March 18, 2026

243 - 20 Informal Settlement Fire Experiments with Sam Stevens

The player is loading ...

A fire in an informal settlement is not just another small building fire. It can be the first domino in a fast-moving neighborhood event, and the little details like the wall material, roof material, door location, even a light breeze, can decide what happens next. I’m joined by Dr. Sam Stevens from Kindling to unpack a massive FSRI funded experimental program carried out in South Africa that burned twenty different informal and humanitarian shelter types to measure real heat flux, flame extension, and fire spread potential.

We start repeating the uncomfortable gap: humanitarian guidance often relies on rules of thumb like a universal separation distance and vague advice to “use fire-resistant materials,” with little experimental evidence behind it. Sam explains how Kindling built a global database of shelter designs and materials, then narrowed it down to common typologies spanning sheet metal, mud, timber, bamboo mats, split bamboo, thatch, and tarpaulin tents. We dig into the field-scale test setup, why wood cribs were chosen, and how external instrumentation like thin skin calorimeters helps quantify the heat transfer that drives structure-to-structure ignition.

Then we get into what the burns actually show. Non-combustible shelters often behave like classic compartment fires where openings dominate, but even modest wind can push flames meters outward. Add combustible walls or roofs and the hazard shifts dramatically: some materials produce short, intense exposure windows while others sustain burning long enough to threaten neighbors at greater distances. The fully thatched shelter stands out for extreme radiant heat and duration, and our discussion on tarpaulin tents reveals why common fire test methods can produce reassuring ratings that still miss real-world behavior.

If you care about informal settlement fire safety, humanitarian shelter design, fire engineering guidance, or modeling large outdoor fires in dense communities, this conversation gives you a data-driven foundation and a clear sense of what questions still need answering.

Learn more about the project and watch their educational videos here: https://kindlingsafety.org/projects/large-scale-fire-experiments-for-humanitarian-shelters/

----
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.

00:00 - Why Informal Settlements Need Fire Data

03:33 - Partnership And Show Support

05:08 - How The Project Started

10:19 - The Gap In Humanitarian Fire Guidance

16:36 - Choosing Twenty Shelter Types

22:05 - Building The Outdoor Test Setup

29:44 - What Non-Combustible Shelters Reveal

33:12 - How Light Wind Changes Flame Paths

37:35 - Combustible Walls And Roof Tradeoffs

48:42 - When Shelters Burn Completely

52:37 - Turning Data Into Practical Guidance

58:54 - Tarpaulin Tents And Misleading Tests

01:07:43 - Open Data Report And Next Steps

01:09:51 - Closing Thoughts And Next Week

WEBVTT

00:00:00.479 --> 00:00:02.479
Hello everybody, welcome to the Fire Science Show.

00:00:02.479 --> 00:00:05.919
Fire physics is complex, beautiful and rich.

00:00:05.919 --> 00:00:13.039
Perhaps the reason I am still a fire scientist and I don't see any better pathway for myself.

00:00:13.039 --> 00:00:26.000
We know a lot about fires and we know that because for more than a hundred years there have been experimental research that was unwrapping these complexities of fires.

00:00:26.000 --> 00:00:30.000
And I would dare to say we know quite a lot about building fires.

00:00:30.000 --> 00:00:46.560
There's some things that we don't know, there's some big things that we don't know, uh but in general we have a lot of knowledge that comes from generations and generations of experimentalists who were looking for answers in the fire.

00:00:46.560 --> 00:01:02.079
And there are spaces in the world of fire which do not have this experimental background at this scale, and uh unfortunately one of those spaces is also the one which perhaps needs it the most.

00:01:02.079 --> 00:01:27.680
That is the informal settlement space and our ability to understand fire behavior at large in those informal or humanitarian settings where people basically build their shelters from whatever they can get, whatever they can source in ways that uh address their current needs and problems and issues.

00:01:27.680 --> 00:01:46.239
They don't have much choice about uh where on how to build them, and unfortunately they often would put themselves in a position where a fire is a hazard, which not necessarily is a hazard that they recognize or choose to to do something about.

00:01:46.239 --> 00:01:49.359
This uh is something that you've already heard in the podcast.

00:01:49.359 --> 00:01:52.480
I had Danielle Antonellis from Kindling twice.

00:01:52.480 --> 00:01:54.719
I had Professor Richard Wolves.

00:01:54.719 --> 00:02:10.080
We've been discussing the the issue with informal settlements and how to provide fire safety to uh had probably one billion of people who who may be living in such settlements as we teased some weeks ago with Danielle.

00:02:10.080 --> 00:02:31.840
This experiments come up because Kinling has completed a very large experimental series burning twenty different informal settlement types of shelters to start and see how does the structure and materials used for the shelter change the overall fire behavior and overall fire outcomes.

00:02:31.840 --> 00:02:44.719
Such a massive project would obviously not be possible without funding and this uh this funding was uh provided by FSRI and uh thanks to FSRI all of this was possible.

00:02:44.719 --> 00:02:50.400
So actually on behalf of Fire Science, I'm really thankful for supporting such an important mission.

00:02:50.400 --> 00:02:56.000
And today in the podcast I have uh a chance to discuss those experiments with Dr.

00:02:56.000 --> 00:03:10.479
Sam Stevens, who's with Kin Ling, who's been there on the ground in South Africa for all of those burns, who's been designing those experiments and uh who for the last uh eighteen something months has been uh playing with the data from those experiments.

00:03:10.479 --> 00:03:15.919
Finally we we can share some of the findings, show some of the outputs.

00:03:15.919 --> 00:03:24.159
We'll discuss how the experiments came to to be and we will discuss where uh this data can go further.

00:03:24.159 --> 00:03:28.319
So without further ado, let's spin the intro and jump into the episode.

00:03:28.319 --> 00:03:34.879
Welcome to the Firescience Show.

00:03:34.879 --> 00:03:38.639
My name is Vojinsky, and I will be your host.

00:03:38.639 --> 00:03:56.639
The Firescience Show podcast is brought to you in partnership with OFR Consultants.

00:03:56.639 --> 00:04:06.639
OFR is the UK's leading independent multi-award-winning fire engineering consultancy with a reputation for delivering innovative safety-driven solutions.

00:04:06.639 --> 00:04:15.759
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.

00:04:15.759 --> 00:04:33.120
So far, we've brought through more than 150 episodes, which translate into nearly 150 hours of educational content available, free, accessible all over the planet without any paywalls, advertisements, or hidden agendas.

00:04:33.120 --> 00:04:40.000
This makes me very proud and I am super thankful to OFR for this long-lasting partnership.

00:04:40.000 --> 00:04:47.360
I'm extremely happy that we've just started the year 4, and I hope there will be many years after that to come.

00:04:47.360 --> 00:04:56.000
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.

00:04:56.000 --> 00:05:03.439
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.

00:05:03.439 --> 00:05:06.160
Check their website at OFRconsultants.com.

00:05:06.160 --> 00:05:07.920
And now let's head back to the episode.

00:05:07.920 --> 00:05:11.680
Hello everybody, I am joined today by Sam Stevens from Kindling.

00:05:11.680 --> 00:05:12.240
Hey Sam.

00:05:12.240 --> 00:05:14.319
Hey Voyak, thanks for having me here today.

