May 28, 2025
203 - The lessons from repeating Jin's experiment on visibility in smoke
The player is loading ...
I've finally done it. We've repeated Jin's experiment! I thought I knew-it-all about that experiment, but boy... knowing and doing it are two different things. I can say, I've finally cleared my mind on some thoughts after this, which I am finally happy to share with all of you!
First things first, massive thanks to my partner in crime Wai-Kit Wilson Cheung, from the group of prof. Xinyan Huang, who was the man on the ground doing the experiments with me. Together we went further into this model, than ever before.
The revelations are far-reaching. We found that Jin used extraordinary lighting conditions—180 lux background brightness and impossibly bright signage—far from realistic building emergency conditions. Background brightness emerges as perhaps the most critical factor in determining what can be seen through smoke, with dramatic differences between light-emitting and light-reflecting signs. Most significantly, the experiment's careful constraint of sign size (using proportionally larger signs at greater distances) created elegant mathematics but removed a crucial real-world variable from the model.
These insights have profound implications. Engineers likely overestimate visibility in many scenarios, particularly with reflective signage. The widely used K-values (3 for reflective signs, 8 for light-emitting signs) appear reasonably conservative for typical building conditions, though higher values might be warranted in darker environments. Most provocatively, simply increasing sign size would almost certainly improve evacuation safety, yet our current models provide no mechanism to quantify this benefit.
Fire safety practitioners will find this episode transformative, offering both practical guidance and theoretical understanding. Should we stick with visibility distance or shift to smoke density as our primary metric? How can we balance lighting conditions to optimize visibility of both obstacles and signage? And most critically, how might next-generation visibility models better serve real-world building safety? These are things we currently work on.
If you look for reading, check the paper on the extinction coefficient by the German colleagues: https://arxiv.org/abs/2306.16182
If you strive for more podcast episodes:
- this one covers historical Jin's experiment: https://www.firescienceshow.com/162-experiments-that-changed-fire-science-pt-9-jins-experiment-on-visibility-in-smoke/
- this one is on soot and smoke: https://www.firescienceshow.com/163-fire-fundamentals-pt-11-soot-in-fire-safety-engineering/
- and this one our general thoughts about modelling visibility and new pathways we see forward: https://www.firescienceshow.com/030-visibility-prediction-framework-with-lukas-arnold/
The research was funded by the National Science Centre, Poland, based on a contract for the implementation and financing of a research project OPUS LAP No 2020/39/I/ST8/03159 and by Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under the project number 465392452 , for the joint project: “Visibility Prediction Framework – a next-generation model for visibility in smoke in built environment”.
----
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 - Introduction to Visibility in Smoke
03:52 - What is Visibility in Smoke?
08:36 - Jin's Historical Experiments Explained
15:37 - Repeating Jin's Experiments Today
21:40 - Background Brightness: A Critical Variable
25:30 - Sign Size: The Overlooked Factor
31:42 - Light Reflecting vs Emitting Signs
39:38 - Engineering Implications and Future Research
42:05 - Conclusions and Design Considerations
WEBVTT
00:00:00.462 --> 00:00:02.508
Hello everybody, welcome to the Fire Science Show.
00:00:02.508 --> 00:00:09.233
Today we are talking visibility in smoke and if you follow the podcast, you know that this topic is very dear to my heart.
00:00:09.233 --> 00:00:19.969
It is something that I research on my own and that has been a very important part of my scientific career thus far and probably will be for ongoing years.
00:00:19.969 --> 00:00:25.524
Many years ago I've done an episode with Lucas Arnold about visibility prediction framework.
00:00:25.524 --> 00:00:36.896
Our new back then grant that we were just starting at that point that we hope that will allow us to revolutionize the way how visibility is assessed in fire safety engineering.
00:00:36.896 --> 00:00:44.173
This project is actually ongoing, but I finally have some findings that I can share with you and that makes me super happy.
00:00:44.173 --> 00:00:47.506
Have some findings that I can share with you and that makes me super happy.
00:00:47.527 --> 00:01:12.887
A year ago I have recorded an episode in the series I've called the experiments that changed fire science and in that episode I've covered experiments by Japanese scientist Jin, which are the basis of this model, and I said back then that Jin's model needs urgent repeat, that we really need to do it, and I knew we were going to repeat it because we've built the rig to do it, but shortly after that episode was published a very happy thing to me happened.
00:01:12.887 --> 00:01:28.605
I got a student from Hong Kong, Wai-Kit Wilson-Chung, from Hong Kong Polytechnic University, from Xinyan Huang's Wai-Kit, stayed with me for six months and he was really focused on doing those experiments.
00:01:28.605 --> 00:01:31.861
So, together with Waikid, we've actually redone the Jin's experiments.
00:01:31.861 --> 00:01:58.135
We've redone them, we've processed the data and today I am happy to show you some findings, because while I think I understood the Jin experiment and you can listen to that in the episode about Gene's experiments, why I think I knew what Gene done actually repeating those experiments, you know, doing that research on your own really opens your eyes on what is important in those experiments.
00:01:58.135 --> 00:02:03.668
And this is a broader reflection the papers don't tell you the full story.
00:02:03.668 --> 00:02:05.507
You need to talk to the scientists to tell you the full story.
00:02:05.507 --> 00:02:20.344
You need to talk to the scientist to really learn the full story what was hard, what was easy, where the challenges were, and I think we've narrowed down where the challenges of the visibility smoke model lie really, and this is what's gonna be said in this episode.
00:02:20.344 --> 00:02:24.112
I've presented this at the sfp, edinburgh.
00:02:24.112 --> 00:02:25.515
So finally, a time.
00:02:25.915 --> 00:02:28.764
This reaches also my dear fire science show audience.
