238 - Fire Fundamentals pt. 19 - Defining fires in your models

Welcome to another fire fundamentals episode! Today we dig into how to place a fire in a model so results reflect real physics. From plume inputs to FDS burners, we show where HRRPUA, radiative fraction, and D* make or break smoke your calculations. Things considered in this episode:

• why defining the design HRR is separate from placing the source
• what a flame is and why we cannot resolve its chemistry
• plume models compared by inputs: perimeter, Q, Qc
• entrainment, virtual origin, and effective diameter
• realistic HRRPUA ranges for building-scale fires
• radiative vs convective fractions and why they matter
• zone model linkage to plumes for smoke control
• volumetric smoke and heat sources for CFD: volume, placement, and limits
• fuel-based fires in CFD and oxygen constraints
• growth modeling via area expansion vs flux ramping
• soot yields, heat of combustion, and visibility
• D* and meshing guidance for credible resolution
• why predictive fire spread modelling for design use does not really exist...

Resources, resources!

• G. Vigne et al. "Review and Validation of the current Smoke Plume Entrainment Models for Large-Volume Buildings"
• W. Węgrzyński & M. Konecki "Influence of the fire location and the size of a compartment on the heat and smoke flow out of the compartment" - (this is a paper from my PhD where I explain the concept of volumetric heat source)
• M. Bonner et al. "Visual Fire Power: An Algorithm for Measuring Heat Release Rate of Visible Flames in Camera Footage, with Applications in Facade Fire Experiments"
• Episode 100 - Smoke plumes! That was a fun one.
• G. Heskestad "Fire Plumes, Flame Height, and Air Entrainment" from SFPE Handbook  (also the source of the overlayed image on the cover showing range at which fires exist)

----
The Fire Science Show is produced by the Fire Science Media in collaboration with OFR Consultants. Thank you to the podcast sponsor for their continuous support towards our mission.

Transcript

Why Fire Placement Matters

Wojciech Wegrzynski 0:00

Hello everybody, welcome to the Fire Science Show. In today's episode, I wanted to take you somewhere practical, uh, something that is highly relevant to my everyday fire safety engineering. Fire safety engineering relies highly on modeling. Uh, we're modeling fires, so in those fire models you have to put in some sort of fire, right? And uh we all do it like every day. We basically model fires, and I'm not sure if everyone reflects on what actually happens when you place a fire in the model, or perhaps rather, how do you correctly place a fire in the model? It's not a very trivial question, actually, and I've probably rejected maybe even too many papers, scientific research papers. Some of them were quite good, to be honest. Papers that shown some physics, some modeling, and in the end, I've noticed that the person done really big mistakes in how they placed fire in the model. And that's enough to reject the paper because if your fire model, if you put something that is not really a fire, it makes no sense. And the same in your everyday engineering calculations, in your everyday engineering CFD, in your everyday engineering practice. I'm pretty sure that if you're trained, fire professional, you do this correctly because we are learned to do it correctly. You learn how to do it correctly in your school, in your university, you learn it from online courses, etc. Uh so I don't doubt that you are capable of doing it correctly. What's interesting is uh is maybe to discover why this particularly is a correct way of doing some things. So in this podcast episode, I wanted to share about what does it mean to model fires or to place fire within a fire model. And we will discuss a lot of different approaches from plume models to CFD. This is the episode where we discuss the HRR per unit areas, convective heat release rates, hits of combustion, fire perimeters, fire areas, etc. This this is an episode where we will cover that, and uh to my surprise, I was wondering if I can uh make a whole podcast episode about this. It seems kind of trivial, and as usual, I've rambled so long I've run out of time. So, yes, there is enough content to fill a whole podcast episode, and I really, really hope it's a good podcast episode, interesting to you, allowing you to reflect on your fire engineering practice. And at the end of the episode, I will tell you why predictive fire modeling that uh predicts the growth of the fire for fire safety engineering applications does not exist, which links to everything I have to say earlier in the podcast episode. If you're interested, stay with me. Let's spin the intro and jump into the episode. Welcome to the Fires Show. My name is Wojciech Wegrzynski, and I will be your host. We've been on this journey together for three years so far, and here it begins the fourth year of collaboration between the Fire Science Show and the OFR. So far, we've brought you more than 150 episodes, which translate into nearly 150 hours of educational content available, free, accessible all over the planet without any payrolls, advertisement, or hidden agendas. This makes me very proud and I am super thankful to OFR for this long-lasting partnership. I'm extremely happy that we've just started the year 4, and I hope there will be many years after that to come. So, big thanks OFR for your support to the fire science show and the support to the fire safety community at large that we can deliver together. And for you, the listener, if you would like to learn more or perhaps even become a part of OFR, they always have opportunities awaiting. Check their website at OFRconsultants.com. And now let's head back to the episode. So you're a fire engineer brave enough to try and answer some important fire engineering questions that your stakeholder asked. How much smoke control do I need? What's the temperature of my structure in the fire? How bad this fire can be, etc. Can people escape my building? We usually resort to modeling to answer those fires due to the inherent complexity of the fires themselves. Also, for the purpose of this broadcast, I assume you've already got the toughest number to get, which is the size of your fire. So you are perfectly aware of how many megawatts of fire will be emitted to the space