The subject of structural fire engineering was long overdue in the podcast schedule. But once I finally got it on my agenda, I made sure to interview one of the very best there are - prof Thomas Gernay of John Hopkins University. Not only a structural engineer and researcher, but also one of the developers of SAFIR® - one of most popular structural fire engineering numerical codes out there.
In this discussion I get to ask some important questions on the role structural fire engineering plays in engineering modern buildings, and Thomas makes the point that it it the way forward in understanding the building performance in a holistic way. I learn about FDS-SAFIR integration (which is superb interesting!), challenges with new materials and the development of design fires (hint - travelling fires get a ton of mention, so be sure to also tag episode 27 with Guillermo Rein on them!). And as usual, in the end, we geek on the future of fire science and technology. God, I love these discussions so much!!!
Make sure to check Thomas webpage at: https://mars.jhu.edu/
If you would like to learn more on structural fire engineering, Thomas has compiled a list of resources you should check:
[00:00:00] Wojciech Wegrzynski: Hello, and welcome to the fire science show. Episode 57. Hope your guys are having a great summertime and some fire science in between will not hurt for sure.
[00:00:10] Today I have a subject that's quite difficult and, me as well, I don't have great expertise in this. So I, I was very happy to learn firsthand from the world, leading the expert, and the topic is structural fire doing and structural fire modeling.
[00:00:24] in fact, I've invited professor Thomas Gernay from John Hopkins university to. Bring me in line , on the, how structural fire engineering works. What's the newest developments in this field. And, what's the future in front of that.
[00:00:38] Thomas is well known for his considerable amount of knowledge put into the world of structural fire engineering literature. And he's involved with one of the primary structural fire engineering codes that is severe.
[00:00:51] so he has first hand knowledge, not only from the perspective of a. Researcher structural engineer, but also as someone who's developing [00:01:00] new tools for us all to use.
[00:01:02] So that's a really great perspective. And you'll see through the interview that we are touching and referring to that all the time, because his perspective is unique.
[00:01:11] So I hope that despite the little summer slack, most of us have, we can join me in this episode and learn a bit about the importance of structural fire engineering. And the role that it plays in a performance based design framework. Okay. I guess that's enough of introduction. It's a good episode. You want to listen it to the end?
[00:01:35] It's well worth it. So let's spin the intro and jump into the episode.
[00:02:02] Wojciech Wegrzynski: Hello, everybody. Welcome to the Fire Science Show. You guys ask. And I deliver here with me, Dr. Thomas Gernay from John Hopkins University. Hey, Thomas.
[00:02:11] Thomas Gernay: Hey, Wojciech, how are you?
[00:02:13] Wojciech Wegrzynski: I'm very good. I'm very happy to have you on the
[00:02:16] Thomas Gernay: just, how are you?
[00:02:16] Wojciech Wegrzynski: And finally talk with someone about numerical modeling of structures. for a nice start. I've heard there was a really nice event lately in Italy, summer school on, structural modeling.
[00:02:28] And I keep hearing about this event from everywhere. Like everyone is telling me how great it was. And, I know the viewpoint of participants. Uh, you were one of the speakers there, so, so maybe you can, uh, bring me in line. What, what happened in Italy or unless what happens in Italy stays in Italy.
[00:02:45] I dunno.
[00:02:46] Thomas Gernay: No, I'm happy to, tell you all about it. And first of all, thank you so much for having me. It's a great pleasure. And I'm a, I'm a fan of your show. So yes, we had this week long, uh, summer school in Como one of the. You know, best places in the [00:03:00] world, really, really nice location. in may, that was organized , by dear colleagues at the Polytech in Milan, Roberto Felicetti Patrick Bamonte and Jean-Marc Franssen was also involved.
[00:03:10] So the goal was to bring together, 10 speakers, , for, each of them delivering you have day lecture. Topic, you know, of their expertise and have, , PhD students and researchers , from all over, get the ability to either come in person or attend remotely. And it was very, interactive and productive.
[00:03:30] It was very nice to be in a room together and being able, not only to, teach about a topic. Close to what I've been interested recently, but also get the feedback and discussion and get to know what all these researchers are, working on. And I mean, definitely lots of, very exciting, topics looking at new problems.
[00:03:51] so just all this structural fire engineering community coming together was great. Um, Yeah, I got , to talk about, the burnout resistance [00:04:00] of structures, which is a topic I've been researching for quite a long time. Actually, that started with my PhD work, where I worked on the development of, numerical model for the behavior of concrete activity, temperature, and.
[00:04:17] Paid particular attention to what happens during the cooling down phases. We wanted to make sure that we were capturing, uh, irreversible processes and damage and transient creep and all these stuff. And so with these models, then we became able to model more accurately what happened during cooling and whether there was a risk of delayed failure.
[00:04:39] And since then it. Evolved into modeling all types of structures during heating and cooling and, developing a framework to understand when there is a risk of delayed collapse. So that, that was about,
[00:04:50] Wojciech Wegrzynski: that's a great topic. And I, I have a feeling we're gonna touch on it a lot in this episode. And I've asked about the summer school because there's a lot of, young fire engineers [00:05:00] listening, a lot of, uh, researchers, uh, students graduates, postgraduates, PhD, students posts. If you ever wonder if it is worth to go summer school, it is worth it.
