135 - Contemplating a design fire for car parks

Hello everybody and welcome to the Far Science Show. Dear friends, I feel like I'm under pressure, under a lot of pressure. As you perhaps noticed, I'm quite involved in the space of car parks and smoke control in the car parks and every now and then, with an increasing intensity, I am being asked how do we design the design fire for the car parks? I've tried to answer this question that may be not in the most direct ways in some previous podcast episodes you may recall the episode 6 on the electric vehicles, or episode 48 with my spear point, where we perhaps went the deepest, but I've never clearly answered you how do I choose my design fires for car parks? And this is a fundamental thing. I had a hope that I can guide you towards probabilistic methods, towards risk engineering, perhaps towards something unconventional, but the question comes back Every now and then how should we choose the design fire for car parks? And it seems to be even more relevant than ever before with our cars growing, with our cars becoming electric and with all of these car parks and ships around that have been victims to launch fires. Boy, what a topic Today. It's time for me to not escape this but try to give you an answer, but of course I won't just give you a number, because that would be a very short podcast episode. I will tell you what I think about experiments and how they are carried and how they experiment itself influences the outcome, which is the fire curve that we get after one. What do I think the design fire concept really is and what are the goals of the design and how that plays together with the choice of design fire. So, hopefully an interesting episode. I'm not sure if every one of you is designing car parks, but it seems to be an awfully popular theme in the podcast. So I hope there are many people interested and if you're not, well, I hope you will learn something interesting about design fires on their own. So I guess we can go. No, wait, there's one more thing. The Fire Science Show community is up online. If you go to communityfireccienceshowcom, you will find my community where my online course, the Book of Fire, lives, the one that I've advertised many times before. So it's online, it's live, it's waiting for you, communityfireccienceshowcom. Check it out. And now let's spin the intro and jump into the design fires for car parks. Welcome to the Fire Science Show. My name is Vojci Wimczynski and I will be your host. This podcast is brought to you in collaboration with OFR consultants. Ofr is the UK's leading fire risk consultancy. 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. 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. 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. 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, get in touch at OFRconsultantscom. And now back to the design fires. Okay, so here we are. I cannot escape this anymore. So let's talk about design fires for vehicle fires, in car parks, perhaps in other settings as well. First, let me tell you the background, why I think it's okay for me to talk about it right now. So you may or may not know, but we are finishing a big wind and fire grant where we are investigating the effects of wind on fires and our case study actually is an open car park. It's a very interesting choice of the case study. Given the lute and fire that happened few months ago, that's a very irrelevant case study, to be honest. I didn't. I did not set lute and fire and I didn't plan on such a huge fire to cure in the late part of our project. But it is what it is. I guess there was some lack or perhaps hindsight in choosing this, this type of case study, but in that project what we were doing was to investigate the effects of fire in multiple wind scenarios. We are talking about hundreds of wind scenarios. We wanted to map out all possible outcomes of fires in open car park in different wind scenarios and, as you can imagine, hundreds of CFD studies, hundreds of simulations. So that was already over a year of simulating. So when we approached that, we of course faced the same challenge everyone faces when they try to run a CFD simulation analysis for any type of unequivocity what kind of fire we put inside. In my previous episodes and I highly recommend going to episode 48 with Mike Spearpoint where we really went deep into that we've discussed that. You could approach fires as a probabilistic problem. You could assign the distribution to some sizes of fires. You can create a cutoff point saying that this kind of design fire covers this percentage of all fires that happen with quite good degrees of certainty. You could use that as a base for risk engineering. You can go even further, apply complex methodologies like JVanley or quality of life index A lot of ways you could incorporate this approach into your analysis. But here we are standing in front of a massive, massive computational effort in front of us, hundreds of wind and fire scenarios to be run, and we don't have this luxury. It's already big enough. We don't have the luxury to run it multiple-aromatic, to run hundreds of different fires in those hundreds of different winds, because we would end up with tens of thousands of scenarios. So we had to pick some, and to do that I've hired a student he's now my full-time employee Bartosz Michówka, and Bartosz was digging very deep into something I've started in 