that you are considering. I mean, getting to that point, getting that design fire, it's a struggle on its own and absolutely deserves not only one but more podcast episodes on their own. And uh this is something we will talk in the future of the Fire Science Show. But today we assume that you are lucky, you have the number, and all is left is to model that. Now, uh why I'm recording a podcast episode, why is this something to be discussed? Because this is not particularly straightforward, or it is straightforward uh for many people because of the way how they are used to do those things, or you know, because of the general approach demonstrated by their colleagues. There are just ways we do things which already went through this logic that I'm gonna present and then provided us with some general guidance and we just do them. But we very rarely think about why we are doing them. Anyway, before we go into details of how the fires are represented in your CFD software, etc. Uh let's perhaps talk about the fire itself or the flames themselves first. So what really is a flame? I think that this is a very interesting question from the perspective of fire safety engineer. What is a fire? What's a flame? So I like to uh ask this or put a perspective on this question by asking what's inside the fire? That's something that when I uh teach people, when I train people uh on on design fires, etc. That's one of the questions that I ask. What is inside the fire? Because you know, we are used uh to this uh image of fire that's in our heads. Perhaps you're now imaging a campfire, perhaps you're imaging a pool fire, perhaps you're imaging a fully fledged compartment fire post-flashover venting into the facade. No matter what you do, there obviously is a physical thing in that, the the fire it exists, you can see, you can see the shape, the size, the scale of it. But inside of it, it's actually hollow, it's empty. Well, it's not empty, it's filled with fuel. Because the flame is basically a region in space where chemical reactions can occur to create energy out of reacting the fuel and the oxidant, which most commonly would be your air. So technically, the flame itself is just the space in which the conditions for those combustion reactions that release energy are pretty much perfect. This is the region where you have more or less stoichiometric mixture of your oxidizer and your fuel. This is the region where the energy in that volume is more than sufficient for the reactions to happen. And this is the region where all the burning processes occur. It's not one reaction, it's not two reactions. It's kind of interesting when you go deep into chemistry of uh combustion because it's hundreds and hundreds of concurrent chemical reactions happening pretty much at the same time in pretty much the same space. And what's interesting, uh which reaction takes place, it's it's kind of a probabilistic distribution. Each reaction depending on the oxygen concentration and the concentration of ingredients and the energy or perhaps the temperature at which uh it occurs has its own probability and they kind of compete with each other. But eventually the fuel in the perfect conditions will burn off completely, it will and all of this reaction and would end will turn into the products. And actually this happens at an extremely narrow space. Um why it's so narrow? Because for those reactions to happen, you have to have a mixture of this oxidizer and the fuel. So there are mixing processes that are happening at a fairly small scale. They're driven by by turbulence, by flow, but also mainly by molecular diffusivity. And depending on the diffusivity and the reaction rate, you will receive a product of that, which basically means this region of space is the region in which the fuel mixes with your oxidizer, and the time that is available in that mixing is also at the similar time scale as it takes for them to burn off. So you end up with with a zone. Why I'm bringing that? Because the zone is tiny. That zone, that that flame sheet would be less than millimeter thick. It would be less than 0.1 millimeter thick in many cases. It's extremely thin and condensed zone of energy which is wrinkled in the space, it exists within a turbulent motion. There are enormous gradients of temperatures existing that cause a very interesting buoyant flow phenomena, turbulent mixing. This is constantly moving, the influx of the fuel, the influx of the oxidizer continuously exists. So within a space the old fuel is burned down and replaced by new fuel and the mixing continues. It's something that happens at a very small scale, and there is an insane abundance of interesting physics that dictate how well those reactants react together and how well in general the thing burns. Now do we model that? No, we we we compare this is the this is the fun part. This if you think about the flame or fire from the perspective of combustion of what it physically is, if you consider it as a combustion reaction between the fuel and oxygen, it it's not something we model in our fire safety engineering. Like there exists no single fire safety engineering tool or model other than doing your full scale fire experiments in which this richness of physics would be captured at the scale at which it exists. It's simply not possible or not feasible. Perhaps you could model a burner or a pool fire with this level of of insight, with this level of capturing the chemistry and turbulence and flows and etc. Even for just modeling, you know, a tray of fuel burning, the investment in modeling that would be insane. Like it it would this simulation would be more complex than any commercial CFD I've ever delivered for any commercial project. So I think we can agree that while we appreciate and understand how complex those phenomena are, we are in a position where we are unable to fully tap into that. We we have to model that, we have to simplify it for our practical use. And here we come to modeling, here we come to fire models, here we come to fire safety engineering, our everyday job. Because we are burdened with answering some very profound questions in our buildings. How much smoke will be produced? How toxic will it be? Will people be able to escape? What's the temperature, etc.? We need to be able to capture those phenomena at a level which makes sense