Podcast: Capturing the invisible: Best practices in gas grab sampling

Join Matthew Dixon, Swagelok’s global technical lead for grab sampling, as he talks about the complexities of gas grab sampling—a critical, yet often misunderstood process in industrial operations. From the science behind phase changes to the design of safe and accurate sampling systems, Matt will share insights that help demystify the challenges of capturing representative gas samples. Whether you're in operations, engineering, or maintenance, this conversation will offer practical takeaways and a glimpse into the future of sampling in a digitalized world.

Below is an excerpt from the podcast:

PS: Tell us a little bit about yourself and how you got into this area of industry.

MD: I’ve been with Swagelok now for 27 years, and most of that time has been spent in the group that we call Custom Solutions, which is where we take the valves and fittings that we make and build them into bigger and better things, including things like grab sampling panels. The last few years, I’ve become what we call our global technical lead—so teaching our organization and our customers about grab sampling and other applications, primarily within the oil, gas, and chemical industries.

PS: What would you say are some of the unique challenges of gas grab sampling compared to other kinds, like liquid sampling, especially when it comes to maintaining sample integrity?

MD: Probably the first thing I would say is different about gas grab sampling is pressure—specifically because gas, being compressible, stores a lot of energy. Compare that to liquid, which is incompressible, and of course doesn’t really have a lot of stored energy.

But because of that compression and that compressibility, we end up with another factor called time delay. That’s where we pack a bunch of molecules into the line, and it takes a long time for the molecules to get from one end of the line to the other—because you can think of our sampling systems as having some sort of maximum flow capacity. Think of it in terms of some number of molecules per minute that we can process. And of course, if you have more molecules packed into the line, then it’s going to take longer to move from one end of the tubing to the other, based on a certain flow rate.

Beyond that, we have to worry about condensation of the sample—the sample actually changing phase on us, especially if we’re sampling a hot gas, like maybe something out of a flare where, if it cools down, now we may condense. Because of that, we try to keep the gas hot, so we use trace lines, a heated enclosure to put our sampling system in, and even drop the pressure, which would help with that time delay problem we were talking about.

Another aspect is sample toxicity. If we’ve got a sample that is acutely toxic, we really have to be concerned about things like spillage from our quick-connects. All quick disconnects will have some sort of spillage when you uncouple them. Even ones marketed as “zero spillage” have some. Some are as little as a fraction of a milliliter; some could be more. But even if you’ve only got a fraction of a milliliter, if the sample is parts-per-million toxic, then that could definitely be a problem when we’re talking about safety and gas sampling.

PS: That safety aspect is really critical then. Maybe we can get down into the nuts and bolts a little bit, and you can walk us through what an ideal sampling system for gases would look like. Are there components that are critical to ensure both the safety of the person doing the sampling and also the accuracy of the sample?

MD: First of all, when we’re talking about sampling, we have to decide what sort of sample container we’re going to use. And with gases, we’re almost always going to be using a sample cylinder. Sample cylinders are typically made out of stainless steel with valves on either end.

When we’re sampling a gas, we would hold it in the vertical orientation when filling and typically fill it from the top down. That way, if there is any of that condensation we talked about—or maybe some particulate—that’ll just get flushed right out the bottom as we’re filling.

We also have to be concerned about overpressure protection when it comes to sample cylinders. So we would have rupture discs or relief valves, something like that, to relieve any excess pressure so that we don’t have any sort of cylinder rupture or anything like that.

I talked a little bit before about hot gases. Normally with liquids, we might have some sort of sample cooler to cool them down. With gases, it’s generally not required. Gases have such a low specific heat that, as we’re transporting the gas from the sample process to our sample container, typically what we’ll see is that the gas is going to get down to ambient temperature within just a foot or two of uninsulated tubing. It’s going to be down to whatever that ambient temperature is, so we don’t have to worry about a cooler, thank goodness.

If we do happen to have any liquids in our sample, we generally want to separate those out because dealing with a sample with multiple phases is pretty difficult to do. There are a few different ways we could do that: a kinetic separator, a knockout pot, maybe a hydrophobic membrane so we can separate the liquids out. But we do have to be very careful, especially with something like a knockout pot, because that becomes what’s called a mixing volume. Basically, it’s a large volume within our sampling system, which takes a long time to flush out and causes excessive time delay in our sample.