00:02:28.764 --> 00:02:32.539
So let's spin the intro and jump into the episode.
00:02:32.539 --> 00:02:38.584
Welcome to the fire science show.
00:02:38.584 --> 00:02:42.026
My name is vojtěj věkřínski and I will be your host.
00:02:42.026 --> 00:03:11.561
The FireSense show is into its third year of continued support from its sponsor of our Consultants, who are an independent, multi-award winning fire engineering consultancy with a reputation for delivering innovative safety-driven solutions.
00:03:11.561 --> 00:03:25.300
As the UK leading independent fire risk consultancy, ofar'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.
00:03:25.300 --> 00:03:41.409
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.
00:03:41.409 --> 00:03:49.968
If you're keen to find out more or join OFR Consultants during this exciting period of growth, visit their website at ofrconsultantscom.
00:03:49.968 --> 00:03:52.074
And now back to the episode.
00:03:53.040 --> 00:04:02.481
So when I talk about visibility in smoke and recently I have a lot of chances to talk about this model I like to start with very difficult questions.
00:04:02.481 --> 00:04:04.622
Easy questions are always the most difficult to answer.
00:04:04.622 --> 00:04:07.865
What is visibility in smoke Like?
00:04:07.865 --> 00:04:09.945
What are we really talking about?
00:04:09.945 --> 00:04:19.591
Because, on one hand, you could consider it being a measure of a distance at which something can be observed.
00:04:19.591 --> 00:04:26.173
I think that would be the simplest, easiest definition of what visibility in smoke is.
00:04:26.173 --> 00:04:34.997
And this definition probably is right, but it has a lot of details in it, like what does it mean you can see something?
00:04:34.997 --> 00:04:39.588
Is it enough that you see a glimpse of light?
00:04:39.588 --> 00:04:41.680
Is it enough that you see a shape, a blurred out shape?
00:04:41.680 --> 00:04:47.723
Or you have to be able to process data that's shown on the thing that you're observing.
00:04:47.723 --> 00:04:57.050
How certain you have to be that the thing you observe is the thing that you observe, and those things change when the visibility changes.
00:04:57.692 --> 00:05:02.725
If you're a physicist, then perhaps this is ill-defined for you.
00:05:02.725 --> 00:05:07.595
Perhaps it's a distance at which the light can still pass through obscuring medium.
00:05:07.595 --> 00:05:21.629
So if you point a ray of light through smoke, fog, whatever aerosol that disturbs it, the light will start to decay, it will scatter to the sides, it will be absorbed by the particles.
00:05:21.629 --> 00:05:32.642
So with every meter you will have less and less light passing through, and if you capture the point at which the light has decayed completely, that's the point where you cannot see the light anymore.
00:05:32.642 --> 00:05:37.153
So perhaps this distance is visibility in smoke In some way.
00:05:37.153 --> 00:05:38.983
This is how we measure it.
00:05:38.983 --> 00:05:39.946
This is how we measure it.
00:05:39.946 --> 00:05:52.233
In laboratories we have densitometer, optical densitometers which emit light, which measure light, and they measure how much light is lost in between the emitter and the target.
00:05:52.233 --> 00:05:56.170
So a physicist could define visibility as such a distance.
00:05:56.701 --> 00:06:11.567
If you think about genes experiments the ones that brought us the visibility in smoke model that we use in fire science, visibility was actually the initial condition or actually the assumption of the study, the distance itself, the meters.
00:06:11.567 --> 00:06:25.158
So Jin built his facility in such a way that you could observe the science from either 5, 10, or 15 meters and in all honesty, we still have not found how he measured the 10 meter distance.
00:06:25.158 --> 00:06:32.458
But anyway, the rig is reported to be able to measure at 5, 10 and 15 meter distance.
00:06:32.458 --> 00:06:36.687
So it's discrete points in space rather than continuous spectrum.
00:06:36.687 --> 00:06:41.723
And in Gene's experiment the participant was placed at this distance.
00:06:41.723 --> 00:07:03.245
They were observing a light source through a smoke and they had a little widget that they could turn around and this widget would make the light dimmer and dimmer and they would just find the dimmest light they could see through the smoke and then Jin would save that data and use that in the further research.
00:07:03.245 --> 00:07:08.074
So in Jin's research the visibility is actually a fixed number.
00:07:08.074 --> 00:07:09.867
It's either 5, 10, or 15.
00:07:09.867 --> 00:07:12.569
There's no intermediate values for that.
00:07:12.569 --> 00:07:14.187
It's quite interesting when you think about it.
00:07:14.187 --> 00:07:25.329
If you're a firefighter, the visibility range would be something that's very natural to you, because that's the distance you can see in your fires, how far into the building, how far into the field you can see.
00:07:25.329 --> 00:07:37.531
So, perhaps closest to the first, most simple definition that I've brought up and allegedly, the observations of firefighters the distances they were saying are 10 meters is probably enough.
00:07:37.531 --> 00:07:44.923
Those were the background for creating the discrete values of visibility that we use today for engineering.
00:07:44.923 --> 00:07:47.964
But it does not mean the same thing for an engineer.
00:07:48.427 --> 00:07:55.387
If you're practicing fire safety engineering, you are not really assessing the ability to observe things through smoke.
00:07:55.387 --> 00:08:05.351
You're not assessing light decay in smoke and you're definitely not assessing the critical brightness of a widget that you can observe through smoke.
00:08:05.351 --> 00:08:10.307
No, you are applying a very simple mathematical correlation.
00:08:10.307 --> 00:08:14.641
In one part of the correlation you put smoke density in.
00:08:14.641 --> 00:08:17.588
How much smoke do you have in your space?
00:08:17.588 --> 00:08:22.461
And this is something you know from your CFD analysis, from your zone model analysis.
