Dec. 3, 2025

229 - Learning from 900 fires with Björn Maiworm

The player is loading ...

What can you learn after processing observations across 900 severe fires? A lot. Actually, I will send you to the paper straight away:

Evaluating 900 Potentially Harming Fires in Germany: Is the Prescriptive Building Code Effective? German Fire Departments Assessed Fire Safety Measures in Buildings Through On-Site Inspections

And now let's dissect this. We sit down with Björn Maiworm of the Munich Fire Department to unpack a decade of structured observations from more than 2,000 significant incidents (900 in the paper but the database already grew!) across Germany—and the results may challenge the assumptions of Fire Safety Engineers. Smoke spread shows up as way more common, despite that legislation should prevent it, and is often seen breaching beyond the apartment of origin when doors are left open, self-closers are defeated, or vertical shafts pull hot gases to the top floor. Meanwhile, true flame spread between units is relatively rare, suggesting that basic compartmentation and detailing are quiet success stories.

We also talk about people. Injuries appear in roughly a third of these consequential fires and fatalities in 6 to 7 percent, with risk concentrated in prisons, elder care, and dense low-income housing. Building age isn’t the driver; height and social factors are. Where self-closing doors are mandated and maintained, smoke infiltration to stairs drops—just not as far as theory predicts, thanks to behavior and upkeep realities. That gap between paper and practice is where small, targeted fixes make the biggest difference.

On emerging risks, the data draws sharp lines. Mass timber’s challenge isn’t fire resistance; it’s the speed and multi-floor spread when exposed surfaces meet window plumes. The result can outpace practical firefighting capacity. By contrast, shifting a typical household to an EV, PV, and home battery can reduce overall fire probability; the true hazards arise from poor products, DIY installs, and dense storage arrangements. The smart response is segmentation and simple physical breaks that buy time, not blanket bans or panic.

We close by reframing fire safety as a complex system problem. Instead of chasing perfect proofs, we can use continuous field feedback to find the leverage points: doors that stay shut, shafts treated as priority risks, vulnerable occupancies protected with tailored measures, and dispatch data that points crews to the right entrance first.

If this resonates, subscribe, share the episode with a colleague, and leave a review telling us which finding surprised you most. Your feedback helps more engineers, firefighters, and policymakers turn real-world lessons into safer buildings.

----
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 Real Fires Matter;

02:43 - Designing For Firefighters’ Capabilities;

08:53 - From Gut Feeling To Data;

15:17 - What The Questionnaire Captures;

19:54 - Stable Patterns In Outcomes;

24:38 - Smoke Spread Paths And Failures;

30:16 - Fire Spread Is Rare, Speed Is Not;

36:08 - Timber’s Fast Propagation Risk;

41:04 - Batteries, EVs, And Real Risk;

48:02 - Scooters, Social Factors, Common Sense;

53:10 - Rethinking Extinguishers And Behavior;

57:08 - When Firefighting Really Starts;

WEBVTT

00:00:00.400 --> 00:00:02.319
Hello everybody, welcome to the Fire Science Show.

00:00:02.319 --> 00:00:09.759
If you want to do a good work, you have to do the work and then you have to figure it out if it works or not.

00:00:09.759 --> 00:00:13.599
That's how you find out the job has been done well.

00:00:13.599 --> 00:00:26.480
Now as a fire safety engineer, the second part finding out if stuff that we've done worked or has not worked is actually quite difficult and to some extent stressful.

00:00:26.480 --> 00:00:39.119
Because when stuff works we don't really know or we rarely find out, and when they don't work it's usually with a big media attention that we do not necessarily want.

00:00:39.119 --> 00:00:44.880
However, there are fires happening, hundreds of fires happening every day.

00:00:44.880 --> 00:00:53.359
And through those fires we could learn a lot about what was done in a building and did it work or not.

00:00:53.359 --> 00:00:59.520
The only issue is that someone would have to capture those statistics and process that data.

00:00:59.520 --> 00:01:06.000
And as you may figure out, this is exactly what we're talking about in this podcast episode.

00:01:06.000 --> 00:01:19.040
I have invited Björn Maiworm from the Munich Fabricate, who has run an exercise like this for years in Munich and other fire stations across Germany, and it actually still continues.

00:01:19.040 --> 00:01:27.599
And a year ago, Bjorn published a paper in fire technology where he summarized 900 potentially harming fires in Germany.

00:01:27.599 --> 00:01:35.519
Today he told me he has 2,000 in his database, and his database is something we will discuss in deep.

00:01:35.519 --> 00:01:38.879
How is he gathering the statistics from the fires?

00:01:38.879 --> 00:01:40.640
What is he looking for?

00:01:40.640 --> 00:01:45.280
And most important, what are the lessons from those those fires?

00:01:45.280 --> 00:01:49.920
We will talk about some general problems with uh gathering statistics.

00:01:49.920 --> 00:01:55.200
We'll we'll talk about collective experience on code writing and law and prescriptions.

00:01:55.200 --> 00:02:05.519
We'll discuss smoke spread, fire spread, and in the end of the podcast we actually this have a very interesting discussion about time to uh start uh firefighting in a building, which is a very, very interesting discussion.

00:02:05.519 --> 00:02:08.000
So I think it's a very nice one.

00:02:08.000 --> 00:02:14.719
I had the chance to record this live because Bjorn was visiting Warsaw for a risk conference that I have attended.

00:02:14.719 --> 00:02:17.840
So let's spin the intro and jump into the episode.

00:02:17.840 --> 00:02:24.080
Welcome to the Firescience Show.

