Rod Ammon: Welcome to the IAAI's CFITrainer.Net Podcast, a feature of the International Association of Arson Investigators premier online learning network, CFITrainer.Net. As buildings from residences to warehouses get bigger, large compartment fire dynamics is taking center stage. How does fire develop and travel in these open concept large void spaces? Is it the same as or different than smaller compartments? And how? What does research show us about fires in large compartments? How can we apply the research and fire science to real world fire investigations? We're about to get into all of this and more. With us today is Michael Rushton, a professional engineer and certified fire and explosion investigation with over 15 years of public and private fire investigation experience. He is currently the supervisor of the Ontario, Canada Office of the Fire Marshal, where he oversees a team of fire investigators, conducts case reviews of fire investigation cases going to court, and leads complex investigations involving fatal, incendiary, and large loss fires. Mr. Rushton is a member of the NFPA 921 Technical Committee and the IAAI. He has spoken and presented regularly on fire investigation topics for over a decade. Most recently at the IAAI's 2025 International Training Conference, where he presented a class in large compartment fire dynamics. He's currently working on a PhD in fire safety engineering where he is developing new models for how traveling fires develop so they can be used in fire analysis, damage analysis, and forensics. We'll talk more about that in a few minutes. Michael, thanks for joining us today on the podcast.

Michael Rushton: Well, thanks for having me. Nice to meet you.

Rod Ammon: Oh, we're really, really happy to have you here and we appreciate your time. I'd like to start out with a little bit about your background as it relates to how you came to study large compartment fire dynamics.

Michael Rushton: Absolutely. When I first joined the 921 Technical Committee, I started working on my PhD, and this was sometime after being in the field of fire investigation. And I was actually chatting with another member, Vito Bebraskis, and I was searching for topics that would represent areas of lack of knowledge or gaps in knowledge for fire investigation, and particularly fire dynamics, that area of fire investigation. And he proposed a topic, and it was actually a bit of a misunderstanding. I was originally interested in studying how fires move towards or are affected by oxygen in post-flashover fires. So for instance, a fire may originate in one location and then maybe propagate to another location and burn more vigorously in that other location due to ventilation. A little bit of miscommunication. He got me down the road of what was termed traveling fires, which is basically, let's call it fire dynamics of large compartment fires. And it was long story short, I ended up following this pathway through Carlton University in their engineering department. And I found a kind of overlap between that area of fire dynamics and fire design, which is what's more interesting academically for my school. So I defined where fire investigation basically met design, and that's where I ended up with large compartment fires.

Rod Ammon: You gave an example. When we did a pre-interview, it was about how the hot gas layer and top and cool at the bottom. Can you talk a little bit about that?

Michael Rushton: Yeah, for sure. When we enter the world of fire investigation, we get that kind of preliminary 1033 type course where they teach fire dynamics, but that fire dynamics is overwhelmingly based upon what we can call small compartment fires. And I say small being like a bedroom or something maybe less than 300 square feet. All the research and understanding of those compartment fires kind of fed into that basic understanding of fire dynamics. So you take that course and then they present with you very simple models of those compartment fires, which are based on experiments. They're based on real life experiments, and they basically show that there is a somewhat uniform hot gas layer up high, and there's a somewhat uniform cooler layer. That's a bit of an approximation. It's not going to be the same temperature, but it's a accepted approximation. It fits well with what was seen. So that idea where we take with us in compartment fires, but the problem is when you take any compartment that moves outside of that bedroom style box, and I mean you can make it higher, you can make it larger, you can even make it a different shape. You can add an extension, a small L-cove, for example, or you make the ventilation dissimilar to a regular bedroom. All those things lead to different fire dynamics. And what I ended up studying was these larger type compartments, ones with big volumes, how the temperatures, well, many aspects don't fit that previous model, but one of the big, big ones is what we call non-homogeneity or not the same uniform temperature up high. And the cooler temperature we can assume to be relatively cool, but it's the hot gas layer that we find differs vastly in large compartments.

Rod Ammon: Interesting. So give us some background. It might be a little bit redundant because you've covered a bit of it, but where does the fire dynamics that is traditionally taught come from? How do things like the definition of flashover or the interpretation of fire effects and fire flow indicators come to be?