00:08:22.461 --> 00:08:24.021
You know your soot yields.
00:08:24.021 --> 00:08:25.483
You know your yields of combustion.
00:08:25.483 --> 00:08:29.464
You can calculate how much soot gets emitted to your room.
00:08:29.464 --> 00:08:41.932
You know the flows in your room, you can calculate how much smoke is there in any given part of my building while performing fire safety engineering.
00:08:42.812 --> 00:08:52.256
And when you do this fire safety engineering, you have to show those results to someone and you usually choose to present them as visibility in smoke.
00:08:52.256 --> 00:08:54.798
So I'll just give a quick recap.
00:08:54.798 --> 00:08:58.868
How does the smoke density turn into visibility in our modeling?
00:08:58.868 --> 00:09:00.592
Because it's a very simple correlation.
00:09:00.592 --> 00:09:23.243
You basically have a factor we call it large K usually and it takes values of 3 for light reflecting signs, 8 for light emitting signs I'll come back to this at the end of the episode and you basically subdivide it by your smoke density multiplied by specific extinction coefficient of smoke.
00:09:23.243 --> 00:09:30.410
The extinction coefficient is something we know from experimental work from Mulholland's and I'll also come back to this at the end of the episode.
00:09:31.059 --> 00:09:37.909
The logmock density you know from your CFD, and so you have all the things that you need to calculate visibility.
00:09:37.909 --> 00:09:45.446
Well, does this mean that you will be able to see for 10 meters in your building if you've calculated 10 meters visibility?
00:09:45.446 --> 00:09:46.470
Of course not.
00:09:46.470 --> 00:10:00.590
It just means that if an entire space was filled with a smoke of density, like you have in this one point of space that you've just measured, then probably in the whole room you would have visibility distance of something like 10 meters.
00:10:00.590 --> 00:10:06.842
That's pretty much what it says, but it doesn't tell you anything about what you can see, what you cannot see.
00:10:06.842 --> 00:10:17.600
It doesn't tell you anything about where the light will stop in your room and it definitely does not tell you at what kind of brightness you can observe things in your compartment.
00:10:17.600 --> 00:10:27.985
It just allows you to translate value that's perhaps a little less understandable the density of smoke into a value that people can understand visibility.
00:10:27.985 --> 00:10:38.113
That's the trick that has been used as the backbone of fire safety engineering for five decades now, I guess, and I cannot say we've been pretty successful with it.
00:10:38.113 --> 00:10:50.293
It's just that it's so profound, so impactful in our engineering that I really find it not great that we have such a coarse approximation that we use for very significant engineering decisions every day.
00:10:50.895 --> 00:10:57.472
So now let's talk a little bit about the story of repeating genes experiments, because it was a very interesting story.
00:10:57.472 --> 00:11:04.043
First we've done I thought I'm clever, you know I'm not gonna, you know use a widget to control the brightness.
00:11:04.043 --> 00:11:06.210
That just gives me one point of data.
00:11:06.210 --> 00:11:13.374
I am a clever scientist and I figured out a way how I can get much more data from the same experiment.
00:11:13.374 --> 00:11:20.688
So in my experiment, contrary to Jin's, I have a huge LED screens at the back of my experiment.
00:11:20.688 --> 00:11:35.201
Jin certainly did not have those in 1970s and instead of projecting one sign, I can project as many as I want and I can make them dimmer and dimmer, and dimmer and control this through my software.
00:11:35.201 --> 00:11:45.765
So when Gene was able to project one sign on his frosted glass, I am capable of projecting however much the hell I want.
00:11:45.765 --> 00:11:47.270
So I definitely did that.
00:11:47.311 --> 00:11:55.456
I did that with my student Pavel, and we were trying to do those signs emitted on a screen of TV and replicate Gene's experiment.
00:11:55.456 --> 00:12:04.043
And in Gene's experiment there is this one way he's presenting results and it's called the dimensionless brightness.
00:12:04.043 --> 00:12:08.360
It's not very useful but it's kind of relevant, so I'll talk you through.
00:12:08.360 --> 00:12:14.466
So dimensionless brightness is that Gene took the brightness of the sign that he was projecting.
00:12:14.466 --> 00:12:25.508
He was measuring that with pretty complicated optical measurements that I cover in the previous podcast episode and he was subdividing that by the brightness of the background.
00:12:25.508 --> 00:12:30.586
So if the dimensionless brightness is above one, that means the sign was brighter than the background.
00:12:30.586 --> 00:12:39.726
If the dimensionless brightness is less than one, it means that the sign was darker than the background and the higher the value, the brighter the sign right.
00:12:39.726 --> 00:12:55.606
So with Paavo we start to calculating dimensionless brightness for our results and we very quickly see that we're nowhere close to Gene's results, like we are nowhere close to the range of dimensionless brightness that Gene's used.
00:12:55.875 --> 00:13:05.256
Our TV is not bright enough, and that was the first shocker because I felt the TVs, the setup that we've built, was imitating the building conditions fairly well.
00:13:05.256 --> 00:13:06.921
I've been in a lot of buildings.
00:13:06.921 --> 00:13:12.080
I've been in a lot of buildings in fires, actually, because we're doing those fire tests in them.
00:13:12.080 --> 00:13:19.605
So I know how evacuation routes in a building in emergency lighting in your evacuation conditions should look like.
00:13:19.605 --> 00:13:31.004
I have a pretty good idea of that and I thought my tvs are fairly well representative of that environment, whereas now we see that we're nowhere close to jeans results.
00:13:31.004 --> 00:13:33.509
That was quite a shocker to us.
00:13:33.811 --> 00:13:46.054
And then wiki chunk came and the first job I gave gave to Wilson was to build me a light emitting source that could match gins and boy, that was a journey.