2013, which is an in-depth analysis of all the data available on vehicle fires, all the experiments that were carried around the world that are in the literature that we could identify. We gathered them up. We've applied some filters that I think were quite unusual. I mean, it's difficult to look at the same data that everyone is looking and see something different than everyone else is seeing. But that was our intention. That's what we were seeking in the data to see something new in something very well-known, because there's a lot of papers, very similar data set. So we were going through this and by this analysis we were able to identify a range of scenarios that we thought, yeah, this is a good range of scenarios for our problem being the open car park and vehicle fire inside of it. And later on, reflecting on what we have accomplished with Bartosz, I thought this is actually good enough to be a self-made study, if I may. So we've actually published this in a post journal. It's about to be released on days. I'll drop a link in the show notes once the paper is published and perhaps you can understand some of it with a Google translate or something. But we're also finalizing our efforts on a bigger paper, a little more in-depth analysis, for a proper fire type journal where we will try to put in everything we found for ourselves. So yeah, that's a background story. That's why I think it's relevant for me to talk about design fires in car parks today and, of course, what I mentioned in the intro, everyone is curious, everyone is asking. You know, in 2015, we've published a book in Poland, our ITB instruction. This is, let's say, some sort of design recommendation for Poland, where people actually are using that to design car parks, and there we have put some recommendations to our design fires, and that triggered almost a war on the internal market. It was quite interesting how much emotion can be behind design fires scenario and we also understand that with all the changes that are happening, we probably have to revisit it. So also a good moment to reflect. I will have to do it for the Polish market, so I may try and talk it over with all of you. So why change is urgent? It's obvious. The market is shifting. We were looking into some sort of European statistics and we found statistics by ACA, which is a French institute, where it was mentioned that in 2011, suv vehicles in Europe were only 14% of the market, while in 2020 there were half of it. In 2020, all the battery-powered vehicles were 5% and in 2022 already 12%. In the same time, the hybrid plug-ins went from 17% to 30%. That are massive shifts in the market, massive shifts towards heavier vehicles, suvs, and massive shifts towards electric power vehicles. These are two things that raise a lot of questions about how relevant are our design fires that were developed in the 80s 90s on quite an old data set. In Poland we don't have such a strong revolution yet, but it is something coming our way. In Poland in 2021, the hybrid vehicles were just 1.6%, electric vehicles 1%, but you can still see the trend towards heavier vehicles, suvs for sure, like everywhere else. So it's kind of obvious. Our fleet is shifting. The cars are different than they were 10, 15, 20 years ago. We cannot escape this question what do we do with it now? How do we approach the design fires? So I promised you I will give you some numbers, but first I promised you three things experiments, what design fire is and what are the goals of the design. So let's try to tackle them one by one. First, the experiments. So let's, for a second, reflect on how do we obtain data on the fire growth of vehicles. We obviously burn vehicles, that's an obvious one. So you take a vehicle, you put it into a calorimeter, you can measure the oxygen depletion and by that figure out how much energy was released in the fire as a function of time, you can measure the mass loss of the vehicle. So how much of the vehicle has burned out and from that figure out what was the rate of fire and how quickly was it growing? And many people would think that's just it. That's a standard procedure. You take a car, put it into a calorimeter, you burn it, you're done. But as I was looking through the literature, as I had the privilege to talk to a lot of people who were doing those tests, those experiments, it's not that easy. The way, how you run the test will significantly influence the outcomes. Even the most basic things like do you have a ceiling above your car or not? How are you collecting the smoke and are you 100% sure that every ounce of smoke is captured? What state is the vehicle in? Have you purchased it new from a saloon or is it something that was put in the market a decade ago? Is it damaged and can that damage influence the outcomes of the test? What state the battery is? How much fuel you have in the tank? Have you fiddled with the tank at all or you just left it as it was in the original vehicle? Perhaps you've opened the windows a little bit? And, most importantly, where did you set the ignition and how big it was. That was a massive one really. When we were looking through the data, we tried to sort it out in different ways and we've identified some key physical properties that you can use to define a design scenario, among them being heat release rate. That's obvious as a function of time, which is the outcome that you could perhaps directly use in your modeling studies Total