The biggest thing, of course, that we’re worried about is safety. We’ve got humans operating these sampling panels, and as a result, there are two consequences of that. One is that errors are more likely. You know, let’s face it, people don’t always follow instructions, or maybe they do something unexpected, and that can cause problems. And when we have a problem, the consequences of those problems are much, much bigger—because again, there’s a person right there who could potentially be exposed to hazards such as high pressure, high temperature, or toxicity.

So the way we combat that is typically through four different points:

• Documentation: Just having simple operating instructions on the panel, and making sure we provide documentation to the engineering and maintenance teams so the panels can remain in good working order.
• Simplification: Making sure the panel is easy to operate. For example, having valves ganged together or geared together so they operate on a single handle. Also having a nice, clean operating interface. One of the things we prefer to do here at Swagelok is to put all of the tubing on the back of the panel, for example. That way, it’s a clean interface on the front, and all the operator sees are the valve handles and pressure gauges they actually interact with, and we don’t have to worry about overthinking it.
• Standardization: Having similar panels throughout a facility. That way, as the operator walks up to one panel, it’s similar to the one that they were just at previously. It makes the system more intuitive and easier to operate.
• Automation: And the last piece is with regard to the human element. What better way to overcome it than to remove it altogether? That would be through something like automation. To the extent that we’ve got a critical application, we might consider automation to completely remove that human element and eliminate those errors altogether.

PS: This conversation is taking me back to two jobs ago, when I worked as an EPA contractor helping field technicians convert data from soil and groundwater sampling into reports. And one of the challenges those technicians had was figuring out whether they had a truly representative sample of what they were collecting. So let me ask you the same question about gas sampling: How does a person ensure that gas samples are truly representative of the process conditions at the time they’re collecting them?

MD: There are a number of different things that go into making a representative sample. Here at Swagelok, we like to start at the beginning, as we say. When people look at a sample panel or a sampling operation, they really tend to focus on the panel itself—but it really does start right back at the sample tap, so where we’re actually extracting the sample from the process, typically from a pipeline. 

Normally, we would use a probe to reach into the center of that pipeline. Typically, we reach about a third of the way in, because the flow in the center of the pipeline is going to be more representative of everything flowing through that pipeline, so using probes is important from that perspective. But also, you can use probes to do things like exclude particles and, in some cases, even coalesce liquids out of our gas sample so that we don’t carry those into our sampling system. So that’s going to be an important starting point—paying attention to that sample tap.

Another thing we need to pay attention to is just understanding what’s going to be in our sample. What are we expecting to see? For example, if we’re expecting to have hydrogen sulfide (H2S) in our sample, then we will typically SilcoNert-coat all of the different surfaces on the inside of our sampling system. Otherwise, molecules like hydrogen sulfide, and other polar molecules like water, can actually stick to the walls of the tubing and the components. As a result, you don’t actually see them coming out the other side of the sampling system—they just get stuck in there and you don’t see them come out until the walls inside our sampling system are completely saturated. Having a silica coating like SilcoNert, or others, helps prevent those molecules from adhering to the wall, and really helps ensure we have a representative sample coming out the other side.

Another aspect is dead legs. A dead leg is an area in our sampling system that isn’t flowing. This could be something like a pressure gauge, or maybe a bypass valve installed on the branch of a T. That pressure gauge, that branch—they’re not flowing, they filled up with a gas sample at some point in the past, and we don’t really know when. What we do know is that as our fresh sample flows past, the molecules are continuously diffusing in and out of those branch connections, and those are continuously contaminating our sample stream. So that’s another factor in representativeness.

We also need to make sure our sample cylinder is appropriately cleaned. I can’t tell you how many times I’ve seen a dirty sample cylinder sent back into the field, or one that still has its previous sample inside. That can be okay as long as you allow it to flush out for a good amount of time—but the operator needs to know they have to do that, and they don’t always.

