103 - The Science and Art of Scale Modeling with James Quintiere

Speaker 1: Hello everybody, welcome to the Fire Science Show. 

Speaker 1: I am here today again with Professor James Quintiere. The previous interviews with James were rather popular over the audience of the Fire Science Show, so I thought let's take a chance and invite James once again to the podcast, and this time I've invited him on a very special topic, a tool that I am using quite a lot in my research, a tool that I know is used and to some extent I've used in the Fire Science, and that is the use of scale modelling. What does it mean to have a physically scaled model of fires? What are the dimensionless groups that allow us to do that? What are the relations that you have to conserve to claim that your model is a true scale model And what can come up from that? And, of course because I am talking to James Quintery a lot of stories from the old times of Fire Science and how all these things were built up in the 70s, 80s and in transfer to the modern times, how scale modelling was used before the CFD and how it can be used today, in the age of CFD modelling in Fire. A very interesting discussion. I've enjoyed it a lot. A little word of warning it's gonna be a little more tough than the usual Fire Science Show episode. There is a lot of physics and dimensionless numbers flying around without a warning in the podcast episode, but please do not get discouraged by this. Just focus on the applications and the possibilities that this physics brings to us. And if you're a scientist, well, you have to realize that if you do not understand this physics well, scale modelling is not gonna work for you, because the scale modelling is the art of finding what physical phenomena drive your particular problem and how to describe them in the best way. That's the art of scale modelling, and I hope this is something that comes up for you from this podcast episode. Anyway, enough, let's spin the intro and jump into the episode. 

Speaker 1: Welcome to the Fire Science Show. My name is Wojciech Wegrzynski and I will be your host, as usual. I would like to say thanks to the sponsor of this podcast, ofr Consultants. This podcast is brought to you in collaboration with OFR Consultants, a multi-award winning independent consultancy dedicated to addressing fire safety challenges. Ofr is the UK's leading fire risk consultancy. Its globally established team has developed a reputation for preeminent fire engineering expertise, with colleagues working across the world to help protect people, property and planet. In the UK that includes the redevelopment of the Printworks building in Canada Water, one of the tallest residential buildings in Birmingham, as well as historical structures like the National Gallery, national History Museum and National Portrait Gallery in London. Internationally, the work ranges from Antarctic to the Atacama Desert in Chile, to a number of projects in Africa 2023,. Ofr is growing its team and it's keen to hear from industry professionals who want to collaborate on the fire safety futures this year. Get in touch at OFRConsultants.com. 

Speaker 1: And now back to the episode on scale modelling. Hello everybody, welcome to the Fire Science Show. I'm here today together with Professor James Quintiere. Again, hello, James, great to see you. Yes, great to be here. Wojciech, so happy to have you In our last interview. We've touched so many things from your career and interests, but there's one thing that we, i think, have not touched at all, and this is so important to me as someone dealing with smoke and fluids and fires that we need to have a whole episode, and that is reduced scale modelling, an experimental technique that has been with the fire science for ages, i guess, and is still being used. It's still fundamental in the field of tunnel fires that I work in. So let's do a scale modelling episode. So how did we start up using scale modelling as a research vehicle in fires? 

Speaker 2: Well, scale modelling starts way before that. The Reynolds number was the key parameter in fluid mechanics and most of those experiments are done in small scale and they carry out into the turbulent domain of Reynolds number. And then, if you go to heat transfer, most of the correlations in heat transfer for turbulent flows were done on small scale And the Nusselt number, Reynolds number, Prandtl number correlations holds and those parameters were identified as the key dimensionless groups Prandtl, Nusselt, Reynolds In natural convection, the Grasshof number instead of the Reynolds number. So people have used this and they've used it effectively. And when today people say it doesn't scale, recognize that most of the correlations in heat transfer came from that place. And so when people make fun of the correlations in fire and say they're empirical, yes, they might be, but they have a basis in scaling. So people need to understand that. The other thing I learned is that in the 40s and 50s, structural people use scale models a lot. Okay, In fact, they used it for different ways in structures, They used it for impact, They used it for loading, until they discovered CFD, where CFD and structural analysis is relatively simple compared to fire. So they took off in the 60s after that, going down that path and forgetting about scale modeling. But it still works. 

Speaker 2: There was a professor, Emory, at UCLA, that did scale modeling in automobile accidents Cars bouncing off poles or bouncing off each other And from the end result of the full-scale accident he could predict what happened based on small-scale experiments. And Kozo Saito was either new Emory or was under him or liked him, And Kozo and Emory sponsored a I think it was under ASME. It was a scale modeling conference in Tokyo back in the 80s And it didn't just bring fire, it brought in structures, it brought in airplanes, winds on buildings. It brought in people studying acoustics in upper houses by scaling the voice down to different frequency. I mean, it was so revealing that it opened up a whole world to me. That said you know, yes, we have computers, we have our algebraic equations, but please don't forget about scale modeling. Don't forget about scale modeling. And so that's where scale modeling comes from. 