for the building. Now, I also go here because the modeling and the way how you define things in a model for me should be very strictly connected with what the output of the model, what the expected output of the model is. Because it's a completely different story when you are interested in general smoke layer height versus when you're interested in perhaps spread of a fire over um solid surface, or maybe you're interested in time to flash over. That would be very challenging to model. And those models would require you to approach with uh with different tools, and those tools, I'm I'm a huge fan of this term that was uh I think coined in Canterbury, the consistent level of crudeness. I mean, if your output is very crude, the models should not be extremely detailed because you gain nothing in this in those details as you lose those details as you get into crude outputs. And uh this consistent level of crudeness makes a lot of sense and it's it's a broad engineering guidance that I follow and really like in my engineering. And in in fire safe engineering, when you consider fires in your buildings, we have lots of models at different levels of crudeness, starting with zone models, starting with uh your plume models, going into zone models, then venturing to CFD at different levels of complexity. And this is the journey that I want to take you in in this fire podcast episode about how we define fires at those different levels of complexities and what factors uh come into play. So we'll start with the plume models and build up the complexity. So for plume models, I can start in a very safe and secure way by referring you to a paper, uh review and validation of the current smoke plume entrainment models for large volume buildings. It's linked in the show notes. It was an interesting work done by a Spanish team, crazy Spanish team, who invited me to participate uh in this with them. Uh it was Gabriela, Alexis, Candido, Guillermo. These guys have done some of the largest fire plume experiments in in the world. So it was it was a huge joy to to collaborate with them. Anyway, in this paper, we went into the fire plume models that exist in the literature or that are used by fire professionals, and we tried to look into the physics of how they are built and then compare them with some large-scale experiments on how they how well they capture uh fire behavior in a very large scale. And the plumes are usually named after their creators, so we have the Thomas Plume, we have Zukosky Plume, we have McCaffrey Plume, we have Haskestad Plume. There are more, there's uh Delicatius Plume, there's uh original Morton Plume, the first one I believe. Uh there possibly are more plume models than that, but those four I've I've named before are the core ones used in fire safety engineering. Plume models are quite interesting because they take as an input the fire, the size of the fire that you have in your building, and give you as an output some of those variables that I've mentioned that are important for us. So they would give you the mass flow at a certain height of the building. This is the main characteristic that you get out of the plume model. You get the mass flow. If you get the mass flow and you know how much energy you have released and you have a way to approximate the losses of heat into heat sinks and radiation, you have the mass, you have the energy, you can calculate the temperature. So basically, out of this single calculation of the of the plume model, you get some very important data that can guide your decision making regarding the smoke control dimensioning or in general regarding what kind of temperatures you could expect in the space that uh you are considering. They take the fire as an input, but they are not unified in the way how they do that. And this this is already quite interesting because there are actually three ways the fires across four four models, there are three different ways the fire is input there, and it already reveals a little bit of interesting physics that we have when we try to simplify fires to include them in our models. So you can input the fire uh through a property called the perimeter of the fire, which is used in the Thomas model. Actually, it's Hinckley's model uh that was derived based on the Thomas model, but we usually refer to that as the Thomas model, and it's very common in uh many standards, especially in UK and Europe. So you would put the fire perimeter. You have models like Zukoski's plume, where you would put the heat release rate Q in the in the model. And you you have uh models like Haskestuds where uh for some of the parameters you would put the convective heat release rate. So three different ways to define fires, and there are some some interesting things about that. So in terms of of those plume models, what you want to get in the end is the mass flow in your smoke plume. I had a smoke plume episode in the podcast as well, it's also should be linked in the show notes. To get that mass flow, you need to know how much it changes with height. So basically, what is the entrainment rate? And the entrainment is a process that happens around the plume. It doesn't happen in them in the core of the plume because there's nothing the smoke can mix with if it's surrounded by by other smoke. The entrainment happens where the smoke meets your surrounding air. So basically around the plume, at the perimeter of the plume, this is where the mixing happens. And this this physical dimension, the size at which this happens is quite important. You can derive that by calculating the virtual origin of the plume. So this is something used by some of those models, where you would assume that the plume is kind of like a reverse triangle, uh standing on its narrowest peak and it just grows in indefinitely uh up, and the the tip of the triangle will usually end up underneath your ground because the fire is larger than a point, so you it has some dimension. So you calculate that number, and from that it goes into the entrainment equation, and from that basically the model figures out the physical dimensions of the fire. So in the end, you supply the heat release rate to the model. Or like in Thomas' equation, you basically supply that. You you give the value of the perimeter, how big fire physically is, and that allows you to approximate how much