Dec. 9, 2025

230 - Wind driven conflagration experiments with Faraz Hedayati

The player is loading ...

A facility with 105 synchronized fans pushing hurricane-class wind across a full-size house while a live fire... This is not science fiction - this is a real research capacity that helps us re-shape our knowledge on the full scale building ignition, fire spread, and failure. That’s the stage at IBHS, where we dig into how wind-driven fire behave differently to small-scale and how tiny choices around a building can decide its fate. Together with my guest - dr Faraz Hedayati, we go from embers generation and fire spread studies, to urban conflagration research.

We start with embers, the quiet culprits behind so many structure losses in the WUI. Embers aren’t a single threat but a spectrum of sizes, temperatures, and lifetimes that ride shifting eddies and stall in stagnation zones. We talk through what full-scale tests reveal: glowing ember lines at the base of walls, roof reattachment zones where deposits spike, and the hard truth that counting particles matters less than controlling where they land. The guidance is clear and actionable—noncombustible vertical clearance, hardened vents, defensible space within the first five feet—because under wind, any component can become the first domino.

Then we tackle conflagration: how a spot fire becomes a neighborhood problem. IBHS’s shed-to-structure and fully furnished burns show exposure arriving in pulses, not a smooth curve. Collapse chokes flames and then reinvigorates them, creating multiple peaks where materials succeed or fail on a timer. We compare 30 mph to 60 mph winds and see how plumes lose buoyancy, flatten into the target, soften vinyl frames, and push glazing inward. Separation distance emerges as a decisive lever: around 10 feet, continuous flame contact dominates; at 20 feet and beyond, exposure becomes intermittent and materials can win—unless “connected fuels” like vehicles, fences, and decks bridge the gap.

The takeaway isn’t a silver bullet. It’s a layered defense: control embers, clean the near-wall zone, harden openings, choose noncombustible claddings, and increase spacing where possible. Small-scale testing and modeling still matter, but wind-driven fire demands validation at full scale to catch the peaks, the collapses, and the failure modes no bench setup can mimic. If you care about wildfire resilience, urban design, or building safety, this conversation offers a rare, data-rich look at how communities ignite—and how we can change the odds.

Learn more about IBHS research at https://ibhs.org/risk-research/wildfire/

Cover picture courtesy of dr Faraz Hedayati.

----
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 Scale Matters In Fire Science

04:28 - Meet IBHS And The Giant Wind Tunnel

08:45 - What Full-Scale Enables And Limits

12:20 - Instrumentation Challenges At Scale

15:42 - Safety, Boundaries, And Test Design

19:15 - Why Embers Drive Most Ignitions

24:10 - The Ember Generator And Wind Interaction

29:05 - Chaos Of Ember Transport And Accumulation

33:40 - Building Aerodynamics And Stagnation Zones

37:30 - Goals: Insurers, Science, And Safer Homes

41:40 - From Embers To Spot Fires To Spread

46:05 - Defining Conflagration And Separation Distances

50:30 - Two Peaks, Collapse, And Exposure Dynamics

WEBVTT

00:00:00.320 --> 00:00:02.959
Hello everybody, welcome to the fire science show.

00:00:02.959 --> 00:00:10.240
Being a fire researcher myself working in the laboratory, I learned one thing while doing those experiments.

00:00:10.240 --> 00:00:16.800
There are phenomena or things in fire science in general for which the scale really matters.

00:00:16.800 --> 00:00:26.800
You can learn a lot from experiments on material scales, and I wonder how many PHDs over the world were done just on the cone calorimetry, and that's that's perfectly fine.

00:00:26.800 --> 00:00:35.039
That's some absolutely beautiful fire science there where you can use those aparatures to come up with clever things.

00:00:35.039 --> 00:00:41.119
But in reality, some phenomena, some things really show up in a grander scale.

00:00:41.119 --> 00:00:47.920
And sometimes it's not even about phenomenon, it's about how they interact with each other or which becomes the dominant one.

00:00:47.920 --> 00:00:53.039
And now if you want to do large scale fire science, it's really hard.

00:00:53.039 --> 00:01:00.479
It's really hard to get those experiments planned, it's really hard to get them approved, it's really hard to get them conducted.

00:01:00.479 --> 00:01:08.319
They usually end up expensive, and if you do them in the field, you often lose control over your variables.

00:01:08.319 --> 00:01:21.680
To have a laboratory that allows you to run full scale fire experiments, I'm talking building scale fire experiments with perfectly controlled pundit conditions, that's the dream.

00:01:21.680 --> 00:01:31.920
And well, there are people living this dream in the world of fire science, and there is one facility in which something unbelievable is possible.

00:01:31.920 --> 00:01:35.200
Full scale wind driven fire experiments.

00:01:35.200 --> 00:01:41.760
Well I mean full scale wind, full scale fire, full scale buildings, just as they are in reality.

00:01:41.760 --> 00:01:48.159
The laboratory is the one of the Insurance Institute for Business and Home Safety, IBHS.

00:01:48.159 --> 00:01:50.560
And I've invited a colleague from there, Dr.

00:01:50.560 --> 00:02:04.799
Faraz Hedayati, to discuss the amazing experiments carried out in the IBHS and what more can you learn in those large scale experiments versus running smaller scale data.

00:02:04.799 --> 00:02:12.000
And importantly, there's space for everything, and small scale is absolutely necessary to move the science forward.

00:02:12.000 --> 00:02:17.360
And at the same time, if we had a bigger facility than IBHS, perhaps we could even observe something more.

00:02:17.360 --> 00:02:28.479
Anyway, this is highly relevant to uh wildfires, this is highly relevant to wildland urban interactions, but it's also important to any type of wind-driven fires.

00:02:28.479 --> 00:02:32.879
And as we know in the building fire safety, we have plenty of those.

00:02:32.879 --> 00:02:38.560
Uh let's learn the insights from the large-scale experiments from Faraz Hadeyati.

00:02:38.560 --> 00:02:41.759
Uh let's spin the intro and jump into the episode.

00:02:41.759 --> 00:02:47.599
Welcome to the Fire Science Show.

00:02:47.599 --> 00:02:51.439
My name is Wojciech Wegrzynski, and I will be your host.