Michael Rushton: Yeah, that's a good question. That's kind of where I started as well. When you do these 1033 courses, a lot of the fire dynamics come from a set of research or work that was, I think, originally meant for the purpose of design, fire engineering, fire design. That was the impetus and the funding reasoning behind it. So we, in fire investigation being a somewhat small subset of, let's call it the larger fire industry, we adopted that. We took that. We didn't have the base knowledge to recreate that. So we took that and we adopted in our basic understanding of fire dynamics. So we looked at the research world and we said, "What are you guys doing with your models?" Which when I say model, I use the word generically like a hot gas layer, like the two zone model, that's a model, it's a basic model. So we adopted those models and we started to teach them. So when we take our fire dynamics and we look at our fire scene, we are more often than not applying these kind of, let's call it, fire protection engineering or design research type models saying it's hot up top, cooler down low. And they work, and I would argue that they work relatively well or strong for many fires, and that application is relevant. But the dangers I would highlight would be when you take them further outside of what we call their area of validity or the research that they came, the type of compartments that they used in the research, if we then apply that model to a different compartment, then it's with outside its normal validity and then you work your way into some potential problems.

Rod Ammon: So the issue I think was they're doing research on small compartments.

Michael Rushton: Yes. Most research that builds these models ... Almost like all of the research that built these basic two zone models were based really simple bedroom style compartments. There is large compartment research out there. First started in the '90s with a series of eight tests in UK. And then there was more recently that last 10 years, about nine to 10 more experiments. But these large compartment fire tests are extremely cost prohibitive and difficult practically and logistically, but we do have a dataset for them now that we can learn a little bit from.

Rod Ammon: So what are the things that you have to be careful about when you're applying traditional understanding and you're dealing with a larger compartment fire?

Michael Rushton: Yeah, absolutely. One thing I'd highlight right away that got me into this topic of interest was in these smaller fire ... I'll use the word small, but really I'm just talking about a bedroom. It's small perspective of what the big compartments I'll be talking about. But these regular bedroom size compartments, we have this concept when we walk into a fire scene, we do comparative fire damage analysis. We look at one location and we look at another location. And if the materials are somewhat similar, we say, "Oh, it's damaged worse in this location than that location. So there's some argument that maybe the fire originated there because it burned longer, burned the materials." That's fine and very applicable and probably used in many fire seams. But the second it goes into flashover or this concept of flashover and oxygen starts to play a big, big role in burning behavior, it has been observed that the materials that burn burn at a faster rate near the windows and the ventilation in these post flashover fires. So that concept was super interesting to me because you can walk into a fire scene, a small compartment fire, and you can say the fire's worse damage to the window, maybe a conclusion could be made that the fire originated near the window when in fact that could be a ventilation product. Now in large compartments, this concept is exaggerated. It is more profound and more impactful, and I say so, I'll give you one example. In one series of experiments, the researchers ignited a fire at one end and enclosed end of a long, large compartment. So we're talking 1,000 square foot and a little bit above 1,000 square foot compartment, and they ignited the fire in eight situations near that kind of enclosed location away from the ventilation, and they allowed it to propagate under some different parameters, let's call it different compartment parameters. And it was very significant how quick and how vigorously the firewood would propagate to the ventilation and burn almost exclusively at the ventilation at that point. So the point where you get the first fuel package to burn maybe 10%, and then it propagates to the ventilation and it will completely exhaust the fuel package at the ventilation. So you'll have ... And you as a fire investigator, you may walk into a fire after it's been extinguished and that will lead to some fire patterns that indicate of a larger mass loss near those windows. And those ones are the ones I think we need to consider in small compartments too, of the effects of ventilation.

Rod Ammon: Does that change the way people might witness a fire?

Michael Rushton: Sure. Yeah. If you say witness, I assume we're talking about someone outside of a building or compartment, which are often the witnesses we're talking to and a source of origin information we used in 921. So their perspective, they are often drawn to those first visible, usually somewhat violent events such as windows popping, or they could be a passerby, someone outside the house. Those windows breaking or that violent fire near those windows may not be in those examples where the fire started, but this is the first manifestation of the fire from an outside perspective. I wouldn't say the same argument for people who observe fires in the building. And I would also mention that in large compartments, I think the overall use of fire patterns is important, but maybe the value of them can be reduced in comparison to other types of information such as witness information or fire alarm data, detector information, especially if you have early fire witness information. And I'd also comment too that anecdotally, I see a lot of large compartment industrial type fires where the fire department will show up at the fire scene and they perceive this industrial warehouse or large compartment of some kind being fully involved. So fully involved is a generic term used by the fire department. Now, that could be interpreted as being the whole thing's burning. But from a research, from a fire science and dynamic side, that is incredibly improbable or unlikely because in large compartments, there's just simply not enough oxygen, not enough ventilation unless the roof is collapsed. But if you assume the compartment still has integrity, there's just simply not enough airflow to support a, quote-unquote, fully developed fire. So the person on the outside can perceive a fully developed fire because the windows are pumping out gas, there's no visibility is obstructed by smoke, for example, but from the fire dynamic side, it would argue and suggest that there's a local fire somewhere in the building that's being limited by the fuel load and the ventilation. And it may move around, which leads to the term traveling fires, which is a term that was coined for this concept of large compartment fires. So the fire starts in one location, it hits a limiting size and then it travels. So it moves, but it does not ever take upon that concept of a flashover flashover being uniform temperature, certain heat flux, certain temperature distribution of the ceiling. This condition will not be met in a large compartment due to the localized burning, the localized nature of the fire and the fuel load.