00:02:24.080 --> 00:02:27.759
My name is Wojciech Wegrzynski, and I will be your host.

00:02:27.759 --> 00:02:57.039
The Firescience 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.

00:02:57.039 --> 00:03:10.800
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.

00:03:10.800 --> 00:03:27.039
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.

00:03:27.039 --> 00:03:35.840
If you're keen to find out more or join OFR consultants during this exciting period of growth, visit their website at OFRConsultants.com.

00:03:35.840 --> 00:03:37.680
And now back to the episode.

00:03:37.680 --> 00:03:38.639
Hello everybody.

00:03:38.639 --> 00:03:43.039
I am joined today by Björn Maiworm from Munich Phi Department.

00:03:43.039 --> 00:03:43.919
High level officer.

00:03:43.919 --> 00:03:46.240
Hey Bjorn, uh good to good to have you in Warsaw.

00:03:46.240 --> 00:03:46.639
Welcome.

00:03:46.639 --> 00:03:48.319
Yeah, thank you very much for having me.

00:03:48.319 --> 00:03:49.599
It's uh it's been a long time.

00:03:49.599 --> 00:03:52.319
I've seen your talk, I think it was at SFP Copenhagen, I think.

00:03:52.319 --> 00:03:53.520
In Copenhagen, yes, yeah.

00:03:53.520 --> 00:03:55.439
That's when we agreed to do an interview.

00:03:55.439 --> 00:04:00.240
We're a few SFP events later, but I'm glad you found your way to Warsaw.

00:04:00.240 --> 00:04:03.439
It's not the greatest weather, so we can at least do a podcast episode.

00:04:03.439 --> 00:04:17.600
So I remember back then you've shown a very impressive study studying the statistics of fires from the perspective of firefighters and trying to make sense out of it of how the measures are effective.

00:04:17.600 --> 00:04:19.360
Do I remember correctly?

00:04:19.920 --> 00:04:31.519
In the end, the idea is well, if you draw a fire safety design of a building, it's a theoretical idea, it's a planning idea, but we don't have any data about the outcome.

00:04:31.519 --> 00:04:34.720
We don't know that what you designed actually worked out.

00:04:34.720 --> 00:04:45.199
So the core question is not how did the firewall fighters work, how was the planning or design, something like that, but did the building and its design work out in real life?

00:04:45.199 --> 00:04:50.319
It's like buying a car, and some fancy engineer told you it's a very safe car.

00:04:50.319 --> 00:04:51.839
We put this and that in.

00:04:51.839 --> 00:04:53.600
We said, Well, how was a crash test?

00:04:53.600 --> 00:04:54.720
We didn't do any crash test.

00:04:54.720 --> 00:05:01.439
So this is actually the summary of many crash tests where the people were to introduce a fire test.

00:05:02.240 --> 00:05:06.240
I mean, it's a very rare perspective when I talk with uh firefighter colleagues.

00:05:06.240 --> 00:05:09.120
It's a common thing, actually, the the issue with data.

00:05:09.120 --> 00:05:12.560
I I had uh Eurostat project people in the podcast.

00:05:12.560 --> 00:05:18.319
Uh I'm due to do one with uh FSRI where they revamped the statistics in North America.

00:05:18.319 --> 00:05:25.120
But it's quite challenging to get meaningful engineering related data out of uh fire statistics.

00:05:25.120 --> 00:05:26.959
What was it the trouble for you before?

00:05:27.439 --> 00:05:42.800
The system the statistics we have from the fire department's perspective is how many fire alarms were there, how many false alarms, how long did it take to go there, maybe if they collected that data, how many hoses did we use, how many engines and trucks?

00:05:42.800 --> 00:05:44.000
How many little little stuff?

00:05:44.000 --> 00:05:49.920
How many so what i if if they even guessed it somehow, but usually they don't have any accurate data.

00:05:49.920 --> 00:05:56.000
But in the end, with that data, you can't do anything concerning building safety and building design.

00:05:56.000 --> 00:06:08.399
So we you must connect the systems because from our point of view, as German fire departments, the building regulations and the building design resembles the ability of the fire department's work.

00:06:08.399 --> 00:06:09.759
So they're interconnected.

00:06:09.759 --> 00:06:16.560
If you don't have a fire department, having a system with, for example, a second escape route wire ladder is helpless.

00:06:16.560 --> 00:06:22.720
You can see that, for example, I was in New Zealand, every smaller hotel has an escape ladder on the outside.

00:06:22.720 --> 00:06:23.920
On the outside.

00:06:23.920 --> 00:06:31.279
In Germany, it's standardized to have a second escape route via the ladders we bring as fire departments.

00:06:31.279 --> 00:06:35.839
But that means you need to have a very powerful fire department everywhere.

00:06:35.839 --> 00:06:44.160
Okay, so this resembles in all the metrics we have in our German building law, resemble the abilities of the fire department.

00:06:44.319 --> 00:06:52.480
I wonder to what extent the technical regulations in Germany were driven by the capabilities of our brigade.

00:06:52.480 --> 00:06:56.800
I I think all safety regulations are written by blood.

00:06:56.959 --> 00:06:57.279
Yes.

00:06:57.519 --> 00:06:58.240
And with blood.

00:06:58.560 --> 00:07:00.800
Due to the experience of loss.

00:07:00.800 --> 00:07:09.920
Middle ages fire, huge city fires, we invented firewalls, non-flammable materials, and all this stuff that's taken for granted now.

00:07:09.920 --> 00:07:19.279
In the time when fire departments developed, like in the late 1890s, 1900s, something, where was the time when the building code was developed.