00:05:14.319 --> 00:05:16.480
You're very welcome here today.

00:05:16.480 --> 00:05:25.519
And today we're gonna finally discuss the massive experimental program that that was uh carried out uh a year ago, I believe, in in South Africa.

00:05:25.759 --> 00:05:27.040
Yeah, about 18 months, I think.

00:05:27.279 --> 00:05:30.319
18 months uh flies so fast, it's ridiculous.

00:05:30.319 --> 00:05:37.199
But anyway, finally, finally, we get to talk about uh the results and finally you can talk about some findings.

00:05:37.199 --> 00:05:39.120
I know the program quite well.

00:05:39.120 --> 00:05:42.160
The listeners don't know it uh probably at all.

00:05:42.160 --> 00:05:46.800
So let's introduce the subject, which is the informal settlements.

00:05:46.800 --> 00:05:55.120
Maybe you can give me the background, uh how this all started, where did it all start, and and what was the first research idea that you went with?

00:05:55.360 --> 00:05:59.920
Yeah, so this this all started back before I joined Kindling.

00:05:59.920 --> 00:06:04.399
Um so I joined in 2023 after I finished my PhD.

00:06:04.399 --> 00:06:15.279
Um, Kindling had a very generous funding package from FSRI to do some experimental work to look at um informal and humanitarian style shelters.

00:06:15.279 --> 00:06:28.879
So historically, and I mean sort of in the past decade or so, there's been a lot of work done on informal settlements in Cape Town and South Africa specifically, just because it's quite accessible from the academic space.

00:06:28.879 --> 00:06:32.959
There was a very good partnership between Stellenbosch University and Edinburgh.

00:06:32.959 --> 00:06:56.560
And there's a lot of sort of momentum or interest in uh the idea of sort of improving fire safety in the developing world, but it it remained very restricted to sort of one kind of shelter typology, which is essentially these like metal boxes, these sort of corrugated metal style of shelters that you get in South Africa, in a lot of sub-Saharan Africa, and in other parts of the world.

00:06:56.560 --> 00:06:59.439
But the focus was very much on Cape Town.

00:06:59.439 --> 00:07:11.199
But out there in the world, there are people using a huge variety of different materials that we don't really have any data on, sort of shelter designs that we don't really have any data on.

00:07:11.199 --> 00:07:15.680
And they're building them in sort of fairly dense settlements.

00:07:15.680 --> 00:07:24.800
In some cases, it's even humanitarian organizations coming along and trying to help in sort of disaster response, providing materials.

00:07:24.800 --> 00:07:27.279
Um, a lot of the time it's tarpaulin materials.

00:07:27.279 --> 00:07:43.120
So we wanted to move from just understanding these sort of very simple metal um style shelters to actually exploring the vast array of materials that are used, predominantly around the question of like large fire spread in these settlements.

00:07:43.120 --> 00:07:56.240
Uh so we're kind of in the space of like large outdoor fires, but we're not exactly woo-e fire, we're not exactly wildfire because we are talking very much short-range structure-to-structure fire spread.

00:07:56.480 --> 00:08:02.879
I know we previously in uh Edinburgh there was uh Project Iris that was focused on informal settlements.

00:08:02.879 --> 00:08:04.720
There was a lot of good work in that.

00:08:04.720 --> 00:08:12.000
I remember uh specifically paper about 20 shelters, yeah, like large outdoor experiment with 20 shelters burned down.

00:08:12.000 --> 00:08:12.959
That was brilliant.

00:08:12.959 --> 00:08:14.959
There was amazing touchover stuff.

00:08:14.959 --> 00:08:17.839
Uh, I think Mohammed Bashir authored that paper.

00:08:17.839 --> 00:08:20.079
Awesome, awesome, awesome fire physics.

00:08:20.079 --> 00:08:24.800
Uh, but iris was mostly those corrugated steel uh type shelters, right?

00:08:24.800 --> 00:08:29.040
And and this is just one topology out of many, right?

00:08:29.279 --> 00:08:40.879
Yeah, so I mean the first the first step in the work before we even got to the experiments was asking the question of actually what is out there, and and there's no easy literature to assess to like find that out.

00:08:40.879 --> 00:08:52.159
We spent a lot of time just trawling through we don't have informal settlement eurocode that dictates it's it's basically yeah, there's the whole point of like people build it as yeah as they can.

00:08:52.159 --> 00:09:08.000
And we were we were having to go through like report after report uh from the UN or the Red Cross or various different organizations that were like, well, in this year, in this flood response, we provided these materials, and there's a few pictures of what the shelters looked like.

00:09:08.000 --> 00:09:35.840
So we're coming, we've got a a database now that's sort of got, I think, about 450 different shelter records entered into it that are you know from I think 90 different countries in the world, and anything, you know, mud shelters, that shelters, that is bamboo, um, but all these different materials that you know people harvest locally or find locally, or that's what they have available to them, that's what they build their homes out of.

00:09:36.159 --> 00:09:41.440
I I think the listeners already uh had a good experience in the fire science show uh hearing about that.

00:09:41.440 --> 00:09:45.120
We we had Danielle Twise, we had Richard Walls uh talking about the background.

00:09:45.120 --> 00:10:01.840
So we I think we the common understanding is that there is potentially large fire hazard related to those informal settlements, and uh unfortunately solving those issues is not the top priority for the people who are under the hazard, they have uh other things they they worry about.

00:10:01.840 --> 00:10:16.960
Anyway, um, in terms of defining those shelters or managing the landscape, do we even have before the project starts, do we even have some sort of like guidance recommendations on how to safely build them?

00:10:16.960 --> 00:10:19.039
Like what does the landscape look like?

00:10:19.039 --> 00:10:24.960
And then hopefully at the end of the podcast episode, we can discuss how the landscape looks after we we finished this.

00:10:25.279 --> 00:10:26.559
Yeah, that's a great question.

00:10:26.559 --> 00:10:29.840
And I guess that sort of frames the research really well.

00:10:29.840 --> 00:10:36.559
Um, there isn't really fire safety guidance for, I'd say we call it the the humanitarian sector.

00:10:36.559 --> 00:10:45.600
Obviously, there are informal settlements existing in countries with regulatory frameworks, it's just the regulation isn't necessarily applied in those spaces.

00:10:45.600 --> 00:10:50.879
But then there is this sort of humanitarian response space where there's lots of organizations working.

00:10:50.879 --> 00:11:07.519
Um, and that can be anything, as I say, from oh, this small disaster has happened, and we just need to supply a few materials to help people recover their homes to massive displacements of people and organizations from the ground up building camps as quick as they can.

00:11:07.519 --> 00:11:11.440
Now, there is in existence or various guidance documents.

00:11:11.440 --> 00:11:15.679
One of them, as far as I'm aware, one of the most commonly used is called the Sphere Handbook.

00:11:15.679 --> 00:11:28.159
It's this massive document, it's like 450 odd pages long, uh, that is supposed to advise camp practitioners on sort of all aspects of how you should build a refugee camp, basically.

00:11:28.159 --> 00:11:37.279
And for you know, all kinds of drivers, like prevent against flooding, to provide adequate lighting, to provide adequate toilets, like literally everything.

00:11:37.279 --> 00:11:37.679
Yep.