Speaker 1: I remember a lot of very fundamental works in the world of smoke control, many of them, when you start tracking back, you end up either with Philip Thomas or Margaret Law, who were applying exactly this methodology to unravel the problem. They unravel the first correlations And then from that working out the physics, like why does the smoke behave like this? 

Speaker 2: And those people with Philip Thomas actually did scaling. They used the parameters to scale their equations, but they also use scale modeling, I think, for roof vents. They even did a form of salt water modeling And in salt water modeling you have very low buoyancy, which is the difference between salt and fresh water, But the flows that are revealed under this low buoyancy are very, very accurate. If heat transfer can be ignored, Just the strength of the fire or the source has to be modeled and the geometry is made geometrically scaled. And if you go back FDS, which is a large eddy simulation code that only deals with viscous effects in a very approximate way, you talk about empirical modeling. The viscous effects in FDS are empirical And if you take them out you have to put them in, otherwise the flow becomes unstable. But if you run FDS against salt water modeling, it should be one-to-one perfect So you can scale fluid mechanics very, very well in fire modeling. 

Speaker 1: And power of scale modeling comes from the difficulties of fire research. If you want to research fires in buildings, there's not that many building owners that would allow you to burn them down for science. It's very difficult cost prohibitively sometimes to run large-scale experiments And having this window to look inside the physics, employing a smaller scale experiments carefully is definitely a powerhouse for discovery, for building the understanding of fires. 

Speaker 2: Let's just talk about that for a moment, because you have smoke movement issues in buildings, high-rise buildings and atrium. Those can be scaled. You could study pressurization effect. You could study smoke movement. You could study the effect of vents and dampers. You could study when vents should open, when they should close. All of that can be done in a scale model. Like you said, people won't give you the whole building, but what you learn from a scale model like that could be really revealing. And it scales Because when the smoke moves away from the fire, the heat transfer diminishes. And now you're on your own with fraud scaling and QSAR and you take it from there. 

Speaker 1: So the basis of scale modeling. The purpose is kind of obvious You want to have a smaller experimental setup in which you can study a certain phenomenon. But first the most important, perhaps the fundamental question how do we know that a fire in my small scale model is the same or representative of a fire? that would happen if this was a real-life building. How can we tell that the fires are the same? 

Speaker 2: Well look, you write down the differential equations or you write down the algebraic equations, you can put them in FDS form and now you produce all the dimensionless group. And now you identify what are the important ones. And I don't think you have to be a super scientist to start figuring out what the important ones are. So that's how it's done and that's how it's justified. Forman Williams wrote papers on this way back in the 60s and said you can't do it all, but you can identify most. And so when people do heat transfer and it's like a very funny geometry they don't say oh, it's a funny geometry, it doesn't hold. As long as the frictional effects are the same in that domain, et cetera, it's going to work. And so you justify it through the equations and your understanding of the chemistry and physics. And those same equations would really have to be solved completely on a CFD. And they're not. When people say I'd rather use CFD, they're not picking all the terms in those equations, they're not fully resolving. 

Speaker 1: So let's talk about the dimensionless groups. Yes, what does the term a dimensionless group even mean, and how many are there? 

Speaker 2: If you look in the summary paper that I put in the scaling modeling conference, it lists them all and it lists them from the differential equations And I don't know. There's probably about 20 to 30 or maybe more. And now, if you put in structures, there's more, and if you put in water effects of droplets, there's more. It keeps expanding, but these things work. Gunnar Heskestad was my inspiration. He scaled fires in compartments back in the 70s and did a very, very good job. But he then turned his focus on scaling with droplets and did a very, very good job. I mean, his standard at FM Global was really high. He had to make those people believe or they weren't going to let him do the work, so he sold it. You know he's an inspiration. 

Speaker 1: I'll try to link the resources, the ones that you sent me and the ones that I know, because, you are right, there's so many parameters that you can scale and this dimensionless groups. There's a lot of them and they describe different aspects of fires and so on. 

Speaker 2: You need to understand what's important and what you're throwing away. But, as I said, people do this with CFD as well. 

Speaker 1: Now, if I understand correctly, the point of the dimensionless number is that if you have two phenomena with different physical scale of the phenomena, like the dimensional scale in meters, and have the exact same dimensionless value for those two, you could, let's say, assume the physics is the same in the small scale and the larger scale. 

Speaker 2: Yes, what you do is let's just take a fluid mechanics problem where the temperature is isothermal, and you put flow down a tube and you know its velocity and you know the diameter of the tube and you know the length of the tube and you measure the pressure drop across the tube And then you could either put those variables together and say you know, we have to include viscosity, we have to include something else, and you arrive at dimensionless groups, or you write down the equations and you make everything dimensionless in a you know, a choice of yours. Usually geometry is well, we'll use a length scale and make everything similar, you know geometrically scaled, and you go from there. So what you get and you get a dimensionless velocity is now a function of geometric dimensionless terms x over l, y over l, stuff like that. And then the Reynolds number pops up. And so you say, if I make the Reynolds number the same and make it geometrically scaled, then my velocity behaves in this way. 