00:07:19.279 --> 00:07:27.839
So these two systems interacted, and of course, fire safety or fire offices, high-level offices were part of the regulations group.

00:07:28.079 --> 00:07:49.439
In Poland, it's kind of like as you say, it it had to be there had to be a tragic event or something that uh issued a big revamp in the system, and basically the system would adjust itself to respond to whatever happened, or perhaps there was a broad big event that happened and we just imported some knowledge, that's arguably the better way.

00:07:49.439 --> 00:07:56.720
But um, I'm not sure to what extent they reflect the capability of the fire brigade, you know.

00:07:56.720 --> 00:08:09.360
If we had an event where we had a big uh spillover of an oil tank in a refinery, Czechovice Jijitse fire, they've implemented a special uh you know industrial fire protection around tanks, etc.

00:08:09.360 --> 00:08:11.120
It had a lot involved.

00:08:11.120 --> 00:08:13.439
I you you told me the story of a firehouse.

00:08:13.439 --> 00:08:20.720
Maybe it's worth it to uh repeat it, how firefighters implement influenced the regulations themselves.

00:08:21.920 --> 00:08:23.759
First, let me call it on your tank fire.

00:08:23.759 --> 00:08:27.519
Yeah, this is a very unique thing from a societal point of view.

00:08:27.519 --> 00:08:28.079
Yes.

00:08:28.079 --> 00:08:32.240
So, first of all, in a building regulation, you want to regulate a standard building.

00:08:32.240 --> 00:08:33.679
What is a standard building?

00:08:33.679 --> 00:08:35.519
It's an apartment, it's an apartment building.

00:08:35.519 --> 00:08:36.480
It's where people live.

00:08:36.480 --> 00:08:39.200
This is the majority of buildings we need.

00:08:39.200 --> 00:08:43.039
Then you maybe need an assembly hall, you need a hotel, you need a shopping center.

00:08:43.039 --> 00:08:45.919
And this is how the regulations are focused on.

00:08:45.919 --> 00:08:52.399
And in in the end, as we we talked about before, let's talk about the escape route length.

00:08:52.399 --> 00:08:54.799
You can do some fancy calculations.

00:08:54.799 --> 00:09:03.679
What's the visibility and the FED model something we heard on the uh conference today, and and and all these aspects.

00:09:03.679 --> 00:09:23.200
But in the end, from our point of view, in as far as what I could find in any historical document, the 35 meters distance between the very end of an apartment towards the door to the stairwell, and doesn't necessarily have to be the door of the apartment, the maximum travel distance is 35 meters.

00:09:23.200 --> 00:09:23.519
Okay.

00:09:23.519 --> 00:09:29.279
Because in the stairwell you can retreat to the next floor below if there's a huge fire as a firefighter.

00:09:29.279 --> 00:09:30.240
So you need to enter.

00:09:30.240 --> 00:09:37.360
And every firefighter, Second World War Time and after and up to today, the length is 15 meters for one hose.

00:09:37.360 --> 00:09:50.320
So there are two guys going up there, every one of them has a hose, and you connect your hose there from this door, and then you enter, crawling under the fire, plus a five-meter distance of spraying the water.

00:09:50.320 --> 00:09:56.320
So the 35 meters in the end resemble the depth of attacking the fire.

00:09:56.320 --> 00:10:00.879
And if you did you ever extinguish a fire, or did you do the training?

00:10:00.879 --> 00:10:01.600
Okay, you did.

00:10:01.600 --> 00:10:04.639
So you know that a hose is quite heavy to pull.

00:10:04.639 --> 00:10:08.879
And if you connect them endlessly, it won't be possible to move.

00:10:08.879 --> 00:10:13.919
So from a practical point of view, craftsmanship, it makes sense.

00:10:13.919 --> 00:10:20.799
And then if you have some guy calculating that you can install an early smoke alarm detection system and everyone runs.

00:10:20.799 --> 00:10:23.360
It takes you 27 seconds to move 35 meters.

00:10:23.360 --> 00:10:28.639
Something like that, so we can extend the length towards to 100 meters.

00:10:28.639 --> 00:10:36.240
No one would ever expect a firefighter to enter a room 100 meters with a house without the possibility of retreat.

00:10:36.240 --> 00:10:37.840
It's far too dangerous.

00:10:37.840 --> 00:10:41.919
And this is what the safety or rescue teams must be taken into consideration.

00:10:42.320 --> 00:10:48.879
Don't quote me on this, dear listeners, but uh I I once was very um intrigued by the concept.

00:10:48.879 --> 00:11:02.559
I'm not sure if you have it in Germany, but if you have a wall that separates rooms in an office building in Poland, let's say you would have an one-hour rated wall, the doors would be 30 minutes, like half the fire resistance of the wall.

00:11:02.559 --> 00:11:14.320
And it was quite funny in Poland because people started installing sliding walls and they were basically just giant doors without any wall component, and suddenly they were like 30 minutes all of them, you know.

00:11:14.320 --> 00:11:16.960
That has obvious business implications, right?

00:11:16.960 --> 00:11:24.960
And I was I was desperate to find where the hell did it come that we do half of fire resistance of the door versus the wall.

00:11:24.960 --> 00:11:29.519
And I found one thing from the 70s, I believe that was in Sweden.

00:11:29.519 --> 00:11:40.159
Like I cannot recall the name or any details, but it basically said that on a door opening, it's so much easier to stop the fire than on a firewall.

