Transcript
WEBVTT
00:00:00.240 --> 00:00:02.628
Hello everybody, welcome to the Fire Science Show.
00:00:02.628 --> 00:00:03.750
Its celebration time.
00:00:03.750 --> 00:00:10.353
Fire Science got another ERC grant that is relevant to fire safety engineering.
00:00:10.353 --> 00:00:20.847
For me as a scientist it's really a thing to cherish, because ERC, the European Research Council grants, are the Formula One of science.
00:00:20.847 --> 00:00:28.544
It's the dream and as you can imagine, those grants being the holy grail for literally every single researcher out there.
00:00:28.544 --> 00:00:33.843
They are so competitive and so difficult to obtain almost impossible.
00:00:33.843 --> 00:00:36.731
And yet we had ERC grant.
00:00:36.731 --> 00:00:46.421
But Professor Ruben van C two years ago awarded ERC starting grant and Ruben is well into his research that we've talked about in the podcast two years ago.
00:00:46.781 --> 00:00:50.229
And today I have a possibility to choose another ERC laureate.
00:00:50.229 --> 00:00:57.732
That is Dr Francesco Restuccia from King's College London, my very good friend, and I am so happy for Francesco.
00:00:57.732 --> 00:01:00.784
So, francesco, you know him from the podcast.
00:01:00.784 --> 00:01:11.251
He was talking here about batteries and a lot of his recent research is about fire safety of batteries and he indeed does some amazing work in that space.
00:01:11.251 --> 00:01:13.242
But the grant is completely different.
00:01:13.242 --> 00:02:07.728
The grant is all about modeling wildfires and his approach to modeling to merge tons of other models that already exist, to find scales at which the transition happens, the regimes change, and create new, better models that would have higher technical complexity or better accuracy than the current ones and yet be applicable and usable, which will definitely create space for new frontiers of fire safety engineering real-time modeling of wildfire, progress, sensitivity studies for fuel management, perhaps assistance for management of WE communities A lot of possibilities that may open after this grant is completed and after Francesco integrates the knowledge that's around and adds his own research on top of that.
00:02:08.479 --> 00:02:11.509
Anyway, this episode has kind of two parts.
00:02:11.509 --> 00:02:16.673
So two-thirds of the episode we discuss about the ERC grant itself.
00:02:16.673 --> 00:02:31.774
So we discuss the models and everything that Francesco wants to do in his grant, and the last one-third of the episode is more towards young researchers who also share the dream of getting an ERC grant on their own.
00:02:31.774 --> 00:02:44.991
So the last part we've spent just talking about the process of obtaining such a massive grant and some, let's say, coaching recommendations for people on what does it mean to write an ESC grant?
00:02:44.991 --> 00:02:45.992
What are you looking for?
00:02:45.992 --> 00:02:47.743
What should you focus on?
00:02:47.743 --> 00:02:50.289
Perhaps, that said, some good mentoring.
00:02:50.289 --> 00:02:57.840
Francesco is one of the best mentors I know in the space of fire science, so I'm sure these are some useful advice to young scholars.
00:02:57.840 --> 00:03:01.542
Anyway, let's stop talking and give the microphone to the man.
00:03:01.542 --> 00:03:04.943
So let's spin the intro and jump into the episode.
00:03:09.846 --> 00:03:11.486
Welcome to the Fire science Show.
00:03:11.486 --> 00:03:14.949
My name is Wojciech Wegrzyński and I will be your host.
00:03:14.949 --> 00:03:34.437
This podcast is brought to you in collaboration with OFR Consultants.
00:03:34.437 --> 00:03:38.343
Ofr is the UK's leading fire risk consultancy.
00:03:38.343 --> 00:03:48.250
Its globally established team has developed a reputation for preeminent fire engineering expertise, with colleagues working across the world to help protect people, property and environment.
00:03:48.250 --> 00:04:04.026
Established in the UK in 2016 as a startup business of two highly experienced fire engineering consultants, the business has grown phenomenally in just seven years, with offices across the country in seven locations, from Edinburgh to Bath, and now employing more than a hundred professionals.