Speaker 2: You want to put temperature in, go to the energy equation, and then you get dimensionless temperature as a function of Reynolds number again. But now it might be a function of something else. If buoyancy is involved, maybe you got to put q star in there, because now you have a fire size, and so that's how it works. If you look at early work by Bernie McCaffrey and Fireplumes, he got all the dimensionless stuff right And he was much more I wouldn't say intuitive, but he wasn't always deriving them, he just knew that that's what they were and he used them effectively. 

Speaker 1: One thing that when I was discovering scale modeling for myself many, many years ago, the first thing that seemed extremely convenient perhaps this is also related why this method is so successful in fire was the fact that the temperatures in a small scale and in a full scale would be the same. Like the temperature does not scale, and many of our experiments would focus on measuring temperatures, especially when you talk about structural domain as well. So it's very convenient that if you scale down your fire, you measure 1000 degrees. It means that the fire of the same size in large scale would produce a 1000 degrees temperature. That is very convenient. Also, concentration scales the same way. So if you have 5% CO2 in small scale, you would expect the same in large scale. 

Speaker 1: Other groups do not scale that easily. Well, other groups to scale, and each of them scales at a different power related to the length. I always found the most fascinating that the time scales as well. So one second in your scale model could be a 10 second in full scale, which actually is why they should scale down the shots in videos in slow motion, because then it looks very real, right See? 

Speaker 2: you could even scale radiation by making the radiation fraction from the flame the same. If you're burning some fuel in your full scale, now you pick a different fuel that makes the radiation fraction the same on the smaller scale, okay, scale down the radiation factor And that then compensates and puts radiation into it. And if you really want to carry it further, look at the soot production and the radiation from the smoke. So in the smaller scale you would want to have more soot because you want to compensate. So you can do things like that. And here's just a side note I sent you a paper also on the World Trade Center. It was this ceremonial one for 20th anniversary, so it talks about people. But buried in there is my critique of this. 

Speaker 2: When this started their investigation, before they even started, i gave them a talk. 50 people from the NIST program showed up for that talk, or more. They were standing. I said you should do this with a scale model. You should do the structures part, the damage from the plane, and you should do the fire part after the damage of the plane. And you should do it in scale modeling because structures, people have done that before. And I said the reason why you should do. It is because you're going to use computer models now and you need to test them against something. What are you going to test them against? 

Speaker 1: Well, they didn't do it, but with Andre Marshall and myself, our laboratory class did it at Merlin from what you were saying you've already mentioned, like scaling radiation and stuff like that, and the image appears that This is not a simple scale, that is not just building a five times smaller model of your compartment. 

Speaker 1: It's about scaling not just dimensions, but some physical properties that are regardless of the size, and And this perhaps is the difference between good scale modeling and not so good scale modeling, where you would just omit things that perhaps have an impact on your results and then we'll definitely go there in later in their interview. I would like to close up this introduction to scale modeling with Short question. We often call scale modeling fruit number modeling, so, and food number is paramount to the scale modeling as a way to describe the fire. So let's talk about food number. If you could explain the listeners like what is Food number and why this particular number is one of the most important ones in the whole scale modeling concept, well, fruit number is important because it contains gravity and in fire you have buoyancy due to gravity. 

Speaker 2: In space you have no gravity. So it's a totally different ball game. You have gravity also affecting ships. The wave motion of the ocean or the river affects the ship, mm-hmm, when people so, a fruit number is basically the ratio of Momentum to buoyancy forces and you get out of that comes the fruit number. If you say viscous to momentum forces, you get the Reynolds number. So in modeling both flows They can't do both Mm-hmm, but they picked the fruit number. So in these models both scales, they forget the Reynolds number. They just say make it large enough so that the flow is turbulent, okay, and and so right away, that's a basis for fire modeling, because that's what people do in fire modeling it Instead of a wave motion it's the buoyancy, and it's the buoyancy, you know, with momentum. That gives you a so-called fruit number and you could turn that into energy release rate By manipulating the equation. So you get, alternatively, q star, which is the energy release rate Over a gravity term and some dimensions to the five-halves power. So that's, that's how that comes about. 

Speaker 1: Yeah, the ratio of momentum to gravity being the fruit number. 

Speaker 2: And the momentum is due to the buoyancy, buoyancy forces and the movement of air momentum is just if you have a ship, it's moving through the the waves. And if the size of the waves are gonna be controlled by gravity, Fantastic. 

Speaker 1: And there's the second number, which we call the dimensionless fire size, and in some of your work it was proposed It's called q star. We often refer to it as the q star and in some of your work It was named Zukoski number, after Ed Zukoski of California Institute of Technology, who introduced it. I actually love calling q star Zukoski i number. We should change it to z you and and we should. Well, i will champion that if you're, oh, ed Zuchowski called the q star. 