smoke will be produced. And those numbers, um when you go into when you use uh the the models which use the virtual origin, there's not that much you can mess up with the definition. Because basically you know your number of of Q, you know your heat release rate, uh you calculate the virtual origin. So you know where it exists and the model will not allow you to mess it up. When you go into the Thomas model, it's different because you have to define perimeter, and perimeter is obviously related to the space in which the fire exists, so so to the area of the fire, if if you consider the fire just as a surface. And when you go into that, there are things that can be broken in here. Uh because fires are phenomena that exists in very specific range of physical conditions. And this applies also to the energy density per unit area. We define this parameter in in modeling as heat release rate per unit area HRRPUA. You've probably seen that parameter a lot. And this is a very critical parameter. If this value is too low, then basically you don't have a fire, you don't have those continuous flames existing within the space because the energy density is too low. You basically have some sort of heater kind of. If you have if this value is too high, then the density of energy contained in a space is way too high for those diffusion processes to happen in a time frame that would allow a fire to exist at At least in the shape that we we know and we expect. So for those extremely high energy densities, you got you would go more into phenomena like pre-mixed flames or maybe m maybe some sort of jet fires, you know. The fires that are not the types of fires that you would normally be modeling as a part of fire safety engineering tasks. So there exist a range of values of this hit release rate per unit area in which it kind of makes sense. The fires exist in that space. Actually there is a beautiful classical plot in Gunnar Heskestad's chapter of the SFP handbook where he shows different typologies of flames uh related to their dimensionless uh fruit number Q star and the ratio between the height and the diameter of the flames. And you can see they all kind of fit or they are plotted in a quite narrow space in which they can exist. I I highly recommend reading through the classical material, and I definitely would consider Gunnar's uh chapter in SFP handbook as a classical material on this. So for the engineering purposes, let's let's discuss what would be reasonable value for the heat release rate per unit area that kind of makes sense. And there's no ultimate answer because this value will one change with your fuels and conditions at which it burns. And two, if you have anything that's more complex than a flat surface, then this value is just a statistical approximation because you have some energy emitted in a space and it's just divided by the space that you have, and you're not accounting for the true surface area of your fuel, which is an important consideration actually. So what what are reasonable values? Um so when we were doing that plume paper, we'd done some calculations using correlating those different plume models. And when we took some assumptions of the Thomas Plume model and calculated the sizes of the flames using those assumptions of Heskestat with the virtual origins, etc., we we came out to the value that the minimum point at which the Thomas model in general is physically makes sense or is valid would be around 375 kilowatts per square meter. This is where it begins being reasonable and uh in technically it it should be much more than that. Uh some researchers have measured H3 straight per unit area in some of those experiments. There were experiments done in which those values would be very low, like 220-ish kilowatts per square meter. But uh I would say for engineering considerations, the fuels that we expect in buildings, those values around 300-200 kilowatts per square meter are extremely, extremely low. When we were doing the comparison between the predictions of the Thomas Plume and our full-scale fire scenario in uh in a very large H room and we were comparing it, how the layer descends in time and how well the steady state is captured by the model. We found that the best agreement was slightly above 500 kilowatts per square meter. That was where we noted that the agreement is the best. There's also a number that is often uh brought up, which is 625 kilowatts per square meter, and I think it it comes from some of the UK standards where this is a value recommended for some of the fires, and I would agree this is quite a reasonable value to be taken uh for a fire. On the lower end, but but quite reasonable. Um a value that I usually try to not exceed would be around uh 1000 kilowatts per square meter, one megawatt per square meter. This is the upper range at which the models, uh the plume models kind of make sense, and this corresponds to to quite a severe fire, to be honest. One uh megawatt per square meter is quite a huge energy generation to give you some reference point if you burn uh heptane. I know I'm not sure if you ever burn burnt heptane, but heptane is one of the most viciously burning fuels that I've seen in my life. It like produces those extremely tall, very narrow uh fire plumes. Heptane is crazy. And uh and heptane gives you the energy density of maybe two megawatts per square meter, maybe a little more, depends on how big the pool is. So I struggle to imagine what would be the fire like if this wall value was higher, if this value could exist in the higher, like five megawatts per square meter. What's interesting is that often uh people uh do mistakes with that. People sometimes would use very large number of this heat release rate per unit area. They would just do one square meter fire source and then just emit whatever amount of fuel they have from that square meter reaching heat release rate per unit area values of I don't know, five megawatts. I've seen that in in scientific papers, and those scientific papers unfortunately did not meet their publication because of that. It's simply unphysical. So you have the perimeter, basically, if you know the heat release rate per unit area, if you need if you know the uh heat release rate that you're emitting, you have the fire area, you have the fire area, you can calculate the diameter