00:11:40.159 --> 00:11:51.279
That if you have just half the fire resistance of the wall, you'll be good because a firefighter uh standing like a few meters away from the door with a spray hose and a spray pattern can hold the fire on there.

00:11:51.279 --> 00:11:55.519
And I was like, oh my god, this is like not how we use it today.

00:11:55.840 --> 00:11:57.120
This may be one aspect.

00:11:57.120 --> 00:12:05.279
In the end, you could have a practical explanation, like on top of the room it's hotter, and then why don't you impact that long and all this stuff?

00:12:05.279 --> 00:12:19.679
But in the end, the cop compartmentation of a system, of a unit, and introducing a door allows us to defend as a tactical term, to defend the neighboring rooms and to defend it from spreading.

00:12:19.679 --> 00:12:22.000
And it's ensures our safety.

00:12:22.000 --> 00:12:27.360
Because if you can retreat behind a wall, you can attack out of coverage, out of colour.

00:12:27.360 --> 00:12:31.759
And then we have the same discussion with large office buildings in Germany.

00:12:31.759 --> 00:12:35.360
Well, you need 30 minutes of fire resistance and all this stuff.

00:12:35.360 --> 00:12:39.840
Yeah, we do know that it can burn through, but we can defend it.

00:12:39.840 --> 00:12:44.960
And in the timeline where we actually can do work, it works out.

00:12:44.960 --> 00:12:46.320
It makes sense.

00:12:46.320 --> 00:12:51.039
Of course, from an engineering or scientific point of view, we don't have any proof.

00:12:51.039 --> 00:12:56.399
It's just a very good gut feeling, a very good engineering judgment, how we call it.

00:12:56.399 --> 00:12:58.399
But in the end, we must trust that.

00:12:58.720 --> 00:13:06.960
Now that that brought us, that segued us beautifully to the subject of this interview because it's about replacing gut gut feeling with data.

00:13:06.960 --> 00:13:20.960
And while a fire engineer, you can spend all the time doing your CFD analysis, doing your finite element model analysis of a wall, how well it will behave in a fire, in what kind of fire, in what circumstances.

00:13:20.960 --> 00:13:25.200
In the end, life kind of verifies this by a real-world fire experiment.

00:13:25.200 --> 00:13:32.559
If a fire happens in that place, and this is an opportunity to figure out how it actually worked.

00:13:32.799 --> 00:13:36.159
Yeah, actually, it just started that for decades.

00:13:36.159 --> 00:13:42.960
It was totally accepted when the fire department's representative said, well, this works and this won't.

00:13:43.200 --> 00:13:44.960
Just by the power of authority?

00:13:45.200 --> 00:13:48.480
No, no, by the power of summarized experience.

00:13:48.480 --> 00:13:48.879
Okay.

00:13:48.879 --> 00:13:52.399
And by trusting in the expert you have there.

00:13:52.399 --> 00:13:57.919
Not just by his own experience, but by the summarized experience of the organization.

00:13:57.919 --> 00:13:59.039
It's not just me.

00:13:59.039 --> 00:14:03.120
I'm representing German fire departments in a lot of committees.

00:14:03.120 --> 00:14:07.919
It's not me talking there, it's a summarized experience and now based on data.

00:14:07.919 --> 00:14:20.559
And then in the last like 10 to 15 years, more 15 years, we had scientists that said, Well, you don't have any scientific proof for that, so we don't need that requirement.

00:14:20.559 --> 00:14:25.840
And this is when my boss and I we said, Well, we we need to collect data, we need to collect the knowledge.

00:14:25.840 --> 00:14:26.960
How do we do that?

00:14:26.960 --> 00:14:37.279
So we developed a questionnaire for the fire departments, and after a fire, with the allowance of the landlords and with the allowance of the police, we go there and we collect the data.

00:14:37.279 --> 00:14:38.639
What kind of building is it?

00:14:38.639 --> 00:14:42.080
Is this a special construction, like a hospital or things like that?

00:14:42.080 --> 00:14:50.480
And then we go through the safety objectives by law, like people's safety, um firefighter safety, and all these aspects.

00:14:50.720 --> 00:14:55.120
By design objectives, you mean you go through what was designed in the building?

00:14:55.120 --> 00:14:56.240
How was it built?

00:14:56.480 --> 00:14:59.279
No, what is the general requirement by law?

00:14:59.279 --> 00:15:09.840
By building it construction products regulation, heritage protection, environmental protection, which are the objectives we have as a fire department by law.

00:15:09.840 --> 00:15:24.159
Our job is to rescue you, get you out, survive the attack and not lose any firefighters there, minimize the damage, take care of the environment and cultural heritage.

00:15:24.159 --> 00:15:25.120
That's in the end.

00:15:25.440 --> 00:15:38.000
I I I like a lot of places do save statistics, but it's it would be very rare to like have a detailed description of what was the building status, like what technical systems did it had, what were they in operation?

00:15:38.480 --> 00:15:42.320
We don't look at the fire def design or the the the building permit.

00:15:42.320 --> 00:15:46.960
It's what's really interesting in this research program, it's non-scientific.

00:15:46.960 --> 00:15:49.679
There are not any scientists going on scene.

00:15:49.679 --> 00:15:54.960
Asking you what was the heat release rate and how do you define what is a significant fire, something like that?

00:15:54.960 --> 00:15:57.840
Because we use the German phrase for significant fire.

00:15:57.840 --> 00:16:05.600
It's a fire that is at least so developed that it can ask the question toward the objectives.

00:16:05.600 --> 00:16:10.480
So a small bin fire is too small to ask a firewall, are you strong enough?