00:02:51.439 --> 00:03:20.479
The Fire Science Show is into its third year of continued support from its sponsor OFR consultants, who are an independent multi-award-winning fire engineering consultancy with a reputation for delivering innovative safety-driven solutions.

00:03:20.479 --> 00:03:34.240
As the UK leading independent fire consultancy, OFR's globally established team have developed a reputation for preeminent fire engineering expertise with colleagues working across the world to help protect people, property, and the planet.

00:03:34.240 --> 00:03:50.479
Established in the UK in 2016 as a startup business by two highly experienced fire engineering consultants, the business continues to grow at a phenomenal rate with offices across the country in eight locations from Edinburgh to Bath and plans for future expansions.

00:03:50.479 --> 00:03:59.280
If you're keen to find out more or join OFR consultants during this exciting period of growth, visit their website at OFRConsultants.com.

00:03:59.280 --> 00:04:01.039
And now back to the episode.

00:04:01.039 --> 00:04:02.080
Hello everybody.

00:04:02.080 --> 00:04:07.360
I am joined today by Faraz Hedayati from Insurance Institute for Business and Home Safety.

00:04:07.360 --> 00:04:09.439
Hey Faraz, good to have you in the podcast.

00:04:09.599 --> 00:04:10.800
Yep, good to be with you.

00:04:11.120 --> 00:04:13.199
Wow man, I am jealous.

00:04:13.199 --> 00:04:23.839
I am jealous of the facilities at IBHS, and uh let's start the interview with that so people understand what I am jealous about.

00:04:23.839 --> 00:04:26.160
Can you introduce me to IBHS, please?

00:04:26.560 --> 00:04:27.120
Absolutely.

00:04:27.120 --> 00:04:32.079
Yeah. Uh the Insurance Institute for Business and Home Safety, or IBHS.

00:04:32.079 --> 00:04:35.519
We are a nonprofit located in Richburg, South Carolina.

00:04:35.519 --> 00:04:38.079
It's the rural side of uh Carolinas.

00:04:38.079 --> 00:04:41.120
Uh the closest gas station is like a 20-minute drive.

00:04:41.120 --> 00:04:41.839
Oh my god.

00:04:41.839 --> 00:04:50.000
As the audience uh Google and look at the facility, you see why we have uh a very large test chamber.

00:04:50.000 --> 00:04:54.800
We are known for our uh the capabilities that we have to study wind-driven fire.

00:04:54.800 --> 00:05:01.360
And uh so IBHS uh is a non-profit organization, uh again uh funded by the insurance industry.

00:05:01.360 --> 00:05:08.079
And uh we have 105 fans, and each of them uh the size of them is about six feet.

00:05:08.079 --> 00:05:09.279
Apologies for the unit.

00:05:09.839 --> 00:05:11.360
We'll we'll manage, no worries.

00:05:11.600 --> 00:05:17.519
So 105 of those, and they can generate we speed from 10 miles an hour to 120 miles an hour.

00:05:17.680 --> 00:05:18.319
Oh shit.

00:05:18.480 --> 00:05:20.000
So a cat three hurricane.

00:05:20.160 --> 00:05:21.600
Wow, I did not know that.

00:05:21.600 --> 00:05:25.759
I just knew they were like large and plenty, but I didn't know that they can go that far.

00:05:26.000 --> 00:05:34.079
Yeah, it's been about 10 years that I've been working at IVHS, and each time still, as of today, I step into the test chamber and look up and I'm like, wow.

00:05:34.240 --> 00:05:36.560
How how big is the is the test chamber itself?

00:05:36.560 --> 00:05:39.759
Because the I'm not sure if if pictures give it justice.

00:05:39.759 --> 00:05:42.399
Is it like 100 feet by 100 feet larger?

00:05:42.639 --> 00:05:53.439
No, it's like four uh basketball feeds uh you can put inside so the wind jet is about 80 feet by 30 feet, so 80 feet wide, 30 feet tall.

00:05:53.680 --> 00:05:54.480
Okay, yeah.

00:05:54.480 --> 00:06:00.800
And when you put a building in there, can you cover you you cannot cover the entire focus field of the wind, right?

00:06:00.800 --> 00:06:03.519
So what was the maximum building size you can put in there?

00:06:03.759 --> 00:06:06.319
So it depends on the blockage ratio that we want to test.

00:06:06.319 --> 00:06:09.839
We typically can put a two-story building in the chamber safely.

00:06:09.839 --> 00:06:10.480
Wow.

00:06:10.639 --> 00:06:19.519
So this like brings uh a complete different level to fire research because I I do not know a single facility that that can do it, yeah, other than you.

00:06:19.920 --> 00:06:26.560
So the wind side of the equation, we have a lot of wind engineers at IBHS that help us to study wind-driven fire.

00:06:26.560 --> 00:06:29.920
And to your point, the interaction between wind and fire.

00:06:29.920 --> 00:06:34.319
I'm one of those scientists that stay on the site that it cannot be scaled down properly.

00:06:34.480 --> 00:06:34.800
Yeah.

00:06:34.959 --> 00:06:38.639
Uh so fully scale testing is always needed to understand the fire dynamics.

00:06:38.879 --> 00:06:39.680
Absolutely, absolutely.

00:06:39.680 --> 00:06:43.199
Well, uh, I wrote that in my reviews about wind and fire.

00:06:43.199 --> 00:06:46.800
It's it's simply not possible, they do not scale in the same way.

00:06:46.800 --> 00:06:53.839
Uh and also the best way of scrolling fires where we use the fruit number, it's kind of good for buoyancy driven.

00:06:53.839 --> 00:07:01.279
But once you have forced wind-driven fires, there's I don't think there's a good capacity of of scaling that.

00:07:01.519 --> 00:07:03.519
Yeah, I I completely agree with that.

00:07:03.519 --> 00:07:06.560
And if there is, to some extent it's possible.

00:07:06.560 --> 00:07:11.199
You cannot go from like a full scale to 10 centimeter fire and expect that that would happen.

00:07:11.439 --> 00:07:13.439
But do you also do a smaller scale?

00:07:13.439 --> 00:07:18.000
Because of course you do the giant scale, the the biggest scale uh experiments.

00:07:18.000 --> 00:07:22.160
But you'd also have like a materials department, a small scale department, or not really?