heat release, which is the entire energy that would be released in a fire of a vehicle when it burns out. So basically the calorific value of your vehicle. But we're looking also in peak heat release rate. So what is the maximum value of the heat release rate that it reached in the experiment? And we were also looking into the time to pick, so how long from the start of the experiment it took to get to this peak. When we were looking into the experiments, we've identified some experiments that had very similar total heat release rates around 7 gigajoules or so, which means the amount of heat produced in all of those fires was almost the same. And looking at the curves there all over the place, one fire you have immediate peak to almost 8 megawatts, immediate, like first three, four minutes of the experiment. The second fire grows very slowly and then in around 20th minute it reaches its peak of 4.5 megawatt and then it slowly declines for the next 20 minutes. Third one again the same total heat release rate. It does almost nothing for 10 minutes, then reaches maybe a megawatt, stays like that for like 30 minutes and then peaks extra about 5 and then quickly decreases. All of them having the same amount of energy released. Three completely different courses of fires. If you're trying to put them directly into your analysis and you want to be annoying to your investor, put the first one. And if you want to demonstrate that your system is the best smoke control system in the world, just pick the third one, which didn't even start until the 10th minute. I mean, of course I'm joking. You cannot do that by the means of this study, by looking through this data. We were really interested. What is the reason for such diversity between fires in similar vehicles of similar type and of a similar size? The same total heat release rate, and it obviously was that. I've told it so many times in the podcast. Vehicle is a collection of compartments and the fire spreads through the vehicle. It's not a crib, it's not a pool fire, although you sometimes end up with one if you have a massive hole in your gasoline tank but it's not going to burn at once. There has to be some events happening during the fire of a vehicle. So in the first one, the one that immediately reached its peak, the fire was set underneath the car and that was a battery vehicle. Very quickly the battery was involved in full in the fire and at six minutes we already had the maximum value of the heat release rate of the vehicle. It was a very large burner. Two megawatts on its own is a very large fire. So a battery violently exposed to a very high source of heat, that's what you could expect. It very violently burns out and, as a side note, similar thing would happen if you had a massive fire and then directly underneath a plastic fuel tank filled with gasoline, which was also observed in many experiments also mentioned in this podcast by Magnus and in some episodes ago and in episode five by Roland Bishop. In the second fire that we've investigated, which also was electric battery vehicle, the first source was directly underneath the battery, again, but this time it was much smaller, 30 kilowatts, and in this case it took 22 minutes for the battery to go on. Once it went off, it was a violent fire again. Third one the fuel source was again underneath the vehicle, but this time not underneath the battery but underneath the left back end of the vehicle and in this case I told you, for 10 minutes not much happening. So this was the time where the fire was breaking into the trunk of the car. Then a megawatt where the trunk was burning and then, around the 40th minute, a violent event the windows fell off, the battery gets involved and the car flash overs, if I may. So, as you can see, three completely different courses of fires, if you look at the plots and you know all of those were battery vehicles. All of those were ignited from underneath. Why would you get such a scatter? Doesn't make sense. How can you figure out the design fire out of that? Should you average them For times, you'd get a median. But once you start learning about the entire process of the tests, how it was carried, it starts to make sense. The peak is when the most flammable, if I may, part of the vehicle gets ignited and fully contributes to the fire. Now the big question is can we predict when a dealt will happen, because a lot of our designed fires have to be a function of time. And that's such a difficult question because, looking at all of this data dealing with this for years, I don't think we can actually do that. It's in a way random Maybe random is not a great choice of words, it's kind of probabilistic. It depends a lot, and the only way we could learn that it would be through burning I don't know 100 vehicles, all of them in a very similar way and seeing how this progresses. But then keep in mind, if you move the fire source to a different place, the fire would go through a different way. So we keep the conclusion that predicting the events in the case of fire in the vehicle, how it will spread through the vehicle, and assigning one ultimate fire curve that defines the growth and the spread of the fire within the vehicle and the heat release rate that corresponds to that, was not truly possible with the data we had. So the best thing we could do with the data we had was to have a statistical overview of the fires that were reported in the literature, and