[00:05:11] Totally. from my experiences with summer schools, it is absolutely amazing way to, to get the knowledge probably the best way, even better than my podcast, I would say but yeah. that. Okay. Uh, so, let's go back. PhD thesis. You said you were developing numerical models, and I also know you are involved in development of SAFIR® of the more popular codes in structural engineering.
[00:05:38] So as we venture into the world of, numerical modeling and structural fire engineering, maybe you can tell me like the story, why there was a need for a. New codes to be built or just wanted a new ya and, and yeah, and the home had Italy coast from that. Well, that's a good motivation too.
[00:05:58] Thomas Gernay: Well, certainly I'm happy , to [00:06:00] tell this story because I think, it's a nice one and a timely one. So, my PGTs is situat. So we are talking about like 2009, 2012, that range and, The, the tensile membrane action was a big topic. since I would say one, one decade earlier, right?
[00:06:15] It had originated with the observations at Cardington in the 1990s with the Bre and all the research in the UK and professor Wang, professor Bailey, you know, doing those. Semi, , findings the beginning of two thousands. And so, it, it became clear that there was really, a need to be able to model, entire structural systems and, the behavior, and not only the structural frame, but that the behavior of concrete and the concrete slabs, became also something that, that matters and that influenced the response.
[00:06:46] Wojciech Wegrzynski: So with the tonsil membrane actions, you mean , the way how slabs will respond to the load, how they would carry the forces in the building and how important part they take in like the general structural stability of the building. So not [00:07:00] just the frame, but also everything that connects the frame.
[00:07:03] Thomas Gernay: Exactly. Yes. Good point. So let me backtrack. Exactly. So at, the Cardington what was done is that, uh, in the 1990s, the Bre, , they tested entire buildings.
[00:07:14] Wojciech Wegrzynski: Yeah.
[00:07:15] Thomas Gernay: Subjected to natural fire, right. To a fire. And so that was the first or one of the first time, you know, that they really investigated the behavior of the structural system as opposed to testing isolated members in, in furnaces.
[00:07:27] And what was observed is that the behavior as a whole was much more favorable and you had this redundancy and those alternate load pass that you could activate. So you end up with structurally. Behavior that's more robust than you would predict based on single members' behavior. For example, you could leave some of the steel beams with no thermal insulation, right?
[00:07:50] They would, have exhibit large deflections, but when you get those large deflections in the floors that, go with them, that deflect with them. , you start activating [00:08:00] tension in the reinforcement. That's embedded in the concrete. And so this steel reinforcement remains cold. It's protected, by , the concrete of the slab.
[00:08:08] But if it is designed properly, now you can carry the loads, with, this alternate load pass and this, uh, in this deformed shape, which becomes really interesting because it means you have a robust behavior. That's really embedded in the structure that doesn't require. Insulation from the thermal election.
[00:08:25] so with this realization, they started to be, you know, development of numerical models to try reproduce this behavior and understand it, and then work toward design methods, analytical methods. but SAFIR was already around and I want , to come back a little bit later on how it had been, uh, developed originally.
[00:08:43] Right. But when I started my PhD, , I would say the behavior of concrete had not been so important before, before 2009. But, , there started to be all this research on Tenile membrane action and, large scale European research projects. And we really wanted to have something rub robust [00:09:00] and accurate to capture what was happening in those concrete floors.
[00:09:04] Uh, when they were undergoing transmembrane action, large displacement heating and cooling and so on. So that was really the motivation to fill this gap in the. Computational, modeling tools at that time to have, actual models for concrete in multi Excel, stress date, subjected to heating and cooling and so on.
[00:09:23] Wojciech Wegrzynski: Hmm.
[00:09:23] Thomas Gernay: so before that it means SAFIR was, was already run and those software as well, of course, to start modeling, you know, structural fire behavior. And I think that started in the late 1980s and mostly in the 1990s. The idea was that, of course, to achieve fire safety of the built environment, you have those different layers, of safety that , you want to rely on, right?
[00:09:46] So you want prevention, you want, active fire protection and so on, but as a last resort, you really have the structural, uh, fire response. For quite some times, Al engineers did not really focus [00:10:00] much on the structural fire response and fire protection engineers probably had the oversimplified way of thinking about structural fire response and my PG advisor, Jean-Marc Franssen.
[00:10:11] Told me the story of first time he, visited to the us, I think. And he, he, he got to, he was a young engineer expert in structural fire behavior, but starting, and he got to speak with, somebody working on the fire dynamics, fire protection engineer, and basically they were asking, okay, can you tell me the critical temperature of your structure?
[00:10:29] Right. Because I can do the heat transfer. I can find the temperature. So if you tell. The temperature at which your restric first, you know, we are all set. We don't need to look more into the structural fire response. So it was this idea that, it's a purely thermal problem, basically, which is, is not, of course.
[00:10:46] Wojciech Wegrzynski: And yeah, but it still is treated like that in many cases, we are drawing isotherms inside, beams or, even the concept of Charing depth loss of, area of, timber. Parts. So it still is, [00:11:00] in a way considered like a thermal problem. And if you think about it, like, let me tap to what you said in Cardington.
[00:11:07] They did for the first time, the full building, but Cardington was like in 1990s and we were doing structural fire engineering for a hundred years back then we've we've been doing it with furnaces. In the furnace in essence, unless you have a loaded element, that's a little different when you have
[00:11:24] Thomas Gernay: So.