Speaker 2: Okay, he made it. He made two dimensionless in fire. He can't just call it q, he could call it q prime, or you know. So he called the q star. So, okay, he gave it that name. And I just mentioned that It should be known affectionately as the Zukoski number, because He was a dear friend and he, he gave me so much That I learned from and he gave so much to the field. And he died young, relatively young, and he had another life too. He was a rocket scientist and he understood instabilities in and rockets. And when Richard Feynman was on the committee to discover how the challenger blew up with the bad old rings, he interviewed Zukoski, because they're both at caltech, and Zukoski told him no, no it, it can't be from the instabilities in the rocket. And Feynman puts that in his book. He doesn't mention Zukoski but he says I went to some fluid Dramas and that did rocket stuff, and so Ed had a position at rocket dine and he had funding from. 

Speaker 1: NIST. Did he ever distinguish it was harder, rocket science or fire science? 

Speaker 2: No, no, no, i I just say that because if you look at it, i mean rocket scientists need to understand the Mach number And and and that's like again that dimensionless number. Yeah, that that's like entropy. You know it's a little mysterious, but if you look at the parameters that govern fire and the fact that it's not a control system, it's uncontrolled, it's got the mind of its own, it's more complicated than rocket science. 

Speaker 1: Yeah, and for the physical meaning of zuchowski number, the q star, it's a ratio of firepower to Some sort of enthalpy. 

Speaker 2: I guess the amount well, well it's, it's really, you might say, firepower To buoyancy power. I mean, okay, that's what it really Amounts to. So it's kin to this Frode number. You could relate it back to the Frode number. If you take the velocity out of the Frode number And combine it with q in a convective term in the energy equation, you'll, you'll out, will pop the, the q star so, in a way, a dimensionless way of saying the how big the fire is? 

Speaker 1: Yes, right, yes, okay. Fantastic, because, um, fires will exist in a certain, let's say, combinations of sizes and shapes and powers. Uh and well, it's physics. I hope we did not lose many listeners due to overly physical explanations of these numbers, but it is. It isn't. For me, it is fascinating. As a scientist, i find it. 

Speaker 2: Here's another thing when I got into this I was a young scientist at NIST and most of us were new to fire and didn't know anything about fire. 

Speaker 2: Okay, and they were doing corridor fire experiments where flames were moving down corridors very fast, and so I had the opportunity to look at the overall dynamics And we made a scale model with a window on it and we could see the smoke dynamics, we could see the temperature distribution and then, curiously, McCaffrey measured The velocity distribution and he saw that instead of just two layers there were four. Okay, there was a ceiling jet and then recirculation in the hot layer And there was a floor jet and a recirculation in the lower layer. And so people had computer models in the 70s. They were based on the Reynolds number averaging type models like Spalding used, and they they would, because they were putting in the viscous effects, they were smearing out this stuff and they were not getting the floor Layers of velocity flow Okay, and they couldn't predict it. And so they were using our experiments and then saying our experiments were wrong. So we redid it in the large scale. The McCaffrey made the measurements and they're almost dead on. So let's. 

Speaker 1: Let's go to the use of of scale modeling Across the fire science. It had so many uses that been fundamental to so so many things. We roughly in Covered compartments. No, let's go back to plumes, maybe. So scaling the Flow itself in the plume. How much smoke is produced? How does it move? How does it bounce off ceilings of walls? How does it flow around obstacles and through openings? I think scale modeling was highly useful to unraveling this phenomena for the first time and then putting an equation. What's your opinion? How did scale modeling been used in this part of fire science? 

Speaker 2: Well, in the early days it was used in the simplest way. So simple, simple geometries. People did scale models of fires hitting ceilings and they used correlations to explain the results. They didn't necessarily say they were scaling to a large scale, they were just plain. They used it for the parameters. 

Speaker 2: That's why in the nij study, when I was pushing modeling, there was a configuration that the atf had in its lab with all sorts of baffles on the ceiling and smoke detectors Buried into some of these cavities, and I said let's see if we could predict that through scaling. And so I had a student, allison, and Allison made a I don't know one quarter scale model or one seventh scale model, and to put the instruments in she actually had to crawl into the tunnel, put her instruments in. So there's some dedicated students behind some of this and technicians behind some of this. It's not just me pushing something around. And so it worked for this complex geometry. So it shows you can push most accurate scale modeling, which you know. Heat transfer is a factor, but not a big factor, it's mostly the fraud number In scaling smoke flows. 

Speaker 1: Okay, and what about larger fires? So let's move to compartment fires, because at that point I guess the heat transfer becomes rather important in what you're modeling And you suddenly are faced with challenges in how your walls transfer the heat outside of your system And, for example, if you have a concrete building, cannot simply use a concrete model of that building right, because it will not be the same. 