00:11:37.679 --> 00:11:46.639
And within this, you know, hundreds and hundreds of pages, there's a total of nine lines of text committed to the subject of fire safety.

00:11:46.639 --> 00:11:48.480
And it's kind of as you say, right?

00:11:48.480 --> 00:12:00.960
So it's it's not that people don't necessarily care, it's that people have potentially 101 other things that they perceive to be, or maybe even are, more important than fire safety.

00:12:00.960 --> 00:12:10.000
So that fire safety kind of is a little add-on at the end where you try and you know do what you can, but it's it's not really the driving factor.

00:12:10.000 --> 00:12:20.320
Um, and with that in mind, the guidance for fire safety ends up kind of being, I would say, pretty rudimental and not very data driven.

00:12:20.320 --> 00:12:31.279
And it's just kind of loose rules of thumb that have been passed into these guidance documents without any kind of data to support their use.

00:12:31.279 --> 00:12:32.879
Can you give ex examples?

00:12:32.879 --> 00:12:35.039
Yeah, I can give a couple of examples.

00:12:35.039 --> 00:12:39.120
Um, the the most obvious is probably like shelter separation distance.

00:12:39.120 --> 00:12:46.720
Because this there's this thing of, well, if you put two shelters X meters apart, then you know that improves fire safety.

00:12:46.720 --> 00:12:54.559
Doesn't say it will stop fire spread, it doesn't say it will slow fire spread, it doesn't say what those two shelters have to be made of.

00:12:54.559 --> 00:13:01.759
It's just for any two shelters, if you have them and the in the shelter guidances, if you have them two meters apart, you're doing well.

00:13:01.759 --> 00:13:06.879
Even better, you should have them at least double the height of the shelters apart.

00:13:06.879 --> 00:13:16.320
And now that's getting, you know, for any reasonably sized shelter for people to live in, that's probably you're looking at four or five meters as your minimum separation distance there.

00:13:16.639 --> 00:13:20.639
Which doesn't really fit to the image I have in my mind of these uh neighborhoods.

00:13:21.039 --> 00:13:30.480
Yeah, so space is at a premium, you're giving people this very high target of what fire safety is by saying you need all your shelters to be four or five meters apart.

00:13:30.480 --> 00:13:33.759
But if you can't do that, then actually you can cut it to two meters.

00:13:33.759 --> 00:13:45.679
But it doesn't then contextualize that with well, actually, if I have a shelter, you know, that we can build out of concrete bricks, uh, that two meters, you know, fine, we don't really care.

00:13:45.679 --> 00:13:47.519
They can be half a meter apart, maybe.

00:13:47.519 --> 00:13:56.399
And you know, as non-combustible material, you're probably okay, versus, oh, actually you're building everything out of timber and thatch and bamboo.

00:13:56.399 --> 00:13:57.519
Yeah, whatever.

00:13:57.519 --> 00:14:05.679
So, like, you know, there's no contextual shift in the guidance for what materials are um or so what materials are being used.

00:14:05.679 --> 00:14:13.360
And you know, that's even kind of summarized in this other line that says use fire-resistant materials where possible, like if you can.

00:14:13.360 --> 00:14:20.000
It doesn't provide any further context as to what a fire-resistant material actually is.

00:14:20.000 --> 00:14:24.879
Like, is there some qualifying factor that makes a material fire resistant?

00:14:24.879 --> 00:14:28.320
No, just use a fire-resistant material if you have one to hand.

00:14:28.320 --> 00:14:40.080
And I think it's reasonable to some extent because you can imagine you know, people working on the ground have some innate concept of like what burns versus what doesn't burn.

00:14:40.080 --> 00:14:54.399
But that's very different to than to having you know a nice bank of experimental data to kind of compare and contrast different materials and say, um, well, if you have bamboo mat, then this is going to be your common fire spread outcome.

00:14:54.399 --> 00:14:58.320
Um, or if you're using thatch, then this is going to be your common fire spread outcome.

00:14:58.320 --> 00:15:08.320
So we've been approaching the analysis of the experimental data from is like, well, what how do people want to approach what fire safety practice looks like on the ground?

00:15:08.320 --> 00:15:14.240
And then how do we speak to each of those, you know, in rules that people are sort of using already?

00:15:14.720 --> 00:15:22.399
I would say, you know, Sam, if I look at that, I also feel like those guidances have to be very simple.

00:15:22.399 --> 00:15:35.279
Those guidances probably cannot relate to a specific testing standard because if people are obtaining materials in whatever way they can, the standardization is not gonna really solve.

00:15:35.279 --> 00:15:40.240
Like you cannot put a E C E marking on anything, on everything for this context.

00:15:40.240 --> 00:15:45.840
So I I think there's general guidance, those are actually not the worst guidances I've heard.

00:15:45.840 --> 00:15:50.080
Like make a two-meter separation, make a use fire resistance.

00:15:50.080 --> 00:16:06.080
But uh, as you said, like they're probably insufficient in terms of what they mean, and and and what you've brought before the lack of experimental or just lack of proof that it actually provides you specific uh result in terms of safety.

00:16:06.720 --> 00:16:15.600
I completely agree, and that's I mean, we can maybe circle back to that in the end because that's been a real challenge in how we've then used the data that we've analyzed.

00:16:15.600 --> 00:16:25.440
We've gone into all this complexity and we want to you know share that complexity, but at the same time, the people we want to share it with and the end users, they don't have time for the complexity.

00:16:25.440 --> 00:16:36.159
So then how do we reframe our analysis to be communicated in simple terms that is better than what exists already but isn't so complex that nobody has the time for it?

00:16:36.240 --> 00:16:48.240
Um so the data you said 450 different shelter topologies uh from what uh Danielle's Paul's here uh last time she's been on the podcast, is it was, I believe, 20 different settlements.

00:16:48.240 --> 00:16:55.120
So, uh, how did you narrow down the 450 down to 20 and what they are broadly representative of?

00:16:55.679 --> 00:17:01.120
There was no fancy statistical methods, and I will say that it wasn't like 450 unique shelter types.

00:17:01.120 --> 00:17:04.880
So it was across the shelter database were then well, what's the most common?

00:17:04.880 --> 00:17:12.559
Okay, and particularly across the way we kind of broke it down was across different countries and then different uh climate regions as well.

00:17:12.559 --> 00:17:22.799
So out of that, I mean you do have the odd wacky shelter of really unorthodox materials that appears once in one report, yeah, um, that you don't then see again.

00:17:22.799 --> 00:17:28.319
But there are materials that are obviously very common uh across lots of different places.

00:17:28.319 --> 00:17:35.920
So we ended up kind of 12 to 15, maybe most common, and we wanted to sort them into categories as well.

00:17:35.920 --> 00:17:58.240
I guess going back to the idea of like how do you make guidance simple is if if we could through experiments show that you can like loosely categorize different materials or different shelters within the same car in the same categories and say, well, there's the sort of a risk level inherent with each of the different categories, then that would potentially simplify things.

00:17:58.240 --> 00:18:05.599
It turns out it has to be a little bit more complex than that, but the initial intention was to group shelters by categories as well.

00:18:05.599 --> 00:18:15.279
So uh the the categories we went for were combust, or so it was grouped by the combustibility of walls um and roofs.