Rod Ammon: So it's basically looking for air and fuel?

Michael Rushton: Exactly. And it will travel towards those two.

Rod Ammon: Okay. So let's step back for just one second and define large compartment. What are we talking about and how is it different from small compartments? I know that may sound common sense, but let's talk about most fire dynamics research, what it's based on, and really how you define a large compartment.

Michael Rushton: Yeah, that's a good question too, because the words are somewhat what ambiguous or even arbitrary, because there is no technical definition in any building code for this. So for a small compartment, let's say that's the one that has the abundance of research behind it. It's what the vast majority of our fire dynamic is, what flashover is based upon. You talk about flashover and it's unequivocally based on these bedroom style compartments, these, let's call them 100 to 180 square foot. Certainly there's going to be a bit larger and but smaller ones within that area, but that's the general range. And then you kind of have this gap, let's call it. It's a gap above 2, 250 up to about 8, 900 square feet. This is an area I'd almost refer to it from my perspective as a bit of a transitionary area. And then large compartment being, I use 1,000 square feet as a defining number, but I could say anything in that region and upwards is going to always, at least through research side, has always shown these traveling fire or localized burning or non-homogeneous types of temperatures. So basically you have one part of the fuel load in that compartment burning and it's traveling around and it's very much controlled by the, like you mentioned before, by the fuel load and the oxygen, where it is, how much it is, the size of the fire, the speed at which it grows. Very, very important topic is fire spread rate in a large compartment fire. We don't talk about it as much as small compartments because flashover is seen to be as a somewhat instantaneous event, although maybe that's a bit a strong term, but it seems as being quick. While in large compartments, that fire spread rate, that numerical value at which the front of the fire moves forward, is very, very important and defined by a couple of parts of the compartment such as a fuel load and location of ventilation. And that speed at which it travels is how quickly that traveling fires moves around the compartment. 1,000 square foot as an arbitrary number, but just to identify that there is a transitionary area below with 100 to 200 square feet being small, then it transitions up to the large compartment. And there's a bit of a gap there. We don't know exactly when and how it transitions, but it does certainly lose those older fire models within that range right there.

Rod Ammon: Okay. So let's focus on fire dynamics basics of large compartment fires. How do they burn? What does the research tell us?

Michael Rushton: We're working on a data set of about, like I mentioned, 30 to maybe slightly more fires. Some were better documented than other ones. Like I said before, they're practically very, very, very difficult to burn. Most effective research looks at a type of a variable, one that they can control. So when you do a large compartment, ideally you're wanting to do a two or three. Sometimes there's one example with eight times, which is quite an outlier because of the cost associated with it. So you typically want to create this compartment, and then you want to burn it twice or more preferably, but let's say at least twice. And you're going to want to modify one of the variables of department. So ventilation is a common one that's modified. Fuel load, another common one, how much fuel load, the size of the fuel load, the distribution of fuel load, the ventilation, the amount of ventilation, location of ventilation, the compartment itself, you try and want to keep rigid for each set of experiments. And the result of that, if we collect this dataset, which is, like I said, not huge, but there is something there, we find that some of the most important controlling parameters are the first one being the fuel and also tied in there very closely is the ventilation. I say the fuel, when I say that the types of fuel, wood cribs are the most common type used in experimental research because of its ability to control it. But the types, even the moisture content of the wood, for example, can have a massive effect on the fire spread. The next one being ventilation. Ventilation, it controls the overall size of the fire. Mind you, I've never seen a fire able to burn the entire compartment unless the entire compartment's opened for a large compartment. But the ventilation will allow it to hit a limiting size, like a certain area of fuel. And then once that area of fuel is burning, it'll start to propagate around the compartment with a kind of leading edge. So you have the edge moving towards the ventilation, and then you have the decay or the burnout edge, which is the one that's exhausted its fuel. So these fires are occurring across a larger time span than a small compartment. So let's say more than 10 minutes, you're looking at 30 minutes, an hour, for example, or maybe two hours or three. And the burnout region will actually move around the compartment and the leading region will tend to move too, but towards the ventilation. So ventilation, fuel load, and then another third one coming in a bit later is the physical size of the compartment. So surprisingly, as the compartment gets bigger, so we're talking 2,000 to 3,000 square feet in area, you'd think that, okay, well, there's more oxygen, so maybe the fire can burn more vigorously. That's not the case. It's actually the large volume allows cooling down of the fire. It allows almost like the heat escapes the local area of origin, and it distributes across the volume of the compartment and actually slows down fire spread as opposed to speeds it up. So when you make the compartment smaller and smaller, you're actually allowing that little box, that insulating box to confine all that energy. And that energy, by the way, gets fed back into the fuel. So one way we break this apart is we call the fire spread is either plume driven. So it's like the area of fuel that's burning and it's driving its own fire spread. And this is true for a small compartment fire at the beginning of the fire. So the fire starts on a isolated piece of material, it creates the fire plume, and the radiation from that fire plume preheats the adjacent material and allows it to spread. That's a very slow, I mean, relatively speaking, it's a slow process. It's a slow fire spread rate. But as you build up the heat energy in the upper gas layer, you get a more dominant effect to do with fire spread. This is the hot gas layer driven or thermal environment driven, there's a couple of term ... Momentum controlled is one term from a set of research. It's to do with the energy being fed back from the larger system of the compartment. That becomes a very dominant and it becomes the dominant effect in large compartment fires. So when you have a smaller compartment, you allow the energy to feedback onto the fuel surface, and then you can get this kind of fire spread rate at accelerating rates, for example. And the compartment gets bigger and bigger, it tends to be a very slow, slow fire.