Speaker 2: You put a different material where your heat transfer is now attempted to be scaled properly. So there's two parameters in that dealing with convection and dealing with conduction in the wall, and they're explained in my book Fundamentals Fire, and so it can be done in. Gunnar Hezkastet laid the groundwork for that. That's what he did. So that can be done. But when you have fire growth with real materials, then you have flame, spread, ignition, radiation, burning of the material. Like I said earlier in a podcast, a different one CFD models can't predict the burning rate of a pool fire, whether it's small scale or giant scale, and if they can have someone present it after this podcast. So you have those challenges and so we've pushed that with a real bedroom fire And what we came up with is that make everything geometrically scaled, except when you make the furniture, keep the thickness in small scale the same as large scale, as a thickness of a fabric or whatever. 

Speaker 2: If it's a mattress, try to make the entire mattress the same thickness. Maybe you can't do that perfectly, but you can aim to get that burning time the same And then, if you can get the burning time the same, radiation and other factors start to harmonize a bit not perfectly, but they start to harmonize. In the NIJ report it's discussed And then we tried it And you could judge for yourself how well it works. There's a student, Mark Campbell that he was a student at Cal Poly going for his master's degree but he was so excited about scale modeling He was doing full scale burns in Denver, Colorado, for the fire service and then doing small scale quarter scale burns. And he kept talking to me, calling me up and asking me different questions. And then I met him one time. 

Speaker 2: He gave a presentation for the fire investigators And then he died suddenly. But he did hundreds of experiments. I sent you his paper that never was published And he shows how it worked and how confident they were. And Dr Linwood is now still doing this. Last week he did some experiments for the ATF. When they teach fire investigators, they burn the room He was burning at small scale to show them that you can actually do this and learn from it. So this is propagating out and even if it's not perfect, you could learn from it. You talked about traveling fires. Yes, I saw traveling fires in those experiments And it was a learning experience for me because it said, when the fire moves away from the back wall and it doesn't have enough oxygen, the fire knows it gets oxygen at the vent, So the flame moves to the vent. 

Speaker 1: The flame detaches from the fuel and moves in different places where there's oxygen in this. yes, Yes. 

Speaker 2: And if you have a big space, it will move from fuel load to fuel load, burning each one out according to how much air you have. So that's it. 

Speaker 1: So modeling fire growth in general, the development of a fire in CFD perhaps that would be one of the most challenging things to do The truly model the growth of the fire capture perfectly, the pyrolysis, the heat transfer, all the mechanisms that occur that make the flame. 

Speaker 2: They can't do it and they won't do it. You know why? Yeah, i'll tell you why they won't do it. They're willing to put in a heat transfer coefficient at the wall to match turbulent convection, but they're not willing to use physical modeling for the fine scale. They'll go to a fancy pyrolysis model that has charring and layering and all this other stuff. But if you go to FM Global, and when they model their rack storage of paper rolls, the primary thing is a sheet of paper unravels, and they did that by the physics of it, not by modeling a single paper by CFD unraveling. That's true, and so CFD is going to have to put intimate submodels into their codes that are based on real science, instead of trying to go more complex, beyond rocket science to model all of it. So they're never going to be successful in this unless they back off. 

Speaker 1: Okay, but to what extent you can be successful in modeling that with a scale model? 

Speaker 2: Look at could look at the NIJ paper. You can look in my book Principles of Fire Behavior. that's written for fire service and investigators And there's a whole discussion of it. So the NIJ paper is key. That thing is kind of lost. If people look at that they can see how well it works. 

Speaker 1: I'm going to push you on this. Let's talk. the most favorite thing of all fire model is the wood creeps. Right, can you scale down a wood creep and obtain the same fire, the fires of the same dimensional? Well, what does it take to scale down the creep? 

Speaker 2: Okay, a crib is a very specialized fire. The person that contributed most to this is James Block, one of Emmons' first students in fire, and he used small cribs And he worked off the work of Gross and Robinson from NBS And what he put in was a model for the flow through the crib that the crib actually had a friction effect, okay. So he put that into his analysis And he also showed in his experiments. The flow doesn't come from the size of the crib, it comes in from the bottom And then the sticks burn as their thickness to the minus one half power. That was something that goes back to Gross and other people maybe. 

Speaker 2: Thomas, in describing how wood sticks burn, it's never been. You talk about pushing. It's never been pushed into the pyrolysis people's minds to try to justify this. It's never been pushed into how massive timber might burn in the fire. So the cribs burn this way And based on Block's modeling and Gunner-Heskastad's experiments with this And if you look in my book on fundamentals from our laboratory classes where the students made their own little wood cribs, you can see how they correlate. 

Speaker 2: In other words, there's a thickness effect and there's some other parameters, but you can model the burning rate. You know, a dimensionless burning rate of a wood crib as a function of another dimensionless parameter And it works. It really, really works. But wood cribs, you know, everything doesn't burn like a wood crib. Maybe a piece of furniture might. That's all made out of wood and has drawers and sides and tops. But it's different. Other things burn, like pull fires or flat surfaces or ceiling. So these things have to be accounted for in any kind of scale modeling So wood cribs can scale. Heskastad showed that in his early work That was, his scaling with wood crib fires. 