00:13:46.054 --> 00:13:52.227
We've 3D printed the box and the box was lined up with light reflecting foil.
00:13:52.227 --> 00:13:58.826
Then we've put a bunch of LEDs into the box, like it took us like five iterations of adding more and more and more and more and more LEDs into the box.
00:13:58.826 --> 00:14:04.503
Like it took us like five iterations of adding more and more and more and more and more LEDs to the box until we've reached the box.
00:14:04.503 --> 00:14:13.442
That's like literally, you know, a lighthouse lantern you cannot look straight into the box because it blinds you for like 30 seconds.
00:14:13.442 --> 00:14:14.899
That's how bright it is.
00:14:14.899 --> 00:14:16.441
And now we match Jin's.
00:14:16.441 --> 00:14:20.524
And now we match the Jin's frosted glass brightness.
00:14:21.235 --> 00:14:24.326
And this was like a shocker, but also an eye-opener.
00:14:24.326 --> 00:14:37.006
Like Jin has used extremely bright sources in their experiments, extremely bright sources, and therefore he could observe those signs through very, very dense smoke.
00:14:37.006 --> 00:14:46.144
If you look at Jin's experiments closely, you'll notice that the ranges of extinction coefficients that Jin is working with they're reaching up to two.
00:14:46.144 --> 00:14:48.724
This is unbelievably dense smoke.
00:14:48.724 --> 00:14:54.664
This is a smoke at which you will not see your hand if you stretch it in front of you.
00:14:54.664 --> 00:14:58.586
You probably will not even see your elbow in smoke of this density.
00:14:58.586 --> 00:15:05.835
It's unbelievably dense smoke and he was carrying observations and calculating those models in such dense smoke conditions.
00:15:05.835 --> 00:15:15.501
So, ah, this is already something that moves Jin's experiments away from the space of real buildings, real engineering.
00:15:15.501 --> 00:15:18.263
The sources are too bright, the smoke is too dense.
00:15:18.263 --> 00:15:25.636
It's not something that we commonly work as fire safety engineers and again, you can read the paper as many times as you like.
00:15:25.697 --> 00:15:41.220
You can, you know, spend ages on studying what has been done in Japan in 1970s, but it really took us to repeat the experiment to very quickly realize what we are dealing with with those light sources.
00:15:41.220 --> 00:15:53.506
Very interesting, and therefore I find this podcast episode of real value, because previously I was speaking about what I think the problems with genes experiments are.
00:15:53.506 --> 00:15:56.224
Now I know what they are because I've run into them.
00:15:56.224 --> 00:16:00.505
But that's not everything that we've realized while repeating Jin's experiments.
00:16:00.505 --> 00:16:05.923
Actually, we've learned a lot more with Wilson, so let's try digesting that.
00:16:07.296 --> 00:16:42.605
When Jin was doing his experiments, the profound finding in his study and I think that's the biggest discovery of Jin really was that the relationship between the extinction coefficient and the distance at which you can observe the sign is fairly constant across a range of sign sizes, distances, background brightness and just the brightness of the sign, which meant that he could collapse a lot of results into straight lines and just find a linear correlation between the variables, therefore creating the model that we are currently using.
00:16:42.605 --> 00:16:44.100
I'm not sure he was creating a model.
00:16:44.100 --> 00:16:51.822
Actually I think he was just looking for a relationship that allows his engineering to be done, whereas we turned it into the model.
00:16:51.822 --> 00:16:58.086
We, as a collective fire safety engineering community, we've started using this as a model to predict visibility.
00:16:58.086 --> 00:17:06.837
He was just representing the outcomes, I think, but anyway, in his experiments he was doing them with different range of lights with different sizes.
00:17:06.837 --> 00:17:23.184
As I said, he was also using dimensionless brightness of a sign to reduce the number of variables in the studies and he was able to actually collapse those results into one very elegant line and I think you can see this plot in sfp handbook.
00:17:23.184 --> 00:17:28.461
We'll see if it's there in the next edition of handbook which is released just in a few days.
00:17:28.461 --> 00:17:31.406
But he was basically able to simplify this a lot.
00:17:31.907 --> 00:17:34.721
Now we've repeated this experiment as closely as we could.
00:17:34.721 --> 00:17:53.957
We really paid a lot of attention to repeat these experiments as closely, as accurately as we could, following whatever is being said in Jun's papers and our results, while they do keep this linear relationship between brightness and extinction coefficient at which you can observe the sign.
00:17:53.957 --> 00:17:59.954
So in some way we confirm that there is this close linear relationship.
00:17:59.954 --> 00:18:01.280
That's the basis of the model.
00:18:01.280 --> 00:18:04.144
That's, I think, a very good finding that we confirmed that.
00:18:04.144 --> 00:18:10.628
Besides that we have not been able to collapse them this elegantly as Jin did.
00:18:10.628 --> 00:18:14.105
So our results just do not collapse that easy.
00:18:14.105 --> 00:18:17.965
They do not collapse that perfectly, and we were looking into why would they not collapse that perfectly?
00:18:17.965 --> 00:18:18.589
And we were looking into why.
00:18:18.589 --> 00:18:20.196
Why would they not collapse that perfectly?
00:18:20.196 --> 00:18:28.157
And we started thinking it could be related to the background conditions, to the background brightness.
00:18:28.157 --> 00:18:43.281
We've used led strips in the in the room so we had a very nice uniform brightness across the room, jim using incandescent lights which he was turning on and off, so he definitely had non-uniform light distribution.
00:18:43.281 --> 00:18:57.722
But as we were looking into that, we realized that perhaps there's a bigger story to be told about the brightness which is not told by Jin and which I think the fire safety community needs to know and understand.