00:04:04.026 --> 00:04:15.682
Colleagues are on a mission to continually explore the challenges that fire creates for clients and society, applying the best research, experience and diligence for effective, tailored fire safety solutions.
00:04:15.682 --> 00:04:26.711
In 2024, ofr will grow its team once more and is always keen to hear from industry professionals who would like to collaborate on fire safety futures this year.
00:04:26.711 --> 00:04:29.307
Get in touch at oafrconsultantscom.
00:04:29.307 --> 00:04:31.607
Hello everybody, welcome to the Fire Science Show.
00:04:31.607 --> 00:04:34.790
I am joined today again by Dr Francesco Restuccia.
00:04:34.790 --> 00:04:36.326
Hey, francesco, good to have you back in the show.
00:04:36.620 --> 00:04:37.764
Thank you again for the invitation.
00:04:37.764 --> 00:04:38.767
Ojec Nice to see you.
00:04:39.240 --> 00:04:45.033
And look, mate, a man who lists two Fire Science Show episodes at the top of his CV got the ERC.
00:04:45.033 --> 00:04:46.584
I'm kind of not surprised.
00:04:46.584 --> 00:04:52.591
Joe Rogan can make you a US president, so I guess this podcast helps you in the career a little bit.
00:04:52.959 --> 00:05:02.906
No, yeah, I always enjoy coming to the Five Science Show, also because you make me think about the problems I work on and having to explain them, and so, yeah, it's always really, really interesting, and I listen to all of your episodes.
00:05:02.906 --> 00:05:07.809
I still don't know how you managed to make so many episodes and where you find the time, but they're fantastic.
00:05:08.180 --> 00:05:09.165
You're in the middle of making one.
00:05:09.165 --> 00:05:15.846
So let's say you have insight to the backend and, anyway, fantastic job on the batteries and everything you've been doing.
00:05:15.846 --> 00:05:22.370
But the ERC you've submitted is so far away from the world of battery that I'm used to interview you in.
00:05:22.370 --> 00:05:26.281
So let me do it properly.
00:05:26.281 --> 00:05:35.509
Let me read out the starting grant name that you've just got Wildfires and Climate Change, physics-based Modeling of Fire Spread in a Changing World Acronym Fire Mod.
00:05:35.509 --> 00:05:39.103
So this is what you just got funded in the ERC scheme.
00:05:39.103 --> 00:05:43.021
For those who don't know, who are listening, this is like this, is it?
00:05:43.021 --> 00:05:44.483
That's the top.
00:05:44.483 --> 00:05:48.872
You cannot go any further, at least in the European Union and the islands around it.
00:05:48.872 --> 00:06:00.052
So, francesco, tell me the genesis of that, like what made you write a wildfire and climate proposal and how long that has been in your head.
00:06:00.380 --> 00:06:08.160
So my research group so I run the research group Heat and Fire Lab, and my group, as you say, predominantly focuses on batteries has kind of three strands.
00:06:08.160 --> 00:06:11.309
So I have the heat side, where I do thermal management.
00:06:11.309 --> 00:06:17.608
I look at biomass, I look a little bit on hydrogen and then I have the fire dynamics part, and the fire dynamics is sort of split into two.
00:06:17.608 --> 00:06:19.307
One is the fire dynamics of batteries.
00:06:19.307 --> 00:06:25.903
I look at fire spread, I look at ignition, I look at suppression.
00:06:25.903 --> 00:06:31.024
Myself and my team and I had currently one, two PhD students who have been working on wildfire, but again it's an area that's smaller in my group.
00:06:31.446 --> 00:06:48.838
I was interested in wildfire for a long time because from a fire dynamics perspective it's one of the most complex phenomena to study at a lab scale and a large scale because there are many, many differences and so there are a lot of similarities and in fact I used a lot of what I learned in my past work.
00:06:48.838 --> 00:06:57.346
Both in the ignition work that I did during my PhD with Guillermo Reina at Imperial, in my battery work, especially the fire spread in batteries.