00:18:15.279 --> 00:18:15.519
Okay.

00:18:15.519 --> 00:18:20.160
By combustibility, I just mean does it burn or does it not, uh, in simple terms.

00:18:20.160 --> 00:18:33.359
So the first category being non-combustible walls, non-combustible roofs, things like sheet metal walls, sheet metal roof, um, or we had mud walls and a sheet metal roof and cement block and a sheet metal roof.

00:18:33.519 --> 00:18:33.839
Okay.

00:18:34.079 --> 00:18:35.119
Um, as an example.

00:18:35.119 --> 00:18:37.839
So non-combustible or non-combustible, more or less.

00:18:37.839 --> 00:18:46.000
And then and then you kind of get the idea the next category is combustible walls with non-combustible roof, so that would be like a timber walled shelter with a sheet metal roof.

00:18:46.000 --> 00:18:50.480
Then we had uh non-combustible wall but combustible roof.

00:18:50.480 --> 00:18:58.000
So an example being uh mud walled with a thatch roof, and then both combustible walls and roofs.

00:18:58.000 --> 00:19:00.160
Uh, how many categories is that for we're at?

00:19:00.160 --> 00:19:06.720
And then the fifth category we ended up with was tents because we didn't really know how where they were gonna sit.

00:19:06.720 --> 00:19:25.599
So tarpaulin tents, there's a lot of interest in the sector, obviously, with tarpaulins, because that's a a huge amount of what the humanitarian sector distributes is basic tarpaulins to you know either repair homes or to build tents completely out of tarpaulin.

00:19:25.599 --> 00:19:28.640
Um, sometimes they provide full tent kits.

00:19:28.640 --> 00:19:34.319
Uh, so we weren't really sure whether they were gonna sit in a combustible category.

00:19:34.319 --> 00:19:38.880
Um, and tarpaulins are very complex and a whole different story that we can maybe come on to later.

00:19:38.880 --> 00:19:43.680
But that was the sort of the loose categories and materials.

00:19:43.680 --> 00:19:46.960
It was actually quite a narrow set of materials in the end.

00:19:46.960 --> 00:19:51.200
We narrowed it down to like timber, sheet metal, bamboo mat, split bamboo.

00:19:51.200 --> 00:19:56.799
So if you like take the bamboo column and split it down the middle, thatch and and then tar pollen.

00:19:56.799 --> 00:20:04.559
So like the common shelter types that exist across the world they are relying on a fairly narrow set of materials, like whatever's naturally occurring.

00:20:04.559 --> 00:20:17.599
You have various different species of bamboo, fine, you have tons of different species of timber, yes, yeah, fine, but as a as a material, timber versus split bamboo versus like these sort of woven bamboo style mats are quite easy distinctions to make.

00:20:17.599 --> 00:20:29.359
But you didn't use any kind of like insulative material like sandwich panels or no, so that that was a typology that that did come up in the database quite a bit.

00:20:29.359 --> 00:20:58.079
Is there are styles of humanitarian humanitarian shelter, pre pre-built kind of container style uh shelters that often the walls and and roof are made out of these insulated sandwich panels, but we didn't really want to go down that route because I don't think we had confidence we could necessarily build one of those ourselves from just sourcing material, and then it was a case of approaching the organizations that build these things and be like, hey, can we have one and burn it?

00:20:58.079 --> 00:21:08.720
And then of course they're reluctant because they don't want their thing to be shown up as potentially dangerous, and that's not to say they necessarily would be compared to other styles of shelter.

00:21:09.039 --> 00:21:15.519
I mean in this particular research problem, I I don't think it's fair calling any of those materials dangerous.

00:21:15.519 --> 00:21:18.240
Well, perhaps some are, but we we'll get back to that later.

00:21:18.240 --> 00:21:33.119
But uh, you know, it's more about none of them are really like fully safe in the ways how how it's built, because like if you would like to have like a fully safe building, you'd build it to a building code, uh, uh and then it would stop being the informal settlement.

00:21:33.119 --> 00:21:48.559
And in here, instead of like pretending like you should do everything according to a code, we'll like I I like what when Danielle said that we um don't really have much to offer to those who don't meet the minimum requirements of the code, like that there's not much you know below.

00:21:48.559 --> 00:21:52.160
Well, in reality, there's a whole spectrum of solutions that can be there.

00:21:52.160 --> 00:21:58.240
Okay, we got the materials, we got the kind of topologies or combinations of those materials.

00:21:58.240 --> 00:21:59.680
So we were testing.

00:22:00.319 --> 00:22:04.960
A single uh settlement or again multiple uh just single, right?

00:22:04.960 --> 00:22:08.319
Yeah, so every experiment was just a single shelter.

00:22:08.319 --> 00:22:16.000
Um we use the the ISO room dimensions, so very similar to how the the Iris fire project set out their experiments.

00:22:16.000 --> 00:22:18.240
So there's what 2.4 by 3.6?

00:22:18.240 --> 00:22:22.559
Yes, 2.4 by 3.6, and then 2.4 high as well.

00:22:22.559 --> 00:22:25.279
Um wood cribs, two wood cribs in.

00:22:25.279 --> 00:22:34.480
And then I sort of I alluded to it at the beginning, but that this is really a a problem of short-range fire spread between shelters.

00:22:34.480 --> 00:22:49.599
So what we're primarily interested in is heat transfer to the immediate surroundings in all directions, which meant uh a huge amount of external instrumentation, particularly to measure the heat fluxes.

00:22:49.599 --> 00:22:56.880
The important, or you know, we wanted to assess the relevance of we had sort of a door and window in every shelter.

00:22:56.880 --> 00:22:57.440
Okay.

00:22:57.440 --> 00:23:02.160
Um so traditionally, compartment fire, you're expecting external frame from the opening.

00:23:02.160 --> 00:23:06.400
So we focused in there, we we had some heat flux gauges or a heat flux gauge at each one.

00:23:06.880 --> 00:23:07.440
On the same facades?

00:23:07.440 --> 00:23:08.240
The door on the window?

00:23:08.480 --> 00:23:10.720
No, so they were on opposite facades.

00:23:10.720 --> 00:23:11.119
Okay.

00:23:11.119 --> 00:23:22.400
Yeah, so we they they were both in the so if you've got a 3.6.2.4 uh floor plan, the door and window were in opposite corners on the 3.6 meter long walls.

00:23:22.480 --> 00:23:22.720
Okay.

00:23:22.960 --> 00:23:26.880
Yeah, so then a huge number of uh thin skin calorimeters.

00:23:26.880 --> 00:23:40.400
I think we had 60 thin skin calorimeters around the shelter in kind of these spokes coming away from the shelter at various positions, with the TSEs being at one, two, and three meters away.

00:23:40.400 --> 00:23:57.759
So we're assessing heat transfer, I guess, at a variety of distances away from the shelter and at all locations to kind of assess well, what's the difference between heat transfer from an opening and heat transfer from an end wall and heat transfer like along the diagonal looking at the shelter?

00:23:57.759 --> 00:24:05.519
Because there just really isn't the data out there to kind of differentiate between different styles of shelter.

00:24:05.519 --> 00:24:09.119
And going back again, I'll reference the Irish fire projects.

