, so glad to have you.
Giancarlo Milano: Very good to be here, Russel.
Russel: Why don’t you tell us a little bit about yourself and how you got into the pipeline leak detection business.
Giancarlo: I’ve been in the industry for about 12 years now. I have a background in electronics from the University of Southern Mississippi. I moved from Mississippi to L.A. and that’s when I found Atmos, or rather Atmos found me.
I started working with leak detection in projects. From the very ground up, just working in leak detection projects and also hydraulic simulation. I had a little bit of background within my studies on hydraulic simulation, communication, fluid dynamics. That gave me a pretty good understanding of how a pipeline works. I was able to take that knowledge and apply it to the technology of leak detection.
Russel: Fluid dynamics — that’s the subject that caused me to want to be a civil engineer.
Russel: I guess that’s one of the things about leak detection, is there’s a lot of math, right?
Giancarlo: There is. There is a lot of math, but it also depends on the different type of methodology, or technology, rather, that you are using in order to do detection of leaks in a pipeline.
Russel: At a high level, what are the various approaches to leak detection? If you would, try to answer that like you’re talking to a novice.
Giancarlo: There’s two methods that are actually used, or implemented, one of them being external methods. This is when the user is observing external factors within the pipeline to see if there’s any product release outside the line.
There’s also internal methods, which actually uses a field instrumentation that is measuring different parameters of the hydraulics of the line, information such as flow rate, pressure, density, or temperature, among many others.
That information is used through a computational algorithm that determines whether there’s a leak in the line or not. Those are the two main type of leak detection systems.
Russel: Let’s talk about external just briefly. I know that’s not the primary domain that you work in, but for listeners and for pipeline operators, a key point to be made here is that this leak detection, it’s a program. It needs to be broader than just any one approach, because ultimately, what we’re trying to do is make sure the product stays inside the pipeline. What are some of the kinds of external leak detection?
Giancarlo: When you’re looking at external type of leak detection systems, what you’re actually doing is you’re using external factors of the pipeline. Some of them could be as simple as monitoring temperature effects on the surroundings of the pipe. These are the ones that are typically known as biological methods.
It could be either the temperature that’s surrounding the pipeline that is changing. It could be a fiber optic cable that is external to the pipeline and the process, where it’s measuring whether there’s vibrations or temperature changes on the surroundings of the pipe.
It could be as simple as someone walking down the line. There’s even dogs that are specifically trained to detect hydrocarbons as they’re walking down the line, just as a drug-sniffing dog but they’re actually trained to detect those hydrocarbons.
Or other systems, such as acoustic sensors that are listening to the hissing sound that a leak makes when that product leaves the pipeline. Obviously, there’s other methods such as infrared cameras to monitor temperature changes, or changes from the surroundings of the pipe.
Russel: What are the limitations of external leak detection? What are some of its constraints?
Giancarlo: External systems are great. It’s a good way to monitor what’s happening outside the pipeline, but one of the downsides is that it’s not consistent. It’s only done at the request of someone going down the line, or someone may be working around an area that they visually or physically experience, or see that there’s a release on that particular area.
It’s not continuous. It’s not a system that is doing leak detection 24 hours a day, 7 days a week. For new pipelines, or pipelines that have just recently start running, and they have not yet been tuned or optimized for a computational type of leak detection system, this is a great way to monitor your pipeline. You don’t have to have a leak detection system installed or running to be able to monitor it.
There’s many companies that do flyovers, or drive through, through a specific area, just to make sure to cover their bases to ensure that there has not been any release, and that the pipeline is operating safely.
Russel: Let’s segue a little bit. Let’s talk more about internal. You mentioned what that is. We probably should ask you to talk about what are the various approaches to internal leak detection.
Giancarlo: When we’re talking about internal leak detection systems, we have to take into consideration that what these type of systems are doing is using field sensors, instrumentation to monitor the operation of the pipeline. Internal pipeline parameters such as the pressure, the temperature, the viscosity, maybe the density of the product, and very importantly, the flow rate.
By measuring these parameters, what the system, or any type of internally based leak detection system is going to be able to do is monitor to see if there are any changes on those variables that are not common, or abnormal. Based on those, an algorithm will kick in and we’ll be able to determine whether it’s a leak in the pipeline or not.