00:18:57.722 --> 00:18:59.787
So what I mean by that?
00:19:00.048 --> 00:19:13.288
If you look at the perhaps most influential graph of all of them in Gene's papers, there's a graph that shows you on the y-axis, the dimensionless number which is smoke density multiplied by visibility.
00:19:13.288 --> 00:19:17.319
On the x-axis, the dimensionless brightness of your sign.
00:19:17.319 --> 00:19:27.144
And in this plot this plot basically, is the basis for the values of k3 and 8, really, that's the original of those values.
00:19:27.144 --> 00:19:30.055
And why it's important?
00:19:30.055 --> 00:19:37.506
Because we've noticed that in the top left-hand of the plot there's a note external light 180 lux.
00:19:37.506 --> 00:19:42.666
Gene has carried his experiments in extremely bright conditions.
00:19:42.666 --> 00:19:50.808
180 lux is not something you normally have in your buildings in the evacuation phase.
00:19:50.808 --> 00:19:59.327
In Poland the minimum is one lux, of course, but in normal buildings like 100 is already a lot, really really a lot.
00:19:59.327 --> 00:20:07.086
180 is a very, very, very bright room and you normally do not evacuate through spaces with such an immense brightness.
00:20:07.086 --> 00:20:10.884
And why I say there's a story to be told about brightness.
00:20:11.476 --> 00:20:38.107
As soon as you go into the laboratory, as soon as you start repeating those experiments, as soon as you start playing with that, you immediately notice how impactful the background brightness is, how important is the brightness of the environment in which the evacuation takes place and how quickly it changes the ability to see or not see the evacuation signage.
00:20:38.107 --> 00:20:48.344
We had experiments in which you would set a brightness of a sign and at some external brightness you would not see even a glimpse of the sign.
00:20:48.344 --> 00:20:52.561
And if you tune the brightness down, you would see the sign perfectly.
00:20:52.561 --> 00:20:56.561
The same smoke, the same sign, just changing the background conditions.
00:20:56.561 --> 00:21:04.144
This is how big impact the brightness of the environment can have on visibility and actually it makes sense.
00:21:04.144 --> 00:21:05.226
It makes perfect sense.
00:21:05.226 --> 00:21:12.708
So if you have a scattering medium in your space, like smoke is, you can think about it.
00:21:12.708 --> 00:21:16.479
It's an averaging filter of light that's going through the room.
00:21:16.479 --> 00:21:23.328
Basically, all the light that passes through that scattering medium bounces off particles, let's say mixes.
00:21:23.328 --> 00:21:35.521
I'm not sure if it's a good word to be used about physics of light, but let's say that the light from different sources mixes and that's why your eyes cannot tell those lights apart, because they're mixed.
00:21:36.143 --> 00:21:39.827
Now you will also see the brighter thing.
00:21:39.827 --> 00:21:46.546
If you have a very bright point next to not such a bright point, you of course see the brighter point first.
00:21:46.546 --> 00:21:58.506
So the sign has to be more bright than the background, and the bigger the difference between the background and the sign is, the easier it's going to be to observe the sign.
00:21:58.506 --> 00:22:00.277
I think that that's quite reasonable.
00:22:00.277 --> 00:22:05.407
Now your signs have a very specific light intensity that they emit.
00:22:05.407 --> 00:22:06.876
It's in their characteristic.
00:22:06.876 --> 00:22:13.698
Therefore, if your background is too bright, you cannot increase the brightness of your signs anymore.
00:22:13.698 --> 00:22:16.522
You will start losing visibility of those signs.
00:22:16.522 --> 00:22:19.728
And, trust me, if you go into the experiment.
00:22:19.728 --> 00:22:28.224
If you go into the laboratory, if you start observing those signs through the glass, and if you're ever in Warsaw, you're very welcome to come to my laboratory.
00:22:28.224 --> 00:22:35.105
I'll show it to you you will notice that the change in the background brightness perhaps has the most profound impact on the outcomes of the study.
00:22:35.595 --> 00:22:41.904
And now, why it's important for fire safety engineering is because of that assumption of Gin's model 180 lux.
00:22:41.904 --> 00:22:44.131
We're not really having 180 lux.
00:22:44.131 --> 00:23:14.501
So therefore, with Wilson, we've tried to create this collapse of results for lower brightnesses to see, because Gin doesn't present a graph like that We've tried to collapse the results for different brightness conditions and for a brightness level of 22 lux or 1 lux I think very reasonable for normal evacuation conditions, we go with K values up to 11.
00:23:14.501 --> 00:23:20.576
That's much higher than Jin did and we generally observe very high k values for those conditions.
00:23:20.576 --> 00:23:29.917
Therefore, reassuring me, in my everyday practice I simply use this k value of 8 as the baseline in my engineering.
00:23:29.917 --> 00:23:32.563
I very rarely engineer for k value of 3.
00:23:32.563 --> 00:23:41.674
And I reassured myself that using high k values is actually more representative of built environment and that's one finding.
00:23:41.674 --> 00:24:00.161
Probably, if you do engineering on that and you go into third party audit, you probably should not quote a podcast on that, but I promise you there's a paper to be submitted very soon and as soon as the paper is published I'll put the link in the show notes and then you will have a peer-reviewed, credible source to quote.
00:24:00.161 --> 00:24:07.042
So I hope I give it to you, my fellow FISA engineer, so you have less troubles in your work.
00:24:07.042 --> 00:24:09.502
But anyway, to summarize, the brightness.
00:24:09.502 --> 00:24:21.780
Brightness is extremely impactful and it seems that higher K values correspond to brightness levels in buildings which are more in line of what we observe in everyday life.
