257 - Fire Fundamentals pt. 21 - Radiation with Simo Hostikka

Why Radiation Matters In Fires

Wojciech Wegrzynski 0:00

Hello, everybody. Welcome to the Fire Science Show. Today, we're venturing into fire safety fundamentals, and, uh, for the first time, I think, in the podcast, we will be covering radiation modeling. I had an enormous privilege to sit down with Professor Simo Hostikka from Aalto University after his fantastic plenary lecture at the IAFSS conference. Uh, I first thought about, uh, you know, interviewing Simo about the contents of his lecture. He went into great depth into the challenges with non-gray gas modeling of radiation in gases. Uh, it was quite fascinating, to be honest, and I probably opened, broadened my mind to new array of potential, uh, issues and challenges with modeling radiation. But then I reflected that I've probably not given, uh, enough to radiation yet in the Fire Science Show, So I've asked Simo to perhaps do a more simple, uh, you know, episode, introductory episode, and he has kindly accepted this concept. So here we are sitting together discussing the fundamentals of radiation. Uh, radiation in fires, what is radiation, how does the heat transfer in that, mode, uh, what's emissivity, what's, what's a black body, what's a gray body, what's a wavelength, how it's wavelength dependent, and all sort of physical stuff around the concept of radiation. But we're also going to do modeling because Simo, is actually the developer of, uh, the radiation model in FDS, so something that probably most of us are using daily, and that's, uh, the perfect person to discuss how radiation is actually modeled in the modern fire safety engineering tool sets. So half of the episode's on the basics, half of the episode on the modeling, one episode full of knowledge on radiation. And this is introductory material, and I... Simo promised me we will do a follow-up with an in-depth jump into all the wavelength dependent stuff and source terms that he has shown in his plenary. I also cannot wait for that, but for now, let's spin the intro and discuss radiation. The Fire Science Show 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 that this year celebrates its 10th anniversary. As experts in fire engineering, they are fully committed to delivering preeminent expertise to protect people, property, and the environment. With over 30 chartered engineers and a team of fire researchers at their core, they continually explore the challenges that fire creates for their clients and society so that the best research, experience, and diligence can be applied for effective tailored solutions. In 2026, OFR will grow its team once again and is keen to hear from industry professionals who want to collaborate on fire safety features this year. Get in touch ofrconsultants.com. And now back to the episode Hello, everybody.

How Simo Entered Radiation Modeling

Wojciech Wegrzynski 3:35

I am joined today by Professor Simo Hostikka from Aalto University. Hey, Simo.

Simo Hostikka 3:40

Wojciech. Thanks for inviting me.

Wojciech Wegrzynski 3:42

Oh, thanks for taking the, the invite. We're here in the beautiful La Rochelle. Yesterday, you gave a fantastic plenary lecture on, um, the role of radiation in, in fire science and engineering. Thanks, thanks for that. And, uh, I hope today you can, uh, give us an interesting view on this, uh, problem for the listeners of the Fire Science Show.

Simo Hostikka 4:00

I hope so. Yeah.

Wojciech Wegrzynski 4:01

I'm sure you can. I, I'm, I'm sure we're on a good position to do that. So, uh, Simo, I never had really radiation covered in the Fire Science Show in terms of, uh, role it plays, like how does it work- Mm how important it is. I think it's a perfect opportunity to help our fellow engineers understand, uh, better something that accounts for so much of the fire problem, right? Like, uh, maybe let's start with how you started in radiation, and how did you drift into fire with radiation? Or was it other way, first fire, then radiation?

Simo Hostikka 4:34

Right. Yes. Well, first of all, radiation is a scary subject.

Wojciech Wegrzynski 4:39

Is a scary subject.

Simo Hostikka 4:40

Yeah. For... It was f- that for me, and it, uh, is for many. I started to do fire modeling and fire research at VTT, kind of right after my master thesis that dealt with fire plumes. So not... Didn't know much about radiation. Oh,

Wojciech Wegrzynski 4:59

so fires first. Fantastic.

Simo Hostikka 5:00

Fires first, and then I very quickly get, uh, involved in, CFD modeling- Mm-hmm using SOFI. I don't know if you have heard- SOFI, okay SOFI from Cr- University of Cranfield at the time. Yeah. And, then at that point, I needed to understand what is this radiation model that they talk about And yeah, then a couple of years later, I, had the chance to go to NIST as a guest researcher, and we didn't... Like, it, it was funny. I met Howard Bauman in a conference and asked, "Can I come join your team as a guest researcher?" Mm. And we didn't have even a plan of what I should do there. Yeah. So I marched to the, to their laboratory and office saying, "Hey, here I am."

Wojciech Wegrzynski 5:41

What I do?

Simo Hostikka 5:41

Yeah. What should I do? And then, okay, Kevin was assigned my kind of host there, and, uh, "Well, here, why, why don't you look at this radiation?" Uh-huh, okay.

Wojciech Wegrzynski 5:51

W- w- was, was, it was like the early days of FDS, right? It was like-

Simo Hostikka 5:54

Yeah, FDS1 had been released.

Wojciech Wegrzynski 5:56

FDS1, okay. Yeah.

Simo Hostikka 5:57

and, you know, FDS1 really didn't have a radiation model. Mm-hmm. Combustion was calculated as a Lagrangian particles that- Mm follow the flow and release energy. And it, it, it's, uh, it was some kind of really, really simple model of then spreading the radiative fraction of heat release to the boundaries.

Wojciech Wegrzynski 6:17

Okay. That's, uh, that, that's an interest- Uh, w- why did you say it's scary? Like, for me, it's scary because the equations are inapproachable and with the... It's not just, it's, uh, complicated, uh, differential equations, but they're also in a spherical coordinates.

Simo Hostikka 6:31

That, that, that's exactly. The figures are complicated. Equations have so many terms that are hard to relate to what we otherwise deal with in, fire. And, like I said to many people here in the conference, that what I was trying to do in my lecture- Mm is to serve as a translator-

Wojciech Wegrzynski 6:50

Yeah, yeah

Simo Hostikka 6:51

When I started to do this lecture, I thought, "Well, how can I do it?" Because there are professors and researchers who have been focusing very deep in radiation the entire- Mm-hmm their career. I owe them a lot.

Wojciech Wegrzynski 7:02

Mm-hmm.

Simo Hostikka 7:03

I, I kind of much of the things we do are applied if we compare to what they are doing. Mm-hmm. But then I realized, okay, but that would be the case with many of us. Mm-hmm. It's very difficult to dive into s- what somebody's doing or has been doing for years. So maybe I can translate some of that more practical or applied people.

Wojciech Wegrzynski 7:24

Fantastic. Uh, well, I, I'm absolutely certain that you've done that excellent. And, uh, the, the way how you build equations with, you know, having the, the un-approachable term, but underneath there was explanation what this term means. Like, uh, this minus this is this. Uh, it was very, very appreciated. And a lot of people are asking how, how do you make your PowerPoint plots- Yeah uh, or if that's a secret, I'll pay for it, but we can discuss that later. Well, anyway, uh, Simo, what is the relevancy of, of radiation in fires? Like, uh, if we look at the fire phenomena where radiation is in the fire science, why are we discussing it? Why are we worried

Radiation’s Two Jobs In CFD

Wojciech Wegrzynski 8:01

about it?

Simo Hostikka 8:01

Well, first of all, my viewpoint is very much of fire modeling and fire CFD.

Wojciech Wegrzynski 8:06

Mm-hmm.

Simo Hostikka 8:06

Some other re- viewpoint like might give a different answer. For fire modelers, radiation plays two key roles. Like, we all want to get heat fluxes. Mm-hmm. That's a task. We need to solve the radiation eq- equation to tell how much radiative heat flux, meets a, a wall or some target, r- steel beam somewhere.

Wojciech Wegrzynski 8:30

Mm-hmm.