The very basics of internal methods, it’s measuring volume balance, how much product is coming in the pipeline versus how much product is leaving the pipeline. This type of check could even be done with a basic SCADA set-up, maybe some scripting or some coding within the SCADA configuration, where one is measuring how much product is coming in versus how much product is coming out the line.
If the pipeline is running nice and steady, and there’s no transience or a startup of pumps or changes in the operation, then everything that comes into the pipeline should leave the pipeline. That being the case, one can do an internal calculation based purely on the flow to make sure that no product have left the pipeline. That is one of the most basic principles for internally based leak detection.
Russel: One of the things you’re laying out, Giancarlo, that I hadn’t thought about before we got on this conversation together is you can grow into leak detection starting with patrols, and then moving to simple mass balance, and then on from there.
Being a SCADA guy and a measurement guy, one of the things I’m always aware of in the mass balance is that you can only detect leaks that are outside of the error of the measurement. Most metering we do is plus or minus one percent. Some of it’s got less error than that, but a one percent leak is a fairly big leak, particularly if you’re talking about a pipeline with a lot of flow.
Giancarlo: That is absolutely correct. When you’re looking at volume balance, you’re typically looking at medium to large size leaks. That’s one of the downsides of this type of leak detection methods. You do have to take into consideration that error of the instrumentation.
Russel: It’s relatively easy to put in if you have SCADA and measurement, but it does have that limitation.
Giancarlo: It does, yes.
Russel: What comes next? If I have those things in place, where might I go next?
Giancarlo: The next method after that would be using a rate of change, either pressure drops or drastic flow changes in the pipeline.
Again, if you have a pipeline that is running nice and steady, and that means no startup of pumps, no shutdown of valves or changes of the flow rates, then everything is running nice and steady. One should expect no changes in the pressure, or no changes in the flow.
If one is monitoring the pipeline just purely on the individual instruments, and notice that all of the sudden, you have a large pressure drop, or maybe a large flow drop at the outlet of the pipeline, one can assume that there could be a leak on the pipeline.
Rate of changes would be the second type of internally based leak detection method that one use to detect leaks in the pipeline.
Russel: Rate of change, and then where do we go from there? I like how this is rolling.
Giancarlo: [laughs] After that, the next step would be to combine the two. Why not, if we are able to do volume balance, measuring how much product is coming into the pipeline, versus how much product leave the pipeline, then we can actually use that principle with the rate of the change methods in order to monitor both.
If your flow changes and your pressure changes while the pipeline is running nice and steady, this is obviously not a change that you’re expecting, then one can say that there’s a leak in the pipeline if those variables change all of the sudden.
Now, the downside of these two, either by themself or combined, is that you’re looking mainly when the pipeline is running nice and steady. Unfortunately, not all pipelines run like that all the time. There’s going to be periods where the flow changes, or pumps are started or shut down. Those introduce transience into the pipeline.
During those particular instances, if you’re just doing leak detection based on volume or pressure, then you’re going to be subject to generating false alarms. By having false alarms, you’re going to lose the confidence of the pipeline operator on that particular system.
That’s actually one of the most important aspects of leak detection. When using or implementing a leak detection system, we want to make sure that we give pipeline operators the confidence that their leak detection system is properly working. When it does alarm, they can actually trust it, and engage in a response.
Russel: We had the comment earlier about finding leaks that are small leaks, and that being a limit of mass balance. When I start doing rate of change, I think one of the things that that creates is the opportunity to find a large leak that happens quickly.
Giancarlo: Correct, yes.
Russel: Maybe you could talk a little bit about rupture detection versus leak detection, how that relates to the things we’ve talked about so far.
Giancarlo: There’s actually three subjects within the leak detection cell that are relevant. When we’re talking to leaks, what we’re looking for is product losses. Products that are being injected into the pipeline that are never really making it to the end of the line.
That product loss could be, as we’ve been talking, due to a leak. Then there’s a rupture, which is the same as a leak, but its reaction and its pattern is different. You have a very large, sudden change on the dynamics of the line of the operation.
All of the sudden, your pressure just dropped to the floor very rapidly. When we are looking at ruptures, we have to take into consideration that maybe if we’re basing the leak detection on the instrument values that we’re reading from our field, maybe when you have a rupture, the value that we’re reading on this instrument is outside their operating scale.