00:06:57.346 --> 00:07:08.213
I saw some similarities and some differences and what really interested me on wildfires for a long, long time has been the fact that they can do so much damage when they're uncontrolled.
00:07:08.213 --> 00:07:09.483
I come from Italy.
00:07:09.483 --> 00:07:11.067
We have wildfires all the time.
00:07:11.067 --> 00:07:24.452
I come from southern Italy, but if you look at Greece, you look at Spain, you look at Portugal, there are a lot of very large scale wildfires and we've been doing a lot of work operationally across the world to try and understand them.
00:07:24.452 --> 00:07:29.702
But there is still a lot to be done, and so I started really looking at this maybe five years ago.
00:07:29.702 --> 00:07:41.444
I started thinking about this and then it took a few years to develop sort of an idea of what I thought I could contribute, and so a lot of my current past work before the CRC has really been on the sensitivity.
00:07:41.444 --> 00:07:52.567
So PhD student Imogen Richards in my group has really focused on the sensitivity of physical parameters for wildfires, and so I was really interested in understanding how different variables affect the output.
00:07:52.627 --> 00:07:54.312
Right, all of our models are models.
00:07:54.312 --> 00:08:06.822
When we do modeling, they have an input, they have an output, they have multiple inputs usually, and often we look at multiple outputs, and what really interested me for the last couple of years was okay, how does my input change my output?
00:08:06.822 --> 00:08:11.351
So I often think of data as if my data is garbage going in.
00:08:11.351 --> 00:08:13.545
The data I'm getting out is definitely going to be garbage.
00:08:13.545 --> 00:08:19.629
And so if I'm doing a model and I have an experimental data point, my experimental data point will have some error.
00:08:19.629 --> 00:08:22.562
And how will that error, how will that change affect my output?
00:08:22.983 --> 00:08:29.529
Now, in the real world, in wildfires, it's not an error, it's the variability of a fuel, variability of nature, right?
00:08:29.529 --> 00:08:38.772
So you can have an area which has a certain moisture content on a certain day, in a certain wind, in a certain slope, and then another day, a week later, completely different boundary conditions.
00:08:38.772 --> 00:08:54.960
And so I really wanted to look at and I looked at the past models, because there's been many models on ignition, on spread, and I want to really look at the fundamentals to understand how we can scale these fire models.
00:08:55.501 --> 00:09:11.697
I had some episodes on wildfires, a lot of them on wildfires in the fire science shows, as you know, and models do kind of exist, like we have this Rothamilz model which lasts for I don't know how many decades already, but it's been the basis of the 1970s.
00:09:11.716 --> 00:09:13.225
Exactly, it's 50 years almost.
00:09:14.162 --> 00:09:15.066
Happy 50th birthday.
00:09:15.066 --> 00:09:23.855
Anyway, we have those tools to allow us to model or predict to some extent the wildfires.
00:09:23.855 --> 00:09:28.001
Everyone is familiar with the maps of fire hazard.
00:09:28.001 --> 00:09:35.624
Right, we have the tools that assist firefighters on the scene with this likely spread of fires.
00:09:35.624 --> 00:09:38.649
So is there really such a gap?
00:09:38.649 --> 00:09:46.520
If you could extract one thing that's lacking from the current models, that justifies doing this massive research project Absolutely.
00:09:46.902 --> 00:09:49.488
Yeah, so it's a sort of two problem, right?
00:09:49.488 --> 00:09:53.423
When you do a model, you have, let's say, y-axis and x-axis.
00:09:53.423 --> 00:09:57.695
If we look at an axis, y-axis physical fidelity, x-axis applicability, right.
00:09:57.695 --> 00:10:02.793
So the further right bottom you are, the most applicable it is, but maybe it has less physics.
00:10:02.793 --> 00:10:18.826
So a Rotterdam model, so any pure empirical wildfire model that maybe has a few physical input parameters but is very empirical or operational models like a Rotterdam model really sit on lower physical fidelity but applicability to many, many scenarios.
00:10:19.280 --> 00:10:23.768
I'll stop you for a second, because I realize the audience may not be that familiar with Rotterdam model.