Simo Hostikka 8:30

And also, when we do model validation, that's what we want to calculate or predict, heat fluxes. That's the first task, and it's very clear to everyone. What was a, a little bit more harder to figure out for many is that in order to calculate, solve the Navier–Stokes-

Wojciech Wegrzynski 8:47

Mm-hmm

Simo Hostikka 8:47

the fluid equation, there's actually term that is related to radiation. that term in the energy equation for gas tells how much energy is, exchanged between the gas and the electromagnetic field. Uh, electromagnetic field is the scary part. Mm. That's the hard part. It's difficult. We cannot see. We cannot really, kind of approach it- Mm-hmm very easily.

Wojciech Wegrzynski 9:13

S- so kind of like Navier-Stokes solves the fluid, uh, part, s- solves the, the fluid motion, and there's a term that translates some of this energy in the fluid motion or like, yeah, in the hot gas- Yeah to a, to an electromagnetic spectrum, which is also in the same space, just not the part of Navier-Stokes. Yeah,

Simo Hostikka 9:30

yeah. L- like, like we have the combustion term. Mm. If you have studied combustion, you know, okay, combustion releases energy. It adds enthalpy. It increases temperature, creates temperature difference, creates buoyancy. Likewise- Mm radiation takes some of that energy away- Mm if the gas radiates. Mm-hmm. So it actually reduces that source term that combustion just created.

Wojciech Wegrzynski 9:56

One thing that, that I find perhaps interesting in radiation is that we're als- always talking about a product of an exchange. Like, if something radiates heat, it also means it, uh, absorbs heat at the same time. And, uh, the, the product- That's right outside is, is the product of this e- exchange. Can you maybe speak on how bodies emit and absorb, uh, heat through radiation?

Simo Hostikka 10:20

That's right. And I must confess that it took me a long time to understand that if, if you took a radiation lecture in, in, in college or university- Mm you might be heard something like Kirchhoff law- Mm-hmm that tells that, okay, how the material emits radiation. It absorbs at the same rate. So they are kind of in balance. And we often assume that for, if we, let's say, solid material emissivity is something that all of us need to specify when we calculate heat transfer- Mm for structures or something. And then we say that, okay, and this is by the way, the same number as we use for how much it absorbs. But it took me years to understand that this is actually not the case. It's only true or it has to be true at a single wavelength at the same time. Mm. It's true spectrally. For if we look at the entire spectrum, the total energy, it-- they don't have to be the same. But yeah General level materials emit and absorb, but they are very different processes. Emission depends on the material itself, its temperature. Uh, the rate can be calculated using what we call Planck function- Mm-hmm or black body radiation. Absorption, on the other hand, depends on what's there outside the material. Mm-hmm. What's somewhere elsewhere. The spectrum of incoming radiation depends on how hot are the other surfaces- Mm-hmm and our environment or other gases.

Wojciech Wegrzynski 11:51

let, let's

Spectrum Basics And Blackbody Rules

Wojciech Wegrzynski 11:52

first discuss perhaps the spectrum because I also find it very- very interesting and perhaps it's something that brings a lot of cha- Well, especially in your yesterday talk, okay, there was something I was not really considering much in my, uh, everyday engineering. Uh, but you were kind enough to show us how complicated this world is. Uh, but, uh, yeah, let's, let's first discuss, how this energy character spectrum. I- my mental model for it is that, uh, the infrared radiation also has colors, you know, and there's different colors of heat that, that could fly to you. That, that's like how I try to imagine why there's like a spectrum, but.

Simo Hostikka 12:27

Yeah. A- actually, that's the very same idea. When I had to draw or explain spectrum to the Finnish firefighters in, in one of magazines.

Wojciech Wegrzynski 12:37

Mm-hmm.

Simo Hostikka 12:38

I draw the spectrum so that I always had the visible colors in the same plot. Okay. Okay? Yeah. They, they are kind of there on the edge at, at the low wavelengths. Mm-hmm. Below one micrometer. That's where is the visible light, and that same spectrum then continues to the longer wavelength.

Wojciech Wegrzynski 12:56

Mm-hmm.

Simo Hostikka 12:57

But it, I think it's very kind of, helpful to see that, that, okay, it's this same stuff that we see that it continues- Mm-hmm to somewhere we, that we don't see. And in terms of energy content for most of the energies there at the higher wavelengths- in the infrared region that we don't see.

Wojciech Wegrzynski 13:15

Perhaps we can, uh, go to Mr. Planck for a little bit and discuss the black body. Uh, what is a black body in terms of, radiation modeling and ra- radiation science, and, uh, how this knowledge of, of black body temperatures relates to a spectrum and the amount of energy we, we transmit through radiation, because I think it's, uh, also a fundamental concept.

Simo Hostikka 13:35

Yeah. Black body, emission or absorption curve is the ideal one. That's the ideal material. How much, in theory, any material can send or send material. The, yeah, emission, it's-

Wojciech Wegrzynski 13:48

Mm

Simo Hostikka 13:48

it's what it talks about. Uh, so We have rules like Wien's law, which says that how the- peak wavelength of that curve depends on temperature. Product of the wavelength and temperature is the constant, and we can easily see that, okay, if we increase the temperature, then the peak wavelength gets smaller. Uh, and so we see that, for example, fire temperatures we have, peak is somewhere between two or five micrometers. And if we go to the temperature of the sun- it's somewhere very close to visible. uh, like meaning around one micrometer or so. yeah. But it's the ideal one. Like, it really gives us the limit of how much we can get out of ener- get energy out of a surface. At,

Wojciech Wegrzynski 14:36

at specific wavelength, at specific temperature.

Simo Hostikka 14:38

Yes.

Wojciech Wegrzynski 14:39

Okay. And, the heat flux itself then is what? Is it a sum of all those-

Simo Hostikka 14:45

Wavelengths, yes. Okay. Sum of all those energies, or as we can say, integral over the wavelength.

Wojciech Wegrzynski 14:51

another thing that makes, radiation perhaps difficult for, for me, like intu- intuitively it's, it's, it's quite challenging, is the fourth power relationship with temperature. I always, you know, it... Like fourth power, I don't think is comprehensible for a human mind. It's, uh... What, what's your experience with that?

Simo Hostikka 15:10

Yeah. It... Well, that's what we learn.

Wojciech Wegrzynski 15:12

Yeah.

Simo Hostikka 15:13

It's the fourth power, and as such, it's, it's easy, even e- easy to remember. I noticed most of my students, if, if anything, they remember that one rule- Yeah. from the bachelor- It's so odd course. Yeah. It... Okay, fourth power. That's easy. But where it comes from, again, this kind of- quantum mechanics explanation of behavior of materials leading to this odd function, which is really strange. Mm. And, and, okay, n- it's typical math task for students. Mm. Okay, now try to integrate this function, and you get the fourth power. But yeah, and, and only f- those who are really good in math then, then end up doing it. But yeah, but I, I, to be honest, I don't also have intuitive explanation for that.

Wojciech Wegrzynski 15:57

I appreciate it because i- if you think about, spectrum of temperatures you would get in fires, like, you can have, let's say, 200 degrees of smoke layer, which is pretty much the edge of what's safe. Why? Because it radiates two, two and a half kilowatts per square meter, and that causes burns already, right? You add a few hundred temp- centigrades to that and you have, like, 600, where radiation is already at the range of, I don't know, 15, 20 kilowatts. You can have a flash over due to that.

Simo Hostikka 16:28

Yeah, you can, I think so.

Wojciech Wegrzynski 16:29

Yeah, but you add another 300 degrees and you're in a fire furnace, and you're talking about 75,

Simo Hostikka 16:37

100, 150

Wojciech Wegrzynski 16:37

kilowatts per square meter. Mm-hmm. You know, go to a tunnel fire and we're talking about 200, 250 kilos, or petrochemical fires, you know? And, uh, it... While y- you're adding very little, uh, you're adding kilowatts on kilowatts of, load, of heat load on, on whatever you are exposing to. So that, that's And I think fire engineers need very, need to build that intuition that, you know, a few degrees could be a very big difference.