00:24:21.780 --> 00:24:25.750
So that's one reassuring thing and really wow.
00:24:25.750 --> 00:24:32.955
We need to study it more because it's really interesting how much the background brightness changes the outcomes of the experiment.
00:24:32.955 --> 00:24:40.970
But this is not the only obvious thing that we knew about Jin's experiment that we did not put enough emphasis on previously.
00:24:41.476 --> 00:24:43.904
The next one is the sign sizing.
00:24:43.904 --> 00:24:55.467
I've mentioned it already in the Jin's experiment podcast episode and I've actually re-listened to the episode and I found ah, I've talked about it but I did not really recognize the impact of that the sign signage.
00:24:55.467 --> 00:25:07.247
So, as I told you, Gene was observing signs from 5, 10, or 15 meters, but he somehow did not want the sign's size to influence the results of the study.
00:25:07.247 --> 00:25:11.000
Therefore he also used signs of different sizes.
00:25:11.000 --> 00:25:18.301
So when he was observing sign from 5 meters, the sign dimension was 5 centimeters.
00:25:18.301 --> 00:25:21.618
When he was observing it from 10 meters, it was 10 centimeters.
00:25:21.618 --> 00:25:24.527
When he was observing it from 15 meters, it was 15 centimeters.
00:25:24.527 --> 00:25:34.039
So the further he was, the bigger the sign was and now, as you can imagine, it has profound impact on the outcomes.
00:25:34.240 --> 00:25:42.099
Profound, of course, in real building, if you are further away from the thing, you observe, the thing looks smaller.
00:25:42.099 --> 00:25:46.856
The smaller it is, the harder it is to observe it through obscuring mediums such as smoke.
00:25:46.856 --> 00:25:52.256
In jinn's experiment the signs were always the same size no matter what distance you observe them.
00:25:52.256 --> 00:26:04.310
So he was really looking into ability to see the light from the source and not really observe a real evacuation signage in a real evacuation scenario.
00:26:04.310 --> 00:26:10.627
And as soon as you go into the lab and you start observing those signs, you immediately see that.
00:26:10.627 --> 00:26:12.722
So our setup has multiple mirrors.
00:26:12.722 --> 00:26:16.806
You can observe the signs from different distances at the same time, really.
00:26:16.806 --> 00:26:22.126
So you can really narrow the time at which you see the sign at 10 meters.
00:26:22.126 --> 00:26:25.874
You see the sign at 5 meters, you cannot see it at 15.
00:26:25.874 --> 00:26:26.960
Or, even better.
00:26:26.960 --> 00:26:29.943
You see it perfectly in 5 meters.
00:26:29.943 --> 00:26:35.781
You almost lose the visibility of it at 10 meters and you cannot see it at all at 15.
00:26:35.781 --> 00:26:37.681
That's probably a better description.
00:26:37.681 --> 00:26:53.484
So yeah, in our case we're able to see those differences and we've took the measurements and it's going to be another paper that's in production with Wilson, which is how much does the size of the sign change the outcomes of the experiment?
00:26:53.484 --> 00:27:03.825
But it's just important to know that in Gene's experiment that was constrained and I think this has really considerable impact on our ability to engineer.
00:27:03.825 --> 00:27:06.229
I think this has really considerable impact on our ability to engineer.
00:27:06.249 --> 00:27:11.638
I always felt it a lackluster that I cannot do any engineering with my science.
00:27:11.638 --> 00:27:21.711
Like, I go into a building, I do my CFD simulations, I do some smoke, I measure the visibility, of course, and I find the reason of visibility.
00:27:21.711 --> 00:27:23.075
Let's say, 9 meters.
00:27:23.075 --> 00:27:32.948
That's the outcome and my HAJ is very unhappy because it was supposed to be above 10 meters and I don't meet my tenability criterion.
00:27:32.948 --> 00:27:49.957
So I have to put a lot of extraction in that room to increase the my capability of removing smoke, to decrease the smoke density, to improve the visibility conditions, and then my building is considered safe because I finally have more than 10 meters visibility.
00:27:50.444 --> 00:27:58.955
But if, instead of placing another fan, I could just put twice the size evacuation signage, would that help In real engineering scenario?
00:27:58.955 --> 00:27:59.861
It definitely would help.
00:27:59.861 --> 00:28:02.536
It would be a difference in fire safety engineering if I could just use a larger signs in my experiments.
00:28:02.536 --> 00:28:07.010
It would be a difference in fire safety engineering if I could just use a larger science in my experiments.
00:28:07.010 --> 00:28:09.035
It would make a big, big difference.
00:28:09.656 --> 00:28:12.589
But with Jin's model it does not recognize the size of the sign.
00:28:12.589 --> 00:28:24.880
Therefore it's impossible to use that, as you know, your design variable and unfortunately I do not think there is an easy way how this could be implemented in Jin's method.
00:28:24.880 --> 00:28:30.184
The reason is that the results collapse into a line because the size is constrained.
00:28:30.184 --> 00:28:36.679
As soon as you remove the size constraint from those relationships, they do not collapse into a line anymore.
00:28:36.679 --> 00:28:41.176
And if there is an empirical relationship, it's going to be very complicated.
00:28:41.176 --> 00:28:48.858
So it's going to be extreme challenge for us to have a model that would include for the sign size in it.
00:28:48.858 --> 00:28:59.204
I'm not saying we're not trying, but it's very hard and I think it's a flow of genes model that we probably will not be able to go over.
00:28:59.806 --> 00:29:31.119
However, what I want to say is that I think in many documents in which fire engineer has ability to use their knowledge for their benefits, it's fairly fair to say in my building, in order to improve the evacuation conditions, I have doubled the size of the evacuation signage, therefore making them more observable from larger distances, therefore making them more observable from larger distances, and just put a claim that the existing visibility in smoke model does not account for size of the signage.