00:16:10.480 --> 00:16:12.720
Or a fire door, 30 minutes resistance.

00:16:12.720 --> 00:16:19.919
So it needs to be at least bigger than a small burned food kitchen fire, which is done a thing at noon you go there.

00:16:19.919 --> 00:16:23.600
And then we spread the news 10 years ago, it's now our anniversary.

00:16:23.600 --> 00:16:25.600
10 years, thank you.

00:16:25.600 --> 00:16:27.600
It's a long, long last long journey.

00:16:27.600 --> 00:16:29.039
Yeah, long, long journey.

00:16:29.039 --> 00:16:36.399
And now we cover with the participating fire departments more than a third of the population in Germany.

00:16:36.399 --> 00:16:36.879
Wow.

00:16:36.879 --> 00:16:44.240
So with the area where they are responsible, of course, their data set grew in the last four years, five years enormously.

00:16:44.240 --> 00:16:46.320
Because it's about trust too.

00:16:46.320 --> 00:16:55.200
Because the fire departments need to trust us as the association of the fire departments, that we don't have a look at the fire department's work.

00:16:55.279 --> 00:16:55.360
Yeah.

00:16:56.080 --> 00:16:57.759
That we only focus on the building.

00:16:57.759 --> 00:17:11.599
So when my colleagues go on a scene, for example, in Munich, where the majority of the data set is coming from, like 30 something percent, I always tell them never ever, never say anything about the fire department's work.

00:17:11.599 --> 00:17:12.160
Don't.

00:17:12.160 --> 00:17:13.279
That's not the point.

00:17:13.279 --> 00:17:14.240
It's not the point.

00:17:14.240 --> 00:17:15.279
Focus on the building.

00:17:15.279 --> 00:17:20.240
And it even led to a situation where the police officers ask us, what happened here?

00:17:20.240 --> 00:17:23.599
And then we explained to him, well, the building's design was poor.

00:17:23.599 --> 00:17:27.680
The landlord's efforts to keep it going were very poor.

00:17:27.680 --> 00:17:33.279
So, from our point of view, the three people died because of the landlord's failure.

00:17:33.279 --> 00:17:35.680
And in the end, he was convicted.

00:17:35.680 --> 00:17:38.720
So this is another focus that can be an outcome.

00:17:38.720 --> 00:17:41.599
But now we have more than 2,000 data sets.

00:17:41.599 --> 00:17:42.559
Okay, that's a lot.

00:17:42.559 --> 00:17:43.839
That's that really a lot.

00:17:43.839 --> 00:17:44.880
He can do a lot of things with it.

00:17:45.039 --> 00:17:51.359
By data sets, you mean you're like real-world fire accidents from which you have the questionnaire filled out by the fire becomes fully full.

00:17:51.599 --> 00:17:53.359
Uh I just drew a missingness map.

00:17:53.359 --> 00:17:54.960
So where's data missing?

00:17:54.960 --> 00:17:56.319
Something that it's really cool.

00:17:56.319 --> 00:18:01.599
It's more than 97% of all the questionnaire parts for the listeners.

00:18:01.680 --> 00:18:09.279
The earlier version of the database, which I'm looking at it right now, it had 900 uh data sets, uh, was published in fire technology.

00:18:09.279 --> 00:18:14.640
So I assume it's it's an ongoing, we'll probably discuss about if this evolves and and how.

00:18:14.640 --> 00:18:17.279
So, can you give me like an example?

00:18:17.279 --> 00:18:20.079
Like people going into let's say warehouse fire.

00:18:20.079 --> 00:18:23.119
What kind of data are they supposed to leave you?

00:18:23.119 --> 00:18:24.319
What are you looking into?

00:18:24.319 --> 00:18:28.960
Like maybe you can maybe have some examples of how a questionnaire looks like a model one.

00:18:29.279 --> 00:18:35.680
A model one would be an apartment, a fully developed apartment file in a residential building.

00:18:35.680 --> 00:18:36.000
Okay.

00:18:36.000 --> 00:18:37.200
In an urban area.

00:18:37.200 --> 00:18:37.519
Okay.

00:18:37.519 --> 00:18:43.920
It's a very urban data set because the smaller rural departments are not that big represented.

00:18:43.920 --> 00:18:45.519
But I will come to that later.

00:18:45.519 --> 00:18:53.440
But you have this apartment fire and then it's um did you see any or did we have any injured fatalities?

00:18:53.440 --> 00:18:55.599
Um what were the means of rescue?

00:18:55.599 --> 00:18:57.920
So did they escape on their own?

00:18:57.920 --> 00:19:00.000
Were they rescued by the fire department?

00:19:00.000 --> 00:19:02.720
Did we see any smoke spreading?

00:19:02.720 --> 00:19:06.400
Because by law it must be pre prevented that smoke spreads.

00:19:06.400 --> 00:19:07.839
Did we see fire spreading?

00:19:07.839 --> 00:19:10.799
Did we see fire spreading towards the neighboring building?

00:19:10.799 --> 00:19:11.839
Firewall?

00:19:11.839 --> 00:19:13.279
Middle-aged knowledge?

00:19:13.279 --> 00:19:15.200
Uh yes, don't laugh, that's it.

00:19:15.200 --> 00:19:17.759
Did we have any injured firefighters?

00:19:17.759 --> 00:19:20.319
Were there was there a smoke alarm installed?

00:19:20.319 --> 00:19:22.640
Were then automatic fire some sprinter system all?

00:19:22.640 --> 00:19:26.720
But it's not in an apartment building, you don't have any sprinter systems in Germany.