Simo Hostikka 17:02

Yeah. Experimentalist kind of has that-

Wojciech Wegrzynski 17:05

Intuition

Simo Hostikka 17:05

intuition from the fact that, okay, heat fluxes add up very quickly- Mm-hmm as you said. But- One of the way of getting intuition is your own feeling. know, how you can kind of feel it with your hand, and it stops there. Very quick. Very, very quickly. Yeah. Very, very quickly.

Wojciech Wegrzynski 17:23

I, uh, this is also, like, uh, interesting, uh, with the distance. It's also... You know, because it's a spherical equation. We, we mentioned the math is complicated because of spherical integration. If you have a point source, if you have a candle, it projects everywhere. So obviously, the closer you get, the more exposed you are. And it also, like, it's not a linear equation. It's not like one meter is, uh, half of what you get at two meters. It's, it's very, very different. And I al- so also feel that in some experiments. And when I have a fire experiment and I feel heat, feel heat on my skin from, like, 20 meters, I know it's like, that's a big fire.

Simo Hostikka 18:00

Yeah, but I, I really like how, how this idea of spherical source has been visualized and made- Mm-hmm made easy to understand in, in the, engineering handbooks and, like- where it's used for radiation calculation because it's explained in terms of the sphere area- Mm which then depends on the, R two squared.

Wojciech Wegrzynski 18:22

I- on this point, we can also, like, perhaps briefly mention view factors, because if you now are a fire engineer, that's also an, an important concept that sometimes you will use, in proxy of that spherical mathematics. You know, just calculate the view factor of how much a surface is, is exposed. Do, do you use, teach them or...?

Simo Hostikka 18:44

I, I do teach them-

Wojciech Wegrzynski 18:45

Yeah

Simo Hostikka 18:45

to, in the engineering course, but in fire modeling, you know. Mm. There we don't actually use them. What the model is supposed to do is to actually predict-

Wojciech Wegrzynski 18:54

Replace everybody or

Simo Hostikka 18:55

replace.

Wojciech Wegrzynski 18:56

Yeah. I like view factors as a mental model of, uh, why some situations are more hazardous than others. Like I mentioned, a smoke layer. Like a, for me, a smoke layer scenario is a very challenging scenario for people underneath the smoke layer because the view factor would be approximately one. Like if you- Yeah, yeah, yeah if you're a parallel to the- Yeah source of radiation, then you're exposed to, like, pretty much as much as it emits. Yeah. While if you stand next to a, campfire, there's a plume of 200 degrees of smoke. Mm. It's not gonna feel that because it's, it vanishes- Yeah in the spherical dimension. Yeah.

Simo Hostikka 19:31

Yeah. It, it, it's easy to visualize if you think about the radiating- Panel or, or piece of the wall somewhere. if you look at it from distance, it, looks small. And when you start to walk towards it, finally your nose is touching the wall- and that's the all you see.

Wojciech Wegrzynski 19:47

Yeah. That's, that's a, that's a good mental model. Um,

Gases Soot And Flame Radiation

Wojciech Wegrzynski 19:50

okay. another question I have is, are there any fundamental differences from the physical perspective in how radiation works for solids and gases and soot and smoke and, and flame? Like, or you can bring those all into the, the same mathematics, it's just a slightly different s-source from your, from your perspective.

Simo Hostikka 20:11

The mathematics is the same. Mm. Physics is the same.

Wojciech Wegrzynski 20:14

Mm-hmm.

Simo Hostikka 20:15

relevance or importance of different phenomena varies. Okay. Like, like, uh, we did a lot of work with Farid Alinejad, who was a PhD student- Mm-hmm on our team, about the solid and liquid radiation for a couple of years. And, and, and Farid was a super productive person, published many papers Mm-hmm. And we showed that, okay, yeah, liquid fuels, they show very strong spectral characteristics, And, and, and so do some solids. Mm-hmm. We, we measured the PMMA and compared to earlier ones, and we developed what are called non-gray models, considering all the fancy- spectral behaviors. But then in the end, when we go to the level of here is the pyrolysis model- and, and I need to plug in whether it's one Kappa, one absorption coefficient there, or something fancier. I-it seems to me that in contest materials, we can simplify things to- Mm to gray or kind of almost gray, which means that we ignore all the detailed wavelength- Mm dependencies and kind of average them out to some, some more simple model. I think this is not the case in gases.

Wojciech Wegrzynski 21:28

Okay.

Simo Hostikka 21:29

So, so it, so it depends on, on the importance. In gases, the variations of material properties, the absorption coefficient between different wavelengths. These variations are like orders of magnitude, several orders of magnitudes. They fluctuate like crazy. So it means that at different wavelengths, the radiation behaves so differently. At b- making this averaging causes much more error than in condensed material. Okay. Something that we should, however, take into account or, or what can make things complex are the interferences. Mm-hmm. We all remember how light behaves at the surface of water. Mm-hmm. If you are above or below the lake surface, you see things very differently. Yeah. And, this also happens in the infrared. Mm. For the radiation that hits the surface of a burning pool. But we never think about these things- in, in fire modeling is it important? Uh, okay. Um, it depends. But if we want to really start to do divine modeling that considers b- all the different phases- Mm both gases, both liquids, then we need to include this kind of- I phenomena

Wojciech Wegrzynski 22:42

we- we're at this crossroads in, fire modeling where, you know, we're successful and good with our design fires, fixed, you know, fire sources, just solving the fluid part in the buildings to, to design smoke control perhaps. That- that's bread of a... bread and butter. Like, a lot of people do that daily. But the jump to, uh, modeling fire spread, for example, like your student Raul did. Like, that's, that's a massive jump, to be done. And for that, you need that finite knowledge, and that is very detailed insight into how heat is transferred-

Simo Hostikka 23:16

Mm-hmm

Wojciech Wegrzynski 23:16

into the pyrolysis medium or into- Mm a pool fire, right?

Simo Hostikka 23:19

Yeah. Fire spread and also, like, some fire phenomena. Like, if you think about spill fire- Mm and try to predict or calculate. The s- spill fire burning rate depends on how much you have the liquid there.

Wojciech Wegrzynski 23:32

How

Simo Hostikka 23:32

far it

Wojciech Wegrzynski 23:33

fell.

Simo Hostikka 23:33

How, how far it fell on, on, uh, on, on the surface of concrete or-

Wojciech Wegrzynski 23:37

Yes, yes. How thick it is

Simo Hostikka 23:39

exactly. Yeah. It- So it would be very hard to construct engineering kind of rule.

Wojciech Wegrzynski 23:45

Yeah. It's always five square meters for a lit- for a liter, and, and we're done.

Simo Hostikka 23:48

Yeah, yeah. So, so we e- easily end up to the question, okay, can you do a little bit more calculation on that? Mm. And, and that was a very example where we showed that the... say, actually the heat transfer things inside the layer matter. That was actually the, in our study about non-Gray modeling of liquid fuels, where things really started to matter were the super thin layers like spill fires.

Wojciech Wegrzynski 24:17

Mm. Another question. Does air, absorb radiation? Like, air is transparent to my eyes right now.

Simo Hostikka 24:24

It is.

Wojciech Wegrzynski 24:24

Yeah. Does it, uh, like, if a fire emits radiation, it radiates away. does the amount of radiation going through air decays with the distance just because air eats it up?

Simo Hostikka 24:35

Yeah. It does, and, and that makes things tricky. Well, if we look at like SFP handbook, we can already see rules for how the pool fire attenuates at distance. Mm. So it has been studied and observed, of course. but yeah Oxygen and nitrogen don't really

Wojciech Wegrzynski 24:54

okay

Simo Hostikka 24:54

these two atomic molecules, they are very passive in our wavelengths with, of interest. They are not fully passive. They, they have some small features that they, exhibit in the spectra, but they are irrelevant.