00:07:22.399 --> 00:07:22.800
Yeah, we do.

00:07:22.800 --> 00:07:26.480
We have quant calorimetry testing at IBHS and smaller scale tests.

00:07:26.480 --> 00:07:30.079
So uh we don't want everything to be fully scale until it's needed.

00:07:30.079 --> 00:07:33.439
So because it's very pricey, as you know, to run these experiments.

00:07:33.439 --> 00:07:40.480
So we do a lot of due diligence on the modeling side, on the smaller scale side, and build up to these full scale experiments.

00:07:40.720 --> 00:07:46.319
Uh okay, like this is probably the most boring uh episode because I'm asking for technical details.

00:07:46.319 --> 00:07:48.000
How much electricity does it take?

00:07:48.000 --> 00:07:49.360
Like, I'm really curious.

00:07:49.680 --> 00:07:50.560
That's a good question.

00:07:50.560 --> 00:07:56.480
So, uh, when the fans are at full power, electricity use can power up 9,000 homes.

00:07:56.639 --> 00:07:58.399
Wow, I can I can imagine it's a lot.

00:07:58.399 --> 00:07:59.199
So sorry, listen.

00:07:59.199 --> 00:08:04.000
I'm like I it's like when I go to sleep, I dream of that wind tunnel.

00:08:04.000 --> 00:08:06.800
Yeah, you're invited.

00:08:06.800 --> 00:08:08.160
Fantastic, fantastic.

00:08:08.160 --> 00:08:14.879
Um, when you do uh fire experiments at that facility, what kind of fire diagnostics you catch?

00:08:14.879 --> 00:08:20.480
Because I imagine the wind is like perfect, but in terms of of fire, it must be quite difficult.

00:08:20.480 --> 00:08:23.360
You are you able even to capture like oxygen.

00:08:23.360 --> 00:08:26.000
I you you you cannot do oxygen chlorometer with that, right?

00:08:26.160 --> 00:08:27.199
No, we cannot, yeah.

00:08:27.199 --> 00:08:30.319
It's uh force conviction that all the smoke goes out.

00:08:30.319 --> 00:08:37.200
Uh, we actually thought about that to sample from the smoke at some point, and WPI uh have some study on that topic.

00:08:37.200 --> 00:08:42.080
Uh it's on a research agenda to do it, but at the moment we don't want it to research it.

00:08:42.240 --> 00:08:45.600
And mass loss rates uh uh and our dynamic scales.

00:08:45.600 --> 00:08:46.399
Do you use that?

00:08:46.639 --> 00:08:49.360
We do we do have uh like large scale mass loss platforms.

00:08:49.360 --> 00:09:01.679
Actually, in a study in 2020 that is published uh with NEST, we put a 10,000 pounds, something like that, structures on a mass loss platform and ignited gold and uh got the mass loss uh rate.

00:09:02.000 --> 00:09:04.720
And I promise this is the last one.

00:09:04.720 --> 00:09:08.480
I'm like, we're gonna go into fun fire engineering things.

00:09:08.480 --> 00:09:10.960
I just need to satisfy my curiosity.

00:09:10.960 --> 00:09:13.840
Are you like limited by the size of the fire?

00:09:13.840 --> 00:09:17.120
At what point do you start getting worried about the fire size?

00:09:17.360 --> 00:09:18.159
That's a good question.

00:09:18.159 --> 00:09:20.080
It depends when you ask that question.

00:09:20.080 --> 00:09:23.919
Four years ago, running the experiments that we are running right now was a dream.

00:09:23.919 --> 00:09:27.519
Uh but now we run fully scale structured residential buildings.

00:09:27.519 --> 00:09:30.159
Uh, I would say there are two factors.

00:09:30.159 --> 00:09:33.200
One is the size of the jet, right?

00:09:33.200 --> 00:09:36.480
It's about 80 feet by 10 feet, uh by 30 feet, right?

00:09:36.480 --> 00:09:38.159
So we can't go larger than that.

00:09:38.159 --> 00:09:40.799
Other one is the safety side of the equation.

00:09:40.799 --> 00:09:45.759
Yeah, we we make sure that everything is as safe as possible before running these experiments.

00:09:45.759 --> 00:09:48.320
So we get a lot of supports from our local fire departments.

00:09:48.320 --> 00:09:49.600
Uh, Richford, kudos to them.

00:09:49.600 --> 00:09:51.200
Richford South Carolina Fire Department.

00:09:51.440 --> 00:09:52.799
Something tells me they love it.

00:09:52.799 --> 00:09:54.240
Uh okay, faras.

00:09:54.240 --> 00:09:59.919
Um, there's a lot of interesting research from IB just to fulfill multiple podcast episodes.

00:09:59.919 --> 00:10:04.960
I would uh love to discuss the stuff that that's probably only possible in your laboratory.

00:10:04.960 --> 00:10:06.720
And those are two things.

00:10:06.720 --> 00:10:18.879
Those are the the fire brands research and how uh how firebrands spread fires, and uh the other ways that that fire spread themselves uh in in larger like configuration style configurations.

00:10:18.879 --> 00:10:22.159
I will like we've agreed to start with the firebrands.

00:10:22.159 --> 00:10:29.759
So maybe maybe we could start with when did you when did you start looking into fire brands cost ignition?

00:10:29.759 --> 00:10:37.360
Like something tells me there's a background story to why this topic was raised and and and basically how how do you do the the firebrand experiments?

00:10:37.600 --> 00:10:42.320
So most of the audience uh of your wonderful podcast know that some of the history about this.

00:10:42.320 --> 00:10:52.720
So when back in the 70s, some Spanish researchers from Europe moved to Missoula Faiwa and worked with uh scientists in uh Missoula and studied fire rands, right?

00:10:52.720 --> 00:10:55.840
So that was one peak in in firebrand study.

00:10:55.840 --> 00:11:01.120
And it was somewhat that was the available knowledge until like 2000 when NIST picked this topic up.

00:11:01.120 --> 00:11:11.600
And when they generated the first generation of ember generators as what we know today, there are other versions of it existed before 2000 when the Missoula Finland was working on it.

00:11:11.600 --> 00:11:26.639
And uh when starting in 2010, uh when the embryo generator uh that NIST uh basically built uh was out there, IBHS realizes that the interaction between these embers and wind is something that cannot be stuck.