[00:11:24] Wojciech Wegrzynski: element and on how you, connect the element to the furnace in terms of what, what the free end is, and then how it can react.
[00:11:32] But in, in a great simplification, the fire test is a thermal test. Like essentially. Test the thermal behavior , of your system and all the criteria you have are in a way related to the terminal response, especially the insulation and, the capability to, to keep the smoke sealed in inside the compartment, not spread through cracks.
[00:11:54] So. We are still , very in the element and, and thermal response [00:12:00] paradigm. so you want to say that that numerical modeling was a way to break this paradigm or, or escape it in a way.
[00:12:08] Thomas Gernay: Absolutely. Yes, you are right. We are still relying on this, uh, simplification in most of the case and, , very largely in the field. And as you say, numerical modeling is actually a way to overcome the simplification and, start investigating the actual, anticipated structural response under.
[00:12:28] Thermal demands. And that is important because we observed that in many case, the thermal approach itself or the element, approach and looking at these ISO terms does not, answer important question or does not really, predict what the NC two behavior is going to be. So the tenal membrane action was kind of the first, a very remarkable example of that, where you can actually leave.
[00:12:54] the temperature go much higher in, some of, in part of your structural frame and still have a very [00:13:00] robust behavior. And so that opens the door to, not only optimization and economic savings in terms of, thermal installation, but also understanding really the safety level of your, building, as well as having a behavior.
[00:13:12] That's very robust because this road bearing Capac is really embedded in the structure. There are many other examples where numerical modeling. Give you, information that is inaccessible with, uh, element testing and, based on the thermal response. So I mentioned briefly that I was talking about burnout resistance, uh, during , the summer school in Como.
[00:13:33] And actually we can rebound on this. , so if you start looking at, the structural response to how the different phases of the fire, and you're interested in maintaining stability to full burn out. Because, uh, structural collapse would not be acceptable because you're concerned about, fire service intervention or, resilience and so on.
[00:13:54] then you, you start, uh, having to model what happens during the different phases, [00:14:00] including during the cooling and the notion of critical temperature really does not. Makes more se much sense anymore, because you will have a delayed temperature increase in different parts of the section. And you have, for example, if we, if we think about a timber section, you might have the Charing that stops, but you still have the heat wave that goes throughout the core of the section, which will lead to additional, reduction of the load bearing capacity.
[00:14:25] maybe. temperatures is lower than the charring temperature, but still higher than what the material can, really support without, being, I would say, damaged or reduced. So in that sense, you really need this capability to model, structures in fire, because either you want to quantify safety, you want to optimize design.
[00:14:45] You want to look at stability to full burnout or many other other situations really.
[00:14:51] Wojciech Wegrzynski: Other thing that immediately comes to my mind, I is traveling fires and, , the concept of, of like four hour fire resistance. Like there, there is [00:15:00] like, no way fire can be in the same place for four hours is like that. That would be. Insane amount of fuel load in one place. Actually you can, um, have fun and try to do the calculations.
[00:15:13] , we were once calculating how much fuel load you would need, to, to create standard condition, on. A ceiling of an airport and we came to conclusion. It would have to be like literally stacks of, trucks filled with like Ikea furniture on top of each other, reaching almost the ceiling to, to have this four hour fire at this super tall position.
[00:15:35] But now, now back to traveling fires. I had an episode with Guillermo Rein about traveling fires and everyone's highly recommended to go back to that one I'll link in the, in the show notes to learn about the method itself. , the thing is that there is tons of possible outcomes of the fires. There are fires that will travel very quickly, which means they will be very high intensity fires, but they will burn through the fuel very quickly.
[00:15:58] There will be fires that are [00:16:00] slow, which burn slowly through the fuel, but they also spread slower. And they can go from left to right from right to left from middle out, from out to middle. there there's hundreds of, or, or there's limitless amount of, ways. How, how a fire can go through the building.
[00:16:16] Even if you're not modeling, if you, if you're just swimming some. Saw the fire behavioral as travelling fires, such as simplified model. And now, now you end up with hundreds or thousands or tens of thousands of scenarios to investigate. And there is absolutely like your fire resistance is just one data point.
[00:16:32] is it. that's your
[00:16:33] Thomas Gernay: That's.
[00:16:34] Wojciech Wegrzynski: point to one time temperature relation. In a certain amount of time within this boundary conditions of this particular furnace in this particular laboratory, that is pretty much the answer you have, and you have 10,000 scenarios to go through and compare.
[00:16:49] So no, no way you can relate that, or we try, but it's hard while in numerical modeling, you can probably just test them all or just figure out which are the. [00:17:00]
[00:17:00] Thomas Gernay: It's.
[00:17:00] Wojciech Wegrzynski: you also have experiences in, in this application , or what was it also something you have considered back then when you joined the SAFIR team?
[00:17:08] I guess traveling fires was not the thing because , I think it was just about to begin, in, in this time, but you short, hot or long cold Euro code fires, localized fires were there for sure.
[00:17:20] Thomas Gernay: Absolutely. It's a great point to bring, because I was telling the story of the
[00:17:25] Wojciech Wegrzynski: Hmm.
[00:17:25] Thomas Gernay: protection engineer, who ask us for, one critical temperature as if it would summarize the response of the structure, but we structural engineers also, uh, may tend to oversimplify and ask, you know, Fire dynamician, uh, to, to provide us with a standard fire curve and think that it'll capture all the possible fire behaviors.