of a circle and you can calculate the perimeter of the circle, which has this area, and this is something you can put into that equation. I also think that in in the British nomenclature, this is standardized, so there are some values to be placed in that model. When you have different bloom models that take heat release rate as an input, it's much simpler because as I said, you have to calculate some parameters based on that heat release rate, so therefore whatever you calculate will correspond to that heat release rate, there's one caveat which is sometimes you put the heat release rate, sometimes you put the convective heat release rate part into the equation. The reason you would like to do that is that heat release rate technically is the entirety of the chemical energy released by the flame, by the reactions happening in the flame, while the convective part is basically the part that goes into heating up the air and causing the upward motion of the smoke. What's not in the convective part is the radiative part, which is what the fire radiates away from itself. Fire is luminous, it's high temperature, there's soot in the flame that creates the very strong emissivity of the flame surface. It radiates heat away, and uh a big chunk of energy of the fire is actually radiated away. It's uh something we call the radiative fraction. It also sometimes is used in CFD. You may not even know that you're defining it, but it's an important parameter to tell how much of the heat is immediately radiated by the fire, by the flames, and how much of the heat goes along with the smoke, creating buoyancy effects, promoting entrainment, mixing, and in the end, eventually influencing how much smoke you have in your room and uh how big temperature that uh smoke is at. The radiant fraction, um a common value used is between 0.3 and 0.35, uh, dependent if you're using zone models, sometimes for for dependent if you're using plume models or going into advanced modeling. For plume models, a typical value would be 0.3, so convective is 0.7 of the of the total. In fire modeling, I believe the default would be 0.35, but you can correct me on that. It kind of stabilizes for large fires, it's it's slightly different for uh small fires, very small fires, which were where it would be important. So those would be the properties, the heat release rate, convective heat release rate, fire perimeter, in a way also the diameter of the fire, the location of the virtual origin, the triangle rule. Those will be the properties that you define when you model a fire with a simple plume model. If you go to zone model, well, it's kind of the same, because the zone models just assume there are two zones in the building, the hot gas layer and the cold uh underneath it. And the fire is usually defined by a plume model in the zone model. So the plume model is the engine that uh moves the gases from the lower region to upper region of your room. It's the engine that creates energy and the mass and it goes into all the equations. So if you know, if you're comfortable with plume models, uh you should also be comfortable with zone models. It's essentially the same math. Now, as much as I like zone models, and that was one of the first uh episodes I've ever recorded in the podcast with uh Colleen Wade about zone modeling not being dead yet, and I still believe zone modeling is not dead yet. The expectations of our peers, the expectations of other stakeholders of our clients is that we deliver CFD. So now let's move into how we define those in CFD. And is it more complex or not? That's a stupid way to put it. It's obviously more complex, that's what the clients are paying for, right? More complexity in the colorful pictures we sell them. Yeah. Jokes aside, uh, you're a good fire engineer, you spent 30 minutes with me learning about fire physics, so you you definitely want to do this well. Uh let's let's talk about how we do it and how to do it well. Uh we have multiple ways we can introduce fires into CFD models, and those ways are also kind of related to the type of software that you are using. Because if you are using FDS, you probably already are committed to a specific path of entering the fire into your domain, which is not the only path. There are there exist different pathways to introduce fires into CFD models. So generally, if you look back in time and if you do look at softwares beyond FDS, there would be two prevalent ways of introducing fires into models. And to discuss them, again, let's bring up why we do this. We do this so we know how much smoke is produced, what's the temperatures, can people evacuate, etc. So there are two prevalent ways. One would be creating a region in your model in which the products of fire are released. So basically you are defining a volume or a surface, but usually that would be a volume. You define a volume of your air in your model, a box, in which you basically release the heat, you release the smoke, you just introduce them as sources to your CFD model, and they just release in there, and that's it. And we call that volumetric uh heat and smoke source model. There's different ways people call them, but in in general it's a volumetric source model. And the other way would be you introduce into your model fuel, and again you release it from a surface, you release it from a volume, but basically you are introducing a fuel into your computational domain, you introduce an ignition condition. So either you assume that the fuel is capable of burning or you provide some kind of uh ignite or spark or some other source of ignition. And in the regions where the chemistry allows for combustion, so there is fuel and oxygen mixed in a ratio that promotes combustion, the combustion occurs. So basically you are releasing fuel and the fire happens or the flames happen, not necessarily exactly at the location of that release, but where the conditions of mixing allow that fuel and oxidizer to mix and burn. Now, why would anyone actually use the volumetric heat source when obviously the more correct way, in a way or more natural way, would be by defining the fuel and modeling the combustion itself. It goes back to the consistent level of crudeness. I mean, modeling combustion is very difficult, modeling the diffusion, turbulence mixing at the scales at which it makes sense is extremely