Then we’re no longer receiving good values for the leak detection system. Now, the third aspect is going to be theft. It’s not much seen here in North America, but that’s also treated as a leak. You will have a product loss that, it’s taken out of the pipeline, with the difference that (this type of) “leak” will be eventually shut down by the person who’s executing this product withdrawal.
It could also be treated as a leak for those three particular type of product losses.
Russel: I haven’t thought about that, because our business is really focused in North America. Outside of North America, hot tapping pipelines is, there’s a lot. Well, I say a lot of it. It certainly happens. It’s not uncommon.
In North America, if you get a leak, eventually you’re going to find it. You’re going to have a discoloration. There’s going to be changes to vegetation. There’s going to be odors or something. When somebody is stealing the product, you’re not going to get any of those indications.
Giancarlo: Yes, that is right. I have to say that, based on our experience, these guys are very smart these days. There’s lots of instruments. The job that they do, sometimes it’s undetected, to the point that one just drive down the line, or walks down the line, and you’re not able to see anything. Everything is very covered and camouflaged.
Again, you do have a product loss. Initially, you do need to treat it as a leak, because that product is not making it to the very end of the line.
Russel: Maybe we could continue to walk this out. Once you’ve got mass balance and rate of change, where do you start going from there?
Giancarlo: After that, I think what you need to do is, we need to get into a little bit more of how we can use the mass balance and the pressure changes with a little bit more of a computational algorithm, a little bit more detail, or deeper look at these parameters to determine whether there is a leak or not.
With that being said, it comes in of one of the first and most implemented leak detections. It’s the real-time transient model, or RTTM, as it’s known. This type of system is using those field instrument values, flows, pressures.
It also uses other parameters such as the temperature, the density, maybe they discuss the other products in order to do a hydraulic simulation of the pipeline. What’s happening inside the pipeline, based on these instrument values, and obviously, a pipeline model?
Based on those results, the system will actually compare those values with other field values. In the case that there is a discrepancy between the two, then the system will go ahead and generate an alarm. That’s what known as real-time transient model, where modeling the behavior on the pipeline, based on some constraints, which are driven by the field instruments, and then we’re comparing that back with the pipeline and the real measurements in order to react to a leak event.
Russel: You’re looking at what the math model says should be happening, versus what you’re actually reading on the instrument. What are the challenges of implementing that kind of leak detection approach?
Giancarlo: I think that one of the most important challenges of this type of methodology is that it takes a lot to model a pipeline. I don’t know if you recall probably when you were in college, when you were doing fluid dynamics, those are one of the toughest classes back in college I remember.
When you’re taking all of that knowledge to model what’s actually happening inside the pipeline, it takes a lot. It’s not basic. It’s not rocket science, either, but it does take a lot in order to model the pipeline accurately.
One of the aspects that I try to emphasize on is you’re modeling what’s actually happening on the pipeline at a given time. Every parameter that you introduce into your model needs to be as close to reality as possible. Many times, those values or those parameters differ from the theoretical values that we’re given by a manufacturer for the pipeline specifics, or by the instruments. We are reading values from the field that, as you mentioned earlier, there’s a level of error. That error is going to have an effect on the real-time transient model.
There’s also going to factors such as the level of details of the product that you’re trying to move, the different parameters that need to be taken into account to accurately, not only model the hydraulic and the behavior of the pipeline, but also how a particular product reacts.
There’s changes in temperature, there’s so many variables out there that could change from day to day that it makes it really hard to maintain an accurate model to be used for the purpose of leak detection.
That’s one of the biggest challenges that the RTTM technology encounters, that you need to have someone that really understands and knows how to build a hydraulic model. You need a lot of configuration and details, calibration and optimization of that model to make sure that the values that are being calculated can actually be accurate enough to be compared with the field instruments in order to generate a leak [alarm].
You really have to stay on top of it, is what it comes down to.
Russel: Certainly, that’s been my experience is, if you think about how difficult it is to get all the records about what you have in the ground and you think that I’ve got to have that and have that as an accurate representation to be able to have a math model that’s going to work for leak detection, that starts becoming a bit mind boggling.
Every valve, every elbow, every material change, every line size change, and on and on all will matter in terms of having a transient model work correctly.