Speaker 1: I know Sara McAllister. she was an early guest of this show, one of my favorite episodes, And I think no one has burned as many wood cribs as they did. 

Speaker 2: Yes, and we communicated because she had an effect that wasn't being accounted for. So I wrote down some stuff for her and she told me it was during COVID times. She said it correlated all her data because she had, like I don't know, it was like two cribs one on top of another or some strange arrangement. And you know I put down some stuff. You know it's basic problems that you get in the back of a book and just using what's known, and so she was successful. Maybe she would publish it someday but I think she was busy off into New Zealand and other places And I haven't talked to her in years. 

Speaker 1: But she's a good lady One of my favorites in interviews in the show. She wrote Carlos's Pello's book. Yes, that's a good one. So another one in the world of scaling one. I have started the podcast with actually the tunnels. So a lot of new knowledge in tunnels comes from small scale experiments. Actually, in the world of tunneling, this costs quite a problem some years ago because there have been this critical velocity correlations. Yeah for stopping the smoke flow. Exactly exactly. They would go back to Thomas. 

Speaker 2: Yeah, all the way back. Everything goes back to Thomas. People forgot. 

Speaker 1: Yeah fantastic. 

Speaker 2: You should have Thomas on the podcast If. 

Speaker 1: I only could Yes. Well, there was also Kennedy who did a lot of research on critical velocity in tunnels, implemented into NFPA 502. It has been a tool to design tunnels for decades. Now, fast forward some years ahead, probably decade, maybe 20 years ago. A lot of new research comes in this field that shows new correlations, new relations between the science of the fire and the critical velocity that you needed in the tunnel. And these velocities are much higher than what you get from Kennedy's equations, for example. This new science is implemented in 502 and FPA 502. And the community backlashes like no, these values are way too high. We don't see that in our projects that there is justified need for this type of velocities And the science behind it, the small-scale science highly criticized. And actually the FPA committee reverts back this change because it seems that the equations were unjustified. 

Speaker 1: And now my question is what extent you can study this type of phenomena. 

Speaker 2: Okay, you put your finger on a serious problem. The committees of NFPA and other code bodies are composed of business representatives and bureaucrats that have some money to go to these meetings. There's really very little science representation. Kathleen Amen, who headed the NFPA Foundation, where they sponsored research, told me that they had a study done on energetic materials and they had a way of classifying energetic materials, and Elizabeth Buck did a revision using science And it went back to the committee And the committee didn't understand it and rejected it and reverted back, and Kathleen told me that with some chagrin and disappointment. And you're telling me the same thing. 

Speaker 1: Well, actually, in FPA 502 committee. I would say this is a group of really highly competent people, Also in scale modeling, because in the committee you have a professor Ingersen, for example. So I would say, in this case, I think the doubts related to the research that led to the emergence of these new correlations for critical velocity. I think in some way they would be justified, Because much of this research, Walter, you're backpedaling. Now You're backpedaling. 

Speaker 1: I want to understand, like, to what extent use of scale modeling and disregarding, let's say, the heat transfer similarities Because in tunnels you often would replace one wall of your model with glass so you can observe, right, Yeah, so You don't have to, You don't have to, but you often would I wonder to what extent you can capture, okay, the general look of the phenomena. Okay, there's back layering, There are flows different at the ceiling, at the ground, Like you know the general things about the flow, And to what extent you can measure the critical velocity to the third significant digit based simply on that exercise. When you disregard so much of the dimensionless numbers because you are not able to constrain them, I wonder where the boundary lies in here. 

Speaker 2: Okay, you can't ignore heat transfer. You can, because Cox sent me something from a real fire that was in a tunnel in the UK and the steel rail melted. Okay, what that said is that this fire burned for days. Ventilation limited underground, probably a strange, you know. It was going down and then up, and so it had strange vent geometry And as the ground heated up it got saturated, so the heat loss became small and you had almost an adiabatic fire. So that drove the temperatures up to the 1500 degrees C. 

Speaker 2: And in compartment fires, when you have burning going on for two or three hours, attacking the structure, if that envelope gets insulated too because it will after all that time you can have much higher temperatures than people even see in a furnace. So all of that can be reproduced in scale modeling by looking at the physics of the envelope on the wall. So you can do that in tunnels too. You can vary the heat transfer by the wall material and its thickness And you can see to what effect it's having on your flows. And so that's one way of looking at it. Another way is to make sure that the experiments were done right. So if the committee thought there was one sole experiment and it was off the beam, then they need to show why it's wrong. You know, if it's published, they need to show why it's wrong. A lot of stuff gets published and it's wrong, so I don't know the answer in this case, but I wouldn't dismiss scale modeling. 