00:00:00.462 --> 00:00:02.508
00:00:02.508 --> 00:00:09.233
00:00:09.233 --> 00:00:19.969
00:00:19.969 --> 00:00:25.524
00:00:25.524 --> 00:00:36.896
00:00:36.896 --> 00:00:44.173
00:00:44.173 --> 00:00:47.506
00:00:47.527 --> 00:01:12.887
00:01:12.887 --> 00:01:28.605
00:01:28.605 --> 00:01:31.861
00:01:31.861 --> 00:01:58.135
00:01:58.135 --> 00:02:03.668
00:02:03.668 --> 00:02:05.507
00:02:05.507 --> 00:02:20.344
00:02:20.344 --> 00:02:24.112
00:02:24.112 --> 00:02:25.515
00:02:25.915 --> 00:02:28.764
00:02:28.764 --> 00:02:32.539
00:02:32.539 --> 00:02:38.584
00:02:38.584 --> 00:02:42.026
00:02:42.026 --> 00:03:11.561
00:03:11.561 --> 00:03:25.300
00:03:25.300 --> 00:03:41.409
00:03:41.409 --> 00:03:49.968
00:03:49.968 --> 00:03:52.074
00:03:53.040 --> 00:04:02.481
00:04:02.481 --> 00:04:04.622
00:04:04.622 --> 00:04:07.865
00:04:07.865 --> 00:04:09.945
00:04:09.945 --> 00:04:19.591
00:04:19.591 --> 00:04:26.173
00:04:26.173 --> 00:04:34.997
00:04:34.997 --> 00:04:39.588
00:04:39.588 --> 00:04:41.680
00:04:41.680 --> 00:04:47.723
00:04:47.723 --> 00:04:57.050
00:04:57.692 --> 00:05:02.725
00:05:02.725 --> 00:05:07.595
00:05:07.595 --> 00:05:21.629
00:05:21.629 --> 00:05:32.642
00:05:32.642 --> 00:05:37.153
00:05:37.153 --> 00:05:38.983
00:05:38.983 --> 00:05:39.946
00:05:39.946 --> 00:05:52.233
00:05:52.233 --> 00:05:56.170
00:05:56.701 --> 00:06:11.567
00:06:11.567 --> 00:06:25.158
00:06:25.158 --> 00:06:32.458
00:06:32.458 --> 00:06:36.687
00:06:36.687 --> 00:06:41.723
00:06:41.723 --> 00:07:03.245
00:07:03.245 --> 00:07:08.074
00:07:08.074 --> 00:07:09.867
00:07:09.867 --> 00:07:12.569
00:07:12.569 --> 00:07:14.187
00:07:14.187 --> 00:07:25.329
00:07:25.329 --> 00:07:37.531
00:07:37.531 --> 00:07:44.923
00:07:44.923 --> 00:07:47.964
00:07:48.427 --> 00:07:55.387
00:07:55.387 --> 00:08:05.351
00:08:05.351 --> 00:08:10.307
00:08:10.307 --> 00:08:14.641
00:08:14.641 --> 00:08:17.588
00:08:17.588 --> 00:08:22.461
00:08:22.461 --> 00:08:24.021
00:08:24.021 --> 00:08:25.483
00:08:25.483 --> 00:08:29.464
00:08:29.464 --> 00:08:41.932
00:08:42.812 --> 00:08:52.256
00:08:52.256 --> 00:08:54.798
00:08:54.798 --> 00:08:58.868
00:08:58.868 --> 00:09:00.592
00:09:00.592 --> 00:09:23.243
00:09:23.243 --> 00:09:30.410
00:09:31.059 --> 00:09:37.909
00:09:37.909 --> 00:09:45.446
00:09:45.446 --> 00:09:46.470
00:09:46.470 --> 00:10:00.590
00:10:00.590 --> 00:10:06.842
00:10:06.842 --> 00:10:17.600
00:10:17.600 --> 00:10:27.985
00:10:27.985 --> 00:10:38.113
00:10:38.113 --> 00:10:50.293
00:10:50.895 --> 00:10:57.472
00:10:57.472 --> 00:11:04.043
00:11:04.043 --> 00:11:06.210
00:11:06.210 --> 00:11:13.374
00:11:13.374 --> 00:11:20.688
00:11:20.688 --> 00:11:35.201
00:11:35.201 --> 00:11:45.765
00:11:45.765 --> 00:11:47.270
00:11:47.311 --> 00:11:55.456
00:11:55.456 --> 00:12:04.043
00:12:04.043 --> 00:12:08.360
00:12:08.360 --> 00:12:14.466
00:12:14.466 --> 00:12:25.508
00:12:25.508 --> 00:12:30.586
00:12:30.586 --> 00:12:39.726
00:12:39.726 --> 00:12:55.606
00:12:55.875 --> 00:13:05.256
00:13:05.256 --> 00:13:06.921
00:13:06.921 --> 00:13:12.080
00:13:12.080 --> 00:13:19.605
00:13:19.605 --> 00:13:31.004
00:13:31.004 --> 00:13:33.509
00:13:33.811 --> 00:13:46.054
00:13:46.054 --> 00:13:52.227
00:13:52.227 --> 00:13:58.826
00:13:58.826 --> 00:14:04.503
00:14:04.503 --> 00:14:13.442
00:14:13.442 --> 00:14:14.899
00:14:14.899 --> 00:14:16.441
00:14:16.441 --> 00:14:20.524
00:14:21.235 --> 00:14:24.326
00:14:24.326 --> 00:14:37.006
00:14:37.006 --> 00:14:46.144
00:14:46.144 --> 00:14:48.724
00:14:48.724 --> 00:14:54.664