that was next step of the analysis. In this dataset we noticed that the origin of fire does not really influence the total heat release rate. Total heat release rate is mostly influenced by when they started extinguishing the vehicles and in some bigger fires that they tend to extinguish them sooner for reasons everyone are running a fire laboratory in the stands. We did not observe a significant difference in things like effective heat of combustion or even peak heat release rate. We see a difference between the origins of fire and the fires where the ignition source was underneath the vehicle tend to have a higher peak, but it's very difficult to say it's that it's a really strong correlation there. However, for the time to the peak, we see a very strong correlation with the origin of fire. If you ignite the vehicle with large source of fire from underneath the vehicle no matter if you have electric or internal combustion engine vehicles you're very quickly exposing the part of the car that has the biggest fire potential, that being the tank full of fuel or your battery, and those fires tend to peak much sooner than the fires where the fire would originate not from underneath the battery or the fuel tank. I mean, to some extent are we measuring the pool fire or the vehicle fire? Really, that's my question. Anyway, we went further and we tried to do some statistical analysis on the person tiles, trying to figure out the average and some higher person tiles of heat release rates, peak heat release rates that you find in different vehicles and for the internal combustion engine vehicles, all the tests that we have accumulated in our database, we ended up with a number around 4.9 megawatt as an average peak of internal combustion engine vehicle, 6.7 for battery electric vehicle, 8 for hybrid. So yeah, this could indicate that they're bigger, but as you go to higher person tiles it kind of flattens out. So at 95th person tile we have 9.4 for internal combustion, 9.1 for battery and 9.7 for hybrid. Of course, keep in mind that number of tests available on batteries and the hybrid vehicles are significantly less than we have on internal combustion engine vehicles. That's obviously skewed. But then we took another look on this dataset. We thought OK, so if igniting the car from underneath makes such a violent difference in the fire behavior, in the way how the fire grows in the vehicle and for me it's perhaps not the most realistic design scenario for fire curing in a car park it kind of indicates arson. And if we use arson as our design fire wire, are we considering a single arson, a failed arson? If I was an arsonist, I would set a fire to multiple vehicles. Please don't take this advice. If you're an arsonist, by the way, you shouldn't listen to spot cause. You're arsonist. I hope I'm not very helpful to that part of fire enthusiast community. Anyway, we've excluded the fires from underneath in the database, and then it was quite interesting because then the average for internal combustion engine vehicles is 4.8. For battery vehicles it's at around 5. Top percentiles like 98th percentile for internal combustion engine vehicles, 9.2 megawatts for battery electric vehicles, 6.2. Not enough data to give conclusions for hybrids. So what do we learn from that? Is that we have more tests done with the fire triggered from underneath for new types of vehicles, which perhaps is helpful if you're studying extinction or suppression, like my guests Elena and Magnus have discussed a few episodes ago, where they were interested in finding out how well you can suppress the fires in electric vehicles. To test that you need to set them on a fire, on a real fire, and you have to do it reliably and repeatedly. So they've defined the methodology that allowed them to get that exposure. But if you're interested in how a natural fire would grow in the vehicle, perhaps setting it on fire from underneath is not the most representative way you can burn a vehicle. I'm very open to your opinions. Perhaps I am in the wrong in here. It's again a disclaimer. These are my opinions, results of our ongoing study, and the way how we view the data that we have accumulated and the way how we have used this data for our own project Doesn't mean that it's a design recommendation for anyone. I mean, if you are basing your design on a podcast, you really should not. This is not a design recommendation. You should do this science by yourself and take responsibility for it. Anyway, disclaimer away Two more perhaps interesting numbers. The total heat release rate that we found for 98th percentile was roughly 10.3 gigajoules, so that represents the bigger end of the cars, and the average was around 5.6 gigajoules for all internal combustion vehicles, seven for electric vehicles. Again, the numbers between the vehicles are pretty much close and this number is perhaps skewed by the any extinguishing actions that were taken in the experiments. Again, very difficult to have a clear number We've also investigated was the effective heat of combustion, which we found to be 24 megajoules per kilogram. You get that from tests which had oxygen calorimetry and the massless rate measurements in them, and this number is very important because it will in a way, drive the smoke production. So 24 megajoules per kilogram was the value that we have eventually found through the statistical analysis. And that's it, my thoughts on the