[00:17:45] And we can work only with that. And of course that's not the case. And as Guillermo is, is shown with, , some papers where he showed that the. Traveling fires can challenge the structure to, to an extent that is different. and in some cases more on than, uh, I would [00:18:00] say compartment fire conditions, and also the opposite, where there are many situations where once you model The fire behavior based on what we call the physically based fire formally called natural fires.
[00:18:14] You find that the conditions may be less on than with the, standard fire or full post flashover fire. And that also opens, opportunities , for design optimization or. Even enabling architectural designs. When, you know, if an architect wants some exposed steel, or exposed timber and, and it really makes sense to look at what are the realistic fire scenarios that could develop in this type of structure and then, , investigate the response of those structure, those fires rather than look.
[00:18:43] What we are using in standard furnace test. Exactly. So, absolutely. To your point. this development of structural fire engineering also, worked alongside developments of fire models for structural engineers and a lot of the. [00:19:00] In the end, impact that it has had on, on practice and where really we could see a value of adopting these analysis was linked not only to a better understanding of the structural fire response, but also, more realistic representation of what, fire scenarios might be.
[00:19:16] So that's, I think that that's key. But I, I think more generally, to your point also, I wanted to compare with what we do in, standardized furnace testing versus what numerical models enable us to do. So in civil engineering and buildings, we are not dealing with, mass production product, right?
[00:19:36] We are not building iPhones or, or cars, uh, where we have the luxury to, build a few specimens that we will test for scale, because we are just building, you know, millions of them. On the opposite buildings are all unique. It'll never be an option to build an entire building aside the actual building we want to use and set it on, on fire.
[00:19:56] Wojciech Wegrzynski: mm-hmm
[00:19:57] Thomas Gernay: what we are doing in a sense is that we are just [00:20:00] taking a few components and, and applying a standardized furnace test and extrapolating that it'll inform us on the behavior of the, of the full, building. As if I would say in a car crash, worse in this test, we are just testing, you know, the, the bumper , and maybe , a few components, and then trying to understand how the car would react in the actual, crush test.
[00:20:19] and besides all buildings are different. So you, you would test the bumper and you would extrapolate that. Maybe it's, it, it informs you on the, the open, last try on the BME BMW series five the same way. Right. so we are never going to do that for ex experimental testing for the building.
[00:20:34] So really computational models make a lot of sense because there you can build these, you know, digital twins and challenge them against, lots of, , types of conditions and all those fire scenarios that you mentioned.
[00:20:45] Wojciech Wegrzynski: I would put one more thing on, on standard testing. Many people don't like this way of talking about standard testing, but, it is in a way, a reality you have to consider standardized. in relation to product it [00:21:00] is a standardized product testing. was never meant to be a test of a building.
[00:21:05] It wasn't because it fits so tightly within the framework of product being certified before it is delivered to the market. So the market understands the, that stands the core characteristics of the product. And that's exactly the need that, fire testing in laboratory, caters for. And therefore, because we have this as a standard product testing, the.
[00:21:30] Manufacturers use this as a ranking tool or, know, it's a way to compete. Ah, my product has this classification. Yours has other ones, so mine is better. It, gives you a fairly easy way to, overlook the market. Others, like only one product that has this specific barter and every other is where.
[00:21:51] So, so I lose this one and it also. It's a very attractive way for nonspecialist stakeholders, to, to put a product because they, they have no [00:22:00] idea, but the code says this product of this characteristic is just enough for your building. So, so like this is a whole framework fit around. Delivering products to market that cater to specific needs and can be distinguished by the characteristics.
[00:22:15] Whereas a building does not have a, a single, uh, response, fire, fire resistance characteristic. uh, like you said, it can burn and, and collapse in probably thousands of ways. of them will be different, very specific. And the job of structural engineer like you is, to make sure that in a probable scenarios, it, does, not collapse or, or that, um, maybe even more that.
[00:22:43] There is no disproportionate collapse. Like very small fire. Doesn't collapse it because if you, if a meter strikes your building, then it's not gonna probably survive. So, here, is this. Fork between the product market and the building market. [00:23:00] It's just in ice of many people. It's the same market where it not necessarily is.
[00:23:05] However, I, I still find but you still don't do structural engineering like that for every single building. Right there, there, there are like, cake type buildings that you just take from the book,
[00:23:16] Thomas Gernay: Absolutely. And so, yeah. Also let me make a few things clear. first of all, I mean, standardized for testing is tremendously important and it has delivered the baseline of safety. Right? It's been a great answer , to a true problem. At the beginning of the 20th century, it has. Product amazing benefits.
[00:23:35] And even for us, you know, working on computational modeling, experimental testing is absolutely crucial component to validate our models. Whenever new material are developed, we need the data, we cannot just, extrapolate material models. We have to have those, those data points. So all this absolutely.
[00:23:53] And so. As we speak. , I would say push on the limitations to make clear, why we do what we do [00:24:00] and what's the value added. But, it doesn't mean that there are lots of, rational. So for continuing doing those, those standardized tests, but I mean, I agree with what you said, it's really in, going further than, certifying these products and understanding or hating, the behavior of, of these, you know, beams and columns and doors and so on.
[00:24:20] How do we scale that up? And how do we re interrogate the true safety level of the assembly of the system? Once we build the building out of these components, and that is something that we don't have the luxury to look at true full scale experimental testing, because in civil engineering, we are building unique.