00:10:23.768 --> 00:10:27.070
It's not something most fire engineers would use.
00:10:27.070 --> 00:10:32.573
So Rotterdam model is basically one equation and you put up things like slope moisture.
00:10:32.573 --> 00:10:33.940
Is there wind in it as well?
00:10:33.980 --> 00:10:36.067
Yes, there's wind, so you look at wind.
00:10:36.067 --> 00:10:37.023
So effectively.
00:10:37.023 --> 00:10:41.066
Rotterdam model takes a flame and it has a contact with the fuel.
00:10:41.066 --> 00:10:46.107
That flame has some radiation, some convection and then some solid mass transport.
00:10:46.107 --> 00:10:47.846
No, no, no, you're over.
00:10:47.846 --> 00:10:53.067
So variables you put in are effectively size, temperature, moisture.
00:10:53.067 --> 00:10:58.809
So you have a few physical variables you put in and then you get out a rate of spread Exactly.
00:10:58.831 --> 00:11:00.422
So super simple, like one equation.
00:11:00.422 --> 00:11:02.822
You put up the variables, you get the rate of spread.
00:11:02.822 --> 00:11:06.028
Then that's the most applicable because it takes you a second to solve.
00:11:06.028 --> 00:11:10.924
But it lacks the physical depth of everything and the other end of scale.
00:11:11.301 --> 00:11:22.125
The other end of scale is so, exactly, if you go to the very high physical fidelity so, for example, physics-based direct numerical simulation of wildfires then you are very, very detailed so you can focus on which aspects.
00:11:22.125 --> 00:11:27.573
So let's say, you really want to get detailed kinetics, you really want to understand how that fuel degradation happens.
00:11:27.573 --> 00:11:41.562
You can get a very, very good understanding, but it's very limited in application because you can't extract those results for many other results and so it's very, very high physical fidelity but very low applicability to diverse scenarios.
00:11:41.562 --> 00:11:43.306
Let's say, so those are the two extremes.
00:11:43.306 --> 00:11:44.927
You have stuff in the middle, right.
00:11:44.927 --> 00:11:46.289
That's called reduced order models.
00:11:46.289 --> 00:11:51.890
And so in the middle between fully empirical and fully physics-based, you have reduced order models.
00:11:51.890 --> 00:11:57.227
Those are either semi-empirical models or there's lots of AI and data-trained models.
00:11:57.227 --> 00:11:58.932
Those are somewhere in the middle, right?
00:11:58.932 --> 00:12:06.663
So you have some physics and some operational aspects, some empirical aspects, and every model has advantages and disadvantages.
00:12:06.663 --> 00:12:07.907
The best you know.
00:12:07.907 --> 00:12:16.482
If you really want to understand how a fuel will degrade and the kinetics of a fuel for a very, very specific scenario, then you do lots of detailed kinetics, right.
00:12:16.482 --> 00:12:20.760
If you don't really care about that, then you say well, I have so many heterogeneous fuels.
00:12:20.760 --> 00:12:25.270
I just want to understand roughly the rate of spread in this with these physical conditions.
00:12:25.270 --> 00:12:27.215
Then you have the fully empirical model.
00:12:27.515 --> 00:12:41.649
But to understand the changing fire regimes and understand the different theories, because, again, when I look at the physics-based fully physics-based I give you the example of kinetics, but I could also say I can take a full physics-based one for atmospheric models, right?
00:12:41.669 --> 00:12:50.592
So I say, well, I want to understand the very large, large scale spread of these clouds of megafires that maybe are spreading very, very largely.
00:12:50.592 --> 00:12:55.889
Then I would use a very detailed atmospheric model to understand that sort of propagation.
00:12:55.889 --> 00:13:12.768
Or let's say I wanted to understand the transition between smoldering and flaming fires, which are really different scales, right, so some are centimeters per hour, the other ones are meters or kilometers per hour, depending how fast it's going, and so then you would need a model that's very good for that flaming or smoldering regime.
