Advanced Pipeline Leak Detection Podcast

Advanced Pipeline Leak Detection podcastPipelines are instrumental in moving energy around the globe. Offshore Technology cites some statistics.

The total length of the global trunk/transmission pipeline network is 2,034,065.0 km (with start years up to 2023), according to GlobalData’s latest report.

Of this total length, crude oil pipelines constitute 379,000.4 km [235,500 miles], petroleum products pipelines constitute 267,289.6 km [166,086 miles], natural gas pipelines constitute 1,295,563.6 km [805,026 miles] and NGL pipelines constitute 92,211.8 km [57,298 miles].

It is paramount to continuously monitor these pipelines for safe, reliable, and sustainable operations. In this Emerson Automation Experts podcast, I speak with Emerson’s Greg Morrow to discuss how real-time transient models (RTTM) in the PipelineManager software application help pipeline companies maintain high-performance levels.

Listen to the podcast and visit the Software for Energy Transportation & Storage section on Emerson.com for more on advanced leak detection technology to help you achieve safer, more sustainable operations.

Transcript

Jim: Hi everyone, this is Jim Cahill with another “Emerson Automation Experts” podcast. Today we’re discussing the importance of models and modeling in making transportation via pipelines safer and more reliable. I’m joined by Emerson’s Greg Morrow to discuss why this is so and how to drive improvements in your operations. Welcome, Greg.

Greg: Hi Jim. Glad to be here.

Jim: Well, it’s great having you here. Well, let’s jump right into it and let me ask you to share your background and path to your current role with us here at Emerson.

Greg: Okay. Well, I got my doctorate in physics a while back. I did my work in particle physics up at Fermilab in Chicago and after that I got a job working in pipelines, pipeline simulation. And I’ve been with Emerson, or our predecessor company working with RTTMs for a long time now, 25 years. It doesn’t feel that long every day, but that’s how long it’s been and my role is really as leak detection subject matter expert.

Jim: Well that’s great background. Thank you so much for that. Well, let’s get into it about, you had mentioned RTTM, What exactly is RTTM?

Greg: Okay, that stands for real-time transient model and let’s unpack that. Model is a representation of something. I can get very philosophical about what all that actually means, but at its core, a model is a representation of something. In this case, it’s a sophisticated physics and mathematical representation of a pipeline, in some context, this is called a digital twin.

Transient means, that it takes into account that things are changing. And this contrasts with steady state, which assumes that everything is constant. And real-time means that it’s based on measurements from instruments in the field, which can be live or recorded. And it’s really important part of a good RTTM, that you’re able to play back that recorded instrument data for tuning purposes and analysis.

Jim: So how does the RTTM fit into our PipelineManager software application?

Greg: So it’s the core of our high-performance leak detection system. And so the key to that is it’s an accurate representation of exactly what’s happening on the pipeline. So in addition to leak detection, it’s also used to manage the pipeline and that’s features like batch tracking or scraper tracking, DRA [drag reducing agents] tracking, hydraulic monitoring like MAOP [maximum allowable operating pressure] alarms, hydrate formation warnings, and natural gas systems, and so on from there.

Jim: So is the PipelineManager application something to be used with gas or liquid pipelines?

Greg: Both. So the key there is the equations of motion are the same regardless. The only difference is the equation of state. The equation of state is a relationship for a fluid where you determine the density in other properties, from the pressure and temperature, and some defined parameters for the fluid. We have different equations of states for all sorts of fluid, from the API correlations for crude and refined products to NGLs, to GERG 2008 for natural gas, that’s well known as the most accurate equation of state for natural gas. So, both liquids and gases, anything you’ve got.

Jim: Well, that’s great. Covers the whole gamut. So how difficult is this application to apply?

Greg: So the RTTM is very complete. So it does have a lot of things you can modify or fine-tune, ground thermal properties, instrument filters and so on, and so forth. But what’s key to understand is that the defaults are widely applicable. So we ship with a bunch of different types of ground and you can use one of those, and it’ll be pretty good. Out of the box, your leak detection system’s going to be accurate, sensitive, reliable and robust, but it also rewards deeper attention to detail. And we call this tuning.

Jim: So for this application, what’s required from an instrumentation standpoint?