00:11:26.639 --> 00:11:37.279
And uh when the uh dragon was shipped into uh Japan and there were some wind driven ember studied there, but it was limited to the size of the facility, right?

00:11:37.279 --> 00:11:39.120
And at IBHS back then, Dr.

00:11:39.120 --> 00:12:00.799
Steve Colbs, uh so my PhD advisor, when he retired, it took two people to fit his shoes at IBHS, took the lead on uh studying embers at IBHS, and we generate or we built eight, ten or more emberators, put them in the wind tunnel, and then looked at the interaction in full scale of how embers interact with that solid body, with the blast body, which is this charge.

00:12:00.799 --> 00:12:12.960
And a lot of those information about how embers penetrate into the buildings or where do they accumulate around buildings, all of those are basically the product of that era of uh research at RDHS.

00:12:13.279 --> 00:12:16.960
That Spanish research was that Tarifa or Yeah, one of them was Tarifa, yeah.

00:12:16.960 --> 00:12:20.080
Okay, that is the name I I I recognize.

00:12:20.080 --> 00:12:24.159
And about the the firebrand generation, you used the name dragon.

00:12:24.159 --> 00:12:27.440
I I guess the listeners may not be familiar with that uh equipment.

00:12:27.440 --> 00:12:30.320
So can you tell me something about the generator itself?

00:12:30.559 --> 00:12:45.840
Yeah, so uh the previous generation of these ember generators were just like a burning body that was like moving around, or like a fire that with a fan that creates a convection convective column that naturally deposits embers and they attack different building components.

00:12:45.840 --> 00:12:47.759
What was the previous version of it?

00:12:47.759 --> 00:12:50.399
This was based on the natural development of fire.

00:12:50.399 --> 00:13:09.039
And then uh what NIST scientists did back in 2000, they uh basically feed uh with uh with an auger system, ignite it in a burrow, and then as the material or as the wood particle lose mass, there is a fan at the base that creates that convective column and they shoot uh embers out.

00:13:09.440 --> 00:13:21.600
So basically, the upgrade is that it's suddenly the stream is controlled and repeatable, and the experiments become much more like uh scientific variable control rather than an outcome of a natural fire?

00:13:21.840 --> 00:13:23.200
Uh yeah, yeah, that's it.

00:13:23.200 --> 00:13:24.399
That's a great way to put it.

00:13:24.559 --> 00:13:28.799
Uh so now we move uh some years ahead, we're 2025.

00:13:28.799 --> 00:13:31.279
Some years have passed since this was discovered.

00:13:31.279 --> 00:13:33.600
What's the importance of this research today?

00:13:33.600 --> 00:13:36.080
Like, why do you keep researching that?

00:13:36.320 --> 00:13:37.519
Yeah, that's a good question.

00:13:37.519 --> 00:13:48.960
So when we go in the field after these disasters and tragedies that we keep seeing disproportionately higher in the US compared to other countries, we see that embers are the leading cause of ignition.

00:13:48.960 --> 00:14:06.080
When we look at direct and indirect ignition, so as these embers accumulate around the structures, shrubs, decks, or fences, whatever you want to call it, when those ignite, the fire can rapidly go from a few inches or a few centimeters fires all the way up, engulf the whole structure.

00:14:06.080 --> 00:14:13.039
And that's where the chain of configuration starts, when those spot fires have the opportunity to become very large.

00:14:13.039 --> 00:14:22.080
So when we see those in the field, we come back to IVHS and then we design experiments to understand how that dynamic takes place.

00:14:22.080 --> 00:14:31.759
And uh we see that like uh we can on the engineering radius, we can have different wind flows or wind speeds, uh, different size of structures in front of it.

00:14:31.759 --> 00:14:49.759
And as ignite uh or as they uh attack the building, one of the key moments in my uh career was I'm I'm deep into image processing side of the equation, and I try to track these embers in a controlled environment of 10 miles an hour wind, not heavily turbulent.

00:14:49.759 --> 00:15:01.279
And it is impossible to track these embers and count how many going through the vents, how many bounce back when they hit the target, how many of them not just fall at the base of the beginning.

00:15:01.279 --> 00:15:10.960
And the conclusion of this was if it's this chaotic in a controlled environment, how can we take opinions or stretch this into the real fee?

00:15:10.960 --> 00:15:29.200
So those concepts were like in through years uh shaped into the development of uh Wi Fi Prepared Home, a program that we have, that we actually, instead of targeting a specific structure uh components of a building, we address a holistic uh structure.

00:15:29.200 --> 00:15:49.519
So defensively space around the building, vents, the vertical cure is on the base of the wall, uh requirements for decks, all of them need to be in place because it's such a chaotic system that we cannot single out a specific component and say decks are more vulnerable than vents, or vents are more vulnerable than base of the wall.

00:15:49.519 --> 00:15:51.279
This whole thing needs to be together.

00:15:51.600 --> 00:15:57.039
I I think I found it on the on the webpage, wildland, fire embers and flames, or mitigation that matter.

00:15:57.039 --> 00:16:04.720
I will link the the materials in the show notes because they must contain a ton of useful uh indications.

00:16:04.720 --> 00:16:06.960
How do you how do you judge that in that case?

00:16:06.960 --> 00:16:17.120
So if you if you're not like so you're not interested in saying how much protection a different strategy gives, you're interested in the complete outcome, like overall as a combination of those.

00:16:17.120 --> 00:16:20.159
How do you judge it then what's better, what's worse?

00:16:20.559 --> 00:16:27.759
In my opinion, for specifically for embers, it is nearly impossible to quantify how many embers attack different components of a bead.

00:16:27.759 --> 00:16:35.679
The way that I look at embers is uh like like a smoke, it's a particle with zero mass, it's just at the mercy of the flow, right?

00:16:35.679 --> 00:16:41.519
Embers do have mass, so they have some inertia to fight back with the wind patterns, right?

00:16:41.519 --> 00:16:52.480
So because embers have different distribution of size, then uh some eddy sizes can move a specific size of embers, and larger ed sizes can move other size of the embers, right?

00:16:52.480 --> 00:17:00.799
So it's impossible to put all of them together and say, like again, like decks are less important than vets and cannot quantify it.