Another aspect is time delay. I’ve mentioned that a few times, but when we’re looking at sampling, people tend to ask the question of what—what is this molecule I’m analyzing? But they often forget to ask when. When was this created? When was it extracted? How long does it take to get from the sample tap to my sample cylinder? Because without answering when, the answer to what doesn’t really do us very much good, right? We need to understand the time frame so that we can understand what’s going on in our process and when.

And then lastly, phase changes. We want to avoid phase changes to the extent we can. In many cases, if you start to change phases—like in a liquid, if you start to condense—you’re going to fractionate your sample, and very likely you won’t be able to get it back into the correct phase, or the correct state it was in before. So not allowing that phase change to happen in the first place is definitely something we want to take into consideration.

PS: Matt, maybe we can shift over and talk about some of the environments themselves, especially demanding environments like liquefied natural gas facilities or refining facilities. What is it about the way that Swagelok approaches these facilities that makes them stand out in the field?

MD: Yeah, so in a rugged industrial environment like that, in a sampling operation, your first consideration has to be safety. It’s really the most important rule—we have to make sure the operator goes home at the end of the day in the same good condition they arrived in that morning.

A general rule of thumb I have is that if you’re relying on the operator’s PPE as the primary safety measure for a sampling operation, then you really need to think about engineering safety into the design of the panel. Some of those things we talked about before—documentation, simplification, standardization, and automation—that’s when we would start to integrate those things.

We also want to make sure our sample is designed for analytical purity. So, making sure that we don’t have those dead legs we talked about before. Making sure that if we’ve got a hose, that it’s downstream of our sample container so that it doesn’t contaminate the sample.

We also want the ability to customize our panels. If you come in with a standard solution and try to make it fit every application, it’s just not going to work. There are unique circumstances in these facilities that require unique solutions, and our ability to come in and adjust—whether it’s materials of construction, the overall design of the panel, different processing or testing, or different methods of construction—that’s something we really specialize in here at Swagelok.

In fact, we’ve trained over 300 of our engineers around the globe to be able to do these sorts of grab sample systems, and to be able to build and design them at the local level as well. And then, of course, fabricate them at the local level so that no matter where our customers are in the world, we’ve got the flexibility to do that and bring those customers in for factory acceptance tests. They don’t have to travel all the way to the Swagelok factory here in Cleveland, OH—they can do it right there at the local level.

PS: Out of all the tactics you've described today, and the strategies too, when it comes to gas grab sampling, what are one or two things that you see people doing in the field that you just sort of recoil at? The things you see happening where you warn people against right away? And what can they do to fix those?

MD: One of the biggest things we see is choosing a poor location for the sample tap. So again, starting back at the beginning, one of the most common things we see is installing a sample panel across a control valve. People do that because you need to have some sort of differential pressure to drive the flow from the process to the sample panel and then back to the process. And a control valve is a pretty convenient way to do that, just because they're everywhere in the facility. They create a differential pressure because, again, they're going to be open or closed by some amount, but generally they’re going to create some sort of differential pressure in the process. And so we can use that to drive the flow.

However, because that valve is going to be open and closed some percentage based off of what the process needs at that moment—and that will change over time—as a result, the differential pressure will change. And that will, of course, change the flow rate through our sampling system, which then changes the timeliness of the sample. In some cases, that's okay. Maybe we don't care. But in a lot of cases, we need to know exactly when that sample was extracted, and that becomes much more difficult if you're doing it across a control valve like that. It'd be much better if you were doing it across, say, a pump or a compressor, where you have a much more consistent pressure increase that you can use to drive that flow.

Another thing we see with regard to the sample tap location is people using process drains in the piping to take that sample as the sample tap. Again, very tempting to do because those are already there. If you're adding a sampling system, we say, hey, look, there's already a tap here, let's go ahead and use it. The problem is that those process drains, as you might imagine, would typically be on the bottom of the pipe. Now, when you sample off the bottom of the pipe, if there's any sort of condensation in the system, guess where it's going to end up? It's going to go into that sample tap. Same thing if there's any particulate in the system—it’s going to go down into that process drain and end up in our sampling system. So using a process drain is also something we want to avoid.