challenging. And you cannot even pretend that you're doing that when you're doing your uh rough engineering simulations. You're not doing that, you're just applying a model. You are not capturing those in implicit little phenomena that drive the combustion processes. So neither of those models actually captures that. And basically going through the consistent level of crudeness. What you care about is the outcomes. How much heat is released, how much smoke is released. And by the way, those outcomes are not an outcomes of the model prediction, usually in engineering, they are outcomes of your engineering assumptions. You come to the analysis with an assumption of what those values will be because you know what is your heat release rate of your fire, you know what is the smoke production of your fire. So you technically don't really need to go through all of those steps involving combustion, uh diffusivity, turbulent mixing to come up with the number. You already know it. Therefore, you could define just the outcomes in your model. You define them so you can just put the number in there. You can say, okay, my farm is five megawatts and and that's it. And I just release five megawatts into my volume and I'm done. And this is very convenient in many ways because it one, it speeds up the analysis. Uh, two, it it goes along with the consistent level of crudeness. And also it's a way that this has been done years, years ago when the computational capacities were not at the modern levels where we are today. Um today, probably the more correct way would be to model combustion either with some flameled models or with models that are implemented in in FDS as defaults. So I agree that in the modern world, perhaps defining volumetric heat sources is a little bit antique. But again, if you're doing rough engineering calculations and you have to deliver them at volume, perhaps this advantage of speed that you gain with this source perhaps is something that is important. The challenge with the volumetric hidden source model is that you have to define the volume at which the burning occurs. Therefore, you kind of define the density of energy in that volume. I hope you follow me. If you have a box which is one cubic meter sized and you release one megawatt of energy in that space, you end up with one megawatt per cubic meter of energy, and if you divide it by the area of the surface, you end up with one megawatt per square meter underneath the source, which we could all consider something that resembles a fire or flame. If you go with crazy values, like five megawatts released in a cubic meter of space, well you end up with extremely high temperatures, you end up with buoyancy that doesn't make sense. It's also hard to capture radiation from that source, so that's another thing. So this definitely makes no sense. If you put 100 kilowatts in a cubic meter, you end up with 100 kilowatts per cube, and that doesn't make sense either. It's not a fire anymore. So the engineer who is driving those calculations, it's their responsibility to figure out the size of that volume so it corresponds with the space in the building in which the fire takes place. It it's located where you want it to be located. And you have to be aware that the density that you supply makes a big difference in the outcomes. So this value has to be corresponding to something reasonable. It cannot be made up. There's actually a value for that. It's quite interesting. Uh we've written about that with uh Matt Bonner and Guillermo. I mean, all the credit goes to Matt and Guillermo. They did the most of the work in here, and Guillermo found this all of the risk constant. So all the credit goes to them. I just supplied them with my experiments where we investigated uh measuring the heat release rates through cameras. We have a paper visual firepower, it's it's quite interesting. Anyway, there is a relationship that correlates linearly the the volume of the turbulent flame and the heat release rate. And that value lands at approximately 1.5 megawatt per cubic meter with plus minus 183 uh kilowatts. And this is uh this is an all-of-the-risk uh value that comes from a classical paper from the 80s where they've shown that for a large range of fires, if you take the heat release rate measured in an experiment and you divide it by the flame volume, you will eventually get a value that is uh close to that, slightly higher than the value of one megawatt per cubic meter, but still within a reasonable range of uh of fires, not five, not 0.1. So if you need a citable number, such a number would exist. Regarding volumetric sources, uh one more interesting piper that paper that exists there is a paper from um Gratz conference, I think that was 21 or 22, by Conversation Michael Bayer, my favorite tunnel uh people. And they have uh shown they were comparing different ways of modeling this with uh within Ansys Fluent, and they've shown basically some limitations of the volumetric heat source model in that paper. It was based on memorial tunnel experiments, so up to 100 megawatt fires, quite an interesting data point. I I would highly recommend looking at that paper if you're modeling with answers because that that that's probably an important uh piece of knowledge for you. Anyway, uh one final uh challenge with the volumetric source is there's also a big importance where you place it and what's at the bottom of it. So you could technically put the cube or the the shape of your flame, the volume, on the floor, where basically all the entrainment will happen from the sides and not the top of the of the volume, or you can elevate it, which creates a completely different dynamics because suddenly the cold air may penetrate your plume from underneath the plume and heat up because there that's where you define your release of energy. And in this case, you end up with some really, really ridiculous fires, and you have to be very well aware that it like does the flow phenomena look like that in your real-world project? It's important to understand that because if you messed that up, you end up with something that definitely is not a correct fire. And for your FDS calculations, it's simpler because here you release the fuel, the fuel enters your domain and burns where the chemistry permits. So in this case, what you are looking at is basically