Rod Ammon: Interesting. Well, I'm trying to think about this from the perspective that I've watched a lot of training fires and compartment fires that are set. You mentioned that these larger compartment fires are difficult to do. And I'm guessing cost is obviously, you mentioned that right off the bat, but is it also about having enough fuel to get things going because this is a slower burn or what makes it difficult?

Michael Rushton: From a research, from a project perspective we talking about?

Rod Ammon: Yeah.

Michael Rushton: So I think, and I tried to do a large compartment fire here in Canada, in Ontario where I'm from, and I can say from firsthand that there is a lot of red tape to do these kinds of ... So first of all, it's just finding a site that in itself is difficult. You typically need a non-combustible building. These are big fire, like large compartment fires, although we talk about them being local, the amounts of heat produced by them are tremendous compared to small compartments. So you're looking at something in the order of 5, 10, or much more megawatts of fire, which is bigger than a regular small bedroom fire. So the compartment, the building it needs to be in needs to be able to handle that fire. You don't want it to collapse. If it collapses right away, then there's no research there. It's just an open burning fire. So the compartment's the first one.Finding an abandoned old industrial building or commercial building that has a non-combustible, a sturdy structure, robust structure, that's the first challenge. The second one is the environmental issues and the community issues. If it's nearby people, or especially one we had nearby that we were going to burn was a couple kilometers from a school. And although the school in the summertime, we were going to do that burn, the problem was even having this large ... This is not a small smoke plume coming up from this building. Like I said, you're talking about many, many megawatts of fire. So these smoke plumes, they're an eyesore to the community around them. And in fact, more than that, they're going to call emergency services. So then the second one needs to be remote or at least remote enough that there's not going to be a upset to the community. And then the third one is just the sheer amount of energy and resources needed to do these because any research from the community perspective is only as good as the instrumentation and the output from it. I've done many burns myself, many fire burns of various compartments, but if you don't instrument and you don't pull the information from them, there's nothing to share with the larger community after. It's like you learn yourself because you're there, you see it, you experience it, which is great and I fully support that and I've done it many times. But, in order to provide that back to the community, you're looking for a large compartment, I mean, the equipment itself, I'm going to say is at least $10,000 minimum for the thermocouples, heat flux sensors, just enough data to pull from that for it to be useful in the larger community.

Rod Ammon: It's a big experiment to put on. I get it. And I can tell you from trying to find locations where we could do burns that would be used for training, I totally understand the difficulty, but I wanted to get in a little bit of that because you had mentioned the difficulty. For a moment there, it made me think about sometimes I hear these large residential compartment fires burn so fast. And when you said it was harder to get them going, I wanted to sort of clarify because it seems as though a lot of these larger compartment fires burn so fast because of the materials that we're putting into homes. Would that be accurate?