00:17:00.799 --> 00:17:04.960
For frames it's different, but for embers, uh in my opinion, it's nearly impossible.

00:17:05.200 --> 00:17:06.160
This is super interesting.

00:17:06.160 --> 00:17:09.599
Uh I think I for the first time I'm having such a conversation.

00:17:09.599 --> 00:17:26.000
So in the end, when the building is subject to distribution of embers, like uh different sizes of embers will be dangerous for the ventilations for HVAC where they can penetrate, and different will be important for the porch, let's say, fire safety, and different will be important for the gutter.

00:17:26.240 --> 00:17:27.039
Yeah, absolutely.

00:17:27.039 --> 00:17:29.680
So uh it's such a chaotic system.

00:17:29.680 --> 00:17:36.880
Uh so uh the interaction between the wind eddy size and the embered ones always dominates red lumber accumulates.

00:17:36.880 --> 00:17:39.519
And both of these are variables, none of them are constant.

00:17:39.759 --> 00:17:42.240
Yeah, but but how about the building aerodynamics?

00:17:42.240 --> 00:17:43.759
Because you said the eddies matter.

00:17:43.759 --> 00:17:55.279
In that case, the the building aerodynamics, the way how eddies shed at the edges of the of the building, uh, how the flow occurs around the building, that must be hell of important as well, right?

00:17:55.519 --> 00:17:56.400
Yeah, absolutely.

00:17:56.400 --> 00:18:05.039
And uh that's one of the important reasons that IBHS and other research organizations talk about zero to five foot clearance around the around the building.

00:18:05.039 --> 00:18:09.440
We come up with the stagnation of the flow, embers accumulate in that area.

00:18:09.440 --> 00:18:15.759
And when it comes to the roof, we can actually clearly see the uh reattachment lengths on the roof.

00:18:15.759 --> 00:18:19.680
So embers jump over that and then you see uh where they land on the on the roof.

00:18:20.079 --> 00:18:36.000
A hypothesis I once had is that you probably could make a link if you if you had a constant flow of wind on a building and it just drawn the static pressure around it, you could pretty much see how big the area of stagnation of air around the building is.

00:18:36.000 --> 00:18:41.119
And I made a hypothesis that a lot of embers would be captured quite well by those pressures.

00:18:41.119 --> 00:18:45.119
I wonder if we've done ever such a simply simplified uh analysis.

00:18:45.440 --> 00:18:47.279
Yeah, like I'm not a wind engineer.

00:18:47.279 --> 00:18:51.759
Uh what I learned from my colleagues who are wind engineers is half of the building height.

00:18:51.759 --> 00:18:54.400
So that that's basically stagnation flow, right?

00:18:54.400 --> 00:19:03.519
So at one point that embers accumulated in that length and also at the base of the wall, a lot of these embers hit the wall and just fall.

00:19:03.519 --> 00:19:13.279
So we see a glowing line at the base of the wall, and that's why a six-inch vertical clearance, non-combustible clearance at the base of the wall is very important.

00:19:13.279 --> 00:19:18.880
We actually have ignited countless buildings unwantedly because of that vulnerability.

00:19:18.880 --> 00:19:26.880
So uh seeing that in at in the facility, then we developed requirements involved for a prepared home to harden that area of the building.

00:19:27.279 --> 00:19:30.480
Okay, so tell me a bit more what drives this research.

00:19:30.480 --> 00:19:34.720
You're in in a non-profit insurance uh institution.

00:19:34.720 --> 00:19:36.799
What's the point of that?

00:19:36.799 --> 00:19:44.319
To to drive the consumers into safer solutions, to uh give uh insurers the ability to uh reject more claims?

00:19:44.319 --> 00:19:45.519
Like I hope not.

00:19:45.519 --> 00:19:49.920
No, I I assume you're looking for how to make the structure safer, right?

00:19:49.920 --> 00:19:51.519
But uh what's the ultimate goals?

00:19:51.759 --> 00:19:55.599
I I'm not sure that there is one ultimate goal or panel competing goals here.

00:19:55.599 --> 00:19:57.599
Uh I'm not necessarily competing, maybe.

00:19:57.599 --> 00:19:59.759
So one of them is to inform the insurance.

00:19:59.759 --> 00:20:04.960
Industry that what are the vulnerabilities and what they should affect for their loss, right?

00:20:04.960 --> 00:20:09.839
We have no say into pricing or their profit, we we are not involved in that process at all.

00:20:09.839 --> 00:20:12.720
Uh, so that's how we serve the insurance industry.

00:20:12.720 --> 00:20:21.759
When it comes to the scientific dynamic, we work closely with academic entities in the US, uh, Australia, European uh universities.

00:20:21.759 --> 00:20:23.200
We are after the same goal.

00:20:23.200 --> 00:20:27.599
How can we make buildings safer or our build environment safer?

00:20:27.599 --> 00:20:28.160
Right.

00:20:28.160 --> 00:20:37.680
So, to some extent, the need of the insurance industry and the need and the scientific needs go up together, and then it uh diverges from a lot of different purposes.

00:20:37.680 --> 00:20:42.240
So during that, the first 80% of it, uh, there is no difference in my view.

00:20:42.240 --> 00:20:51.680
And then as we get to the end, we actually instead of having a deeper dive into fundamental science, we try to hand it over to universities.

00:20:51.680 --> 00:20:57.759
As you know, I we work with UC Berkeley, University of Maryland, University of Melbourne, all of those are uh we work closely with.

00:20:58.240 --> 00:21:12.319
When you design those experiments, are the experiments very like US-centric or they are generalizable to, I don't know, fireband shows in Portugal, in Greece, in in Poland, if we ever have those.

00:21:12.319 --> 00:21:15.039
I hope not, but most likely we will at some point.

00:21:15.359 --> 00:21:18.960
Uh yeah, so the building materials are North America focused for sure.

00:21:18.960 --> 00:21:29.359
So when it comes to the exposure side, which is like the ember side of the equation and the interaction with wind, I don't have a reason to say that there is a difference between European embers.

00:21:29.359 --> 00:21:31.519
I guess like the fuel is different, right?

00:21:31.519 --> 00:21:35.200
But to to some extent there is a good overlap between the two.

00:21:35.519 --> 00:21:39.759
And is the quantification of the ember size distribution or amber life distribution?