Wojciech Wegrzynski 25:09

Mm.

Simo Hostikka 25:09

They can be considered like, uh, non-radiating. But CO2, and of course water vapor in the atmosphere, cold atmosphere, they absorb a lot. And they absorb in a very non-gray manner, meaning that, that it strongly depends on wavelength.

Wojciech Wegrzynski 25:25

Y- you've used the term gray gas or, or gray manner a lot, a lot. Like, let- let's perhaps explain that. Mm. Wha- what is the difference between black model and, and a gray model?

Simo Hostikka 25:34

Well, black model is the perfect-

Wojciech Wegrzynski 25:36

The sum of everything

Simo Hostikka 25:37

the perfect rate or, or ideal rate at which material- Okay can send radiation. Gray model is something, okay, it's not perfect, but the level of imperfection is the same- at all wavelengths Non-gray means that, okay, your-- let's think about surface emissivity. Now, it depends on wavelength. At- Mm. At some wavelength, it's close to black surface. At some wavelengths, it's more shiny. Like, like, uh, uh, it, it reflects, and at the same time, it maybe doesn't absorb- or emit.

Wojciech Wegrzynski 26:11

So when I put, uh, emissivity less than one, I use a gray model.

Simo Hostikka 26:15

Then you are doing gray. That's correct. Okay. And, and that's what we usually do.

Wojciech Wegrzynski 26:19

And when I put like, uh, for this wavelength zero nine, for this wavelength zero two, for this wavelength zero three, then I'm moving into a non-gray.

Simo Hostikka 26:27

Correct.

Wojciech Wegrzynski 26:27

Okay. I hope I'm doing it right. so back to the, uh, CO2, H2O, h-h-how do they absorb? Like, uh, is there specific wavelengths they like and they eat out- Yeah and the rest is, and the rest is transparent? Yeah.

Simo Hostikka 26:41

Well, if we look at the databases of molecular spectral transitions, first, first of all, these databases are based on-- They come from quantum dynamic- Mm-hmm or quantum mechanics explanation of the states of the molecules. Mm. Molecules vibrate, rotate, whatever things happen there. And all these changes in these states can be caused by an absorption-

Wojciech Wegrzynski 27:08

Mm

Simo Hostikka 27:08

or emission of a photon. One, one quantum of radiation. And these different transitions have different energies, leading to a photon with different wavelength. Okay? So it means that these transitions, energies, and wavelengths are quantized.

Wojciech Wegrzynski 27:26

Mm-hmm.

Simo Hostikka 27:27

They cannot happen at any wavelength, but only- Very- certain, very specific. However, there are a lot of those.

Wojciech Wegrzynski 27:34

Lots, yeah.

Simo Hostikka 27:35

Yeah. They, like-

Wojciech Wegrzynski 27:36

Even for simple molecules

Simo Hostikka 27:37

like- For a simple molecule like CO2- Mm you would think that, okay, it cannot be changed in so many way. But, but, but water vapor and CO2 in, in our wavelength range in relevant for fire, there are I don't know, thousands, tens of thousands, or maybe hundreds of thousands of those lines. I don't remember.

Wojciech Wegrzynski 27:56

Mm-hmm.

Simo Hostikka 27:57

And, To construct the absorption spectrum, from those databases. It's, it's big task of calculation. Mm. But, but it can be done, and it's very basic thing that radiation researchers actually do- Mm nowadays.

Wojciech Wegrzynski 28:10

And if you have a very complex mixture and, uh, like, uh, what you would find in the products of a fire. And let's for a second put the, the soot aside. Mm. So let's just talk about gas- gases that are produced in fire, and you will have a whole range of gases. Th- th- this also means, like, there will be more lines in there that will, like, cover more of the spectrum eventually, up to probably all of it. or, or not.

Simo Hostikka 28:34

That's right, yeah. And, well, most gases, if we plot the, spectrum, if you see a figure in a, in a-

Wojciech Wegrzynski 28:41

Mm-hmm

Simo Hostikka 28:42

a article or a textbook, it, it looks like the gases cover the entire spectrum. Mm. Even water vapor and CO2. But what often fools me is that those figures are plotted in a logarithmic axis.

Wojciech Wegrzynski 28:56

Okay.

Simo Hostikka 28:57

If you did the same plot- Uh-huh in a linear axis- Okay, okay you would see that, okay, water vapor seems to have, like, two or three- important regions in the spectrum- Mm where it absorbs. And, and, and that's how it can be simplified.

Wojciech Wegrzynski 29:10

Mm-hmm.

Simo Hostikka 29:10

But then if we want to do fancy detailed things, then we actually want to consider all those smaller levels as well, and we need to plot things in the log axis. Right. But yeah, for the sake of kind of engineering calculations, we could say that, okay, CO2 has maybe two regions in the spectrum- Mm which are important. Water vapor, maybe three. And, and with those we could describe most of the energy transfer. I

Wojciech Wegrzynski 29:39

from a different perspective, it's also, like, an interesting fingerprint of each gas because it also allows for remote sensing, right? Like, y- you can, knowing where those peaks are in the spectrum- Yeah you can also, like, kind of-

Simo Hostikka 29:52

Remote sensing and fire detection, of

Wojciech Wegrzynski 29:54

course. Yeah. Of course, yeah. Yeah, yeah. fantastic. And, what about soot? Like, uh, does soot fundamentally differ from, from all of that? I saw on your plot yesterday-

Simo Hostikka 30:02

Yeah

Wojciech Wegrzynski 30:02

it was different.

Simo Hostikka 30:03

Yeah. We see that the soot spectrum is very smooth, and it, most people actually approximate it as a constant over the spectrum- Mm which is fine. Or, very simple linear relationship- Mm-hmm between its absorption coefficient and, and, wavelength, which is good approximation.

Wojciech Wegrzynski 30:22

Mm.

Simo Hostikka 30:22

Why is that? Like, we know that soot particles, they are actually really small. They are tiny kind of pieces of carbon and some other atoms. But it seems that, okay, although they are so, while they are so small, they are still from the perspective of light or, uh- Okay radiation wavelengths, they are big and they radiate like solid-

Wojciech Wegrzynski 30:46

Okay

Simo Hostikka 30:46

material.

Wojciech Wegrzynski 30:47

Yeah. I, in my mental model is they're like flakes of carbon flying around just- Yeah and the, uh, the, the size matter much here? Because, like, the visible light is one micron and less, all right, around that.

Simo Hostikka 30:59

Mm.

Wojciech Wegrzynski 31:00

Soot particles, the smoke particles we measure in, in, in fire experiment also, like that's around a micron range. Yeah,

Simo Hostikka 31:06

the smallest one. It's an interesting- The inner flame. Yeah. Yeah, that's true. And to be honest, I don't understand soot very well.

Wojciech Wegrzynski 31:11

Yeah. And,

Simo Hostikka 31:11

uh- I

Wojciech Wegrzynski 31:12

not understand

Simo Hostikka 31:13

I, I have been scared of soot and, and avoid touching it for all of my career.

Wojciech Wegrzynski 31:18

I would r- under- I would not recommend touching soot to anyone. Yeah. That's a, that's a We've learned a lot, uh, since, uh, people were comfortable touching soot and getting dirty with soot. how about flame? How does a flame emit? What, like, what in flame emits and, is it gases? Is it products? Is it soot? Everything?

Simo Hostikka 31:36

Yeah. Radiation researchers love the flames that we fire engineers don't usually do.

Wojciech Wegrzynski 31:44

Okay.

Simo Hostikka 31:45

Radiation researchers loves to study flames that are optically thin, so that they are non-sooting. Okay. Like, like, like methanol flames and- Okay things like that, that just have that nice blue color, because then we can focus on the, all the- m- complexities of our radiating gases- Mm and, and kind of take, take the soot out of picture.