[00:24:41] Objects. Right. So, I argue that computational modeling is really the tool to, bridge this gap and to go to that, , higher level. And as you said, there are many ways that, a fire can challenge a structure. And we need to make sure that what we are building is, robust when it's in, I mean, the [00:25:00] true conditions and subjected to physically based fire.
[00:25:03] And when we are looking at performance subjectives that might go beyond, standardized rating or one or two standard for resistance, but we might need to demonstrate survival to full burnout or, the fact that the damage remains localized. And we don't have this kind of. Progressive collapses you
[00:25:23] Wojciech Wegrzynski: . And, in the interface between fire testing and, , full scale behavior. You use fire testing
[00:25:30] Thomas Gernay: were, uh, referring to and so on. Yes. And.
[00:25:31] Wojciech Wegrzynski: and, , as a way to obtain, , data on, on the materials on structural behavior. But in the end, , when you go into numerical modeling, of a building. If we can jump in into that part, you still like using the standard curve as your boundary condition?
[00:25:49] Or would you go like with your code, your localized fire, how do you like you, you have your properties of your materials and stuff. How do you define [00:26:00] fire in, in structural for engineering then? .
[00:26:02] Thomas Gernay: So when we, apply computational modeling for, , predicting the behavioral infrastructure and fire. And as you said before, it's not for. Every, small building or every case, but when we do so usually we frame it in , the context of performance based, fire design, right? So it means that we start by, uh, setting up objectives, performance objectives for design.
[00:26:25] So explicitly we have to
[00:26:27] Wojciech Wegrzynski: Mm.
[00:26:27] Thomas Gernay: define what we are trying to achieve or what we want, what conditions we want, the, to survive , and. Failure means and what, a successful design is, because that will be different, for whole structures. As I said before, for some, we may want to demonstrate survival to full burn out of the fires and for those a certain duration of stability or.
[00:26:50] Yet other objectives. So then at the, the second step, we will work on design fires. And in this context of performance based fire design, usually we don't [00:27:00] rely on the standard, fire curve, which really has another purpose. So once we are going with modeling, we are interested in, Trying to predict realistic or physically based fires that that would develop given the conditions and what we know of the inputs for the building.
[00:27:16] And so we might have to check both post slash over fires and localized slash traveling fires, depending on the geometry of the compartment, ventilation conditions, the fuel and. it is not enough to consider one design fire because there are so, so many uncertainties involved, a robust performance based design will rely on, uh, a series of design fires and, and look at the effects of variation of these inputs.
[00:27:42] Usually it's good practice to, in any case, check a localized. That might be very close to a column, for example, which will always occur at the beginning, right. At the early stage of the fire and then the, the full growth and development of that fire, possibly transitioning to flashover. we know of course, for example, [00:28:00] ventilation assumptions will have a big impact also.
[00:28:03] So, there is some guidance. to look at when, windows will, break. And so how, you can have your, ventilation condition suddenly change when the gas temperature reaches a certain value, these type of things. Um, so first performance objectives, second design fires. And then we go with, the tur marts to response of, of the structure
[00:28:23] Wojciech Wegrzynski: you have the assumptions, but you don't have them fire modeling yet. So you would go with like some simple, hand calculations, the model. I don't think CFD, but because I think that would be very rare. So how you go from assumptions to thermal?
[00:28:39] Thomas Gernay: You're right. I didn't mention it, but no, we have access to the full suite of tools there so we can do. Analytical methods, kind of OC code barometric fire that we have commonly used. For example, in probabilistic, risk assessment analysis. We want a lot of scenarios and really look at this variability of conditions [00:29:00] for localized fire.
[00:29:01] Wojciech Wegrzynski: go traveling fire, like we've mentioned before, where you would have a plethora of scenarios, which are like, pregenerated for your geometry with a very simple model of different complexity, depending which you use. Yeah.
[00:29:14] Thomas Gernay: Absolutely. So analytical model for post flashover or traveling fire or localized fire. So in the Euro code, , we are usually relying on models in the Euro code. you have Hasemi you have, uh, Heskestad you have now locafi will be included in the next, version of Euro code, which is a solid flame model to capture the, heat, radiation to a member that is outside of the fire.
[00:29:37] we have, sort of traveling fire models that you mentioned also with some very elegant and, and relatively simple solutions, but we also use FDS and, actually one of the. Not so recent anymore, but relatively recent development in severe was to facilitate the transfer between, output data from FDS and the thermal structural analysis.
[00:29:59] So [00:30:00] usually there are two techniques that we would use, to couple, CFD simulation. And I'm talking about FDS interchangeably because that's really the most common, with, uh, FVM analysis. We can use the concept of about surface temperature developed by, by Wickstrom and we've done that. So we have all those sensors around or switch.
[00:30:19] We need to make sure we have enough of them to capture thermal gradients to capture, shadow effect and so on.
[00:30:24] Wojciech Wegrzynski: Hmm.
[00:30:24] Thomas Gernay: And, uh, but if, when we are looking at. Modeling a larger structure. For example, we've used in open car park simulation and these type of things where we have lots of structural members and, we would need a lot of, of sensors.