00:21:39.759 --> 00:21:45.759
I I don't even know what would be the critical variables to describe the ember from your perspective.

00:21:45.759 --> 00:21:47.119
What's important for you?

00:21:47.119 --> 00:21:48.319
Mass, size?

00:21:48.640 --> 00:21:51.200
Ultimately the thermal energy that they can carry, right?

00:21:51.200 --> 00:21:57.440
Okay, which is a function of mass, size, traveling, distance, temperature, what is the material that they depart from?

00:21:57.440 --> 00:22:01.599
So all of those come to equation and uh define that thermal inertia.

00:22:01.920 --> 00:22:09.759
Okay, now going back to the uh this pathway from firebrands to conflagration, there's a spot ignition in between.

00:22:09.759 --> 00:22:13.279
How how do you quantify the ignition in this case?

00:22:13.279 --> 00:22:15.599
Like how do you investigate it in that case?

00:22:15.839 --> 00:22:22.000
So when we are in the field, uh we see first responders jumping on the spot fires and suppressing them.

00:22:22.000 --> 00:22:23.519
A lot of examples of those.

00:22:23.519 --> 00:22:40.319
And the next or stretching that observation is that it's very likely that when uh first responders were not there to suppress the fire, the fire grew and involved like larger fuels like cars or sheds, and then ultimately structures, right?

00:22:40.319 --> 00:22:54.640
That if they can if they have some food.

00:22:54.640 --> 00:23:07.359
And we clearly see the roll-up embers, both flying and rolling embers, as they just get stuck behind the vertical surface like a wire fence, overwhelm that, ignite that, and then the chain of conflagration gets going from there.

00:23:07.599 --> 00:23:11.920
Are they also like uh do they play a role in any like heat transfer?

00:23:11.920 --> 00:23:18.640
Is there like uh radiation heat flux from uh cloud of embers, or are they too small to cause those phenomena?

00:23:18.880 --> 00:23:25.200
I'm sure some level of heat transfer happens because we see the ultimate product, which is uh ignition, right?

00:23:25.200 --> 00:23:31.359
There are some lab experiments on the pine of embers or like singular ears of like things of that nature.

00:23:31.599 --> 00:23:32.240
Yeah, I think.

00:23:32.240 --> 00:23:36.720
Yeah, I am Do you need one ember or two embers to ignite something, five embers?

00:23:36.720 --> 00:23:37.119
Yeah, yeah.

00:23:37.440 --> 00:23:38.000
Yeah, yeah.

00:23:38.000 --> 00:23:41.759
I am not aware of a an experiment that was done based on this.

00:23:41.759 --> 00:23:51.599
Michael Golner and Wuchan at UC Breakley worked with IBHS during the past year, and they actually took the lead on collecting embers during our full extra full scale experiments.

00:23:51.599 --> 00:23:59.759
And to my knowledge, that is the only study that looks at a full scale ember exposure, wind-driven ember exposure, uh, like what we did.

00:23:59.759 --> 00:24:02.559
And that study is under development yet, it's not published.

00:24:02.960 --> 00:24:03.920
One more thing about embers.

00:24:03.920 --> 00:24:10.079
Do you also try to quantify how many embers are produced in uh house fires in your experiments?

00:24:10.400 --> 00:24:12.880
Mostly on the accumulation side of the equation.

00:24:12.880 --> 00:24:15.519
So where do they accumulate?

00:24:15.519 --> 00:24:19.519
And uh, I don't want to give too much information because uh UC Baker is working on it.

00:24:19.519 --> 00:24:23.920
Of course, estimated how many embers can depart from a burning structure.

00:24:24.240 --> 00:24:25.599
Okay, that's very interesting.

00:24:25.599 --> 00:24:32.400
Because I also I I think I saw experiments are at IBHS where there would be multiple structures, like one next to another.

00:24:32.640 --> 00:24:33.279
Yeah, yeah, yeah.

00:24:33.279 --> 00:24:41.759
We do study those, and uh big part of the equation that we are working on right now, or like research agenda that we have, is how configuration takes place.

00:24:41.759 --> 00:24:55.839
So when the first building ignites for whatever reason, what is the influence of wind speed, separation distance, fuel load, and like the relative location of these two buildings together that defines junk from the first to the second one?

00:24:56.079 --> 00:25:03.119
Can you define like so so configuration would be for you any significant fire spread within a community, or how would you define it?

00:25:03.279 --> 00:25:07.920
Yeah, when we see uncontrollable flame-driven fire spread, we call that configuration.

00:25:08.559 --> 00:25:09.359
Flame-driven, okay.

00:25:09.599 --> 00:25:10.000
Yeah, yeah.

00:25:10.000 --> 00:25:11.680
There are a few criteria for that.

00:25:11.680 --> 00:25:24.160
In my view, structure separation is the driving factor because when you have dense neighborhoods, it's likely that those homes are connected to each other by cars, by sheds, by fences, things of that nature.

00:25:24.160 --> 00:25:29.599
And uh when there is more fuel, there is rain, and then unlimited fuel, unlimited oxygen.

00:25:29.680 --> 00:25:30.960
That's a recipe for disaster.

00:25:30.960 --> 00:25:34.160
Well, no, not fantastic, but uh that's that's that's very interesting.

00:25:34.160 --> 00:25:37.039
Okay, uh, let's discuss the conflagrations.

00:25:37.039 --> 00:25:51.359
So uh maybe again give me give me a story of how did you start working on large-scale building fires and and uh how does uh a fire turn from uh just a spot ignition into uh something that can create a conflagration effect?

00:25:51.599 --> 00:26:17.440
Yeah, back in 2019-2018, uh when we were looking at ender accumulation around buildings, and then we we put different fuels in the zero to five foot uh building, as that ignites and the flame starts to touch the cladding, typically uh combustible cladding, it's in the order of 45 seconds to one minute and a half for the fire from a few inches tall to like tens of feet tall uh flame.

00:26:17.440 --> 00:26:29.119
So when we saw that fast transition, those were those were the first questions that uh were generated back then that when this takes place uh fast, how does the fire jump from this building to another?

00:26:29.119 --> 00:26:29.519
Right.

00:26:29.519 --> 00:26:35.039
And then in 2020, we worked with NIST, uh US Forest Service Service, NIST, IBHS.