Wojciech Wegrzynski 32:07

So it's like, uh, sorry, it's like flames from my gas stove when I turn it on. Yeah. They're beautiful blue and-

Simo Hostikka 32:12

Yeah

Wojciech Wegrzynski 32:12

there's no smoke in my kitchen.

Simo Hostikka 32:14

Yeah, yeah. and then we can study the gas radiation behaviors. Then other sooting flame, like, like ethylene, typically in experiments or any big carbon, based fire in, in a, in a fire experiment. So yeah, they are almost all showing the yellow flames- Mm which comes from the... It's a s- a sign for us that, okay, now there's soot- because we see, in the visible wavelength, we see the kind of smooth, a nice color. But al- it's also a sign that, okay, in the infrared range, now there's a uniform spectrum- Okay soot almost entirely-

Wojciech Wegrzynski 32:52

A-

Simo Hostikka 32:52

dominates

Wojciech Wegrzynski 32:53

a- and measuring the spectra, you can see significant difference between a sooting flame and, uh, um-

Simo Hostikka 32:57

Absolutely. Yeah. Yeah. Yeah. For those gas flames, we would see these peaks that we just mentioned- Mm for CO2 and water vapor, maybe some other gases. Like you said, uh, if there are fuel gases or unburnt- things, maybe CO, we some, see some additional spikes. For the yellow flame, what we would see was the Planck function. Mm.

Wojciech Wegrzynski 33:18

W- what's the reason for flame changing colors? I saw those ex- Rory loves to do those experiments where- Mm he, he does all colors of flame.

Simo Hostikka 33:27

Yeah. Well, we see these transitions of molecules that we talked about- appearing in our visible range, meaning from 0.3 to 0.7 micron. Doing

Wojciech Wegrzynski 33:40

spectral analysis with your own eye by just witnessing the color of the flame. Yeah,

Simo Hostikka 33:44

yeah.

Wojciech Wegrzynski 33:44

It's kind of beautiful, isn't it?

Simo Hostikka 33:45

Yeah, it's really nice. Yeah.

Wojciech Wegrzynski 33:47

Okay.

How FDS Solves Radiation Transport

Wojciech Wegrzynski 33:48

Um, let's move, uh, let's move further. Perhaps now let- let's try and touch on modeling radiation. We've talked a lot about, uh, you know, how radiation works, about, uh, the gases, the solids, the, the Planck equations, et cetera. But, now you're a CFD engineer. Now you, you just, uh, approached, uh, NIST, and Kevin tells you, "Look into that radiation." what are our choices as modelers to account for that in our engineering, in CFD first?

Simo Hostikka 34:18

Yeah. Well- When I started to do it, I, I noticed that people were using something like, ray tracing- methods. And I think that was very intuitive way of doing it because then we see the similarity between model and the actual physics. Mm-hmm. Photon leaving a surface or leaving- Hitting something a molecule, and then the pho- ultimately hitting somewhere. Yeah. So we follow the straight path of the mo- photon until it's absorbed. And that made a lot of sense, but it was very inefficient or hard way of doing it in the numerical terms. And then we started to do, modeling or consider energy fluxes. So basically say that, okay, instead of doing a huge number of these rays, let's integrate this thing first over a range of directions. Let's- Mm-hmm okay. in simple terms, I could say let's integrate the equation so that we calculate radiation in just two directions: upwards and downwards. That would be a flux-based method.

Wojciech Wegrzynski 35:21

So how much, uh, you have a sum. Uh, it emits a 10 kilowatts, seven go up, three go down, and here you go.

Simo Hostikka 35:27

Yeah. Yeah. And that could be spectral or gray that we discussed a- Mm a moment ago. It doesn't matter, but it, now we talk about how the energy moves in space So the flux-based methods provide kind of smoother solution- Mm and can be implemented also fairly easily. Uh, what's interesting in radiation modeling is that we are almost every time using the same mesh as the CFD-

Wojciech Wegrzynski 35:53

Okay

Simo Hostikka 35:53

model is using.

Wojciech Wegrzynski 35:54

Is it necessary?

Simo Hostikka 35:55

No, but I think it's just a sake of convenience- Okay to use the same mesh because then- implementing this exchange from the radiation- Okay to CFD equations is made easy.

Wojciech Wegrzynski 36:08

Closing the neighbor cell experiment. Yeah.

Simo Hostikka 36:09

Yeah. We are exactly the same locations- Mm-hmm same wall cells.

Wojciech Wegrzynski 36:13

Every cell is your absorber. Yeah. Every cell is your- Yeah emitter. You close the neighbor locally.

Simo Hostikka 36:18

Exactly.

Wojciech Wegrzynski 36:19

but our mesh is, for example, in FDS, they are, like, rectangular, so-

Simo Hostikka 36:22

Yeah

Wojciech Wegrzynski 36:22

it, it Like, if you would just exchange it at the walls, you have six direction on every cube. But, uh, radiation goes in a What if it, I want to hit it by an angle, not, you know, perpendicularly to the surface? Yeah.

Simo Hostikka 36:34

Well, that's what we do in this current, flux-based methods that, that are called, like, discrete ordinates and finite volume method or finite angle method at- Mm-hmm as it should be called nowadays, which you can find in FDS.

Wojciech Wegrzynski 36:49

Mm-hmm.

Simo Hostikka 36:49

We Like, in FDS, we divide the Full solid angles. Mm-hmm. They're all the directions that you can think of- A ball. Yeah, the ball or sphere. A sphere. We c- we divide that in, I think the default is 104- Different directions, meaning that, okay, in each octant of this sphere- how many angles you would have? 100 divided by eight. A bit more than 10. Yeah.

Wojciech Wegrzynski 37:14

I know there's a story how, how, how, how... Because this is the FDS default, and the user can change that- Yes. Yeah that, that number. H- how did the number c- c- come to life? Like, it's a golden number. Explain, explain yourself

Simo Hostikka 37:25

now. Ooh, yeah. Well, yeah, we, like to make this joke that it, it's kind of the task to achieve certain, computational cost. Uh- Mm-hmm Kevin likes to tell this story- Yeah that, okay, if radiation is responsible for, uh, 30% of the fire energy transfer, so the calculating the radiation should not take much more. It

Wojciech Wegrzynski 37:45

should cost the

Simo Hostikka 37:45

same. It should cost at, at, at most. So we target it. Okay, it's fully arbitrary choice- Mm-hmm but we, we noticed in those early days of FDS2 that, okay, with this kind of number, we hit that ballpark.

Wojciech Wegrzynski 38:00

Okay.

Simo Hostikka 38:01

but it could be something else. I, I think, uh, default versions or typical versions of discrete ordinates, which is mathematically almost identical- Mm-hmm to what we do in FDS. the typical direction, weight sets are maybe similar order, maybe a l- a few- Mm-hmm directions less, but-

Wojciech Wegrzynski 38:21

Mm-hmm

Simo Hostikka 38:22

but, but very, very similar.

Wojciech Wegrzynski 38:24

let's perhaps reiterate on how the model works. So you have a cell that, is your source of radiation. What, what, what happens there? You emit a heat flux divided by 104 angles, and then you track what it happens- Yeah in every of those angles?

Simo Hostikka 38:42

Then I track the energy starting from a corner of the domain. Okay. I basically say if I, if my energy o- o- or, or the control angle is now upwards, and let's say- in 2D case, let's say it's to the right.

Wojciech Wegrzynski 38:57

Yeah.

Simo Hostikka 38:58

Then I would go to the lower left-hand corner of my domain- because then I say that all the radiation is now coming from the walls of- Mm-hmm of my corner. And from there, I start marching forward and updating my flow based on how much energy was absorbed by the gas- Ah and how much was emitted. And I can march through my domain to the- upper right-hand corner, and then I have covered every cell

Wojciech Wegrzynski 39:27

That's, that's interesting. Uh, my, my mental model was completely different. I thought you just take a cell, shoot 104 of those, and then go to next cell. No. And, uh, but, uh, okay. So you, you, you do the same calculation 104 times-

Simo Hostikka 39:41

Yes

Wojciech Wegrzynski 39:41

just every time starting from a different position.