00:24:09.119 --> 00:24:12.160
Again, they have these sheet metal shelters.

00:24:12.160 --> 00:24:18.880
The most common place to be measuring heat flunks is from the opening because that's where your external flames are going to be.

00:24:18.880 --> 00:24:31.200
But a real question around the experiments we did was well, beyond just external flames from the openings, are there particular hazards associated with different materials and different styles of shelter?

00:24:31.200 --> 00:24:36.400
And can we get fire spread from in different locations, not just at the openings?

00:24:36.400 --> 00:24:47.279
I think the very obvious answer to that question when we talk about the results is yes, because if you have a combustible wall that's completely uh in flames, then there's going to be a huge amount of heat transfer from that wall.

00:24:47.279 --> 00:24:50.319
Same with a roof, if you have a roof opening up.

00:24:50.319 --> 00:24:50.960
Yeah.

00:24:50.960 --> 00:25:05.359
And I guess uh again, I'll talk about this more in the results, but there's a question of like to what extent can you stick to conventional compartment fire dynamics as your understanding of how a shelter burns.

00:25:05.359 --> 00:25:16.960
But to come back to uh what we're measuring, I should expand a little bit and say these are um these experiments are all happening in a field on a farm in the middle of rural South Africa.

00:25:16.960 --> 00:25:27.119
So we had to come up with some reasonably innovative ways of measuring things that you know might be a lot easier under a calorimetry hood in a lab.

00:25:27.119 --> 00:25:32.880
So the entire shelter was built on a very large platform scale.

00:25:32.880 --> 00:25:38.000
So I think it was sort of slightly larger than the floor plan of the shelter itself.

00:25:38.000 --> 00:25:40.240
I think it was about three by four meters.

00:25:40.240 --> 00:25:46.079
Um so we but we were measuring mass loss to I would say do a robust measurement of heat release rate.

00:25:46.079 --> 00:25:53.119
Of course, it's not going to be robust if you have parts of the shelter falling off and various different things like this, and no calorimetry.

00:25:53.119 --> 00:25:56.799
But it was it was very useful when we look at the data sort of on the back end.

00:25:56.799 --> 00:26:04.400
It was very useful in gauging the different orders of magnitude of fire size between different shelters and different uh materials.

00:26:04.880 --> 00:26:06.400
What's the relevancy of the crib?

00:26:06.400 --> 00:26:09.839
You mentioned you put two cribs inside as your source.

00:26:09.839 --> 00:26:11.279
What was it based on?

00:26:11.279 --> 00:26:18.559
Just to get the threshold, or is it was it representative of some specific fuels in the in the in those types of settlings?

00:26:19.039 --> 00:26:25.920
It is, I think, going in what we use as a kind of a standard experimental fuel load, right?

00:26:25.920 --> 00:26:32.480
So they were sized and designed to be an approximate fuel load of uh like an informal home.

00:26:32.480 --> 00:26:38.000
There is some fairly limited data describing fuel loads in informal homes.

00:26:38.000 --> 00:26:44.079
I think we were sort of on the slightly higher end of that range to generate worst-case scenario.

00:26:44.079 --> 00:26:51.279
I remember conversations we had before, because because I mean ITB were very much involved in the some of the planning of these experiments.

00:26:51.279 --> 00:27:07.039
I remember conversations we had where you were theorizing about having gas burners, and I think the particular decision around cribs over gas burners that was that we wanted the fuel load to be able to interact with the shelter as it's burning.

00:27:07.039 --> 00:27:08.319
That's what you can do.

00:27:08.319 --> 00:27:09.119
Yeah.

00:27:09.119 --> 00:27:19.519
So it wasn't just a constant controlled heat release rate from the cribs, but that the feedback between the fuel load and the the shelter was significant.

00:27:19.519 --> 00:27:32.799
We could potentially have maybe gone a bit further down the route of more realistic fuel loads, but at the end of the day, cribs are just more repeatable in a sense than putting you know whatever realistic looking furniture.

00:27:32.799 --> 00:27:49.519
And there was, I mean, there was a conversation within Kindling about making experiments relatable to the people that we want them to affect, because people that live in these environments they won't they won't necessarily look at an experiment and see something that looks like their home.

00:27:49.519 --> 00:28:01.119
So if you fill it with you know a sofa and a bed and some furniture instead of cribs, then it's more relatable to the average human that isn't a fire scientist.

00:28:01.119 --> 00:28:09.599
Whereas, you know, for for us and maybe for listeners, it's kind of intuitive that you look at an experiment and there's a wood crib there, you're like, yeah, that makes sense to me.

00:28:09.599 --> 00:28:10.480
That's making sense.

00:28:10.480 --> 00:28:25.680
So there was there was everything discussed on the spectrum from like concentrate gas burners to wood cribs to maybe we actually do put real fuels in, but I think cribs was the sort of natural selling point for us.

00:28:26.000 --> 00:28:29.599
In the end, you already have 20 different you know experiments to run.

00:28:29.599 --> 00:28:36.079
And if you add variability of fuel into the mix, that makes uh the design of experiments very challenging.

00:28:36.079 --> 00:28:50.559
I asked this follow-up question, you know, because uh not that many people do large-scale fire experiments, and but a lot of fire, awfully lot of fire engineers, every fire engineer is dependent on the results of those experiments because that's what what we get to design stuff later on.

00:28:50.559 --> 00:29:06.640
So I really love uh you know to pull uh fire researchers on their logic behind those seemingly like basic choices for the experimental design, but uh they have they sometimes have a very long-lasting ripple effect in the in the community where they are later used.

00:29:06.640 --> 00:29:10.640
So it's very interesting to to understand where where did those kick came from.

00:29:10.640 --> 00:29:13.279
Um let's perhaps discuss some uh findings.

00:29:13.279 --> 00:29:21.680
I I wonder if I I think we've covered the measurements, so you said a lot of thin skin colorometers plus some diagnostics around the along the openings.

00:29:21.680 --> 00:29:26.559
I know you had a lot of cameras because you've said like a terabyte of videos, yeah, which I appreciate.

00:29:26.559 --> 00:29:40.319
I'm not sure you do appreciate it, but uh no, I I am I am but but there there was a video recording system so we could also figure out some stuff regarding the flame lengths and and and flame exposures.

00:29:40.319 --> 00:29:44.880
I know from those videos that those cases were very different, so let's jump into the data.

00:29:44.880 --> 00:29:47.519
So, uh, what did you actually find?

00:29:47.519 --> 00:29:57.200
Maybe let's start with the simplest one, the non-combustible ones, and perhaps let's relate them against what was already known about those types and and then build up on the flammability.

00:29:58.160 --> 00:30:03.839
Yeah, so uh and the non-combustible ones are a very nice base case because we had the data from Iris.

00:30:03.839 --> 00:30:24.160
The first shelter we burned is uh a sheet metal shelter, and then our our next two were sort of mud walled, so we had this sort of wattle and daub style construction where we have a you know wattle wood frame, but then big thick mud layer plastered over it, and then these sort of cement blocks as well as our non-combustible walls, in every case, a sheet metal roof.

00:30:24.160 --> 00:30:31.839
And the I guess the nice and easy finding from these is they behave as you would expect a compartment fire to behave.