Michael Rushton: Yeah. Actually, you hit it on a very one little unique topic that I think it would be worth mentioning is combustible linings. And what you said is true generally speaking, the amount of fuel load is a huge ... The type of fuel, the amount of fuel load is huge for any compartment fire, including a large compartment. But I'd also make a side comment too, is that combustible linings is being researched in large compartments, in part due to these large timber style buildings that are ... I know they're happening in Canada, and I think it's probably in the US too, they're starting to experiment with massive timber, cross-laminated timber sometimes is referred to as CLT. And this is a unique type of building construction, and it's causing combustible linings to be exposed. The timber itself is a combustible lining. There's also, I've seen cases ... Actually, I saw a double fatal a while back in a compartment, the occupant put fiber board, like a medium density fiber board all around the walls and the ceiling in the walls. So I have done myself burn experiments with this material, and there is research that's been produced on this topic. And I would say that combustible linings is incredible the effects on the fire spread rate of a fire, so the severity, how quickly it burns. So before we talked about plume-driven and hot gas layer driven, when you add a combustible layer to the compartment ceiling and the walls, you're no longer looking at the model we talked about of a hot gas layer, because now it's a burning layer. There's a fire on the ceiling that's in a hot gas layer, and that fire itself will have a significant heat flux, like heat driven back down onto the fuel load. So now it's not just the hot gas, but now you have active live combustion. What we've seen is that that will ... Almost, you have to take these experiments and put them in a group of their own. They're that special and unique and violent and aggressive. So to speaking in general, if you get a fire where investigation where you observe combustible linings, take that as a very significant characteristic of that compartment that may have accelerated the fire spread.

Rod Ammon: Interesting. I wasn't thinking about that and I'm glad you've spent the time to explain it. That does seem to change things quite a bit. I can understand from the flammability perspective of everything around and the fuel. Yeah, I guess that would create a lot of changes. So how can the fire effects and fire flow indicators commonly used in these small compartment fires, how can that analysis work differently in large compartments?

Michael Rushton: Yeah, that's how I challenge myself too with these large fire investigations that I'm conducting. I say to myself, if the models aren't clear and they're still being developed on how we understand fires and combustion compartments, how then I as an investigator can look at the evidence from a compartment fire and make some comments as to potential origin? The first comment I would make is unfortunately it's often in fire science and any sciences, the more you learn about something, the less you realize you knew about it. And this is very true for large compartment fires. So even in a small compartment, whenever I'm dealing with post flashover fires, I'll always say to myself, well, okay, now ventilation and fuel load configuration need to be accounted for. So when I'm looking at the damage, if there's a concentrated amount of damage by a window, I need to consider that the window could be associated with that or caused the increase in damage. Maybe not, but it's consideration. That same logic I think needs to be applied more significantly to large compartment fires. You should expect the fire to propagate. The evidence we have on the research side suggests that the fire will tend to propagate towards a window and it'll also burn more vigorously. So when I say burn more vigorously, I mean mass loss. I mean that the material's consumed, which creates a fire pattern, which we as investors go in and look at fire patterns. So you should expect strong, deep fire patterns by windows that may not be an origin pattern ... It may be, but just to consider that that effect of ventilation should be at the forefront. And pre flashover small compartment fires, you don't think about it until later. You should think of it right away in a large compartment fire. Ventilation, fuel configuration. If you have an opportunity to sit down, find out how the fuels were configured. If it's not evident within the evidence you see in the scene, it's too damaged, for example, you may have to sit down with the building owner, find out how things were laid out. I also find that witness information and data or digital surveillance video data becomes more valuable because it becomes more objective. It's not as biased and affected by those fire patterns. The overall fire patterns themselves, while they should be documented and used, they should be used with caution.

Rod Ammon: Okay. So one of the things I had from notes here was that partly this is because the hot gas layer may not build at the same speed in a large compartment related to the size and volume. It's harder for the hot gases to build and descend. Is that sort of the reason? Because like you said, it's searching for fuel and oxygen towards that window, but we're not getting the flashover as you described at the beginning.