00:13:12.768 --> 00:13:32.326
Or let's say you wanted to understand the risk okay, this is less for wildfires but in general for fire modeling the risk of deflagration, right, then you focus on very specific type of models where pressure becomes very important, and so it really depends on which physical variable and to establish fire spread transport thresholds for diverse environments.
00:13:32.326 --> 00:13:34.828
So let's say you're looking at centimeters versus kilometers.
00:13:34.828 --> 00:13:42.860
We use different models for physics-based, and that's actually the problem that I encounter when I use a lot of models.
00:13:42.860 --> 00:13:47.143
Every model is useful for some scenario, right Okay?
00:13:47.163 --> 00:13:51.770
so you would use FDS for, let's say, compartment scale fire.
00:13:51.770 --> 00:14:00.804
But FDS is tricky when you want to do urban configuration and it's tricky if you want to investigate a matchstick, for example, because the scale is too small.
00:14:00.804 --> 00:14:18.221
So you have models for different scale and in your applicability versus technical complexity thing I would assume that some sort of crown achievement would be technically advanced model which would be as applicable as Rotterman model, exactly.
00:14:18.663 --> 00:14:19.705
I think that's the future.
00:14:19.705 --> 00:14:27.291
Yeah, so when I say what is my ideal model, it would be a physics-based model that has the same applicability range as a raw thermal model.
00:14:27.291 --> 00:14:35.254
Right, because then you can get very detailed physics included and incorporated into something that's very diversely applicable.
00:14:35.720 --> 00:14:45.075
But in here applicability could be also understood as user-friendliness or the difficulty in use of the model.
00:14:45.075 --> 00:14:53.831
So Rotterdam model is extremely easy to use because you basically can make a spreadsheet and you just put in your slope, your moisture, your wind and you get the outcome.
00:14:53.831 --> 00:15:05.428
Or you have already made packages that can import a topographical map of your terrain and just release the fire and it already knows the wind, already knows the snow.
00:15:05.428 --> 00:15:08.629
It's so easy to apply.
00:15:08.629 --> 00:15:14.480
Yet the extreme complex models you've mentioned DNS modeling of the fuel package Jesus Christ that's.
00:15:14.480 --> 00:15:19.504
I got chills when you told me that because I know how much work would be setting up that model.
00:15:19.504 --> 00:15:22.168
So is there any shortcut that we could apply those complex models in a simple way, like perhaps automate them to some way?
00:15:22.168 --> 00:15:27.107
Is there any shortcut that we could apply those complex models in a simple way, like perhaps automate them to some way?
00:15:27.500 --> 00:15:41.509
So yes, if we can find the thresholds for applicability, which is actually so if you look at the summary page of my grant, I had to write for the EU like a half a pager of you know why is this important as a project.
00:15:41.509 --> 00:15:55.389
And I think if we can find the scales where changing fire regimes actually change, right, so if we can look at temporal and spatial scales to understand those changing regimes, fires are driven by different heat.
00:15:55.389 --> 00:16:00.587
The dominant heat transfer mechanism changes as the fire changes, right, so you could have convection driven, radiation driven.
00:16:00.587 --> 00:16:19.269
So if we can find those changing fire regimes, when we include vegetation dynamics which is really the tricky part when we look at this from wildfires, then we can start scaling up maybe those fine mesh models that you were talking about, where we have all the complexity to larger grid sizes and not lose accuracy.
00:16:19.269 --> 00:16:21.893
And then sort of the third aspect.
00:16:21.893 --> 00:16:32.985
So that's all heat and modeling-based, but the third aspect which is fundamental from fire is also understanding the effects of the different fire spread regimes, right.
00:16:33.210 --> 00:16:38.216
So if I have a regime that's a fast fire I really enjoyed last two weeks ago.
00:16:38.437 --> 00:16:45.495
It was a paper in Nature and it was the science and it was the cover and it was fast fires right, and I said, oh fast, what do they mean by fast?
00:16:45.495 --> 00:16:46.696
Then they meant fast spreading.
00:16:46.696 --> 00:16:57.863
But when we can look at those let's say, fast or slow fires and we can integrate the different behaviors of both in our model, then we can make a very predictive tool right.