00:30:31.839 --> 00:30:43.119
So you have a bit of growth in the cribs, and then you get flash over, fully involved compartment fire, and your dominant fire spread hazard is flames from the openings.

00:30:43.119 --> 00:30:49.759
The slight caveat to that is with these corrugated sheet metals, the corrugation you've obviously got.

00:30:49.759 --> 00:30:54.960
I think in the past we've called them sort of flutes, where it's it's not a perfectly sealed compartment.

00:30:54.960 --> 00:31:08.559
There is a little bit of flame leakage, especially with the sheet metal walls, as they're sort of heating, they can warp a bit, and you can get gaps elsewhere where there is some flame leakage.

00:31:08.559 --> 00:31:19.920
So if we're theorizing about a real settlement and when you have shelters very, very close together, it is possible that some of that flame leakage might impact upon neighboring shelters.

00:31:19.920 --> 00:31:31.279
We were measuring heat fluxes at one meter away as a as sort of the minimum distance that we were measuring, and we didn't really pick up the effects of flame leakages at that distance.

00:31:31.279 --> 00:31:36.960
But we did have a we were using perfectly brand new, nice, you know, good quality corrugated metal.

00:31:36.960 --> 00:31:48.720
I appreciate in real settlements, these sheets get weathered over time, or they'll reuse old ones, and the the quality of construction might be uh less than ideal, so you might get a lot more of this flame leakage.

00:31:48.720 --> 00:31:53.039
But that aside, we're talking fairly conventional compartment fire dynamics.

00:31:53.279 --> 00:31:57.200
But you still pick up quite significant fluxes from the door and window plumes, right?

00:31:57.359 --> 00:31:58.319
Uh the opening, yes.

00:31:58.319 --> 00:32:02.000
And this is also, you mentioned the the cameras and the video we took.

00:32:02.000 --> 00:32:05.519
We're also looking at flame extension from shelters.

00:32:05.519 --> 00:32:13.359
So across what distance are we projecting flames, um, that we might get flame impingement from one shelter to the next.

00:32:13.359 --> 00:32:22.319
And we're also picking up heat fluxes within the flame if that the flame is impinging on our instrumentation, which it was in a number of cases.

00:32:22.319 --> 00:32:28.000
But generally it's about characterizing yes, what is that kind of field of heat flux around the openings.

00:32:28.400 --> 00:32:35.680
As a comment, perhaps we didn't mention that, but those uh flux gouges were not just like lying on the ground, they were also like elevated.

00:32:36.079 --> 00:32:42.960
So we had them uh two meters above the ground level at the floor level of the shelter and 2.4 meters.

00:32:42.960 --> 00:32:54.960
So that was equivalent to the the height of the top of the openings and the height of the roof, and we're measuring at those heights at one, two, and three meters away from the shelter.

00:32:55.200 --> 00:33:07.599
Uh okay, so if they behave like a normal far compartment would, and you have this uh complexity of window and the door at the same time, was there anyone any experiment in which you would have a wind-influenced flow?

00:33:07.599 --> 00:33:12.160
Like it would turn into like one-directional flow or something?

00:33:12.480 --> 00:33:14.000
Yep, that's a great question.

00:33:14.000 --> 00:33:21.839
And I mean, we aimed to test in as low wind conditions as possible, but you know, large outdoor testing, it's it's not unavoidable.

00:33:21.839 --> 00:33:27.119
And actually, some of our more interesting findings is that it really, really doesn't take much wind at all.

00:33:27.119 --> 00:33:37.119
We're talking sort of sub one meter per second to be able to notice the influence of wind, particularly the case when you start adding combustible materials into the mix.

00:33:37.119 --> 00:33:47.440
For our sheet metal shelter, certainly there was more pronounced flame from the door than the window, just the direction of wind that day.

00:33:47.440 --> 00:33:49.279
Well, it wasn't particularly high.

00:33:49.279 --> 00:33:56.240
I think it was about 1.5 meters per second on average, which we didn't regard high.

00:33:56.240 --> 00:33:57.119
That's not much at all.

00:33:57.119 --> 00:34:06.000
And you know, it was enough to kind of kill the flames on the window side, but then push the flames to kind of two or three meters from the door on the door side.

00:34:06.000 --> 00:34:18.000
The other two were at lower wind speed and were more balanced, but you know, even in the course of a what 10-minute experiment or 10-minute fully developed phase, we've got slightly blustery wind outside.

00:34:18.000 --> 00:34:22.400
We'll have you know moments where it's more pronounced at the window, then more pronounced at the door.

00:34:22.400 --> 00:34:29.440
And so if you look into the the heat flux data, you can kind of see that where you have periods of slightly elevated heat flux.

00:34:29.440 --> 00:34:33.760
So, um, yes, wind is a factor, quite an important factor.

00:34:33.760 --> 00:34:43.360
And we did do a couple of experiments with in higher wind speeds as well, one of those being a repeat of our sheet metal shelter.

00:34:43.360 --> 00:34:57.360
We basically had uh a few days of uh bad weather during the experiments, and there were days where we were like, we have to do something, and there was days where we had really high winds to the point we couldn't set our instrumentation out safely.

00:34:57.360 --> 00:35:01.920
Okay, like it was just falling over, and it was you know, it was all kinds of carnage on site.

00:35:01.920 --> 00:35:06.079
But um, we're like, well, we'll just get some wind data, even if it's just visual.

00:35:06.079 --> 00:35:12.000
So we settled the we set up the sheet metal shelter again in what were we at?

00:35:12.000 --> 00:35:14.559
I think we were sort of five meters per second on average.

00:35:14.559 --> 00:35:24.079
So when I was doing going back over all the over the or all the data, I was trying to convert my wind speeds to what it what is it on the Beaufort scale.

00:35:24.079 --> 00:35:24.559
Okay.

00:35:24.559 --> 00:35:27.679
And it's still something like, oh, it's a it's a light breeze.

00:35:27.679 --> 00:35:35.679
Yeah, when you're there, it really felt like you know, you're properly nice gusty winds, but it still wasn't significant.

00:35:35.679 --> 00:35:37.920
We were sort of five meters per second on average.

00:35:37.920 --> 00:35:41.360
I think it we had gusts up to 12 meters per second.

00:35:41.360 --> 00:35:46.559
And when the days we had stronger winds, it was also very heavily unidirectional.

00:35:46.559 --> 00:35:56.320
So we would I think when we did the sheet metal, we had just this sort of jet of flame out of the windows, probably three to four meters in length.

00:35:56.559 --> 00:36:04.159
So the practicality of this those findings for non-combustible one is is that the location of those openings would highly matter.

00:36:04.480 --> 00:36:04.960
Yep.

00:36:04.960 --> 00:36:08.800
Super, super important if you've got non-combustible materials.

00:36:08.800 --> 00:36:18.159
If you've got, you know, as we'll come on to, if you've got combustible materials, you can't necessarily trust that your highest hazard location is going to be your near your openings.

00:36:18.159 --> 00:36:33.039
But you know, that's an immediate, immediate, easy finding for us, is that yes, we can show you the data that says opening placement is extremely important and is something that is not considered in any sort of humanitarian guidance yet.

00:36:33.039 --> 00:36:38.400
For reference, for those non-combustible ones, how how long did it take the the experiment?