Michael Rushton: Yeah, definitely. You can take the concept of flashover and you can basically just remove it from the equation I think is a very safe thing to do. So there may be examples that do meet flashover that are unique, but I say generally speaking, I'm yet to see a large compartment example go into what would be commonly qualified as a flashover. So it's probably going to be the fire starts in a location. It hits a certain ... It grows, of course, it has to start and grow. It starts to burn in an area, a local area. And then the compartment fills with gas, but the gas offsets the oxygen, but it's not like a small compartment that hot gas starts to obliterate everything in its region. It could be less affected, far away, more locally. It's hot, but that burning location of fuel will continue to spread. You'll have an advancing side of the fire and it will tend to move towards the oxygen. What you cannot do though is you cannot use comparative damage analysis along the pathway of that fire. So if you have one section, let's say that's burnt of a large compartment, and you look along that section, you document it, and you find that clearly the fire traveled along this fuel load and then it eventually maybe hit a ventilation, which would be typical. You look at the damage within that trajectory, that pathway of fire, you may have areas that are more damaged and less damaged. I would say that the depth and degree of that damage isn't typically related to where it starts as opposed to a small compartment fire, which is often the concept we apply that a concentration of damage could just be the concentrated area of fuel load or the ventilation that's immediately adjacent to it. It's seldom having to do with where the fire originates. So when you document, but like I said, that the value of the individual fire pattern used against compared to the other one, I would call it less valuable. Or if it's used, it needs to be used in consideration of ventilation and fuel load configuration. If you factor those two things in, you can get a stronger ability to analyze those individual fire patterns.

Rod Ammon: So putting it simply, going back to some of my notes, decay process can be different in large compartment because of how the fire is moving around.

Michael Rushton: Yep. And we're seeing that it typically burns the, in most cases of large compartments, it actually exhausts the fuel load in a local region. So it'll start here and then it'll actually exhaust that fuel load as it propagates. Now, keep in mind that that could be affected by the window. You could get a fire start at one location, it propagates to a nearby window and then exhausts the fuel in that location and then comes back. Oh, that's confusing. It starts in one spot, propagates over the window. And I would say that this has been observed in not less than five or six full large compartment fires where it starts one location, goes to the window, completely burns locally, regionally, in that window, and then propagates back into the compartment. So that's going to leave me, us, as an investigators, very ... I would say, I mean, fire patterns are fire patterns. They only become a problem when you start to analyze them, but at least in terms of origin analysis can lead to potentially deceiving fire patterns.

Rod Ammon: Yeah, that's interesting. I mean, I'm visually going through this in my head. It's just the way I roll and I'm seeing this fire traveling towards this window. And then what you're saying is after it takes its fuel and does what it can do right there, it starts to go back into the compartment. What's causing that?

Michael Rushton: It's the, well, in general terms, I'd call it fire dynamics, but in specific terms I'd call it oxygen. It's air. It's oxygen. It's actually, I don't think it's unique to large compartments. I think it happens at a very different scale in small compartments. I hope in the future we can have a set of research that looks at this in small compartments, but I can describe it more accurately in large compartments because it's been observed from the research side. So in a large compartment and a small compartment, but exaggerated in a large compartment, you're only able to burn with the oxygen you have available to you. And that oxygen will be provided by ventilation or openings in a large compartment. There's going to be leakage through the compartment. This is true for all compartments, but this is not necessary amount of oxygen to feed a 5, 10 megawatt fire. It needs a window, it needs a door, it needs something, or else it'll just stuff itself out. So that window, that door will become the only, or maybe there's multiple ones become the only oxygen sources. So as the oxygen's entering the room, it's going to be used at the first location of combustion, which will almost always be leading edge of the fire. So it gets consumed at that leading edge, but that's also driving because the combustion is the most violent heat producing event in the fire as opposed to the hot gas, which is a secondary kind of heat feedback. So that combustion is going to keep propagating towards the oxygen because that's where that, let's call it combustion zone, let's call it. It's where that fuel meets that air. And as it moves towards the oxygen, because that zone's following the oxygen, we observe that it hits the window and it just burns locally at the window. So there's no ability for the oxygen coming through those openings to migrate into the compartment. So even if the fire starts farther away, it's simply there's just nothing there to support its combustion process.

Rod Ammon: Okay. I was trying to get at, and I might just be missing it, but what I was trying to get at is what would make the fire come back? What I thought I heard you saying was is that that fire, it goes to that window, it builds up there and then starts to come back into the compartment again.

Michael Rushton: Yes, exactly. When the fuel's burned out. So it will burn locally. And these examples are most significant when there's controlled oxygen intake. If the roof is gone, it's more violent and whatever. But let's say there's a large window or a couple of large windows, the fire will burn locally at the window. So the fuel will ... There's a discrete amount of fuel there. There's only as much fuel as whatever's at the window. Once that fuel burns away, the fire will then start to trace back into the room, maybe indiscriminately, just in any direction, just looking for the next fuel to burn.

Rod Ammon: Okay, got it. Thank you for clarifying for me. So there's a lot going on. What should fire investigators take away from this discussion? The next time they respond to a fire that's a large compartment, what should they think and do differently?