Speaker 1: That was not the point to dismiss it completely. It's useless. I would rather Like when you are talking about scale modeling, you are projecting confidence in it. You understand the complex physics. You understand there is 20-something dimensionless numbers And each of them has a certain role and a certain meaning and represents a certain physics. What I struggle with observing modern fire science, especially in the field of tunnels, is I see people going into scale modeling, not for it being a great tool to study physics, but for it being a convenient vehicle to publish research, because it's, i mean, you can do a lot of scale model research You can quite cheaply build a tunnel. 

Speaker 2: Yeah, yes, and I know where some of that work comes from, and those groups are still learning. At least they're doing something. Okay, maybe they have to publish, and so that's one of their goals. So they're doing as best they can, but they're learning, and that's all I can say. I mean to me. I think you could scale model tunnel flows with salt water modeling And you could do that very effectively And you could change the inclination of the tunnel and you could change the slope and up and down And you could put barriers in there and cars inside And you would see flows that are very realistic, very realistic. And then run FDS And you'll probably see something similar. 

Speaker 1: We'll definitely go there. I mean, I have a scale model in which I love to study the effects of slope and other aspects on my fires, Though I also have a respect on how much it takes to really get a good result out of that. For the final part, you've worked a lot with scale modeling as a tool in investigative research in reconstructing fires. To what extent the scaling approach is useful in figuring out the course of a fire that happens and understanding origins or scale. You've already mentioned WTC, but maybe there are some other interesting uses of scale research in investigative fire science. 

Speaker 2: Well, WTC is very complex. But in the 90s I was approached by Mike Dillon from California, who was actually not an engineer but an English major. He got his job by writing down his vocation as ENG And then he became an engineer after that, dealing with smoke control in California, and so he got involved in a case where there was a fire in an atrium and it had a smoke control system And the smoke control system brought in fresh air and the vents at the top were supposed to open and exhaust the smoke. So you had a nice layer and you had your atrium design for smoke and people could get out And the layer would be very high, so most of the stuff in the building would be clean. Well, it didn't happen. The smoke got all mixed up, it went through the entire department store, ruined everything on Christmas time and bankrupted the store that was in business for 50 years. And they were looking to see who they could blame And they blamed the people that had the air conditioning vents and they said their dampers didn't close. Well, mike Dillon said the whole smoke control design was wrong because it had the intake. Air was a vertical jet at the floor that sent this momentum jet to the ceiling. It was in the 90s. 

Speaker 2: Fts wasn't invented, then People would see if these were still using Reynolds number types models It wouldn't do the job. So I said, mike, we make a scale model And according to my instructions he had it built. We went out to Montana where the fire was and we were in a big warehouse where actually someone lived. They bought the warehouse and they lived in the warehouse. But they made the scale model and they were helping us conduct the tests And it took a long time for the test to get underway. So we had the jet and we had artificial smoke and we had temperatures And they didn't get underway until it was 3 AM in the morning in March in Montana And they wouldn't do it in the warehouse because people lived there. So they pushed it outside And I was asleep by then on the sofa And then I woke up and went outside and it was freezing And they did the experiments And it showed that everything got mixed up due to that jet. 

Speaker 2: They brought it into court and they cleared the vent. People And the other side said, judge, we accept this, but we don't want that model used by anyone else. Could you just put it on the wraps until our case is closed. And the judge did, and then the case was closed And they said Jim, do you want the model? You could use it at University of Maryland. I said yes, let's see if I can get someone to fund some research on it. I went to NIST, i went to The ashray, the American Society of Idiom Ventilating Engineers. I went to NSF. 

Speaker 2: No one would touch it Because even scientists have some bias on scale models now that proved the court case, so the lawyers didn't dismiss it out of hand in Europe. 

Speaker 1: There was these. Yes, germany, professor Gerard, did you ever meet them? 

Speaker 2: No, not not only that, but you reminded me, vickstrom did did the fire in in Sweden, the nightclub fire, and they did that by scale modeling, okay, and they learned and proved the point from that fire. That was that it's now rise, but whatever the name it was before they did the work there And there was a fire in an upstairs nightclub where people died and Vickstrom was so excited that they could do it He was director at the time and they could do it with scale modeling and learn from it. So, and then you have another example I interrupted you from Germany. 

Speaker 1: Yeah, germany, there was professor Gerhard in Institute in Aachen and he was very famous for helium plume modeling and he's very. 

Speaker 1: He was very influential in the European standardization in smoke control and From him emerged the whole school of thought in using scale models, helium based scale models, in the design of smoke control systems and at some point in Germany There was a very popular thing, even in times when CFD was already available, and popular people will be still building this huge 120 scale models of whole shopping malls. 

Speaker 1: And, coming from this school of thought, perhaps the most stunning Example of how this was used to develop a smoke control system is the Stuttgart Museum of Messages Benz, which is essentially a one giant rum that goes many, many floors above in a spiral or like in a building, like the grouping huh yeah, and they have. The inlets are placed in a way that they create one giant like Whirl in the middle of that atrium and the giant fan that extracts the smoke on the top and It was actually demonstrated in in scale modeling that it works and it was demonstrated in in full Experiments in that building that it works and they actually run it as a show on Evenings. They, they start the smoke control system. They put some smoke and people can observe the tornado. For me, this is how good it is. I will put the link to that to the should get Mercedes museum. 