00:36:38.400 --> 00:36:56.079
After ignition, we were usually about three to four minutes to flash over, and then probably 10 to 15 minutes in fully developed, inevitably things would start collapsing, roofs would fall in, the the mud shelter going down was interesting.

00:36:56.079 --> 00:37:02.000
We had mud walls pancaking in all kinds of directions, which was was fun to clean up.

00:37:02.000 --> 00:37:12.400
I can imagine, but but yeah, I guess there's a there's a question there of does collapse affect fire spread hazard?

00:37:12.400 --> 00:37:21.599
And from these experiments, I would say no, not particularly, is the the highest hazard conditions that are most likely to affect spread are you know immediately post-flashover.

00:37:21.599 --> 00:37:30.719
And you know, if your fire is gonna spread, it's gonna spread significantly sooner than it would usually take for one of these shelters to collapse.

00:37:30.719 --> 00:37:34.800
They were you know fairly robust for 10, 10, 15 minutes, as I say.

00:37:35.519 --> 00:37:37.360
Let's move to to some more complex.

00:37:37.360 --> 00:37:44.480
So I uh you said you had non-combustible walls and combustible roof, you had combustible walls, non-combustible roofs.

00:37:44.480 --> 00:37:49.760
How how these combinations of of partial combustibility look like and what changed?

00:37:50.239 --> 00:37:52.880
So I'll start with combustible walls.

00:37:52.880 --> 00:38:01.679
The significant variable, which is fairly intuitive, really, is the thickness of whatever the combustible material is.

00:38:01.679 --> 00:38:08.559
So we did all with sheet metal roofs, we did timber, split bamboo, and bamboo mat walls.

00:38:08.559 --> 00:38:22.960
Um, and the fire develop into the early stages look very much like a compartment fire transitioning through flashover, except as you're going through flashover, all your walls will ignite and burn externally.

00:38:22.960 --> 00:38:30.480
But the key difference is the duration for which they remain burning or remain sort of structurally sound.

00:38:30.480 --> 00:38:33.920
So I think that the timber walls lasted the longest.

00:38:33.920 --> 00:38:35.199
It wasn't particularly thick.

00:38:35.199 --> 00:38:42.960
We had like 12 mil thick planks, so I think you know, even it's likely that thicker planks are used in reality as well.

00:38:42.960 --> 00:38:58.400
But they lasted probably like four to five minutes, where we have this our shelters burning as I would say as a post-flash over compartment fire, with the additional impact of the external walls all being a light and burning away as well.

00:38:58.400 --> 00:39:01.760
So the openings are still important in those cases.

00:39:01.760 --> 00:39:20.800
We find that the heat fluxes were still significantly higher at the openings because you have a lot of the fuel gas which is you know being released internally, still venting out and burning as it's mixing with the oxygen at the door and window, but with the addition of all the external walls also burning.

00:39:21.119 --> 00:39:25.920
And and the light wood materials like the bamboo mats, I would assume they would be gone fairly quickly.

00:39:26.239 --> 00:39:31.440
Yeah, and so that's that's the important thing is the bamboo mat was something like 20 to 30 seconds.

00:39:31.440 --> 00:39:32.480
Seconds.

00:39:32.480 --> 00:39:34.079
Yeah, yeah, it was gone.

00:39:34.079 --> 00:39:38.079
Like ignited very, very easily, uh burned away very easily.

00:39:38.079 --> 00:39:42.800
And the the the kind of you end at the same place with all these experiments was quite a nice thing.

00:39:42.800 --> 00:39:51.519
You end with your this sort of freestanding sheet metal roof over a couple of wood crypts that are burning because the walls are are gone.

00:39:51.519 --> 00:39:59.199
And it was just the the rate at which those materials burned away was sort of the decisive factor in the hazard of the shelter.

00:39:59.199 --> 00:40:19.519
Now, this becomes a more interesting question when we start to think about modeling the ignition of a hypothetical neighboring shelter because the thickness of your burning material or your your sheltered material not only influences how long it burns, but if it was on an exposed neighboring shelter, how quickly that material can ignite.

00:40:19.519 --> 00:40:35.039
And so at the same time we were doing these shelter scale experiments, all the materials that we were using, we had there was sort of a commercial uh fire lab in Cape Town that was doing cone calorimetry on all our materials for us so we could get some ignition data.

00:40:35.039 --> 00:40:55.199
Now, if I look at something like uh the timber, if I pair my experimental like heat flux exposures that we got from all our RTSEs with the sort of ignition data and try and model the ignition time of the timber at those exposures, you end up with you know, at varying distance, it's fairly fairly rapid fire spread.

00:40:55.199 --> 00:41:04.159
But say across the region from one to three meters away from the shelter, I have like a steadily decreasing predicted ignition time.

00:41:04.159 --> 00:41:12.639
But you compare that to, I think, so like the bamboo mat is a very good example of you have this very short, sharp hazard of the shelter burning.

00:41:12.639 --> 00:41:24.719
So it's maybe 20 or 30 seconds long, but that's all that it takes for a neighboring bamboo mat shelter to ignite, but only if it's within like a meter, a meter and a half.

00:41:24.719 --> 00:41:47.119
So for two different materialities of shelter, if you if we're looking at separation distance as a as sort of an isolated problem, is that for timber, say, we have this nice gradient of as you increase separation distance, you are steadily increasing ignition time to a point where the separation distance is sufficient that no fire spread can happen.

00:41:47.119 --> 00:41:50.000
But that distance is like three, maybe more meters.

00:41:50.000 --> 00:42:06.400
But with bamboo mat, you have this very sort of binary problem of either it's within this sort of one meter perimeter and it's gonna ignite very, very all its neighbors super quickly, or it's gonna be just outside that zone and it's not gonna ignite at all.

00:42:06.400 --> 00:42:18.880
And so it's not just that materiality changes the required separation distance, it's that materiality changes the whole context of the way you have to think about separation distance.

00:42:18.880 --> 00:42:30.480
So is it a binary yes, no, fire can spread or fire can't spread, or is it a sort of a whole spectrum of risk that you have to be thinking about across a large range of separation distances?

00:42:30.719 --> 00:42:40.000
Yeah, well, this is a brilliant uh topic for a follow-up, I guess, to see how what would happen if you actually had a row of those, you know.

00:42:40.000 --> 00:42:45.440
Like I could imagine like Project Iris 20 settlement thing bad with different types of those.

00:42:45.440 --> 00:42:53.519
Well, that probably would we would most likely prefer to do it with uh with CFD, not uh with a full-scale experiment that would be awfully expensive.

00:42:53.519 --> 00:43:07.840
But yeah, this is a very interesting, like almost like what was better to to have like a longer steady fire but less ignitable material, or a short-lived one, but very prone to any source of ignition.

00:43:08.079 --> 00:43:15.199
And that's a I mean, that's a great point to raise because I was just about to go on to talking about combustible roofs but non-combustible walls.

00:43:15.199 --> 00:43:15.679
Okay.

00:43:15.679 --> 00:43:19.599
And it's another it's another perfect example of one of these trade-offs.

00:43:19.599 --> 00:43:30.000
So anytime we had a combustible roof, even if it was thatch, which burned like absolute crazy, we could completely eliminated external flames pretty much every time.