00:16:57.863 --> 00:17:12.637
Because if we have a predictive tool that can also be used operationally I think that's sort of the long-term vision of where I think we eventually need a FHIR model to be Then you can say, okay, those FHIR risk maps that you mentioned, actually we can do something a lot more accurate for this specific scenario.
00:17:12.637 --> 00:17:21.577
So let's say, we have a risk of FHIR here, we plug it into a quick model, hopefully, and that quick model will tell us, yes, no, and this spread or this spread right.
00:17:21.577 --> 00:17:28.945
That's sort of the long-term goal, I think, where fire modeling can go, because, as you said, there's lots of fire spread models and they're very useful in different scenarios.
00:17:33.430 --> 00:17:40.378
I've been calling smoldering fires the slow fires since always and I'm very happy that science has taken the fire science nomenclature for fires and now applies them on the covers.
00:17:40.378 --> 00:17:41.601
That's good on the covers.
00:17:41.601 --> 00:17:43.844
That's good, do you think?
00:17:43.844 --> 00:17:55.196
Because in your paper I find a lot of like really interesting things, and then you could expect that from ESC grants up to like using satellite readings, remote sensing and everything.
00:17:55.196 --> 00:18:19.194
So immediately in my head I see this applicability of a model where you basically are in a part of a terrain and you know the topography from the GIS maps, you know the current weather from the satellites, you know where the fire is from your remote sensing units, and then everything happens on the back end and just provides you with an accurate, quick, robust prediction.
00:18:19.194 --> 00:18:22.843
Is this this high end of applicability that you're?
00:18:22.863 --> 00:18:24.171
talking about Exactly that.
00:18:24.171 --> 00:18:34.242
So if I was to summarize that in a sentence right, you want a holistic, real-time risk prediction and fuel management tool that's based on identifiable fuel and landscape, right?
00:18:34.242 --> 00:18:41.243
Because if you know the landscape, as you said, and you know the fuel, then you can either have the risk prediction or the fuel management prediction.
00:18:41.243 --> 00:18:51.578
If we want to manage that fuel instead all in one tool, yeah, that's, I think, the goal, that's, I think, where we want to be with a fire model, let's say, in 10 years' time.
00:18:51.578 --> 00:18:54.598
And what's stopping us from having it right now?
00:18:55.190 --> 00:18:59.097
Identifying the driving, heat transfer and chemical thresholds for the varying scales.
00:18:59.097 --> 00:19:03.897
So, when we look at satellites right, so you mentioned Rotterdam in the 1970s.
00:19:03.897 --> 00:19:07.655
Then we had a lot of satellites used for wildfire protection since 2005.
00:19:07.655 --> 00:19:08.657
So it's been 20 years.
00:19:08.657 --> 00:19:11.813
But there's a size limit, right.
00:19:11.813 --> 00:19:18.737
There's a grid and a satellite image will have a certain scale, and then there's even the larger ones, the atmospheric ones, for very, very large scale.
00:19:18.737 --> 00:19:25.400
But identifying the thresholds for the varying scales is what's missing at the moment, so Finiacal.
00:19:25.400 --> 00:19:34.282
If you remember, there was a paper in the Presidio of the National Academy of Sciences about 10 years ago where they really looked at that convection-driven fire spread.
00:19:34.282 --> 00:19:40.740
So that's identifying a threshold, right, that's identifying the heat transfer mechanism for that specific type.
00:19:40.740 --> 00:19:49.196
If we had that for more types and not just convection, and not just one type of fuel, that's kind of what's missing at the moment, in my opinion.
00:19:49.196 --> 00:19:55.516
Guillermo wrote an accolade to nature about that work Fantastic work, fantastic experimental and theoretical work in my opinion.
00:19:55.516 --> 00:19:56.558
Yeah, it was a great paper.
00:19:57.634 --> 00:20:02.075
And those gaps, so you must have proposed a way forward to solve it.
00:20:02.075 --> 00:20:06.650
If you got this five-year funding, yeah, what are the next steps for you and your group?