Michael Rushton: Yep. I think that ... So the basic methodology that we are applying to a fire investigation as we go, we document like any fire and you're going to collect your data. The part that is the analysis part ... So what used to be the four pillars of origin determination became, let's go to three pillars in the newer versions of 921. There's the fire patterns, dynamics, and witness and other digital sources. The analysis part, the part that goes into the dynamics, when you take all those fire patterns and you start to analyze them with respect to how fires work, I think you need to put on a different hat, a different thinking. And I wish I could refer to a singular model that best describes large compartment fires, but I can only tell you what we know and what we know is we're kind of building the models and understanding right now. But I could say that we should not apply small bedroom style fire dynamics. We shouldn't think of it as a hot gas layer, a cool ... Well, at least not a uniform hot gas layer, not a uniform two layer zone model. And you shouldn't think of any concept related to flashover. So that's kind of that idea that the fire quickly burns across the exposed fuel load, the surface of the fuel load within compartment, and then, poof, everything's simultaneously burning. This I have not seen ever happen in a typical large compartment, so that concept is out the window. So what we do know is that the concept of traveling fires is applicable in many large compartments, so the fire moves about. So one thing I'd say is that even though the fire department may have observations that the fire was fully developed, the compartment was completely burning, probably not the case, it's probably locally burning. Finding out information that maybe where it was locally burning could be valuable because you could get some time lapse or time data as to specific locations of local burning, and that could be supportive of a pathway of a fire. And then of course, the fire patterns themselves, you can document yourself where those fuels are exhausted and burned away, and those locations will be part of that pathway and then add a layer of ventilation and then fuel load distribution onto it. And you start to get a little bit of a picture of how a fire normally propagates in a large compartment.

Rod Ammon: Does it change the way you're looking for a point of origin?

Michael Rushton: Absolutely. When you do that fire dynamic analysis, I think if you use a large compartment type of fire dynamics, the individual fire patterns are going to be probably a little bit more difficult, a little bit, I'll say less valuable at certain times, especially if you have window effects around a window very strong, a lot of fire damage around a window. So like I said, in large compartments, I tend to lean on a lot of those other data sources a bit more too, the witness observations, and then the digital surveillance video, smoke alarms, big one, smoke detectors. If you get a zone specific or even a device specific fire alarm system, that kind of information is invaluable for large compartment fires.

Rod Ammon: Excellent. So what's next? What are you going to do now with this topic of large compartment fire dynamics research?

Michael Rushton: So right now, because I'm currently doing some ... Well, the topic of fire dynamic simulator is somewhat shown its face in fire investigation. We are all familiar with it. This is the area that I'm currently working in. I'm using fire modeling to recreate and maybe help define large compartment fire dynamics. This is a way to curb and reduce the cost of those large scale tests and also maybe to isolate some of those parameters like we talked about, you try and change one thing between experiments to change a whole bunch of things between them and compare them to each other. So I'm working quite a bit with FDS, so fire modeling, basically.

Rod Ammon: Yeah. Cathy mentioned in one of her notes here that AI may be useful, but you wanted to be real careful about that.

Michael Rushton: Oh yeah, absolutely. So I'm a bit of a nerd myself. When new technology comes about, I really like to play with it and explore it. So AI being ... I mean, it's hard to not see the impact it's having on the world and the changes it's having. It's actually shown up quite significantly in fire modeling, for me, with AI now, you can help build the code associated with FDS scripts like the models you create. Very interesting. You can take a design of a building and you can have AI help you develop the obstructions and development of the geometry of that code from that blueprint. There's actually AI programming specific to FDS, but I also caution ... There's, by the way, a new section coming into 921 on the topic of AI is almost like we kind of need this new technology to define how it can or should and should not be used. And that right now, the drafting in that new section is suggesting that anytime AI is used, it should be independently verified or validated. So I see it as a great tool. I'm really excited about its ability to support the modeling efforts, but at the same time, I draw strong caution in the sense that anything you produce from it needs to be independently validated and verified. So don't trust the systems. And I'll tell you firsthand the hallucinations and misinformation they'll provide. They don't tell you how much they're lying, they just give you every answer with extreme confidence. So there's no gauge at the top of an AI response saying 50% confidence, 80%. It just tells you the answer with pure confidence. And I can promise you, especially when you get really, really technical with fire dynamics and these very precise things, you move to the leading edge of these topics, it breaks down and it presents it as real information. So be very, very careful.