Giancarlo: Correct. When I think about hydraulic simulation, I think about virtualizing the physical pipe that’s on the field. We try to enter all the little details as much as possible. That’s very time-consuming. It does take a lot time to make a hydraulic model be as accurate as it could be.
Russel: Then any time I make a change, I’ve got to go back through all that again.
Giancarlo: Correct. You need to go ahead and calibrate it, high flow rates, low flow rates, different operation. Maybe the product that you’re injecting today is slightly different than yesterday, has a slightly different composition. That needs to be taken into account. Otherwise your results are going to be inaccurate. Then that will be prone to false alarms.
The best way that I like to describe RTTM is garbage in equals garbage out. If you don’t have a model that is accurately or realistically modeling your pipeline to the best possible way, then your results are not going to be very good, and therefore your leak detection is not going to be very good.
It does take a lot of time and a lot of effort to maintain them and make them work as they should and as many of them do, provided that you have the expertise and the resources to maintain it.
Russel: Again, you just said a mouthful there. That’s a whole deep conversation in and of itself.
Giancarlo: Oh yes.
Russel: We’ve talked about real-time transient. We probably ought to move on from there and talk about what’s next after real-time transient.
Giancarlo: The next method would be a statistical volume balance. With this type of methodology, what we’re doing is, again, we’re measuring the volume in and out of the line. We’re taking into account the pressure, when pressure changes on the pipeline, to keep track of the inventory of the line. Then we’re using that to determine whether there’s a discrepancy, but the system just doesn’t alarm there.
There’s a statistical aspect that evaluates the probability of there being a leak in the pipeline. The system just doesn’t react right away as far as generating an alarm. It starts reacting as that probability increases over time. Once it gets to a comfortable level, let’s say 99 percent, then at that moment, it goes ahead and generates an alarm that’s eventually sent to the operator.
For this particular type of system, the operator can see that the probability is increasing. These type of systems do have that output available. Even before the alarm gets there, to the 99 percent, he can start picking up the phone and making some inquiries to inquire that everything is okay, if there has been some change that is out of the ordinary or maybe some operation that is not normal.
With that being said, it can be ready by the time the leak alert comes in or be ready for when that happens.
Russel: I think that leads us to the last method. Maybe we could talk about that shortly.
Giancarlo: The last method would be negative pressure wave. With the negative pressure wave, what we’re looking at, we’re looking at that change in pressure when there is a leak in the line.
As you probably know and everyone in the industry should know, when you have a leak in the pipeline, that’s going to cause a drop in the pressure on your line. That pressure wave is going to travel outwards from the location of the leak.
When that pressure drops at that leak location, it starts to travel. That pressure wave will eventually be picked up by pressure sensors that are installed on the pipeline. For these particular system, you need to have a highly sensitive, highly accurate, fast sampling pressure sensor because you’ll be reading very small changes in the pressure.
You need to make sure that you sample them fast enough in order to capture a good pressure drop as that pressure waves travel through the sensor itself. Based on those two detections, meaning two pressure sensors installed on the pipeline, one can confirm that there’s a leak that comes from inside the pipeline and not some that’s coming from the station into the pipeline.
One can also confirm where is the location of that leak by doing what’s called a time of flight method to identify where the location is. Negative pressure, it’s subject to lower leak sizes, faster detection times. But then again it does require that additional hardware where you’re measuring this pressure variables very accurately very fast.
Russel: Interesting. How widely adopted is negative pressure wave? Isn’t that relatively new technology?
Giancarlo: It’s been out for quite some time. It’s actually a technology that I think it’s been around for, I would say, a good 10 years probably, since the ’90s. About seven, eight years ago is when we started implementing it into our portfolio.
Russel: It’s like a lot of other technologies in pipelining. We’re relatively risk averse so we’re slow to adopt. We want to have a high degree of confidence before we actually start putting things into production.
Giancarlo: Correct. It’s still evolving by using slightly different algorithms to get rid of the false alarms. Little by little, the confidence of the users is growing on it.
It’s very important to say that there’s not one leak detection technology that works for all the pipelines in the world. There’s no two pipelines that are alike. They could have exact same product. They could have similar elevation profile or maybe pipeline diameter. But the operation and the way they would react to certain conditions, it’s going to be slightly different.
That being said, you could have two pipelines that are very similar, but then one of them is going to be better implemented with negative pressure wave while other will have better result with an RTTM or maybe a statistical volume balance.