00:26:35.039 --> 00:26:40.000
If you I hope I'm not missing any any other organization, it was a conversive effort, right?

00:26:40.000 --> 00:26:55.440
So we started looking at fire, uh, how fire jumps from a burning shed to another structure, and we divide and conquer uh this task by uh NIST taking care of smaller sheds and no wind, IBHS did the larger sheds with wind.

00:26:55.440 --> 00:27:05.920
And during those experiments, we realized that uh interestingly, as the fire develops in the source compartment, which is just a shed, the fire gets hotter and hotter.

00:27:05.920 --> 00:27:10.319
Before it impacts the target building, it actually impacts the source structure itself.

00:27:10.319 --> 00:27:11.759
Okay, and collapse happens.

00:27:11.759 --> 00:27:22.720
So the roof collapse on the on the fire limits oxygen, and at the peak of the fire, we see an 80% or like 50% drop in the fire intensity or the heat flux that we measure down with.

00:27:22.720 --> 00:27:23.279
Right.

00:27:23.279 --> 00:27:31.920
So those were like interesting observations that we start seeing how fire departs from the classical growth steady decay fire curve.

00:27:31.920 --> 00:27:36.960
And interestingly, a lot of the damage modes actually happen beyond the growth part.

00:27:36.960 --> 00:27:43.599
So similar to the cone calorimeter test that we see, two peaks, we see multiple peaks in these experiments.

00:27:43.599 --> 00:27:49.440
In some of them, ignition or damage happens on the first peak, but that's not a must.

00:27:49.440 --> 00:27:55.759
In many cases, ignition and damage happens on the secondary or third uh peaks decay phase.

00:27:56.000 --> 00:28:08.319
So in this case, it's like two parallel timelines: one when the fire devours the initial structure and and how that damage interacts with the fires itself, and and how when it will jump to another structure.

00:28:08.319 --> 00:28:09.440
That's how I understand it.

00:28:09.680 --> 00:28:10.480
Yeah, absolutely.

00:28:10.480 --> 00:28:18.880
So there are two knobs, the exposure side, which is driven by how hot the fire burns, and then the resistance side, which is basically the material on the target page.

00:28:18.880 --> 00:28:23.200
The first one overpowers the second one when fire spreads.

00:28:23.200 --> 00:28:27.359
And if the other one has a fighting chance, then the fire does not spread, right?

00:28:27.359 --> 00:28:29.440
And that's how we put these two together.

00:28:29.440 --> 00:28:40.240
When the exposure is so high, meaning the structure separation of three meters or ten feet, then under those cases, ignition happens on the first peak because there is no finding chance.

00:28:40.240 --> 00:28:50.319
But when the structure separation is more of like 20 feet or uh larger than that, then the exposure is reduced to some extent that the building material have a fighting chance.

00:28:50.319 --> 00:28:53.839
So they don't respond necessarily to the first peak.

00:28:53.839 --> 00:28:57.359
They might respond to the second one, the third one, or even in the decay.

00:28:57.680 --> 00:29:04.480
By the exposure, like I guess it's all modes of transfer plus firebrands in this case all together at one time.

00:29:04.799 --> 00:29:07.759
So uh you didn't look at firebrands in this exposure side.

00:29:07.759 --> 00:29:09.440
What makes it correct?

00:29:09.440 --> 00:29:11.599
Like in real world, that's how it takes place.

00:29:11.599 --> 00:29:18.079
We are only looking at the heat flux measurement uh by convective and uh convective heating and frame contact.

00:29:18.559 --> 00:29:20.880
How big is the convective heating in this case?

00:29:21.119 --> 00:29:25.839
Uh we saw 140 kilowatts per meter square fire.

00:29:25.839 --> 00:29:26.240
Okay.

00:29:26.240 --> 00:29:28.480
That did not ignite wood.

00:29:28.480 --> 00:29:29.039
Okay.

00:29:29.039 --> 00:29:29.680
Yeah.

00:29:29.680 --> 00:29:41.359
So uh when later on, uh maybe like three minutes later we see ignition happening, or like 10 minutes later we saw ignition happening at much lower heat fluxes, the more like like 40, 50 uh kilowatts per meter square.

00:29:41.680 --> 00:29:46.559
But this 140 kilowatt was the convective heat transfer or combined?

00:29:47.039 --> 00:29:47.440
Combined.

00:29:47.440 --> 00:29:53.359
Like the total uh uh heat flux meter, uh, we measure 140, 140 kilowatts, yeah.

00:29:53.519 --> 00:29:53.839
Mm-hmm.

00:29:53.839 --> 00:30:05.119
And uh by flame contact, in that case you distinguish it by observation, or there's a like a uh value of heat flux where you just consider it was a flame contact.

00:30:05.359 --> 00:30:11.759
So uh if we see a sudden drop in the measurement, then we look at the videos and we say, okay, that's what the direct flame contact, right?

00:30:11.759 --> 00:30:18.079
So most of the flame direct flame contact in a continuous way happened when the structure separation was at 10 foot.

00:30:18.079 --> 00:30:24.160
At 20 feet separation, about six meters, then it becomes an intermittent flame contact and convectivity.

00:30:24.799 --> 00:30:26.960
Okay, well, now you're bringing the time component.

00:30:26.960 --> 00:30:27.920
Thank you very much.

00:30:27.920 --> 00:30:30.640
Uh very, very interesting.

00:30:30.640 --> 00:30:34.400
So, okay, uh we'll we'll take the time component in a second.

00:30:34.400 --> 00:30:36.480
Uh let's let's look at the resistance side.

00:30:36.480 --> 00:30:41.039
How do you define the properties of the materials in that case?

00:30:41.039 --> 00:30:42.319
What are you looking into?

00:30:42.559 --> 00:30:47.839
Yeah, so uh we tested common in North America, like vinyl windows, double pane temper.

00:30:47.839 --> 00:30:50.559
They have about 25%, 20% market share.

00:30:50.559 --> 00:30:55.680
Different species in combustible, non-combustible, all of them representative of what we have uh in this country.

00:30:55.680 --> 00:30:59.759
And then we realize that uh they have a response time, right?