Rod Ammon: Totally hear you on that. I'm glad you brought it up. I used AI to try to recreate some fire doors for passive fire systems. And it was very interesting just to see, as you say, the hallucinations. I'm looking at the door and I'm like, okay, the door's going to close. Well, it closed from the wrong side,. Closed from the hinge side or the entire door just went away. It was amazing to watch what you would think would be a simple prompt needing quite a bit of detail to get it to be accurate. So I think your focus on the modeling sounds like that would be the first, most useful place for AI. Add a window here, put a wall there, that kind of thing, I guess that's what you're saying.

Michael Rushton: Absolutely. It will build you ... So it's very interesting you can use some of the big prompt models to build FDS codes. You can give it the manual, the user manual. "Here's the manual, build me a room with this." It'll do that. It'll build you a room. Whether that room will be correct, that's another question altogether. So the tool is becoming there. It can do something, it can build something really quickly and effectively. But you cannot accept that result ... Like I said, it needs to be independently verified. So you still need to learn what it's building. It's going to do it for you, but you also have to learn it yourself. So it's a great tool, but you just use it with caution.

Rod Ammon: Great. Thank you. There's a lot of discussion about AI, so I just wanted to make sure we tapped on that. What else do you want to tell anybody before we go? Any of the fire investigation folks that are out there, people that are listening and could learn something else. Anything?

Michael Rushton: Yeah. Anyone who has come across some of these, especially fire modeling topics or even large compartment fire modeling topics, love to have them reach out and we can coordinate. I'm always interested in fire investigation specific research, which is tough to do sometimes because we're a bit of a small group with respect to the larger science fields. So anyone, especially those topics like compartment fires looking at oxygen effects to do in the compartment, like how oxygen affects burning rates. So I'm not talking qualitatively, I'm talking quantitatively. How can we start to quantify and better understand how these windows are affecting burn patterns and stuff? Yeah, that's very interesting to me, and I'm hoping we can solve some of those problems in the future.

Rod Ammon: Wonderful. Thank you, Michael. There's a lot to think about here, and it's interesting to challenge some of the assumptions and talk about where some of these fire dynamics or rules come from so we can continue to evolve how fire effects and fire patterns are interpreted in origin and cause investigation. And I think you're helping us keep up with the changes in some of these modern built environments. Once again, very, very grateful for your time.

Michael Rushton: Absolutely. Thanks a lot for having me.

Rod Ammon: CFITrainer.Net recently revised and updated the NFPA 1033 in your career module for the 2022 edition of NFPA 1033, standard for professional qualifications for fire investigator. If you have taken and passed this module before, you'll want to take this updated version to refresh your knowledge based on the current NFPA 1033 edition. If you haven't taken this module, now's a great time to check it out. It will help you plan your professional development and organize your credentials to show how you have fulfilled NFPA 1033 requirements. Just completed at the end of 2025, CFITrainer.Net launched Instructor Mode, a feature that allows instructors to use CFITrainer.Net modules during in class instruction. Instructors can play lesson videos from the module, step through knowledge checks to guide discussion, and access the supporting resources to extend the learning. These capabilities encourage instructors to integrate CFITrainer.Net module content into their classroom instruction to teach key concepts, spark discussion, support group work, and go into greater depth with content from supporting resources. After using CFITrainer.Net module content in class, instructors can authorize their class attendees to take the module skills challenge test to earn their certificate of completion without having to watch the modules lesson videos again or complete the knowledge checks on their own. To apply for access to Instructor Mode, you'll need proof of status as an instructor in fire investigation or a related field. The article announcing the launch of Instructor Mode has more information on the feature and the application process, and you'll find a link to that on this podcast episode page on CFITrainer.Net. Eagle eyed CFITrainer.Net registered users probably have already seen the available program search feature, which has been added at the top right. With close to 100 modules, typing a keyword in the search box can help you find the program you're looking for. Give it a try, let us know what you think. Finally, CFITrainer.Net has added subtitles for three new languages to the lesson videos of a set of core fire investigation modules. Canadian French, Brazilian Portuguese, and Arabic subtitles are available for select modules using the CC button in the lesson video toolbar. This IAAI CFITrainer.Net podcast is brought to you by the International Association of Arson Investigators and is made possible with funding provided by the Fire Prevention and Safety Grant from the Assistance to Firefighters Grant Program administered by the Federal Emergency Management Agency of the US Department of Homeland Security. Support also comes from the Global IAAI Membership Network, the Bureau of Alcohol, Tobacco, Firearms and Explosives, Strategic Partners, and voluntary online donations from CFITrainer.Net users and podcast listeners. Thanks for joining us today on the podcast. Stay safe. We'll see you next month for the International Association of Arson Investigators and CFITrainer.Net. I'm Rod Ammon.

NFPA 1033 and Your Career

Instructor Mode

Available Programs Search Feature