[00:30:38] We have also, the option to use, automatic, interface file. So, it was developed with the developers of FDS. Actually, we can ask FDS to write a transfer file as a output of the simulation that will. Output the, the gas, temperature and radiant intensity along lots of directions, in space.[00:31:00] I mean, special coordinates in a domain that we define and SAFIR can read this transfer file and, uh, use that as input thermal boundary conditions for the heat transfer analysis in the sections. And that is the advantage that it's automatized. So it's well suited when you have a big structure and, uh, it captures the shadow effect.
[00:31:22] It, it sees the view angles and, and everything to capture radiation, , appropriately.
[00:31:26] Wojciech Wegrzynski: This is brilliant. I didn't know about that. I actually, uh, I'm like in ANSYS ecosystem, so, I know. A bit on FDS because, my I have to follow
[00:31:37] Thomas Gernay: All of
[00:31:37] Wojciech Wegrzynski: but I'm not like everyday user of, FDS outside of scientific research. And this, sounds like, really like, Really nice development sometime ago.
[00:31:45] I was
[00:31:45] Thomas Gernay: that.
[00:31:46] Wojciech Wegrzynski: like, what has changed in like last 10 years when I work in this field and not that much has changed, but this is something like really a huge quality of life, improvement that really allows you for less, uh, [00:32:00] connection between your fluid and structure That that's very interesting. And have you tried anything. in the, in, I know in ANSYS ecosystem, I, I have not have not mastered this functionality yet. One of my colleagues in the office is playing, but we have like, two-way coupled interactions. So if like, the structure, let's say you have a steel sheet roof, and if the roof fails, you have a.
[00:32:28] And that hole is then transferred from my structural model to fluid model. So the fluid model in, in incorporates the, the way how, uh, this hole would, change the flow fields. but I guess that that's quite complicated, probably difficult, especially if you have, solvers, not within the one ecosystem.
[00:32:46] Thomas Gernay: It is complicated. That's something we, we have not addressed yet, but it is relevant for certain applications. You mentioned an example. There are, of course other example where this, Two way feedback or two way coupling, makes [00:33:00] sense. I would say fundamentally, when I'm describing this, these developments, it really comes from a, a realization our behalf and people working in structural fire engineering, that a lot of the advances and the gaps were at the intersection of the disciplines as is often the case in, you know, in engineering and science.
[00:33:18] So. I told you earlier, the story of the fire protection engineers, who asked for the critical temperature and the, the structural engineer who asked for the, the standup, Fire resistance or fire curve. So, we saw that value was in, connecting those two words and making it easier in terms of the numerical modeling ecosystem to bridge and couple the advanced fire models with the advanced, , structural simulations.
[00:33:43] So we are putting a lot of efforts on these, and right now we have all these options for the one way, coupling from the fire to the structure. What you describe is, is very interesting. future development that can possibly, you know, is making even more important. [00:34:00] and I would say timely now with, the use of compostable building materials and this tension between sustainability and, fire safety that, that we are observing.
[00:34:10] And professor Luke Biby had the, had the great, uh, Position paper on this, the fact that we are innovating fast to address the climate emergency, and we are doing those changes in the way we build structure and we use material and, and we have, the example with facets, the example with mass timber and so on, and they are implications for, fire safety.
[00:34:31] And we need to make sure while we address this very important. Momentous, right. A challenge of climate crisis that also, it's not at the expense of public safety. And I think computation modelling is a tremendous hole to play there because innovation goes so fast. We need to predict what, the implications are going to be and what challenges it poses for fire, behavior, including structural fire behavior.
[00:34:55] But in some cases, there might be a question of two-way coupling that you were [00:35:00] describing it's abuse, that exposed timber is going to fuel the fire and affect the fire dynamic as well as having, you know, the load bearing function that it has. And so on. And there are, there are other examples where, you might have the switch response affect , the biodynamics , and thermal environment.
[00:35:15] It's also important in, um, other. Questions, for example, I would say compartmentation issues and even, , failure of non load bearing, uh, walls and compartment boundaries, that this is also something that right now we cannot really predict with those computational models, but. As we were talking about traveling fire, it might be interesting to have the capability to, predict if due to large deflections of the road bearing structure, it might damage some, non road bearing wards, and then the fire can travel to the next component.
[00:35:49] So there are still lots of, very interesting and, uh, challenging questions to explore.
[00:35:53] Wojciech Wegrzynski: For me, one of the most interesting is the behavior of, of glazing because it's something that, we [00:36:00] are very highly reliant on in our. and, and not always, we're very explicit about it because, uh, we just assume an opening factor, which is usually related to the glazed area, assuming that it's gonna fall off, which not necessarily is the case.
[00:36:17] And, uh, modern glazing is, is showing more changes. Did you have any solutions for that in, in SAFIR like special models or, or something?
[00:36:25] Thomas Gernay: We don't explicitly model, usually the glazing, we don't, we don't model, non road bearing, members. So tradition is a very
[00:36:32] practical answer.
[00:36:33] Wojciech Wegrzynski: Sorry. So that goes back to the, your assumptions and , the thermal model that is run before , the thermal structural model. Yeah.
[00:36:42] Thomas Gernay: That's right. So, a practical answer is that there, there is, , a guidance in the Luxembourg annex to the Eurocode, which is. Just pragmatic, way of approaching this question of, glazing and, changing ventilation. So in the Luxem book annex, they give [00:37:00] a temperature of the gas at which a certain ratio of the windows get broken, depending on the type, if it's simple, double, triple glazing or reinforced glazing.