00:20:06.650 --> 00:20:13.996
How do you think we can move towards incorporating all of those models and identifying those scale challenges?
00:20:14.410 --> 00:20:16.557
So obviously, yeah, we can't work on everything right.
00:20:16.557 --> 00:20:25.018
So I've given you lots and lots of problems and lots and lots of variables that can be studied, but obviously we can't focus on all of them.
00:20:25.018 --> 00:20:32.176
And so, as you can read on my fact sheet for the grant, if you have a look, I sort of focus on three different methods in parallel.
00:20:32.176 --> 00:20:41.269
And so, again, I'm interested in the prediction and understanding of the occurrence of uncontrolled fires and I'm interested in a physical, fundamental physical model.
00:20:41.269 --> 00:20:43.157
So again, there's lots of different types of models.
00:20:43.157 --> 00:20:50.999
I really am interested in advancing the physical model of fire spread process for different conditions and different fuel types and so sort of.
00:20:50.999 --> 00:20:52.843
I look at three methods in parallel.
00:20:52.843 --> 00:20:55.980
That's what my group would like to look at in this project.
00:20:55.980 --> 00:21:04.721
The first one is studying fires across temporal and spatial scales, right To understand those, changing fire regimes and bringing in vegetation dynamics.
00:21:04.721 --> 00:21:14.455
The second one is understanding fire on multiple scales and that will help us scale up from those, you know, small scale fine mesh models to large grid sizes.
00:21:14.455 --> 00:21:28.005
And then the last one is to integrate that effect of you call them slow and fast fire, so integrating the effect of smoldering, combustion into the modeling of fire spread, just something that is not done often, and that's sort of the methodology and tools.
00:21:28.425 --> 00:21:30.090
And again, the wider context.
00:21:30.090 --> 00:21:34.221
If I look back, you know lots of people already work on wildfires, and you know this already.
00:21:34.221 --> 00:21:37.801
But the larger aspect is really the human side as well.
00:21:37.801 --> 00:21:41.616
You know we have over a billion people a year impacted by wildfire smoke.
00:21:41.616 --> 00:21:51.103
If I look at European Union alone you know I'm again from Italy we have over 4 billion euros of wildfire cost a year right Worldwide.
00:21:51.103 --> 00:21:57.788
We emit from wildfires over 2000 megatons of CO2 a year, and so it's a larger scale problem.
00:21:57.788 --> 00:22:02.631
It's not just the engineering and the fire science one that we're interested in, it's also the health aspects and so on.
00:22:02.631 --> 00:22:08.703
Again, I focus on the engineering because that's my expertise, but the wildfire problem is much larger than just this one.
00:22:09.131 --> 00:22:17.050
I also like that you've included the climate aspect in your proposal and in the B1 that you've sent me.
00:22:17.050 --> 00:22:33.432
You've shown the Arctic region and different types of of wildfire, wildlife or vegetation interactions that can be happening there, and also smoldering, like transition from climbing to smoldering, smoldering to flaming.
00:22:33.432 --> 00:22:34.974
Can you comment on that?
00:22:34.974 --> 00:22:41.497
Why you've picked this part of the world and how does that connect you with the climate thing.
00:22:41.517 --> 00:22:42.619
Great question, question again.
00:22:42.619 --> 00:22:44.946
So different vegetation.
00:22:44.946 --> 00:22:50.324
So I picked different fuel areas and different fuel properties and you pointed out one.
00:22:50.324 --> 00:22:56.000
So the sort of Arctic ones was an example I used because there's been a lot of recent data on Arctic ones.
00:22:56.000 --> 00:23:02.388
I also picked ones more Mediterranean because I was interested in seeing how there is data on fuel.
00:23:02.388 --> 00:23:10.057
So I wanted to look at the data on the different fuel properties of different fuels in different areas to have a better understanding of ignition properties of them.
00:23:10.250 --> 00:23:13.656
So the first part of fire spread is the fire to ignite right.
00:23:13.656 --> 00:23:27.816
And so there's been a lot of work on ignition and I wanted to bring together that work that already exists on ignition, especially North America, Arctic and Europe, to differentiate if I could see which physical variable is most important there.