00:19:26.720 --> 00:19:29.359
So this is more in the special construction branch.

00:19:29.359 --> 00:19:30.880
But you still save the data.

00:19:30.880 --> 00:19:32.160
We saved the data.

00:19:32.160 --> 00:19:35.759
And did you did we have any environmental impact and all this stuff?

00:19:35.759 --> 00:19:40.960
And from the very beginning, like we had 150 to 200 data sets.

00:19:40.960 --> 00:19:43.200
The numbers kept being the same.

00:19:43.200 --> 00:19:46.079
And I was very surprised that they didn't change anymore.

00:19:46.079 --> 00:19:53.759
And even if I have a look at sub-datasets from other states in Germany, maybe from Bavaria, from Northern Westphalia, and everywhere.

00:19:53.759 --> 00:19:55.359
I see the same numbers.

00:19:55.359 --> 00:20:05.920
I see the same numbers and the same percentages if the data is coming from Berlin or from a small rural fire department who sent me like 10 questionnaires.

00:20:05.920 --> 00:20:08.319
That's really an interesting aspect.

00:20:08.319 --> 00:20:10.400
How stable the data set is.

00:20:10.720 --> 00:20:19.680
By stable you mean in terms of uh the likelihood of the fires, like how many times they happened in different times, or no, no, I don't have a look at the likelihood.

00:20:19.839 --> 00:20:20.079
Okay.

00:20:20.079 --> 00:20:24.640
Because the questionnaire focus if there is a fire, what happened?

00:20:24.640 --> 00:20:28.000
And it's kind of biased because I don't look at the small fires.

00:20:28.000 --> 00:20:42.079
So the likeliness, I don't I don't care for the probability about the probability because from a fire department's point of view, I only come to your home, to your apartment, if this probability is summarized in a fire.

00:20:42.319 --> 00:20:51.039
So then you but by similarities you mean the outcomes, the numbers of injured per residential would it's always the same pattern.

00:20:51.359 --> 00:20:56.640
So the the likelihood that we see a spreading of smoke is always the same.

00:20:56.640 --> 00:20:57.200
Okay.

00:20:57.200 --> 00:21:00.559
About 60% of the cases, two-thirds.

00:21:00.559 --> 00:21:01.759
I see a smoke.

00:21:01.759 --> 00:21:08.400
Well, so we we can see a smoke spreading that violates the objective by building law.

00:21:08.720 --> 00:21:13.920
Are you capable of distinguishing this by the age of the building or oh yes, we did that.

00:21:14.079 --> 00:21:14.880
Very interesting.

00:21:14.880 --> 00:21:18.640
This was one of the few results where we were surprised.

00:21:18.640 --> 00:21:21.359
So you usually you have a very good gut feeling.

00:21:21.359 --> 00:21:25.119
Well, let's have a look at the numbers, for example, for injured people of fatalities.

00:21:25.119 --> 00:21:32.400
Injured people, if there is really a developed fire and you regularly see smoke spreading, as we learned, you will see injured people.

00:21:32.400 --> 00:21:34.000
And injured means coughing.

00:21:34.000 --> 00:21:38.000
You inhale smoke, and they were injured, they were transported to the hospital.

00:21:38.000 --> 00:21:44.240
It's not a very sophisticated ICD 10 something medical evaluation, but they were injured.

00:21:44.240 --> 00:21:45.839
30% in the cases.

00:21:45.839 --> 00:21:53.599
If there is a fire, the number of fatalities is from my experience comparatively high at approximately six to seven percent.

00:21:53.599 --> 00:21:55.839
I think six or seven percent of events.

00:21:55.839 --> 00:22:00.640
If there are 100 fires with six or seven fires, you will see a fatality.

00:22:00.640 --> 00:22:01.039
Okay.

00:22:01.039 --> 00:22:05.200
That's very high compared to the low fatality numbers we have.

00:22:05.200 --> 00:22:12.559
Because we have a lot of small fires that will never ever be big enough to ask the question toward the building.

00:22:12.559 --> 00:22:13.920
So they are not in our data set.

00:22:13.920 --> 00:22:15.200
Yeah, you're guys, yeah, of course.

00:22:15.200 --> 00:22:19.359
So and we see a lot of rescue efforts by the fire departments.

00:22:19.599 --> 00:22:22.480
As in the fire department has to rescue people.

00:22:22.960 --> 00:22:29.279
We have to rescue they are in the apartment, maybe standing at the windows, and then they can't escape on their own.

00:22:29.279 --> 00:22:31.839
So in 40% of the cases.

00:22:31.839 --> 00:22:32.960
That's quite high.

00:22:32.960 --> 00:22:36.000
And then we did just the last master thesis.

00:22:36.000 --> 00:22:44.480
We work closely together with the Technical University of Munich, and we now have 30 thesis, 13 evaluating the data set.

00:22:44.480 --> 00:22:45.359
And that that's it.

00:22:45.359 --> 00:22:51.519
And then, of course, what we always see is smoke spreading, rarely fire spreading towards another unit.

00:22:51.839 --> 00:22:56.880
Okay, but the the the statistics you said, like six percent fatalities, 30% injured.

00:22:56.880 --> 00:22:59.279
This is for the entire dataset.

00:22:59.279 --> 00:23:02.960
How about when you look by occupation?

00:23:02.960 --> 00:23:04.640
It's all residential.

00:23:04.960 --> 00:23:15.119
What's really interesting is that we only see three types of buildings where where we have an elevated level of injured and fatalities.