the density of energy that you supply to your burner surface. In FDS, the fuel will burn. If you set up the simulation correctly, it will ignite it, it will burn. You implement a burner surface, which basically emits that influx of fuel. And here a property that you put in is the heat release rate per unit area again. So all the way back to the um calc to the discussions we had about the plumes, this is also highly relevant. You need the heat release rate per unit area to be supplied to FDS. Again, you supply this value to low, you're gonna have a tiny flame which is not a continuous flame as we see them in fires. You put this value to high, you're gonna end up with some sort of extreme jet fire or an unphysical fire, really. So this value has to be has to make physical sense. One interesting thing about FDS and modeling those fires, and also I guess that goes back to all CFD models, is that the hit release rate value is usually not constant across time. Because we all also model growing fires, that's an important part of our engineering calculations. When you are modeling the growing fires, the size of the source that you're releasing that fire from is usually constant because you predefine that before you start your analysis. So let's assume you have a fire that will reach 4 megawatts, you have four square meters of your fire source, the fire starts in the middle, and now you want it to grow over time from 0 to 4 megawatts. If you have two ways you can define it. You can define it such that the heat release rate per unit area increases on the surface until the value that gives you the final heat release rate is reached. So, in this way, if it's 4 megawatts out of 4 square meters, you're growing up to 1000 kilowatt per square meter. You start, let's say, with one kilowatt, two, three, five, ten hundred. And as this value grows, the size of the fire grows. Which means that for a very long part of this process, the fire will be smaller than a fire. You follow me? It's it's the heat release rate will be too small to call it a fire. And it actually could be the whole length of your evacuation, for example. This doesn't make sense. This does this does not make physical sense. You've made a calculation, CFD calculation to prove that people can evacuate in the fire, and in the end, you were not modeling a fire at all. So this is something to be aware of. There's another way where you increase the size of the of the source and have a static uh steady state value of the heat release rate per unit area or per unit volume. And in that case, basically you say, okay, my fire has a density of 1000 kilowatts per square meter, and I want it to grow to 4 megawatts, so it's growing from let's say one single cell 20 by 20, 10 by 10 centimeters, up to the whole surface of the of the of the burner. Uh and it at every point of time I have 1000 kilowatts per square meter. I I think this is a much, much better definition of the fire growth in in modeling, and this is something that FDS uh uh allows you to model. In ansys, it's more complicated. You need to have some crazy UDFs to account for that. So we have to deal with much more complicated sources of fires if we want to do this uh properly. Uh one more thing, if you emit fuel into your space, then you can capture the effects of oxygen. If you don't have oxygen, it's not gonna burn. It's an important distinction between the uh the volumetric heat source models and the and the fire models, because the volumetric model, if I release 10 megawatts, I don't care about oxygen. I cannot even measure how much oxygen I've lost. Uh in modeling with uh with uh releasing the fuel and modeling the burning, I can get that. I if there's not enough oxygen, the fire will not burn, and the megawatts, despite the fact that I've released my fuel, will not be released in space. This is very important. And this this relationship is kind of important when you try model things like suppression, actually. Because uh you also then need to capture some actions of, let's say, water cooling on the stuff that's happening within your uh cells. So having a model that actually accounts for combustion allows you to do a lot more things than one in which you only have a volume in which the heat is released, no matter how convenient or easy to set up that that previous one is. Um, from the things that the designer also defines is the suit yields, is the heat of combustion, do not necessarily translate into the size of the fire in terms of its geometric size. They are the products of the comical reactions, but they will heavily influence how much suit you have in your model, which will influence the visibility in smoke, which will influence the uh amount of uh tenability you have in your room. You can put it like that. Um another thing that correlates to the size of the of the fire, the physical dimension of the fire, is something that a lot of engineers are doing, which is the calculation of this star. This star is you could consider it like as a dimensionless size of the fire, it's uh the root of the food number. It gives you a value which pretty much gives you the the dimension of the fire. And there is a very well-known relation which can be traced to some Nurek guidelines which correlate the size of the D star to the size of the cell element. So there's a famous relationship that the D star divided by the size of your element shall be between four and sixteen. That actually comes from the Nurek study, where they were testing different relationships and they found feasible outcomes at this kind of range. It's not universally true, it's not true for every fire, for every source of the fire, but it's an engineering like guidance that can guide you to getting a correct mesh. Keep in mind that this star, if you have a growing fire, is also changing with time. So a D star for your 10 megawatt fire will be very different than the D-star you would have for a 500 kilowatt fire when it's growing. And in the growing phase, perhaps that is the important one for your evacuation. So uh calculating this number once and saying okay, 20 centimeter mesh is fine for my project. May not cut it when you have a growing fire. You have to consider that at least. And in general, uh because you can never get with the mesh to the perfect size, because that would be like under millimeter