[00:37:10] They also prescribe doing a few scenarios. So for example, they would say we're in a scenario where 90% of your windows are opened from the start and then another scenario where they are closed. But I don't know. I remember 30% that opens when the gas go to 200 degrees, C and then 90% when it goes to 300 degrees C it just gives you a framework such that you, first of all, you check different scenarios because we know the fire.
[00:37:36] conditions, the demand will be very different depending on those inputs. So again, one design fire is not enough. and also while you have some guidance, everybody does, you know, the same and, and it's, it is linked. It is correlated to, failure temperature of glazing based on thermal gradients on the, on the surface.
[00:37:53] But it's not a detailed
[00:37:54] of the, of, of the class behavior.
[00:37:56] Wojciech Wegrzynski: for this robust, answer this. was also not aware [00:38:00] of the Luxemburg annext. I, I need to check that out because it, it is something that I'm recently very interested at. I'm also, uh, about to organize the nice guest on, on the glazing in, in the podcaster. There's also an episode to look for.
[00:38:14] now I wanted to ask you, like, where is it going? Is there any particular direction that, structural FSC modeling is
[00:38:22] Thomas Gernay: How you
[00:38:22] Wojciech Wegrzynski: or SAFIR is, is heading? Like how, how do you guys
[00:38:26] Thomas Gernay: guys more
[00:38:26] Wojciech Wegrzynski: the future? Will it be like more simulations, more answers more probabilistic approach, or maybe rather, into root of optimization and trying to find.
[00:38:40] The cheapest, non collapsible structure you can build
[00:38:44] Thomas Gernay: you can.,
[00:38:45] Wojciech Wegrzynski: your boundaries. I wonder what, what's the hot topic now in the world of structural fire engineers.
[00:38:51] Thomas Gernay: that's great question. Of course, we are working really on a range of problems and always trying to move that forward. I think high [00:39:00] level I'm, I'm really convinced that numerical modeling is, is a key enabler of, you know, everything we do in, in, in civil engineering to engineering is, we want. Provide two things. We want to provide the tools, really the ecosystem that people can go. And it's, it's robust, it's validated it's state of the art and you can run those software to predict the response. And we also work , on the framework, which is to understand how. Computational modeling, can support and is linked to performance based design to probabilistic risk assessment and these type of, very important approaches that are used in not only fire safety engineering, but in general in structural design.
[00:39:38] So. In the future, what, what we are really focusing on now, I would say maybe, , three things that I, I want to highlight. The first one is that there are always new, , materials and innovations that come on the market that are considered and understanding the fire safety implications of these are really, critical and.[00:40:00]
[00:40:00] Because if we don't, it can, hinder the implementation of these in the market. So, I mean, we have, you know, amazing people doing developments to, bring these new materials to, uh, maybe be more, environmentally friendly, but we, we need to be able to, to apply them. So we need to understand, What the implication is for fire safety.
[00:40:17] And so that, implies being able to model the thermal mechanical properties, the behavior of those materials. In some cases, when it's, combustible material, it is, I mean, other challenges as well. And then how do we model the, these in structural system and, building simulations, right? So that's, that's really the fir the first point.
[00:40:36] the second one is. As I was just hitting at, working on the frameworks. So I'm very interested in performance based fire design, which means how do we go, about this, this process of defining performance objectives. So already lots of questions. What is, appropriate? Performance level, what kind of, targets reliability should we strive for?
[00:40:58] What, are the implications [00:41:00] of looking at resilience of the built environment against fire, multi hazard? So lots of questions they already, then once we have the performance objective, we have to define the design fires. And we've talked about that. There are also lots of scenarios and how, uh, do we know which one to select and then how do we make.
[00:41:16] Easy to, develop these and interface them with the SIM with the Al
[00:41:21] Wojciech Wegrzynski: I love how the challenge is like at the philosophical level, not necessarily related to like, transferring loads or, or modeling, uh, non-linear thermal behavior materials. No, we really have to sort out the scenarios and objectives. This is the number
[00:41:37] Thomas Gernay: Well,
[00:41:37] Wojciech Wegrzynski: like the need. I, I really love how you positioned it because it echos so well with so much that has been said in the, in the podcast.
[00:41:44] I absolutely love it, but sorry for interrupting. You continue this.
[00:41:47] Thomas Gernay: No, no, thanks APRI. But yeah, I think it's key. Absolutely. It's both the tools and the framework. How do we use the tool? We want to think very hard about that as well because providing the tools is not enough. We want it to be robust. We want it to, [00:42:00] make sense in terms of the risk assessment and the performance and so on.
[00:42:03] So working on the design fire, still on, in my performance base for design, right. And then on the structural modeling.
[00:42:10] Wojciech Wegrzynski: Mm.
[00:42:10] Thomas Gernay: There is a lot we can do. We've worked a lot, this last decade and before, but there are, there, there are still things we cannot do, modeling a CLT floor. we don't really have the element to do that, right now.
[00:42:21] So, but it's, it's used people build buildings with that. So if we want to model this structural system, That will be something more practical. We, we need a PhD to look at a PhD student. I made to look for three, four years at how do we develop the type of, shell elements and the material model and, and C on for this or Tropic behavior.