00:43:30.000 --> 00:43:35.519
Because very early on, you're creating a pathway for all your hot gases to vent upwards rather than outwards.

00:43:35.519 --> 00:43:38.480
Oh, so no more window plume, no more door bloom.

00:43:38.480 --> 00:43:39.039
Yep.

00:43:39.039 --> 00:43:45.119
Everything becomes a very sort of vertical wind wind knot sort of included in this analysis.

00:43:45.119 --> 00:43:53.360
But you get this lovely vertical plume, your roof catches fire, nice established burning on your roof, and you get this big tall plume above the shelter.

00:43:53.360 --> 00:43:56.159
But of course, this is a very lateral fire spread problem.

00:43:56.159 --> 00:43:59.599
We're thinking about shelters next to each other spreading fire to one another.

00:43:59.599 --> 00:44:02.639
Not a shelter spreading fire up somewhere, right?

00:44:02.639 --> 00:44:20.079
So we go from having these very intensely concentrated hazards at the openings that are spitting fire like right in the direction of where our neighboring shelter could be, to all our heat is actually being released above the shelter, but then radiating back to the shelters below it.

00:44:20.079 --> 00:44:27.360
Now, which is a more desirable or less so that's bad language, but which which which do you prefer in a settlement context?

00:44:27.360 --> 00:44:36.480
Do you want roofs that are venting flames upwards, or do you want roofs that are non-combustible and make sure that your openings are spitting flames sideways?

00:44:36.480 --> 00:44:45.599
And again, the only way you can really evaluate an answer to that question is if you know what hypothetical material you're igniting.

00:44:45.599 --> 00:44:56.400
And you can add the fact that to have to prevent your flames of venting out through the openings, you need a combustible material on your roof, which might be otherwise replacing a non-combustible material.

00:44:56.400 --> 00:45:02.320
You know, you've got a whole other layer of, well, then that combustible material could become an ignition pathway for that shelter.

00:45:02.320 --> 00:45:04.559
So there's trade-offs like this everywhere.

00:45:04.559 --> 00:45:08.000
There's trade-offs of do you want a material that burns fast or slow?

00:45:08.000 --> 00:45:12.000
Do you want a roof that burns through and vents everything upwards?

00:45:12.000 --> 00:45:22.159
And because the data for those combustible roof experiments showed us that exactly that, that you get lower on average, you're getting lower heat fluxes.

00:45:22.159 --> 00:45:25.199
So lower peak heat fluxes, but you're getting them everywhere around the shelter.

00:45:25.199 --> 00:45:31.920
Okay, versus the doors where you're getting much higher peak heat fluxes, but you're getting them in very specific locations.

00:45:32.320 --> 00:45:44.960
I mean, it is very challenging to also make some like final conclusion out of this data set, because obviously, if you would have strong wind and open the roof, then you have no idea how bad it would be.

00:45:44.960 --> 00:45:46.960
Yeah, potentially it could be very bad.

00:45:46.960 --> 00:45:47.360
Yeah.

00:45:47.360 --> 00:45:51.440
And two, firebrand production and how firebrands spread.

00:45:51.440 --> 00:46:02.000
Because I would also imagine that if you have like just a vertical plume emerging from a house, I've created a lot of firebrands in my life by accident in fire experiments.

00:46:02.000 --> 00:46:11.760
I know every little piece of foam, a fabric in your compartment, a little uh splinters of timber that's gonna separate from anything.

00:46:11.760 --> 00:46:18.239
There's a lot of things that can be captured by the plume and carried out of the building and they become a firebrand.

00:46:18.239 --> 00:46:22.880
Not necessarily just your no timber firebrands that we're used to in multifires.

00:46:22.880 --> 00:46:27.039
So so basically, every compartment is a generator of firebrands.

00:46:27.039 --> 00:46:39.679
And I think that this hazard could be amplified if it's just you know vertical, because it's easier to carry stuff out, whether rather when you have you know plumes projecting to some extent horizontally from your opening.

00:46:39.679 --> 00:46:52.480
It's a hypothesis, it's not that it's not that I know that, but I would hypothesize that this could be a potential factor, especially if you consider it in the as a part of a dense neighborhood where you know that those things can land, right?

00:46:52.880 --> 00:46:58.559
Yeah, and that's really that kind of takes us to the uh limitation of of the work we've done so far.

00:46:58.559 --> 00:47:05.599
And I like one of the questions I'm really interested in is that we burn all these individual shelters, but what does a fire really look like when it's multiple?

00:47:05.599 --> 00:47:10.239
And you've you've you've referenced uh Iris's 20 shelter experiment.

00:47:10.239 --> 00:47:26.480
I think it's maybe with the closest we've got, but what does what's what's the difference between some shelters on fire and a conflagration where the the behavior of the fire is independent of the individual behavior of any single shelter, and you just have a mass of fuel burning because the thing's so big.

00:47:26.480 --> 00:47:30.639
Um, and I think firebrands might be an important part of that question.

00:47:30.639 --> 00:47:42.880
Certainly, you see that I think with wildfire where you get firebrand deposition uh or the the combination of firebrand deposition and radiation from the firefront sort of combining to progress that firefront.

00:47:42.880 --> 00:47:44.320
It's a it's a tricky one.

00:47:44.320 --> 00:47:51.920
In it there's there's very few sort of known instances of like spot fires occurring in these kind of settlements.

00:47:51.920 --> 00:47:58.800
I'm not saying that's because they they don't happen, I think they do, but it's very specific to which materials are in involved.

00:47:58.800 --> 00:48:12.000
Okay, and certainly we weren't we weren't doing anything to measure firebrand production in our experiments, but you could see which like for instance the thatch and the bamboo mat, you know, produce huge numbers of firebrands.

00:48:12.000 --> 00:48:15.599
I can't be more quantitative than that, but like visually you could see it.

00:48:15.679 --> 00:48:19.840
Yeah, but also your choice of using the crib as your fuel source.

00:48:19.840 --> 00:48:34.960
Like, I would be convinced that the firebrands are not just products of you know your construction materials, but literally anything in your room could potentially yield firebrands, you know, paper, clothes, any any type of foam material, etc.

00:48:34.960 --> 00:48:41.920
But uh anyway, I'm I'm taking you away from from the important things that we've we've observed in this project.

00:48:41.920 --> 00:48:47.840
How about the the the fully combustible structures, the the the ones you've mentioned touch on thatch?

00:48:48.079 --> 00:48:49.119
Again, very interesting.

00:48:49.119 --> 00:49:02.719
I think it's the best example of where I talked earlier about the idea of categorizing shelters to make it simple for everyone, is where categorization really falls apart, is because this is where the the most diverse behavior sort of occurred.

00:49:02.719 --> 00:49:16.480
So we had three in this category that were bamboo mat with a thatch roof, bamboo mat with a split bamboo roof, and thatch walls with a thatch roof.

00:49:16.480 --> 00:49:24.559
So with both the bamboo mat walled ones, obviously, as I've mentioned before, the bamboo mat ignited very quickly, burned away very quickly.

00:49:24.559 --> 00:49:28.400

243 - 20 Informal Settlement Fire Experiments with Sam Stevens

Listen On

Recent Episodes

Support On