00:23:27.816 --> 00:23:29.661
And that brings us back to the sensitivity work.
00:23:29.661 --> 00:23:31.757
My PhD student, Imogen Richards, does.
00:23:31.757 --> 00:23:42.875
So a lot of this work we have already done sort of a feasibility work before in her PhD thesis is to sort of get a better understanding of the physical parameters, sensitivity to change.
00:23:43.196 --> 00:23:49.974
So if I have, let's say, that Arctic fuel and it had a 10% difference in moisture, how will that change my ignition?
00:23:49.974 --> 00:23:59.376
How will that change my spread, If it's 20%, or if it's a wind change of 10%, or if it's a terrain change, slope change of 10%, how will that change my output?
00:23:59.376 --> 00:24:06.650
And so sort of having a database of fuel properties from all those different areas will help me then make a model that's more applicable.
00:24:06.650 --> 00:24:15.982
Because if I can say, well, actually, Arctic okay, let's simplify it infinitely let's say Arctic, Polish and Italian terrain, they're all behaving identically, even though they're very different.
00:24:15.982 --> 00:24:18.439
Now they're not, but hypothetically they are right.
00:24:18.439 --> 00:24:29.474
Then the input in my model let's say you're the user who wants to use an operational model when you go click input, you can choose any terrain.
00:24:29.474 --> 00:24:29.936
Right, it won't matter.
00:24:29.936 --> 00:24:38.471
Obviously there are differences, and so then you'll have to choose the terrain that matches most the problem you're trying to solve, and so you'll have a database there that you can use for inputs and you'll know what the sensitivity of that input is.
00:24:38.872 --> 00:24:39.492
Yeah.
00:24:39.492 --> 00:24:51.586
So in other words, you could expect that a fire in Siberia, a fire in southern Italy, in the middle of Poland and in Berkeley, those are four completely different wildfires.
00:24:51.586 --> 00:24:59.284
And today I don't think the Rothamil model could predict the boreal fire at all because it did not have the smoldering component in it.
00:24:59.284 --> 00:25:04.781
So, by definition, you would most likely have different models for each of those.
00:25:04.781 --> 00:25:39.817
Yes, and in a way, those differences in your variables would be hidden within the empirical constraints of those models, in each of them, somewhere Exactly of gravity for fires, you know where, instead of defining a model for each of them, you understand what makes the difference between a fire in Calabria versus a fire in Siberia and because you know the difference, you know when the regime changes and when different variables start to be the key players.
00:25:39.817 --> 00:25:40.961
That's brilliant.
00:25:40.961 --> 00:25:42.002
Yeah, exactly that's the goal.
00:25:43.051 --> 00:25:44.416
That's all of the problem to solve.
00:25:44.416 --> 00:25:47.337
But then you obviously need more inputs to do that right.
00:25:47.337 --> 00:25:52.260
And so let's say that you let's use that remote sensing you mentioned since you mentioned satellites and remote sensing.
00:25:52.260 --> 00:26:11.728
If we can then have enhanced remote sensing to say, actually we know what, these are the differences that we're looking for so we can use an input data, that aspect that we, the model, we modelers, say this is what we need and hopefully you can make it automated, as you say, so we can extract some of that data as it comes in right, then we can use enhanced remote sensing to inform our model.
00:26:11.728 --> 00:26:23.519
And that's, I think, how you eventually you'll reach a real time risk prediction, because then you'll know some variables about the area you're in, but you don't need to know every single variable, because if you needed to know every single variable your model would take.
00:26:23.519 --> 00:26:30.816
It's fantastic from a scientific point of view, but it would take months to run and so it's not very useful for a prediction or real-time prediction.
00:26:31.490 --> 00:26:36.460
One thing that I've missed in the project, and it becomes very apparent to me that it's necessary.
00:26:36.460 --> 00:26:40.096
Perhaps it's there, I just have not seen this how do you quantify the fuel packages?
00:26:40.096 --> 00:26:43.576
Do you have any idea for automated quantifying the fuel packages?