Simo Hostikka 39:43

Yeah. Basically, one of those eight corners of the domain, we, uh, we start- Okay based on, uh, on which octant of the, our sphere we are moving to.

Wojciech Wegrzynski 39:55

And does, does, uh, does it happen every single iteration of the CFD

Simo Hostikka 40:00

simulation? Well, it could, yeah. And, and if you want to do things really the, uh, uh High resolution in time- Mm then you probably have to. But, but the default number, you know, is, uh, it's like every third time step we do a little bit, and every third time step we update every fifth of those directions. Okay. Okay. That's the default. So it takes, in total, 15 time step for all the directions to become updated. But that, those again are something that the user can change.

Wojciech Wegrzynski 40:30

Yeah. Well, it's, uh, in FDS, in a large ground fire, you're, you're looking at time steps of, uh, 100, 1000s of a second anyway. So it means, like, in 10 milliseconds you probably go through all of them anyway. So, so that- Yeah if you, if you think about the- Yeah like, iteration perspective, it may be short- Yeah but, uh-

Simo Hostikka 40:47

and, and when people have noticed that, okay, they, for some reason, need to increase the number of angles from 104 to something, let's say 1000 or a few thousand, well, it hasn't been any problem then to relax this time- resolution. All right.

Wojciech Wegrzynski 41:02

All right. Yeah. Yeah. So, so the user has also the power... Okay. So-

Simo Hostikka 41:05

Yeah

Wojciech Wegrzynski 41:05

uh, that's interesting. So you could, uh, sacrifice temporal, uh, resolution for a more spatial resolution.

Simo Hostikka 41:11

Absolutely. Yeah.

Wojciech Wegrzynski 41:12

It, it's

Ray Effect Diffusion And Monte Carlo

Wojciech Wegrzynski 41:13

important because, you know, a, a, engineering is a game, uh, in terms of, client's expectations, deadlines, prices, computational power, and everything. It's easy to, "Okay, I just want 5,000 angles," and, uh, and it is gonna increase my computational 50 times because it appears it's, it's just linearly adding to the-

Simo Hostikka 41:31

Right. And, and, and, uh, there is a one kind of trick in, in, in the modeling that we do to avoid that. Like, if you think that you would update radiation, let's say every 100th-

Wojciech Wegrzynski 41:44

Yeah

Simo Hostikka 41:45

time step, you would probably start to see kind of jumps-

Wojciech Wegrzynski 41:48

Mm-hmm

Simo Hostikka 41:48

in your energy. Okay. You would see the flame heating up because there was no radiation e- emission. Ah, okay.

Wojciech Wegrzynski 41:55

So

Simo Hostikka 41:55

you- It accumulates heat, and then suddenly t-

Wojciech Wegrzynski 41:57

The m- Uh- You miss the, the Navier–Stokes update that you mentioned.

Simo Hostikka 42:00

Yeah. But that's not happening because regardless of the choice of this time increment, we update the, this source term emission every time step anyway. Uh-huh. What- whatever the user chose for the updates of the transfer equation.

Wojciech Wegrzynski 42:16

Mm.

Simo Hostikka 42:16

So we still see smoothly developing temperature fields.

Wojciech Wegrzynski 42:21

But okay, this does also have a practical consequence of modeling, one that is immediately, visible in some simulations. Uh, you referred to this, I believe yesterday, as a ray effect and, uh, I also seen that in, in, in some cases as a product of this. May- maybe let's, let's talk about ray effect here.

Simo Hostikka 42:40

Yeah. I think ray effect has been now demonstrated- quite many times, and most engineers I have noticed are already aware of it. Mm-hmm. It's simply the fact that if our radiation source- is relatively small compared to the distances or size of our domain, from the lo- long distance, it starts to look like a tiny part of our, visible or, or The view factor-

Wojciech Wegrzynski 43:06

Yeah,

Simo Hostikka 43:06

yeah that we discussed. It's the same. It's- Yeah the view factor becomes small. Uh, it means- It's like a

Wojciech Wegrzynski 43:10

pixel on a large-

Simo Hostikka 43:11

Yeah matrix. Yeah. And then it means that those 104 directions are, are actually, if you think them as a ray, at far from that source, they are really far from each other.

Wojciech Wegrzynski 43:21

Mm.

Simo Hostikka 43:22

They get further and further when we go away and it seems then we start to see heat fluxes at, as blobs- Mm on the, uh- Yeah, I mean- opposite wall

Wojciech Wegrzynski 43:32

y- you have one on your, one wall on which one pixel is shining. Mm.

Simo Hostikka 43:37

Yeah.

Wojciech Wegrzynski 43:37

And you have a second wall at where- Mm 1,000 pixels could receive the light, but you only have 104- Mm angles, so you're gonna hit 104 out of the 1,000- Yeah pixels. That's- Yeah, yeah.

Simo Hostikka 43:47

And to be honest, I s- still don't fully understand why it happens. I understand why it happens in the ray tracing- Okay models, because then we really follow- You're shooting exactly- We're shooting the ray, and I know the energy is not going anywhere else. But in this finite angle method, where we first integrate over the angle, in theory, the energy distribution there should be smooth. Mm-hmm. We shouldn't see hotspot in the direction of one of those 104 angles, and then cold region between them. The energy should be carried in all of them, but it's something in the numerical calculation, how these angles interact with our CFD mesh.

Wojciech Wegrzynski 44:31

Yeah,

Simo Hostikka 44:31

but- Sometimes the angle kind of overlaps the cell a little bit, something, sometimes it overlaps l- less, and we calculate the fluxes based on the cell areas. And it's kind of very complicated. I

Wojciech Wegrzynski 44:42

mean, the, the ray could be barely touching a CFD cell- Yeah but then again, you would deploy some of that energy to that cell. And suddenly- Yeah you have a whole cell radiator- Yeah which in reality was barely touched. Maybe, maybe something like that. It's interesting.

Simo Hostikka 44:55

It, yeah, and it's very kind of much in the how the code was implemented- Mm how the code developer decided to calculate those exchanges. And, it's possible that some other way of doing it w- would reduce-

Wojciech Wegrzynski 45:09

Mm

Simo Hostikka 45:09

the ray effect. But I'm not worried about ray effect.

Wojciech Wegrzynski 45:12

I mean, it's I, I would say, okay, I had one in my entire career- Yeah we had one case- Mm where we would be really, really, uh, interested. It was like, a fire safety in a, in a restaurant kitchen, and there was a, a hob, uh, and we were looking. It was kind of close to an evacuation exit- Mm and we needed to understand if there's an oil fire on that hob, what kind of, you know, radiation we would have in the nearby proximity of that hob and, and at- Mm the evacuation exit with kind of a bit of shadows, et cetera. So in that case, like, We started our simulations was with discrete ordinates model, but we obviously didn't have enough, angles in the discrete or- because it, it was obvious that you have, like, a, a very high heat flux in one location- Mm and just 30 centimeters to the left, you don't have it. And I was, "Okay, we need to increase the angles." And then we got to, uh, something that looked plausible. But it was like really I was interested in the target radiation from a point source where this was my r- concern

Simo Hostikka 46:13

Well, what you just said- Yeah is the reason why I'm not worried about it.

Wojciech Wegrzynski 46:17

Yeah, you can see it.

Simo Hostikka 46:18

Because you can see it. You were already aware of its existence. Ray effect, which, that we just discussed, it's so kind of easy.

Wojciech Wegrzynski 46:26

Mm.

Simo Hostikka 46:26

Fairly easy to- Doesn't make sense to, to, uh, to explain everyone and, and, and every kind of educated, experienced user of these codes already knows about it. So, uh, or, or they very quickly become aware as they see it with their own eyes.

Wojciech Wegrzynski 46:41

Mm.