00:23:15.119 --> 00:23:18.559
And by gut feeling we knew that, but we couldn't prove it.

00:23:18.559 --> 00:23:18.960
Okay.

00:23:18.960 --> 00:23:20.720
It's jail.

00:23:20.720 --> 00:23:21.279
Okay.

00:23:21.279 --> 00:23:24.880
Of course, they have that's a challenging evacuation system.

00:23:24.880 --> 00:23:27.599
Yeah, they don't have an escape plan, actually.

00:23:27.599 --> 00:23:32.319
Some of them have, but obviously, yeah, some of them have, but not in that case.

00:23:32.319 --> 00:23:37.680
So jail has higher values, not a hospital, what you would expect.

00:23:37.680 --> 00:23:39.839
I would expect retirement house.

00:23:39.839 --> 00:23:40.559
It is.

00:23:40.559 --> 00:23:42.000
Yeah, it exactly is.

00:23:42.000 --> 00:23:48.400
And what we see in the data, and we try to prove that with the next thesis, it's a social aspect.

00:23:48.559 --> 00:23:48.880
Okay.

00:23:49.200 --> 00:23:55.440
Because if you put a lot of people, poor people, regularly, call the poorer, in.

00:23:55.440 --> 00:24:06.640
And what we can see in the data set with another PhD thesis, if you are married, the likeness that you die from a fire is half even less likely than if you're a widow.

00:24:06.640 --> 00:24:07.920
So widows rise.

00:24:08.079 --> 00:24:08.240
Okay.

00:24:08.480 --> 00:24:18.319
And that's very interesting because it gives us a perspective of a social aspect and a residential building like student apartments and all this where a lot of people pull people up put together.

00:24:18.319 --> 00:24:22.000
We see more likeness of being injured or four fatalities.

00:24:22.240 --> 00:24:34.079
Well, this is well, this is challenging that I I keeps popping out, and by common sense, I think it's it's reasonable that you're most likely to be injured in a fire in your own house or or in general in a residential building.

00:24:34.400 --> 00:24:37.119
But on a scale where you see that many fires, yeah.

00:24:37.119 --> 00:24:49.119
What I think it's what is really reasonable as a system of explanation is that if you don't pay that much rent, your landlord can't put that much money in the building.

00:24:49.119 --> 00:24:57.359
And if you're rich, you have a very thick door towards the stable, for example, because you don't want the noise from the neighbor, don't you?

00:24:57.359 --> 00:24:58.000
Yeah, yeah.

00:24:58.000 --> 00:24:59.519
You have modern windows.

00:24:59.519 --> 00:25:05.039
Okay, so our So in an old building you have the thin doors and the thin windows.

00:25:05.519 --> 00:25:12.079
Okay, so you want to say that within the residential subset, you have also full subsets that are socially linked.

00:25:12.319 --> 00:25:12.480
Okay.

00:25:12.480 --> 00:25:13.359
Okay, to the risk.

00:25:13.359 --> 00:25:15.599
So this is one one of the explanations.

00:25:15.599 --> 00:25:18.559
And what was really interesting, now I'm coming back to the surprise.

00:25:18.559 --> 00:25:23.920
We thought the older the building, the more likely it will be that you see smoke spreading.

00:25:23.920 --> 00:25:25.359
That's what I would say as well.

00:25:25.359 --> 00:25:28.160
None really, absolutely no correlation.

00:25:28.160 --> 00:25:29.839
Lock it, regression.

00:25:29.839 --> 00:25:38.720
So we really did some math on it, some statistical doings, and in the end, we saw it connect with the how many floors are there.

00:25:38.720 --> 00:25:39.200
Okay.

00:25:39.200 --> 00:25:48.640
And the bigger buildings in urban areas.

00:25:48.640 --> 00:25:50.720
Usually poorer people live there.

00:25:50.720 --> 00:25:51.200
Okay.

00:25:51.200 --> 00:26:03.039
And what we saw is in in in the last thesis, if there's a fire in your kitchen or your uh bedroom and all this, 40 to 50 percent likeliness that we see smoke in the stairwell.

00:26:03.039 --> 00:26:06.480
If there is a fire in the cellar, more than 80%.

00:26:06.480 --> 00:26:07.759
But guess what?

00:26:07.759 --> 00:26:13.440
By German building law, you need a self-closing 30 minutes fire resistor towards the cellar.

00:26:13.440 --> 00:26:14.799
But usually, what is it?

00:26:14.799 --> 00:26:16.799
Kept open because it's annoying.

00:26:16.799 --> 00:26:17.359
Yes, yes.

00:26:17.359 --> 00:26:18.319
If you go down the center.

00:26:18.319 --> 00:26:19.920
Mine is all also open.

00:26:19.920 --> 00:26:21.599
And we see that in the data.

00:26:21.599 --> 00:26:23.200
Uh we see this common sense.

00:26:23.519 --> 00:26:24.880
We're moving to smoke spreading.

00:26:24.880 --> 00:26:40.000
Uh, could you define what does the firefighter who's doing your inquiry, your your uh questionnaire, when do they say it's the smoke is any proof of smoke outside of the compartment of origin is it's very obvious.

00:26:40.160 --> 00:26:45.200
Yeah, it's very it's not about some well now it smells in the neighboring unit.

00:26:45.200 --> 00:26:49.039
Usually if we say it's smoke spreading, it's dark.

00:26:49.039 --> 00:26:50.319
Okay, so you can see it.

00:26:50.319 --> 00:26:55.920
So this is one of the things when I say it's not very scientific, and we're doing science with it.

00:26:56.000 --> 00:26:56.319
Yeah.