experiments. The most important one is that the way how you do the fire experiment in your vehicle will definitely influence the outcomes of this experiment. And oh, this goes further than just where you set the fire. If there's a roof above the vehicle, how big the roof is, how you extract the smoke, are there walls around the vehicle? Are there secondary vehicles near to it? And can there? Will there be any heat feedbacks from fires of those secondary vehicles? So many feedback loops that will occur in a real fire that we try to exclude from fire experiments. Sometimes it makes it very, very difficult to really capture a real car fire in your experiments. Final stop on this is that sometimes people tell me that you should just take an experimental curve and drop it into your CFD and just justify that that's a real experiment. Well, yeah, they are, but I'm not sure if I would take the responsibility to take a single data point, a single curve coming from one experiment, if I was not sure that the experiment was carried in a way that represents the design fire that I envisaged for my car park. And here we come to the second thing related to how we design design fires for car parks, what the design fire really is. So if you think about DesignFire as an input to an analysis, then you are treating it in a very similar way that we treat the ISO curve for furnace testing. That's a parallel that I like to make. If DesignFire is a rigid input and you just define how heat release rate changes in the function of time, be it alpha t squared curve, be it a prescribed growth rate, whatever way you used to define this evolution in time, you're basically describing a rigid fire. A rigid fire that I considered a test condition, not a real event In a real building. There is too many variables that will influence that fire. First, what sparked around your vehicle Can the fire spread? Second, the heat feedbacks that you will have in a real car park. That will be very different in different places of that car park. All of those will influence how the fire grows. The location where the fire started, the ventilation conditions. I mean, if you talk with experimental people, a lot of them will tell you that windows were a little bit open when they've run the fire experiments in vehicles, because if you do start a fire inside a passenger cabin with closed windows, very often you will end up with a fire that self extinguishes before anything interesting happens. So such a tiny detail, like if the windows were open or not and we usually would park our vehicles with windows closed, right A lot of tiny things that would determine how the fire grows, decays, spreads and what happens with it. Another one is the human factor. Will anyone attempt extinguishing the fire? Will the fire brigade be called? Will it be automatically informed? Will it arrive after 15 or 25 minutes? And by arrival I mean they start the actions, not the fact that a shiny vehicle is outside your building. That's not the start of extinguishing yet. So a lot of factors that pretty much make it impossible to say okay, this is how fire will develop. In my car park that I'm analyzing, we're always resorting to a test condition, to a fixed scenario that represents a threat that to some extent we understand, a threat that to some extent we are comfortable designing for, a threat that we can use on multiple design occasions and we can build understanding how different car parks react to it. I really like this parallel to do ISOCURVE, because now my team has done CFD for, let's say, 100 car parks in Poland. That's a big chunk of experience that we've built up. And now I'm doing a car park number 101, and I can relate what I see to the results of the previous ones because I'm using the same test condition, the same design fire I've been using forever and I can see if this car park is the same, different, better, worse. It allows me to bring my judgment into action. So a design fire will never be a representation of what a fire will look in the building really. If you want that, the closest way to get it is by listening to episode 48 and going probabilistic. Perhaps that's where you could go the closest to what could be the outcomes of a fire in the building. But don't be naive. The description we use in our everyday design practice, the thing that we call the design fire, it's not a real fire in your building, it's just a test condition that we apply to make judgment based on the tool that we have and we feel comfortable with. Sometimes you would base the test condition on a standard that's very helpful. Like we know, bs7346. That would give you 4, 8 megawatts. I was also repeated in 12-1-1, part 5, en. There's also a new standard, 12-1-1, part 11, that I participated in creation and in there you can find 4 and 10 megawatt design fires which you can use for design. So referring to a standard it's obviously more comfortable than defining the value yourself. Those exist and that's it. They are test conditions. They are not real fires and if you want to seek for real fires, I don't think you will ever be able to find it. That's also why it's so difficult to now include for bigger vehicles, to include for electric vehicles in this design fire paradigm, because it was artificial from the start. I'm not sure if we can update it easily for the new challenges. Another point that we are getting is promising to an extent, seeing that those fires, their peak heat