Another aspect is the pressure. I mentioned that before. If we have a high-pressure gas system, that can really cause significant time delay in our system because of the compression and having so many molecules in there. The way we would typically handle that is: if we've got a high-pressure gas, right at the sample tap—as close as possible—we would drop that pressure. Drop it really low, maybe a bar or less. Just enough to drive the flow, but not enough to where we're really going to have a ton of molecules packed in that sample transport line.

Another common mistake we see is that people will often choose the wrong sample container. Maybe they’re used to sampling liquids and they'll try to sample a gas in a bottle as a result of their experience. Even if you've got a rubber septum on the top of that glass bottle, you really can't put any pressure in there. And in order to get any sample size of any significance, you do need to have some pressure on your gas. Bottles are just not pressure-containing, so you can't really do that. So sampling in a bottle is generally not going to be preferred. I won’t say it's not possible, but it's definitely not preferred. You're likely going to have some leakage through the rubber septum if you do try to do that.

Another consideration is when you sample in a cylinder. A lot of times, you'll see people orient the cylinder in a horizontal orientation as opposed to vertically. And that can cause problems again with condensation and particulate. With the cylinder horizontal, if there's any sort of condensation, it's going to collect in the cylinder and not be able to be flushed out. Same with particulate—it’s going to be carried into the sample cylinder by the flow of gas. Then, as the velocity slows down, the particulate will settle out. And of course, it will settle out into the horizontal cylinder and end up there. So we'll end up transporting that back to the lab.

PS: For the last question today, I'd like to move from the physical world more to the digital. You mentioned before that automation is changing the face of grab sampling. How do you see grab sampling evolving with that and other technologies like digitalization and real-time analytics?

MD: Online analyzers and the like—you know, those are really kind of the gold standard in terms of being able to control our process. They give us essentially a real-time result on the analysis of our process. And that's really important in some cases, especially where you have processes that are changing quickly and we need to get a fast analysis done so that we can adjust our process and keep it within the goalposts, keep it within the specification.

However, there is an economic aspect as well. Analyzers are not inexpensive instruments—especially when you consider all of the support structure that needs to go around them. Generally, an online analyzer needs to be installed inside some sort of shelter, and of course, that takes real estate within the facility that can’t always be found right next to where you’re taking your sample. So if you’ve got to extract your sample and the shelter is, say, 100 meters away, now you’ve got 100 meters of tubing transporting the sample out to the shelter and 100 meters returning it back to the process. There’s the cost of the shelter itself, the cost of installing all that tubing, and the cost of the associated utilities.

So online analyzers absolutely have their place, but grab sampling is certainly something that will always be around because of the economics of it. If you don’t need to get that answer right away, then a grab sampling operation is definitely going to be the more economical choice. It makes a lot more sense in those cases.

Beyond that, even where you have online analyzers, you’re still going to have grab sampling. There’s always going to be some sort of redundant sampling operation going on, as a backup and also as validation. For those online analyzers, you might take a grab sample on some periodic basis—maybe weekly, maybe monthly—and then compare the result of that grab sampling analysis to the result the online analyzer is giving you. That way, you can validate that it’s giving a true reading.

So grab sampling is not going anywhere. I think that even in cases where you’ve got analyzers, it’s still going to play a major role. I would say, though, that as we find more and more analyzer applications, the role will diminish slightly—but overall, it’s still going to be significant.

PS: For listeners who want more information on grab sampling, where would you point them?

MD: I would say start out at Swagelok.com. We've got a lot of good information there—blogs, different articles, things of that nature. Beyond that, we’ve got our network of local sales and service centers. As I mentioned, at our local centers, we’ve trained over 300 engineers worldwide. So chances are, at your local center, there will have been at least one, probably multiple engineers who have been through this training course and know the ins and outs of grab sampling—not just the standard systems, but also how to modify them, how to make them work in different applications, and really tailor them to whatever unique application the customer happens to have.

About the Podcast
Great Question: A Manufacturing Podcast offers news and information for the people who make, store and move things and those who manage and maintain the facilities where that work gets done. Manufacturers from chemical producers to automakers to machine shops can listen for critical insights into the technologies, economic conditions and best practices that can influence how to best run facilities to reach operational excellence.

Listen to another episode and subscribe on your favorite podcast app