Greg: So the RTTM has a minimum set of required instruments and for our purposes, it’s basically pressure, and flow at every inlet, and outlet, and pressures it pumps, and regulators that can change the pressure from the other side of the pipeline, but more instrumentation is better. So in the case of pressures, putting a pressure instrument on an intermediate valve at a river crossing or something like that is helpful. Smaller segments tend to make the model more sensitive. Sensor quality is also a factor. If your pressures report by exception at 0.1 bar, which is a pretty big step, you’re not really going to get good results, modeling slow drifts and sharp transients. Temperatures are also very important depending on the fluids you transport. Some of them like gas can be extremely sensitive to temperature. So good temperature measurement is important.

Can also have fluid property sensors like density or viscosity, or real-time composition measurements like gas chromatographs that tell you, for example, this batch of fluid has a lot of extra ethane in it or something like that. Then you also can use instruments to support other features. Like if you’re doing scraper tracking, you’re gonna need scraper sensors to tell you when the scraper is launched and where it passes, and so forth. So all of those things play into the development of the model. And one of my remarks is often, I won’t turn down extra instrumentation. More instrumentation is almost always better, but again, out of the box, the RTTM is gonna work with the instrumentation you have as long as you meet that minimum set.

Jim: Now, you had mentioned that more measurements can help with the tuning of these models. So what role does tuning play in the models?

Greg: So, again, I could wax philosophical about this. You can catch me offline and I’ll be happy to talk with you about it. But for just very briefly, tuning is changing the parameters of the model to achieve greater correspondence with reality. So, you look for places where the model doesn’t show what’s happening in reality. So for, you know, something that’s not unusual is when you turn on a pump, get a pressure surge, and the fluid reacts in the model slightly differently than the fluid reacts in reality. So, you do something like modify the fluid composition or the pipe wall thickness or the thermal conductivity of the ground or some other parameter to make the model correspond better in that place where it’s not.

So, as a practice, a lot of tuning is actually understanding the instrumentation. That’s both the instrument quality itself, as well as the SCADA data acquisition properties like the “Report by Exception” that I mentioned, and also, what part of the pipeline the instrument applies to. For example, if a pressure’s upstream of pipeline equipment versus downstream of some pipeline equipment, like I said, the valves or something like that, that can make a difference in part.

And understanding that helps you get the model accuracy to match reality. Because the RTTM is the most sophisticated model of the pipeline possible, you can get it extremely close to reality, which helps your leak detection achieve the best sensitivity, reliability, robustness, and accuracy. And this can be an ongoing process built into your process of continual improvement.

Jim: Well, that’s interesting. I noticed that that’s the second time you talked about sensitivity, reliability, and some of those other things. Is there a reason for that?

Greg: So, those are terms from API 1130, which is a document that discusses computational pipeline models or CPM, like the RTTM. And they’re particularly useful terms for evaluating leak detection systems. Sensitivity is how small a leak you can detect and how fast. Robustness is your ability to detect a leak no matter what pipeline conditions there are, like, steady-state or transient, shut-in, slack, supercritical, and so forth.

Reliability is how much you trust the system’s alarms. If it gives a lot of false positive alarms or misses true events it should have caught, it’s not as reliable as if it’s an alarm is always completely correct. And accuracy is how accurate the system is in determining things like leak size and location. The transient part of RTTM, Real-Time Transient Model, is the biggest part of the robustness of the leak detection system because modeling things as they’re changing is the biggest part of ensuring that you get sensitivity regardless of what the pipeline is doing.

How sophisticated and well-tuned the model is, determines how sensitive it can be. Sensitivity and reliability are often in a bit of a trade-off so if there’s a certain amount of uncertainty built into the model that just because the models cannot be perfectly accurate. If you push on the sensitivity, you’re gonna push yourself in the direction of that uncertainty and you’re going to generally get more false positives.

On the other side, if you say, I cannot tolerate any false positives and some customers, that’s the rule, then in order to make sure that you’re above that sort of underlying irreducible systematic uncertainty, then your sensitivity is gonna be somewhat higher. And this trade-off is really exacerbated by the statistics of small numbers. So, if a leak is one in a million and you’re 99.99% reliable, you’re still going to have 100 false positives for every time you have a real leak. Reliability is a big part of our ongoing R&D. Machine learning and statistical models can really help here.