release rates etc are not significantly higher than the previous ones. I'm talking about electric vehicles for larger SUVs. I've seen electric Hummer, that's a thing, and I have no idea what would be the peak heat release rate of that monster. It obviously will be larger. I mean it's more fuel, more plastic, more material to burn, bigger area to burn, so obviously a larger fire. Where the number will be, I have no idea, but the thing that gives me comfort. A lot of comfort is that if you design for a specific design fire and the real fire is, let's say, a bit bigger, let's say 30% bigger than the one that you've used to test the systems, it doesn't mean that systems stop working in your building. It means they will work less efficiently. Your smoke control will be still removing smoke, your detection system will still be detecting fires, your sprinkler system will still be putting water into fire and smoke, cooling it down effectively, even though it's larger than assumed and perhaps the performance will not be as expected. If your system has met your design goals in your design fire in an event which is slightly bigger, you should still be giving the occupants the chance to escape, the firefighters chance to enter. Of course this has its limits and perhaps you should run a sensitivity study. That's also an interesting research direction. But in general, it's not that the systems completely stop working the moment that the fire size exceeds the design one. There would be still some residual capacity in the systems' high design. That would improve the conditions in the fire, in my opinion, significantly. And now we go to part three what are the goals of our design? So I strongly believe that your design fire should reflect the purpose of your analysis. And for car parks, I can easily identify three main goals of analysis. One would be the life safety evacuation. Can people escape the car park before the conditions inside overwhelm them? Second would be firefight access. Can firefighters have a reasonable chance of entering the car park and performing effective firefighting action before the fire is overwhelming for them? And the third one if the fire grows to a large size, what would be its impact on the structure and after what time it would lead to a structure collapse? So those three things would really need three different design fires. If I'm talking about human safety, let's face it. If you put that fire I've mentioned a few minutes ago 8 megawatts growing in first 3-4 minutes of your simulation, if you put that into your simulation, unless your car park is 5-6 meters tall there is no way you can show their safe evacuation in that. But also I recognize that this experimental curve comes from an experiment with a very large heat source underneath the vehicle. If I narrow the design fires to events not starting underneath the battery or the fuel tank but all the other fires that were reported in the literature, you get much slower growth rates or perhaps would reach a smaller value of the peak heat release rate in the very earliest stage. I'm talking about first 3-5 minutes of your fire. Then you can probably quite comfortably choose a design fire with a size of 1.5-2 megawatts, either growing with some alpha t-squared like fast curve, or just having a linear growth rate like in Professor Schleich's curve that's part of TNO report that we've also been using in Poland previously. Or perhaps you could just start with a flat value like kind of steady state. I can't recommend you the exact value, but 2 megawatt ish is the range that I see for the early phases of the fire, really the first minutes. See, if you're in such a scenario your smoke control system can reasonably take this smoke. Can you provide escape routes sufficiently clear of smoke? Can you actually deliver safe evacuation in a setting like that? And I think that would be quite a reliable assumption for an early phase of a fire, so early stage. I'm now gravitating towards those steady state fires and we will probably pick a number between 1.5 and 2 for firefighters access. Now we are talking about fire that has been growing for a good while. As mentioned before, it can as much take a few minutes as 40 minutes for a fire to reach to its peak value. It's very difficult to define when that peak will be definitely reached because it will be different for different cases and also the time to intervention will differ from case to case, depending on what information the fire service receives. Are they automatically informed? Does someone have to call them for them to come? That that will be completely different times for intervention. Anyway, you can perhaps quite reliably assume that the fire will be able to reach its peak at the point where the firefighters will come. That would mean something different for sprinkled carpaks. That would mean something different for unsprinkled carpaks. In sprinkled ones you could assume safely that sprinklers have activated at that point and the fire power was to some extent reduced by the action of sprinklers. Of course it's difficult to assume that sprinklers will take out the fire. They will not. But the feedback loops will be disturbed. The heat from the smoke will be removed. They will have a positive impact. A lot of standards go back to the British standard where 4MW was the design fire for sprinkler scenario and I think that's a reasonable point if you have sprinklers, if you have