Speaker 2: No, it, it. It verse the point that helium, though I consider that I think you have to Overdue the amount of helium you put in. It doesn't match the Q star, mm-hmm, but it will reveal similar flows, yes, so of course you lose all the heat transfer stuff because it doesn't lose. 

Speaker 1: Buyers with heat loss. But salt water could do almost the same thing. We with helium. To finish this experiment, i will show you one funny thing that we did with helium. That was quite hilarious. So we've purchased a bottle of helium for some scale experiments. It was like big plastic bottle in which the helium was and it had an electronic counter on top of it Which said 370, and we were wondering like what this counter? like, what is it like? it's definitely not pressure, because this Bottle is definitely not that 400 bars. It's not volume in cubic meters, it's not volume in liters. It didn't fit anything. But when we were releasing it, the the counter would go down 269, 268, and after some point it was like This is balloons, like Hunter tells you how many balloons you have left in your helium bottle. Because that That was what the bottle was meant to to fill balloons at parties. 

Speaker 2: And we're well, that was brilliant. Here's something for you which which can be used as scale modeling. I filled balloons with methane okay and the string that I had on the balloon I soaked in heptane okay and then I ignited the bottom of the string and let the balloon go. This was at a fire training course in Malaysia and it was to illustrate fireballs Okay. 

Speaker 1: Wow, okay, cool, that's a good idea. 

Speaker 2: But you could actually study fireballs that way and at MIT they did very small-scale fireball experiments That hold. For large-scale the equations hold. If you take the fireballs that occurred in the World Trade Center, within 10% you could estimate how much fuel burned outside the World Trade Center buildings based on the size of those fireballs. Fantastic took nests a year to try to model that with FDS and you. 

Speaker 1: Well, we have to wrap up. Yes, we were over the time already, but it was such a nice Conversation. I, i thought like you, really spark confidence. You project confidence in in scale modeling, but but I would still like to say that it's because of the careful understanding of physics and then careful approach to this. Scale modeling is definitely a powerful, beautiful technique that has been in science for much longer than the fire science exists and Fire science found its way to use it. Thank you. 

Speaker 2: Yeah, yeah for doing that. Go back to Phil Thomas and formant Williams. They believed in it, they believe profile and then that's very who other people in fluid mechanics. So come on. 

Speaker 1: We can use it. Fantastic, james. Thank you so much. I hope some young fire scientists here's this conversation and goes. Maybe I could apply scale modeling in my research and maybe they will find out something They didn't even hope for, so that I would be. If you're a scientist and you do that after this episode, let us know. We want to know. 

Speaker 2: Thank, you so much, thank you. Thank you for check. It's such a pleasure talking to you. You make me remember all times and happy to do this. 

Speaker 1: Fantastic. Thank you, and that's it. Thank you so much for listening. I was actually quite surprised with this episode. I was not expecting James to be so optimistic with the scale modeling. I I thought this is gonna be more what not to do with skill modeling rather than how can we praise it. But I actually enjoyed it a lot and I love how professor Quinn theory sparks confidence in skill modeling. He understands it. He understands the role of dimensionless numbers. He understands the complicated physics that goes behind them. He understands the amount of effort you have to put to get the skill modeling and he had the best experiences with it. 

Speaker 1: I mean, come on working with people like Gunnar Haskis, that on this, and participating in the greatest Movement in fire science in us, which you can hear in the other podcast episode with James. I guess the use of scale modeling in their time was really the way to go and I understand why they enjoyed so much. I do some scale modeling in my laboratory as well And I also find it a useful tool. At the same time, i find scale modeling abused in some cases by some laboratories who just use it to mass-produce Papers, because you can put anything in a scale model measure something new that no one has ever measured and just correlated. Put some polynomial Correlation on results and well, i you have a impact factor paper. That that's how much of the science today comes. But yeah, that's why I'm here, that's why fire science exists to give you a filtered and and Catered version of fire science around, so you won't find this type of research here. Anyway, i hope you've enjoyed it. 

Speaker 1: I understand this episode was not the easiest one and there was a lot of very hardcore science involved in it. And yeah, scale modeling is not easy and you have to understand physics, so that's why this episode was full of physics. Regardless, i hope it sparked confidence in you. You find scale modeling an interesting technique And if you are a researcher, give it a try. Find out if you can in the identifying dimensionless correlations for your problems. Even that itself on its own will be a very nice exercise to do and to develop your your skills to the understanding of the fire phenomena That you are studying. So thank you very much for being here with me this Wednesday. I hope you have a great day and See you here again next Wednesday. Thank you, this was the fire science show. Thank you for listening and see you soon.