One of the interesting challenges is, of course, because of those statistics of small numbers where false positives are just sort of inherently more likely than true positives. You can, if you do it naively, accidentally train your system that any potential alarm is false and just throw it away. You’ll be right 99 times out of 100, but you also won’t have a functioning leak detection system. So, this is one of the interesting problems that we’re investigating. As scientists, interesting problems are the ones that really get us into work in the morning.

Jim: Well, it’s fascinating that trade-off and that the numbers involved in that where you can still have all those false positives even with that level of reliability. So, given the criteria that established in some of those standards, how does the RTTM measure up?

Greg: Very good question. Among CPM systems, the RTTM is sort of universally understood to have the overall best score. That’s not to say that other systems don’t have their advantages. Other systems may excel in one area, negative pressure wave or NPW systems are by far the most accurate for leak location, for example. But NPW systems are not particularly robust or reliable. The RTTM is a standout for covering all four corners thoroughly. It’s also good to have complimentary systems. In fact, it’s often required by regulatory bodies.

And PLM [PipelineManager] includes several complimentary systems like RuptureSentinel, and it can also integrate with third-party complementary systems. So, whatever your needs are, we’re likely to help meet them. PLM also helps you develop key performance indicators, KPIs, for your system as well. One of our apps is specifically aimed at this, the leak detection performance analyzer. So, we’re really there to support your Leak Detection Program Management.

Jim: Now, I know we’re not the only RTTM available in the marketplace. So, how does our PipelineManager application stand out from others in the industry?

Greg: Not all RTTMs are the same. They differ in what’s included in the model and the mathematical method used to solve the model equations. Basically, the details that I’m not going into here. A really simple one is just how you calculate the transfer of heat from the pipeline to the ground. You can do that a bunch of different ways and the RTTMs vary in that sort of details. They offer differ in what you do with the output of the model.

So, the output of the hydraulic model is an exact representation of what’s happening on the pipeline. And what you do with that can also vary how, for example, you determine whether there’s a leak alarm or how the batches are moving through the system, etc. So, RTTM stands out for its thoroughness. We’ve used as much physical rigor as possible to ensure the greatest accuracy possible in the hydraulics.

It stands out for our experience and expertise. We’ve been doing this for a long time, and that’s given us a lot of opportunity to hone the performance of the model as well as the services we provide implementing PipelineManager to meet your needs. And it stands out in terms of performance. We’ve won a number of head-to-head competitions where we’ve demonstrated greater performance in those four API criteria.

And it stands out in terms of features. It’s not enough to have “batch tracking” you have to provide batch tracking that meets the pipeline operator’s requirements, whether that’s calculation of interface volume or providing actual volume corrections for commercial applications. It’s not just batch tracking, of course, we have the widest selection of pipeline management features no matter what your needs might be.

Jim: Well, that’s a really good overview. I liked how it started that it could be applied right out of the box and then you have the flexibility to improve the accuracy of the model just based on how you’ve instrumented it. I guess before we close things down, do you have any final perspectives to share with our listeners about RTTMs?

Greg: So, as you just said, out of the box, PLM’s gonna give you good leak detection. And the reward of a continual improvement process is world-class leak addiction performance in all four sectors of sensitivity, reliability, robustness, and accuracy. And even though leak detection is very important, you’re not going to be dealing with leaks every day, or if you are, you have other problems than the software.

You’re going to be operating your pipeline every day. Your controllers are gonna be busy doing controller things, shipping fluid to this location or that location. PipelineManager helps you do that. Whatever it is that you’re doing, we’ll help you understand what’s going on your pipeline better than anybody else in the market.

Jim: Well, I know we’re only able to scratch the surface in this podcast. So where can our listeners go to learn more?

Greg: So, we have a four-part webinar series and anybody can go to www.emerson.com/leakdetectionwebinars. And you’ll be able to go through all of those sessions and get a more in-depth view of our products and what they can do for you. You can also visit www.emerson.com/scadaforenergy to access more information on PipelineManager, as well as our entire suite of energy, transportation, and storage software offerings.

Jim: Well, that’s great. And I’ll add hyperlinks around those in our transcript that we include with the podcast. Well, Greg, I wanna thank you so much for sharing your thoughts and your expertise with our listeners today.

Greg: It’s always a great pleasure to talk to people about this. And again, feel free to contact me offline. I’m happy to discuss all of this. And you can see me at conferences and especially our Flow User conference every year.

-End of Transcript-

The post Advanced Pipeline Leak Detection Podcast appeared first on the Emerson Automation Experts blog.