unsprinkled carpark, a data point that we were usually using was 8MW at that point, as I've just told you, a minute before, we observed larger fires for 98th percentile, reaching 9.2MW if you exclude the fires from underneath the vehicle and 9.6 if you include those for 98th percentile. So much larger fires of a single vehicle than 8MW exist. Given the fact that vehicles are larger, I would tend towards saying that the fires I would expect would be larger. So perhaps the 10MW data point that comes from 12.11, part 11, is not a bad assumption and this is a single vehicle fire. Now, interpreting your results, it's not just seeing if the firefighters can enter when there's a 10MW fire around. It's, for me, it's also understanding if the fire can spread or not. And that can be done by analyzing the radiation, the heat fluxes around your vehicle, by understanding where the smoke and gases can accumulate. Does the architecture of the carpark support efficient removal of the hidden smoke or there are places in the carpark where the smoke can accumulate, create a very hot smoke layer and accelerate the ignition of subsequent vehicles? Because if it does, then you could expect a growing fire and that should be one of your concerns. This is a type of analysis that we are usually pursuing in our engineering. We analyze can the fire spread easily or not? I cannot really answer the question will the fire spread or not? Because that again, to some extent is random and you would need different heat flux for a tire, different heat flux for your upholstery inside the vehicle, different heat flux for gaskets, for example. It will depend how far the secondary vehicle is. What are the feedback loops around it? The heat feedback loops around it. So not easy to tell if for sure the fire will or will not spread. However, you can determine if the fire will spread easily. If you have 25 kilowatt per square meter around your vehicle, in the radius of few meters around it, you can be sure that the fire will spread and your systems may not be enough to provide a sufficient fire fighter access. Because that's what you're looking into. If the firefighters can enter. Is there a path through which the firefighters can enter the car park, identify where the fire is and pursue action? I really should invite a firefighter maybe Shuman again to the podcast to tell you about how the action in the car park is done. He already was mentioning this a bit in the previous podcast episode which I'll link in the show as perhaps you will be interested in that. The third one, the response of structure to the fire. Now that's a big one, because if you assume that the firefighters came and you have had only 8 megawatt fire and then the firefighters came and stopped it, then the risks to your structure are very low. There is risk for local damage to the neighboring members, like columns or beams above the location of the fire. That perhaps could actually be damage destroyed. But from a single vehicle fire I would not expect much bigger damage. However, we all recall the big fires Luton, liverpool, eco, warsaw, staffinger Airport. All of those fires were associated with high structural damage because these were very long lasting events with very severe fires unable to be controlled by the fire department Out came. This is very similar to the traveling fire concept that is used in structural fire engineering and I actually think a traveling fire concept could be adapted to represent vehicle fires. This is also something we will be working on in the future to investigate the spread between vehicles and how the fires can grow if you are unable to control them by actions of fire department or the systems you have in your building. So that's, let's say, in the future department. However, here again, if you just put a heat release rate curve of a single vehicle and you omit the fact that the fire could violently spread to neighboring vehicles because your car park is perhaps very low, or perhaps you have multiple levels of vehicles parked, one above each other, and no sprinklers. Scenarios like that. It's naive to just take a single vehicle fire curve, drop it as your design fire and say my structure, it will not be damaged by the fire, because you don't know that the fire can grow in a different way and Can potentially be the reason why your structure is significantly damaged or even destroyed. That's why I say that design fire should reflect the goals of your design. When picking a design fire for human evacuation, for firefighter access, for structural damage, you are looking on different ends of fires, you're looking at different challenges and your design should account for that. So now some final thoughts. I've already gave you some numbers from our statistical analysis. It's published in Polish I'll. Once the paper comes out, I'll link it in the description so you can look up the numbers. The data came from the well-known, well identified research that has been published in the past up to, I think, 2014. Its best described in the PhD thesis of Mochtow here. I'll link to that PhD thesis in the description and since that time We've been accumulating the scarce experiments that were published around the world. So that's not a huge data set and if anyone knows a good data set on vehicle fires, please send it to me because I'm super interested in. My final thought is that for my entire professional career I was working with the growing