Simo Hostikka 46:42

Mm. Some other effects are much more difficult to comprehend. Mm. And that's what I'm kind of worried. Okay. That we have errors or uncertainties that we cannot fix because we don't even notice them. We don't have a reference.

Wojciech Wegrzynski 46:55

Well, give, give me an expl- a, a, an example.

Simo Hostikka 46:57

It, it is this, for example, this gray versus- Okay non-gray that we just talked about. It's very difficult to demonstrate, uh, in a, in CFD code or see your own eyes- Mm how much error we do if we smooth out the- Mm absorption coefficient- Mm-hmm to a single number instead of using a couple of different values for different, different wavelengths.

Wojciech Wegrzynski 47:19

Yeah. F- uh, that, that was a big, big part of your yesterday's lecture. It was very, very interesting. what about different approaches? So finite angle model is, uh, something that's default in FDS right now?

Simo Hostikka 47:31

Yes. And there it's called finite volume- Finite volume.

Wojciech Wegrzynski 47:33

Mm

Simo Hostikka 47:33

and, and, and that's It, it was a very good point made by, Professor Michael Modest in 2023 in his speech that, okay, now for 30 years we have been calling this fine- finite volume method, but what we actually are doing, we are discretizing- The angle the, the angle. So let's change the name.

Wojciech Wegrzynski 47:52

Yeah. That's f- fair point. Yeah. Fair, fair point, especially that the finite volume method is used for other things and perhaps in, in, in, in-

Simo Hostikka 47:59

Yeah is- And it's, it's more confusing to call it finite- Mm. Okay. in the, there kind of in, in the, in the roots of the finite volume or finite angle method, we do the, we have the same operation of volume integration and transforming those, uh, integrals into surface integrals using divergence theorem, blah, blah, blah, blah, which we do in the finite volume method- Mm for Navier–Stokes. The, the, so it made sense in the early days, but it would be m- it, it is more descriptive- Mm to call it finite angle

Wojciech Wegrzynski 48:30

method. What, you, you've also brought up ray tracing method, which would, which would I guess be the ultimate one if I had infinite

Simo Hostikka 48:37

comp- Ray tracing as, as w- engineers use it is not the ultimate because it's deterministic method. Uh-huh. We choose the directions in advance, and that's the case.

Wojciech Wegrzynski 48:48

Okay. So it's again, user dependent. You have to choose, uh-

Simo Hostikka 48:52

Yeah. What we, I would say is, is the ultimate case is the Monte Carlo method. Okay. And like, uh, it is o- obviously very close to ray tracing, but now we do the choice of the directions- and any other parameter in a stochastic manner- Mm which gives us the freedom or possibility to refine our calculation- Mm-hmm by just increasing the number of samples until we approach actually what is, what we can consider exact solution. H-

Wojciech Wegrzynski 49:21

how, how much you need to, of those, uh, how expensive it is to reach almost the exact solution?

Simo Hostikka 49:27

Well, in the IFSS MACFP benchmarks-

Wojciech Wegrzynski 49:30

Mm-hmm

Simo Hostikka 49:31

the, Chandan Pal and Somesh Roy, who actually did these Monte Carlo calculations, they, said that, okay, after the convergence study, they ended up sending, like for every time step that we did, they sent one, like something like 100 photons from every cell- whether being a wall cell or a gas cell And with those 100 photons, we could reach a kind of convergence. For every cell. It, yeah. It's, it's not a perfect level still. I, I'm sure that we st- still can see kind of a little bit fluctuation and variability here and there. But with the 100 photons, we ended up doing, like, 100 million or 200 million photons for a single- point in time.

Wojciech Wegrzynski 50:14

Well, that, that doesn't sound bad. I mean, when you do sprinklers, we sometimes go to that amount of Lagrangian particles in a model for sprinklers. Yeah, yeah. So that- Yeah that doesn't sound like, impossible. It sounds expensive, but not impossible.

Simo Hostikka 50:25

Yeah. It takes some time, but it's not something you need to wait- But again- for, for a week. It's, it's like a several hours.

Wojciech Wegrzynski 50:31

When, when we, when we did the kitchen hob and we got to, like, 2,000 or 4,000 angles in the discrete ordinates, it was still not cheap either. Like, it And the files were unimaginably large.

Simo Hostikka 50:41

No. And actually, what has been found that if you need to go to the level of thousands of angles in your- finite angle method, then actually the Monte Carlo method starts to be competitive. And I think it's well possible in the future that we replace the finite angle method with the Monte Carlo method because it's giving more flexibility, and it's about the same order of cost.

Wojciech Wegrzynski 51:07

Mm.

Simo Hostikka 51:07

Now, the super high cost in, in, uh, these benchmark calculations come from the fact that in addition to the, this directional sampling, we also do sampling in wavelength. Mm. So at the same time, we consider- Ah, okay the spectral things. And then things become costly. But if we- Kind of do gray or some- simplified treatment of the spectrum, but stochastic thing for the directions- Mm and place, then they are about the same order.

Wojciech Wegrzynski 51:37

ye- yesterday you also mentioned about, um, the, uh, numerical diffusion and effects of the radiation, like kind of going around the corners. That's, that's how i- Yeah like you shoot the light a- and there's an obstacle, the light does not pass it. It doesn't go around it. Yeah. But in- But in CFD it can. Yeah. I think it's very interesting because I know also a lot of people, a lot of our engineers would be studying stuff like petrochemical, uh, plants, et cetera, where, uh, shade is important from complex, you know, pipes and, and lines and everything. So this kind of messes up with that.

Simo Hostikka 52:12

Yeah. It, it would, remove your shades.

Wojciech Wegrzynski 52:15

It does re- yeah.

Simo Hostikka 52:16

Yeah, yeah. uh, but on the other hand, it has helped us to reduce the ray effect.

Wojciech Wegrzynski 52:22

Okay.

Simo Hostikka 52:22

Because it- Okay. Okay kind of smooths out your, our hotspots from the wall. Yes.

Wojciech Wegrzynski 52:27

Yes.

Simo Hostikka 52:27

So that's why we haven't really been eager to reduce this diffusion or, or false scattering- Mm as it's often called.

Wojciech Wegrzynski 52:37

Is, is that, i- is it a product of the finite angle model or w- w- what's causing this? Just-

Simo Hostikka 52:43

I, I think, uh- All flux-based methods like discrete ordinates, finite angle have to have the- Will, will have some Two main error sources, the ray effect and- Mm uh, and diffusion. Because they are numerical solvers. Mm. All numerical solvers have these issues. For the sake of computational speed, we have chosen to use very low order numerics there. Okay. Like, like something, if you know the term first order- Yeah upwind, which is the most simple way of telling, "Oh, okay, what is the value in the next cell?" Well, it's kind of saying that, that just up- In, uh- one, one cell upwind plus something. yeah, then the diffusion is strong. Uh, but as I said, we haven't worried about, about it because it helped us.

Wojciech Wegrzynski 53:28

Mm-hmm.

Simo Hostikka 53:28

But now we have started to see cases where it actually ruins everything. Mm. And, and, and that's something when the radiation needs to attenuate, you know, some medium at very short distance.

Wojciech Wegrzynski 53:42

Mm-hmm.

Simo Hostikka 53:42

Like if you imagine a, a fuel bed o- of vegetation particles.

Wojciech Wegrzynski 53:46

Okay.

Simo Hostikka 53:47

Uh, and you are trying to do wildfire, and you have a, a, a couple of cells, two, three cells, maybe five cells covering your fuel bed, and you should calculate how the radiation attenuates there and then heats up the particles. The low order upwind methods just ruins everything. Mm. And the, the radiation goes through or something without heating of the particles. So there, there we have noticed that we need to do- Mm do better.