size, where you start to capture this complex physics. Uh mesh is always a simplification, and FTS doesn't really converge that easily with the mesh size. The changes happen when you start to capture different fire phenomena. There's a whole episode on Turbulence with Randy where we captured that. I think I had an episode with Jason Floyd where we discussed this a little bit more. But this is also something important in the definition of your fire model in your numerical software that should be considered. Wow, boy, I was wondering before pressing record if I'm capable about rambling on fire dimensions for an hour, but here we are. I'm running out of time, so one last thing that I think is quite important is the definition of a growing fires, but in a way that the growth of the fire kind of simulates the fire growth. The ultimate goal of modeling, we don't define fire in megawatts. We receive the fire in megawatts as an outcome of our analysis. We provide ignition, we provide fuel, and it grows. And uh this type of fire modeling, I still say this is something that is not possible to do yet. No, it is not. Like there were limited successes in the world of academia with modeling wood creeps. Yes, some people done papers showing wood creeps spreading fires, and those results would be in line with experimental. This take an extreme amount of effort. This required coming up with very clever submodels or simplifications that allow for this type of work. And I am not sure how much those are generalizable. Like if you take that model and put it in a shopping mall, will it work or not? I'm not sure. The challenges, why it is so damn hard. Like, okay, you can put a box in FDS, you can say the box has one megawatt heat release rate and the temperature of ignition of the box is 450 degrees, then box ignites, bam, you have another megawatt in your fire, the fire has spread. Yes, you can mimic that behavior, you can mimic this, but it's not a predictive spread of fire simulation. Because if you consider solid fuels, what happens is that the fuel is emitted in the process of pyrolysis. And the process of pyrolysis is super tightly correlated with the radiative heat flux coming to the surface. And at the length scales that we consider in the CFD of 10, 20, 30 centimeters of your meshes, you are unable to capture the phenomena at the level of detail at which there exists. This is what we started the episode with. You cannot capture those phenomena as they exist in reality, and in the end, you come up with some very crude approximation of how would pyrolysis look across a 20 centimeter sample, which is very, very, very crude for this for this goal. And what's even worse is that if you just have you know a flat surface and you have you know that the fire spreads, let's say uh one meter per minute, let's say you have a velocity of the fire front. That's easy. That that's not a problem. But if you have a compartment and there's a tipping point, a flashover to happen, if there's you know a point in time where exponentially the the feedbacks will increase and the growth would exponentially grow in a very short period of time, you're not able to capture this. There's no way you're not able to capture this phenomenon in CFD with the current models and the ways we do it them, period. You're not. If you tell me you are, I invite you to the podcast explain to the world how you are doing that, and I'm quite sure you're not doing that. Because the some of the best people in the academia that I know who are working on this, they face struggles. This is very common. If you go into this at the beginning, it looks easy. Oh, yeah, temperature recognition, it just releases heat. Let's go. But then you start to do pyrolysis properly, you see that inverse modeling of the cone results into CFD is very hard. You start seeing that those things are very transient, they are very heat flux related, the spatial and time constraints that you do are too coarse to solve them. Like this is a common pathway that you discover this, and in the end, it appears to be extremely, extremely difficult to model fire spread. So as a fire engineer, if you go into the problem and you tell me you model the growth of the fire, the spread of the fire, and you actually have not defined the kilowatts or megawatts as your assumption, uh maybe it's an interesting exercise, but I would not say it's a predictive fire modeling. So I think I ran out of time. Let me wrap it up. There was a lot of interesting things I hope for you on this very simple topic. You have your megawatts and you put them into your model. We've gone through some properties of fires that are modeled, the dimension, the physical dimensions, be it the uh area, be it the perimeter beat, the heat release rate per unit area. Those things are very important. If you get them wrong, you have a very wrong representation of the fire in your model. We went through approaches in CFD using volumetric heat sources, using boxes to emit uh heat, to release heat. You can also emit fuel in your CFD models in three dimensions, from boxes you can emit them from surfaces. This is how you would do it in FDS. And then use combustion modeling, simplified combustion modeling to get your uh chemical heat release rate. You have to care about the convective part, you have to care about radiative part, you have to be aware that those parts are separate and and uh you have to capture those phenomena. And if you do it all well, if you give enough consideration into building those, I think you'll end up with some really good models, and I I really hope that you do. So uh that's it for the beautiful world of placing a fire into your model. I hope that was interesting for you. It was interesting for me to revisit all the things that I I knew and and and kind of work it up in my head. Also, an interesting exercise for myself that is most likely beneficial for my future and career as a fire safety engineer. Every day I'm trying to be a better fire safety engineer and I take you on this journey with me. I hope you will stay with me. And uh if you do, then next Wednesday you'll have another opportunity to become a better fire safety engineer, and so will I, because we will learn some new things. Thanks for being here with me in the fire science show. See you here next Wednesday. Cheers. Bye.