00:26:56.480 --> 00:26:58.559
Because it just what is obvious.

00:26:58.799 --> 00:27:04.799
So you found smoke spread in um more than half of all your cases.

00:27:04.799 --> 00:27:11.039
You said you're it's it's a contradiction of the German code which explicitly gives you measure to prevent.

00:27:11.039 --> 00:27:13.039
Do you know what failed?

00:27:13.039 --> 00:27:16.559
Like was it mostly the doors kept open manually?

00:27:16.960 --> 00:27:18.559
No, it's not keeping the doors open.

00:27:18.559 --> 00:27:20.319
It's it's generally the apartment door.

00:27:20.319 --> 00:27:28.559
If you don't have a self, if you don't have a self-closing mechanism applied to your door, and you try to get out, you leave the door.

00:27:28.559 --> 00:27:40.960
What we regularly see is if someone's trying to flee and and and he comes to the border between hot and cold, his system, his body collapses and he's lying right in the door, even though it's, for example, self-clothing.

00:27:40.960 --> 00:27:47.279
And what we saw in a subset very new data is like almost 20 years ago, they introduced self-closing doors.

00:27:47.279 --> 00:27:49.039
No, 30 years ago in Bavaria.

00:27:49.039 --> 00:28:02.480
And what we now see to compare to the state that don't have that, that the preding likeness reduced to the stairwell, but not down to the level what would be expected from a technical point of view.

00:28:02.480 --> 00:28:07.279
Like a self-closing mechanism fails in 1% of the case, something like that.

00:28:07.279 --> 00:28:08.240
Why is that?

00:28:08.240 --> 00:28:09.920
Multitude of explanations.

00:28:09.920 --> 00:28:14.720
But one could be they disengage the self-closing mechanism because it's annoying.

00:28:14.880 --> 00:28:15.039
Yeah.

00:28:15.200 --> 00:28:20.880
You have your baby on your arm, you shoot in your in the back, and you try to get in, and the door is self-closing, all this crap.

00:28:20.880 --> 00:28:22.799
So it doesn't work.

00:28:22.799 --> 00:28:24.559
And it needs maintenance, of course.

00:28:24.559 --> 00:28:29.680
And it's not checked like any other fire-resistant door at another place in the building.

00:28:30.000 --> 00:28:33.200
What about the pathways like the building ventilation system?

00:28:33.200 --> 00:28:35.519
Like I'm a smoke control engineer.

00:28:35.519 --> 00:28:40.079
A lot of my work is towards smoke control, which is obviously a ventilation system.

00:28:40.079 --> 00:28:53.200
But the the first step of successful smoke control is to shut down the HVC in the building, which means I have to turn everything off and close in the big building a thousand fire dampers.

00:28:53.200 --> 00:28:57.680
Have you seen any uh signs of that system?

00:28:58.000 --> 00:29:02.559
Yeah, what we see is apartment door, most likely, then the window, obviously.

00:29:02.559 --> 00:29:07.440
You mean true exterior, so the smoke exits and enters towards another unit.

00:29:07.440 --> 00:29:07.839
Okay.

00:29:07.839 --> 00:29:14.559
So if I talk about smoke spreading, I mean violating the border of the very apartment or the so-called unit.

00:29:14.559 --> 00:29:18.960
So you're fine with the fire being in one unit and if it goes to another unit.

00:29:18.960 --> 00:29:23.599
If you have a fire in your kitchen, we expect there to be smoke in your bedroom.

00:29:23.599 --> 00:29:23.839
Yeah.

00:29:23.839 --> 00:29:24.720
I don't care.

00:29:24.720 --> 00:29:27.119
Questionnaire we say smoke spreading, no.

00:29:27.119 --> 00:29:31.680
But if your neighbor so building law is mainly protecting you from your neighbor.

00:29:31.680 --> 00:29:38.160
So there we see the window, and what we see are shafts inside the apartment building.

00:29:38.160 --> 00:29:41.279
And what we see is the very experience we have.

00:29:41.279 --> 00:29:50.319
For example, you have a fire on the first floor, and then at the top floor, you regularly see smoke spreading by the shaft.

00:29:50.319 --> 00:29:51.599
This is very common sense.

00:29:51.599 --> 00:29:56.960
You have a kitchen shaft, for example, or from the bathroom, and then it tries to leave the building.

00:29:56.960 --> 00:29:58.799
It draws back.

00:29:58.799 --> 00:29:59.599
So it's double.

00:29:59.599 --> 00:30:02.559
The likeliness for the top floor.

00:30:02.559 --> 00:30:06.079
So, from a firefighting perspective, that's interesting too.

00:30:06.079 --> 00:30:08.640
Because usually you would check the next floor.

00:30:08.640 --> 00:30:16.640
And I myself, knowing the data, I regularly send someone to look at the top floor in the bathroom, in the kitchen, if there's smoke.

00:30:16.640 --> 00:30:18.480
And I even have a fatality once.

00:30:18.480 --> 00:30:23.200
An ordinary kitchen fire on the first floor led to a fatality in the top floor.

00:30:23.599 --> 00:30:25.839
It was not me, but my colleague from office, Jagosh.

00:30:25.839 --> 00:30:42.160
Uh, he was once investigating a case where there was a fatality in a building, and allegedly the wind conditions in that day were so that CO produced by someone else's heater was released through the chimney, and the wind pushed that CO into the chimney.

00:30:42.559 --> 00:30:43.200

229 - Learning from 900 fires with Björn Maiworm

Listen On

Recent Episodes

Support On