fire paradigm. We were using the design fire described by professor Schleich in the TNO report in the Netherlands standard. Then we've adopted that as a design fire that we were you Recommending in our Polish standard and I'm still using it. I'm still comfortable using it. I like it because it has those distinct phases of fire being Medium-sized 1.4 megawatt to be exact in this design standard and then growing to a size of approximately six or eight megawatts, depending if you assume sprinklers or not, and that for me represents the conditions where the firefighters can enter. So this test Design fire allows me to assess my structure versus evacuation, versus firefighters entrance all the things that I am looking for in my analysis. We had to adjust it a little bit to accommodate for electric vehicles, something that I've covered in episode 6. So we, when we are dealing with electric vehicles, we make the 1.4 megawatt peak appear a little sooner into the fire to reflect the potential of burst fire in a battery that started in a battery. Those fires can develop very quickly and and we've noticed that the fire evolves literally within seconds. It's a big challenge for smoke control systems in the car park, so that Modification of the design fire accounts for that. But the more I think about it, the more I'm thinking we perhaps should just move to steady state fires, just just put one number on the design fire, perform analysis that Doesn't take time in their account. Perhaps it could take time into account, Perhaps you could run it for five, ten minutes, whatever, but Scenario in which we can quite reliably tell what is the performance of the system against this design fire. I strongly believe that running such simulations in multiple spaces of a car park would yield much more information than Running one very long transient simulation simulating the growth and decay of the fire. It's a massive shift. For me it's Literally a paradigm shift. It's even more interesting. It's a shift back to how this was done many, many years ago, where you only had steady state fires and that's how we would assess the performance of systems in your buildings. So for the end of this episode, my final thought is. I'm really reconsidering going back to steady state fires, giving up on this transient evolution of design fires in the car parks. I simply don't see a pathway where I could really say that this particular one Universal design fire reflects the growth and decay of fire in any car park. It's based on evidence. I see it's simply not possible to say that this design fire is a Representation of a real fire, and they never are. They are test scenarios. And if it's not a real fire, if it's a test scenario, I see a pathway for more efficient test scenarios which allow me different approach to my engineering with, which will allow me to run multiple simulations at the same time, which will allow me to draw conclusions from multiple simulations, large number of simulations and Perhaps will allow me to engineer my systems better with less uncertainty. So yeah, that's it. I wonder if this resonates with you, if you expected one single curve from this episode. I guess I end up this disappointing you, but I think it was a bunch of considerations that go through my head that perhaps will be interesting for you If you have a different opinions, if you design it in a different way, if you approach this problem from a completely different side that I am. Please show your thoughts and opinions, because it's still a work in progress for us. It's still something I like to think about develop, improve. It's not the only thing that we are working regards car parks. It's a part of a much bigger, bigger set of developments that are happening at it be which I, hopefully, will be able to share with you. Anyway, I'm open to suggestions. I hope you've enjoyed this episode, even though I'm doing the best, to diligence, I can, to provide you with Reliable data. Keep in mind this is a podcast episode, not to be reviewed your paper, so perhaps you don't want to quote me on your design. Do your own investigations. The data is available. There are great sources that accumulated this research, so you can actually go through it on your own. And perhaps soon, if the journal likes it and the reviewers agree, we will publish a peer-reviewed paper on that and that would be a very good thing to cite in your design. And if that happens, I'll let you know. For sure, and, as usual, I will be here with more content for you in next Wednesday. Oh, and remember, community dot fire science show dot com. That's a place where you can reach me, where you can talk with me, where you can meet other fans of the fire science show, enthusiasts of fire science Hopefully, not arsonists. It's a place where the book of fire lives, the online curse for people who are just starting their fire career, who wish to find resources, who need some guidance on what fire science is and how can you learn more about it and where can you end up being a fire scientist or fire researcher or fire engineer. All of that is in the book of fire. I highly recommend it to you. If, if, fire science is something new for you. Once again, community dot fire science show dot com, that's where you will find it. All it's free to access. Just register your account and join us, and I hope to see you there, and I hope to see you here Next Wednesday. Thank you very much. Bye, oh you.