00:14:54.664 --> 00:14:58.586
00:14:58.586 --> 00:15:05.835
00:15:05.835 --> 00:15:15.501
00:15:15.501 --> 00:15:18.263
00:15:18.263 --> 00:15:25.636
00:15:25.697 --> 00:15:41.220
00:15:41.220 --> 00:15:53.506
00:15:53.506 --> 00:15:56.224
00:15:56.224 --> 00:16:00.505
00:16:00.505 --> 00:16:05.923
00:16:07.296 --> 00:16:42.605
00:16:42.605 --> 00:16:44.100
00:16:44.100 --> 00:16:51.822
00:16:51.822 --> 00:16:58.086
00:16:58.086 --> 00:17:06.837
00:17:06.837 --> 00:17:23.184
00:17:23.184 --> 00:17:28.461
00:17:28.461 --> 00:17:31.406
00:17:31.907 --> 00:17:34.721
00:17:34.721 --> 00:17:53.957
00:17:53.957 --> 00:17:59.954
00:17:59.954 --> 00:18:01.280
00:18:01.280 --> 00:18:04.144
00:18:04.144 --> 00:18:10.628
00:18:10.628 --> 00:18:14.105
00:18:14.105 --> 00:18:17.965
00:18:17.965 --> 00:18:18.589
00:18:18.589 --> 00:18:20.196
00:18:20.196 --> 00:18:28.157
00:18:28.157 --> 00:18:43.281
00:18:43.281 --> 00:18:57.722
00:18:57.722 --> 00:18:59.787
00:19:00.048 --> 00:19:13.288
00:19:13.288 --> 00:19:17.319
00:19:17.319 --> 00:19:27.144
00:19:27.144 --> 00:19:30.055
00:19:30.055 --> 00:19:37.506
00:19:37.506 --> 00:19:42.666
00:19:42.666 --> 00:19:50.808
00:19:50.808 --> 00:19:59.327
00:19:59.327 --> 00:20:07.086
00:20:07.086 --> 00:20:10.884
00:20:11.476 --> 00:20:38.107
00:20:38.107 --> 00:20:48.344
00:20:48.344 --> 00:20:52.561
00:20:52.561 --> 00:20:56.561
00:20:56.561 --> 00:21:04.144
00:21:04.144 --> 00:21:05.226
00:21:05.226 --> 00:21:12.708
00:21:12.708 --> 00:21:16.479
00:21:16.479 --> 00:21:23.328
00:21:23.328 --> 00:21:35.521
00:21:36.143 --> 00:21:39.827
00:21:39.827 --> 00:21:46.546
00:21:46.546 --> 00:21:58.506
00:21:58.506 --> 00:22:00.277
00:22:00.277 --> 00:22:05.407
00:22:05.407 --> 00:22:06.876
00:22:06.876 --> 00:22:13.698
00:22:13.698 --> 00:22:16.522
00:22:16.522 --> 00:22:19.728
00:22:19.728 --> 00:22:28.224
00:22:28.224 --> 00:22:35.105
00:22:35.595 --> 00:22:41.904
00:22:41.904 --> 00:22:44.131
00:22:44.131 --> 00:23:14.501
00:23:14.501 --> 00:23:20.576
00:23:20.576 --> 00:23:29.917
00:23:29.917 --> 00:23:32.563
00:23:32.563 --> 00:23:41.674
00:23:41.674 --> 00:24:00.161
00:24:00.161 --> 00:24:07.042
00:24:07.042 --> 00:24:09.502
00:24:09.502 --> 00:24:21.780
00:24:21.780 --> 00:24:25.750
00:24:25.750 --> 00:24:32.955
00:24:32.955 --> 00:24:40.970
00:24:41.476 --> 00:24:43.904
00:24:43.904 --> 00:24:55.467
00:24:55.467 --> 00:25:07.247
00:25:07.247 --> 00:25:11.000
00:25:11.000 --> 00:25:18.301
00:25:18.301 --> 00:25:21.618
00:25:21.618 --> 00:25:24.527
00:25:24.527 --> 00:25:34.039
00:25:34.240 --> 00:25:42.099
00:25:42.099 --> 00:25:46.856
00:25:46.856 --> 00:25:52.256
00:25:52.256 --> 00:26:04.310
00:26:04.310 --> 00:26:10.627
00:26:10.627 --> 00:26:12.722
00:26:12.722 --> 00:26:16.806
00:26:16.806 --> 00:26:22.126
00:26:22.126 --> 00:26:25.874
00:26:25.874 --> 00:26:26.960
00:26:26.960 --> 00:26:29.943
00:26:29.943 --> 00:26:35.781
00:26:35.781 --> 00:26:37.681
00:26:37.681 --> 00:26:53.484
00:26:53.484 --> 00:27:03.825
00:27:03.825 --> 00:27:06.229
00:27:06.249 --> 00:27:11.638
00:27:11.638 --> 00:27:21.711
00:27:21.711 --> 00:27:23.075
00:27:23.075 --> 00:27:32.948
00:27:32.948 --> 00:27:49.957
00:27:50.444 --> 00:27:58.955
00:27:58.955 --> 00:27:59.861
00:27:59.861 --> 00:28:02.536
00:28:02.536 --> 00:28:07.010
00:28:07.010 --> 00:28:09.035
00:28:09.656 --> 00:28:12.589
00:28:12.589 --> 00:28:24.880
00:28:24.880 --> 00:28:30.184
00:28:30.184 --> 00:28:36.679
00:28:36.679 --> 00:28:41.176
00:28:41.176 --> 00:28:48.858
00:28:48.858 --> 00:28:59.204
00:28:59.806 --> 00:29:31.119