When you’re looking at leak detection, you need to keep an open mind as to which one will be the best technology for your particular pipeline and your operations.
Russel: I think, Giancarlo, that leads into a nice segue to talk a little bit about I think it’s API 1175 which is the leak detection program management recommended practice. That is a relatively new standard. Can you give us an overview of what’s in that standard and how somebody new to leak detection might engage with that content?
Giancarlo: It’s more about a high-level view of how pipeline operators should maintain a program on how to do leak detection within their companies, recommendations based on U.S. regulations, and not just install a leak detection system and be done with it. It talks about really doing everything you can in your company to make sure that you have the best leak detection system possible.
It talks about the culture, the strategies, KPIs, performance targets, and very importantly, testing.
If you have a system that is not generating alarms, it’s great, but then will it actually react when there’s a real alarm?
Control room procedures, responses, all the management, it gives the users a good idea on how to have that leak detection mentality within your company, not just implement the leak detection system on your pipeline, commission it, and close the door.
This type of systems need to be maintained, they need to be babysat. All you want to make sure is that they’re always working. They’re always doing their best. That’s where API 1175 really focus in. Are you doing everything you possibly can to have a good leak detection system?
Russel: I think that’s a great place to wrap up. One of the things I like to do, Giancarlo, is that at the end of one of these conversations is try to sum it to a few key takeaways. A couple of key points I think you made.
The first is that leak detection is a program. There’s a lot of different ways to do it. The approach you use for leak detection is ideally going to be appropriate to the kind of pipeline and the kind of operation that you have.
I think another key thing that you made as a point is that as you’re implementing this, you could walk into it, but you need to do it in a way that the people who are getting the alarms have confidence that those alarms need to be reacted to.
Then lastly, leak detection is not “buy a system, put it in, you’re done.” It’s really about a capability that you’re building in a company. You’re constantly evaluating how good a job you’re doing and what you need to do to improve. You think I nailed it?
Giancarlo: I think you did, yes. Those are all very important points. It’s very important that when you look at leak detection, you think about it as a second pair of eyes and something that will help the operator, making sure that you’re giving the operator every tool that he can possibly have to make the best call when that emergency comes in.
Russel: We talk about that a lot, Giancarlo, when we’re talking about pipeline control rooms and what the role of the control room is. Ultimately, issues are not normally or very rarely are they caused by what’s being done in the control room. Any issue that does occur, the control room is material to the response.
The whole role of the control room is to respond quickly and accurately to mitigate any negative outcomes. I think you’re making that point exactly with leak detection, is leak detection is just a way to give you an indication, so you can do a better job of responding.
Russel: I want to make one plug before we go. Two of the founders of your company, Michael Twomey and Jun Zhang, recently published what I think is a great book called “Introduction to Pipeline Leak Detection.” It’s available on Amazon. I bought and read a copy.
It’s fairly basic but it’s very comprehensive, and I think it covers all the things that we talked about today in a bit more detail. Plus, it has pictures, and I like pictures, because pictures help.
Giancarlo: They certainly do. I would definitely recommend to everyone who has an interest on pipeline leak detection or in the industry to take a look at this book. It provides a lot of good information. As you said, it’s very basic, but it’s about putting down the principles of each methodology and how we look at leak detection. Definitely a good book that I recommend to everyone.
Russel: With that, Giancarlo, thank you so much for being our guest. I very much would like to have you back and take a deeper dive into some of the other aspects of pipeline leak detection.
Giancarlo: Absolutely. Thank you for having me. It’s been a pleasure having a talk with you, Russel. Looking forward to the next time.
Russel: Thanks for listening to this week’s episode of the Pipeliners Podcast. I certainly enjoyed the opportunity to talk with Giancarlo about leak detection. I hope that you enjoyed it as well.
Just a reminder before you go, you can register to win our customized Pipeliners Podcast YETI tumbler by visiting pipelinerspodcast.com/win and filling out the form to enter yourself in the drawing.
Russel: If you have ideas, questions, or topics that you’d be interested in, please let us know by going to the Contact Us page at Pipeliners Podcast, or reach out directly to me on LinkedIn. My profile is Russel Treat. That’s R-U-S-S-E-L, Treat, T-R-E-A-T. Thanks again for listening. I’ll talk to you next week.
Transcription by CastingWords