00:30:59.759 --> 00:31:22.240
So when we consider these as uh semi-infinite solids, and then we solve the equation for it, and then uh based on different heating ramps, we can see that the time that it takes for different building components to go from say ambient temperature 20 degrees Celsius all the way to the failure temperature for wood, about 300 degrees Celsius for glass, maybe 180 or something like that.

00:31:22.799 --> 00:31:24.240
How do you define failure temperature?

00:31:24.480 --> 00:31:29.200
So for for wood is ignition for glasses, breakage, for uh vinyl is melting.

00:31:29.279 --> 00:31:33.519
It depends on the the significant significant event with uh given material.

00:31:33.759 --> 00:31:34.799
Well, absolutely, yeah.

00:31:34.799 --> 00:31:40.000
Okay, so that time frame is in the order of five seconds to a minute.

00:31:40.000 --> 00:31:45.599
It is not instantaneous heat and it's not total heat flows in the order of 10 minutes or something like that.

00:31:45.599 --> 00:31:59.920
So when we chose that, and then you look at the statistical distribution of how much energy is received during that time, we started seeing some statistical patterns that match exposure and uh resistance of the building materials.

00:32:00.160 --> 00:32:05.440
Did you have the chance to connect them with some uh modeling of heat transfer within the structures?

00:32:05.440 --> 00:32:16.960
I wonder how the one the thermal bulk will definitely be in a factor of here and then perhaps the capacity to to like convectively cool on the other side of the of the element, perhaps.

00:32:17.200 --> 00:32:19.920
Yeah, we are working on some modeling side of this right now.

00:32:19.920 --> 00:32:25.039
Uh Michael Wilder, his team is doing that right now, so that will come out hopefully sometime next year.

00:32:25.279 --> 00:32:26.079
Okay, fantastic.

00:32:26.079 --> 00:32:28.160
And now uh for the exposures.

00:32:28.160 --> 00:32:35.119
So when you are exposing those building materials to those uh what what's your initial source of fire?

00:32:35.119 --> 00:32:39.440
Do you set up a big fire in front of the building and then see the flames?

00:32:39.440 --> 00:32:41.519
How do you carry the experiment actually?

00:32:41.680 --> 00:32:49.599
Yeah, so the first phase of these experiments, which were sheds, we tested a UL standard or nominally UL standard uh 6A wood cribs.

00:32:49.599 --> 00:32:53.920
Each of them is about 300 pounds, 140 kilograms, something like that.

00:32:53.920 --> 00:33:04.319
And we put like 15 of those in each building inside inside the shed, yeah, and then ignite that from like an internal ignition, and then look at the fire go.

00:33:04.319 --> 00:33:12.720
In the second phase of the experiment, which just finished a few weeks ago, we tested auxiliary dueling unit, uh residential structure.

00:33:12.720 --> 00:33:21.279
And uh the size of those were 25 feet by 25 feet by 17, so a one bedroom with an office fully furnished with furniture.

00:33:21.519 --> 00:33:21.759
Okay.

00:33:21.920 --> 00:33:26.559
And then in these experiments, we ignited those by external ignition.

00:33:26.559 --> 00:33:30.559
So we put wood crepes uh five feet away from them, ignited those.

00:33:30.559 --> 00:33:36.880
So we created a configuration condition that the fire impulse or impacts the building from an external.

00:33:37.279 --> 00:33:41.519
So in this case, you see how the fire enters the building and uh what happens later on.

00:33:41.759 --> 00:33:42.480
Yeah, absolutely.

00:33:42.480 --> 00:33:48.640
And like the very clear observation that we had is uh the building is as as strong as its weakest link.

00:33:48.640 --> 00:34:00.960
You can have a non-combustible siding on your own, but the window and door can fade much earlier than the other components, and that triggers fire spread and how fires enters into the living area.

00:34:00.960 --> 00:34:10.000
And with wind, flashover is once the fire enters about two minutes after that, and then the train of configuration keeps going.

00:34:10.159 --> 00:34:15.679
Uh like you're going into wild places with uh with my false process right now.

00:34:15.679 --> 00:34:19.920
But no, no, I have to stop uh and ask some clarification questions.

00:34:19.920 --> 00:34:28.159
Did you have a chance to look into a completely non-combustible structure but with the combustible uh materials inside?

00:34:28.400 --> 00:34:30.480
So non-combustible gliding, yes.

00:34:30.480 --> 00:34:33.760
We tested stock oil, fiber cement, like those materials.

00:34:33.760 --> 00:34:37.760
But all of those do have vinyl windows, which is a combustible component.

00:34:37.760 --> 00:34:39.440
Okay, or doors, which is fiber.

00:34:39.440 --> 00:34:48.000
These buildings are built to requirements of California's WUI building codes, but you do have uh combustible uh components on there.

00:34:48.320 --> 00:34:51.760
Okay, and and and still through those weakest links you see spread.

00:34:51.760 --> 00:34:55.760
But I assume there must have been a difference in the timelines between those.

00:34:56.320 --> 00:35:02.079
Well, not necessarily when the building itself fails, and that's a very great uh question.

00:35:02.079 --> 00:35:07.840
The difference kicks in when we look at the response of the source fire on the target building.

00:35:07.840 --> 00:35:17.920
When we have a combustible uh building with combustible cladding, the flame length is significantly higher compared to other buildings with non-combustible, sorry.

00:35:17.920 --> 00:35:23.280
So of course the flame length and flame surface is much higher than another wood structure.

00:35:23.280 --> 00:35:32.400
But when you have a building like a uh fiber cement structure non-combustible, the flame surface area is limited to windows and doors, just openings.

00:35:32.400 --> 00:35:41.519
So radiation is significantly lower, and then flame length is also lower, and that's how resistance and exposure are merged.

00:35:41.519 --> 00:35:48.559
So we talked about resistance on one side of evacuation, exposure on the other side, but through this dynamic, they are connected.

00:35:48.559 --> 00:35:51.920
So resistance actually plays a role in exposure.

00:35:52.159 --> 00:35:52.800
Fantastic.

00:35:52.800 --> 00:35:57.920
How does a fleshover in the wind-driven fire look like as a phenomenon?

00:35:58.320 --> 00:36:03.599

230 - Wind driven conflagration experiments with Faraz Hedayati

Listen On

Recent Episodes

Support On