Wojciech Wegrzynski 54:11

One

Radiative Fraction Emissivity And Wrap Up

Wojciech Wegrzynski 54:12

final thing I wanted to ask you because, uh, you're an FDS developer, and I think, um, m- most of the audience members are FDS users, so it's very valuable to learn, uh, from you how FDS works. Um, tell me about how FDS figures out the emissivity of smoke in a particular place. Like I, I know I have one gram of soot in this cubic meter of air in my numerical domain. Mm-hmm. How does FDS go from that into knowing how much heat it's gonna emit? I know the temperature, I know the, the amount of soot.

Simo Hostikka 54:44

Yeah. Well, the fundamental physics is just that, okay? We have the absorption coefficient-

Wojciech Wegrzynski 54:50

Mm

Simo Hostikka 54:50

and we have the temperature.

Wojciech Wegrzynski 54:52

Mm.

Simo Hostikka 54:53

We t- multiply absorption coefficient and t- T to the fourth.

Wojciech Wegrzynski 54:57

Mm.

Simo Hostikka 54:57

And yeah, we have kind of the emission power, but yeah, there, there is the practical catch- Mm because we know that our temperature can be very wrong. You know, if you have 15% error in your temperature, you will cause 50 per- Oh, power of

Wojciech Wegrzynski 55:10

four

Simo Hostikka 55:11

50% error in, in T to the fourth. So, uh, all, all the emissions would be 50% off. And that's why I have this concept of radiate fraction that almost all of us use in- Mm these engineering calculations. Instead of trying to predict that T to the fourth- we specify this number- Uh, hope that code reproduces that

Wojciech Wegrzynski 55:37

amount of radiation W- w- h- h- h- how, h- did we measure that number? Like-

Simo Hostikka 55:39

Yeah, we take the number from, handbook or some measurements. Somebody made burn this pool fire or burner in the laboratory, set up an array of heat flux gauges around it, integrated the heat flow, and, and calculated what is, what was the share of radiation compared to the heat release rate.

Wojciech Wegrzynski 55:57

Mm-hmm. And there's approximately that 30% that can let you use for-

Simo Hostikka 55:59

Yeah, yeah, yeah, which, okay, we know it's a kind of rough number.

Wojciech Wegrzynski 56:03

Rough number, but

Simo Hostikka 56:04

yeah.

Wojciech Wegrzynski 56:04

Yeah. And, and s- so what, ev- let's say the, from the temperature calculation you would get like, okay, 60% of energy would be radiated away. So FDS is by this radiant fraction, uh, setting by the user. Mm-hmm. It knows, oh, it should be 30%, not 60. So then it fiddles with the absorption coefficient and, and, uh, reduces it so it matches or?

Simo Hostikka 56:26

Yeah, it makes a correction there that- Correction, okay for the volume which we consider as flame. Mm-hmm. That volume we apply a correction. Elsewhere in the model we wouldn't. Right. And there your soot and temperatures would still be used- Uh, will still- for calculating this energy source or radiation source. I

Wojciech Wegrzynski 56:44

mean, that, that, that, that's fair because also, you know, the treatment of a flame, flame is very tiny compared to the numerical cell. I just interviewed- Mm a part on that. We have a lot on that in the- Yeah previous episode, so.

Simo Hostikka 56:56

Yeah. But here is something that if you are using some other tools than FDS- Mm then what we do in FDS is actually a bit different. The users of Fire Foam or, uh, probably other CFD codes too, they also have the concept of radiant fraction, but it's used in a different way. Mm. It's as far as I understand, it's used in a combination with optically thin radiation modeling, which says that, okay, we s- specify, enforce this kind of loss of heat release rate as radiation, and then all the, uh, kind of participating or, or the, uh, interplay between the gas properties and the intensities is not calculated. We just let this 30% go away. Mm-hmm. And maybe it hits the wall and we get heat fluxes, but there's no any, any, uh, interaction.

Wojciech Wegrzynski 57:49

Mm-hmm.

Simo Hostikka 57:50

What FDS is doing is kind of somehow mixed approach where we do enforce radiance fraction for part of the domain because we don't trust our temperatures. Mm-hmm. But in the rest, we- You allow it will, it, it will be allowed to change, and that is causing sometimes a little bit trouble because even though you specify 30%, what you see in the FDS output file may be something different. Mm-hmm. You see that, okay, actually according to the calculation, the fraction was 25. Mm. So wh- where's the problem? Uh, this now goes to the detail of these interactions.

Wojciech Wegrzynski 58:25

Mm-hmm. Very, very nice. Very nice. Uh, I hope the FDS users appreciate, uh, these additional explanations. And, if you want more, I think y- I think your lectures from the Jülich Summer School are also online available. I'm pretty sure. I, I was checking a few days ago. I, I pr- I, I was on the first edition and I, I still found your lecture out there, so.

Simo Hostikka 58:46

Okay. Very- I didn't know that.

Wojciech Wegrzynski 58:47

I think s- I think they are. So, so, uh, gr- great job Jülich team, and, and, and thanks for you. I, I remember that the-- I still have notes from that lecture. It was really, really good. Simo, uh, there's so much more to cover in the world of radiation, and I hope we'll come back to this discussion to discuss these, um, non-Gray models and all the interesting aspects of, how, we figure out this, this spectral, uh, wavelength, uh, dependency. And we also have to talk about source terms. Uh, so still a lot on the table of radiation, but I hope for this, uh, FAR Fundamentals episode, uh, w- we've covered a lot for the listeners.

Simo Hostikka 59:22

We did. Yeah. I, I, I would love to continue for these subjects. This is something which I'm now hoping to focus rest of my career.

Wojciech Wegrzynski 59:31

And we hope that, uh, we will benefit from effects of that career and that it will bring a lot of, uh, plenary lectures and awards and recognitions for that. Thank you so much, Simo.

Simo Hostikka 59:41

Thank you, Wojciech.

Wojciech Wegrzynski 59:42

And that's it. Thank you for listening. I especially appreciate those tidbits on, uh, how FDS solver works and, uh, you know, all the tricks of the trade that go into us having a reliable solution for a radiation problem in our everyday modeling. This is an effect of two decades of hard work by Simo and others who, who developed these, these tools for us and this is amazing that, uh, those people are not only willing to, you know, build those models for us, but also happy to talk about them in detail and explain them in quite an approachable way. Radiation is like Simo said, it scares him and I can understand why to an extent and I especially appreciate that someone can talk about something this difficult yet in such an approachable way. Um, there's plenty of resources on radiation out there if you want to broaden your knowledge. I hope that in this Fire Fundamentals we were able to at least define the most important parts of, of what constitutes the radiation model. We went from how radiation is emitted by different bodies, how radiation is absorbed by d- different bodies, by gases, by liquids, by solid fuels. We've discussed what's a wavelength spectrum and how with different temperatures you're gonna have peaks at different levels of that spectrum. We've discussed how we are able to simplify the spectrum into black body model, how you can put an emission factor to account for non-perfect emission and absorption, into a gray model and, uh, some even more complicated models where this is wavelength dependent and I promise you we'll come back to that in the future. We've discussed the basis of finite angle model that is commonly used in CFD software, but also discussed discrete ordinates, ray tracing, and Monte Carlo approaches to that. We've briefly touched the view factors. So yeah, that's a lot of stuff on radiation and, uh, I think we've just covered the really, really baseline out there. I hope this was useful for my fellow fire safety engineers. I have certainly appreciated the ability to be able to sit down with Simo and discuss this fascinating matter with the man who just delivered the, you know, invited plenary talk at IFSS on this exact subject. Um, that's it for today's Fire Science episode. I hope you are happy with the content in the recent weeks, and I can only promise you more fantastic episodes with really, really good guests are coming up in the upcoming weeks. I'm working hard to deliver you great content for your summertime so you can enjoy, uh, perhaps at the beach or, uh, at the pub. Well, probably no. Don't, don't do that. Don't enjoy Fire Science show at the beach. Enjoy beach at the beach and enjoy Fire Science show when it's time for that and when you, when like that. Thanks for being here with me today and see you here next Wednesday. More Fire Science coming your way. Cheers. Bye.