[00:42:39] So they are also this kind of questions of course, that we are working on. And so first was. You know, new materials, enabling that S on the framework, the per performance based design, also probabilistic. I've not talked that much about it, but I've worked a lot with some incredibly smart colleagues, just Ruben van Coile and Danny Hopkin, whom had here Negar [00:43:00] Elhami-Khorasni a good friend and amazing researcher from university of Buffalo and others.
[00:43:06] And together we, we teamed up because , it's a very. Big challenge to move, structural fire engineering, into the probability risk assessment framework and make it, very, consistent in terms of how we think about risk, what type of reliability we design for and on. And because it requires exploring a range of scenarios, also computational modeling is the natural tools to, to go around it.
[00:43:28] Right. And so the first thing I would say, , when you ask me about what's the future. It's really more forward looking now, maybe I'm talking about 20, 30 years down the road, but I'm, I think we have this capability for modeling the non-linear response thermals to try in multiphysics of buildings against fire.
[00:43:51] And that is great, but right now it still requires a lot of expertise and it's only done for some kind of iconic [00:44:00] structures when you have the time and money and, and expertise. Many people are working on other aspects of, uh, the built environment with digital tools, with numerical tools. So we have linear, real, uh, software for the response, but more broadly we have BIM and we have, the ability to track, the construction process and moving forward, we need to do more about circular economy and sustainability reusability.
[00:44:26] So probably there will be. Digital ecosystems to track really all the components of these buildings and how they go together. And so on. I think that in the future, we should be able to. Again, interface and couple and put together these different types of, numerical models. And basically I think we will move toward having digital twins for all, complex objects, Aircrafts buildings and so on.
[00:44:53] And maybe we'll have, in the metaverse or, or whatever. We have the same cities that we have. Here. but [00:45:00] with, this digital twin object that allows us to track all the resources, components and so on, but also to inigate the response against extreme hazards, fire, earthquakes, and so on. And as the condition are changing, climate is changing.
[00:45:14] We can change the stress source and we can change the demand and so on. But I. Structural fire modeling cannot stay apart from all those other digital efforts. And basically we should have a, a digital ecosystem where we put everything together for modeling buildings and ensuring they are sustainable and they are resilient.
[00:45:33] Wojciech Wegrzynski: Based on my experience with how BIM is incorporated actually the, the structural engineering part makes a lot of sense to, to couple that. So maybe it's not even, maybe it's not even decades. Maybe it's just years ahead of us to, to have a, at least this part coupled because. Coupling with, with CD modeling is probably a little harder because of other challenges.
[00:45:55] the main being, you build model for a certain [00:46:00] purpose architect drawing the building or engineer drawing, their structural model has a different purpose than CFD engineer. It's just that simple and that problematic.
[00:46:11] Thomas Gernay: Absolutely. It still requires a lot of human thinking to know how to model a structure for the purpose of structural for engineering, because you have to think about the heat transfer problem, then the structural response. So you, discretized your structuring in a certain way, and it's not just lines on, on a sheet of paper that you can extra, you have to have a kind of information and data that today we don't have in, in other.
[00:46:37] Pieces of this digital workflow, but as I look, you know, back decades and then I look forward several decades, I'm sure we'll find solutions for this kind of challenges, but indeed it's, it's not trivial because as you say you right now, each of these tools have very specific purpose. And so you have to have an engineer who think about what's the right modeling approach.[00:47:00]
[00:47:00] To target this purpose. And so to replace or to, to harmonize that, which requires some, lots of research.
[00:47:06] Wojciech Wegrzynski: And how you feel about emergence of AI as a support tool in the structural fire engineering.
[00:47:11] Thomas Gernay: I'm glad you bring that up. That that could definitely help. I think, we are still, Playing with the concept and trying to see where opportunities lie. But if we start, being systematic about the data and collecting and classifying the data, I don't see why AI could not help in this field the way it has helped in, in others.
[00:47:33] Right. So, uh,
[00:47:35] I think we are still relatively in, the early stage of the, the field of structural fire engineering. And we are still, looking at these philosophical questions that you were mentioning, right. That at the high level, we still don't have really the answers about oh, We do have the answers, but we have to think, hard about the scenarios and the performance objectives.
[00:47:54] In all case, it's not really codified yet, but as this become embraced more and more, and [00:48:00] as we move to a more systematic there, I think that, AI could also, uh, come in
[00:48:04] and help.
[00:48:05] Wojciech Wegrzynski: I hate doing this podcast. Like, you know, I ask people to talk about their fancy tools and the most impressive stuff they've done and it always ends up. Like we don't understand our objectives. Well, fire. Safety's horrible. We need to start our objectives guys. Come on. , I want more. Exciting stuff in the podcast just bragging about how we don't.
[00:48:25] But yeah, that is the sad reality.. Like that is truth that we spend so much money and effort on many, many fancy things and minuscule improvements in secondary
[00:48:41] Thomas Gernay: Where.
[00:48:42] Wojciech Wegrzynski: where. We don't have a fundamental question answered. Like, do we want our structures to resist burnout or not? And, and how to choose if we want it.
[00:48:54] we have not answered the question of what design fire and that what probability is, is, [00:49:00] for this building and how it would be for another building. It's Overwhelming, but I'm happy. We have lots of work. Uh, future of fire engineers is bright.
[00:49:09] Thomas Gernay: Well, these are complex questions, but to some extent, because there is such a multiplicity of, of buildings and projects and, uh, jurisdictions and, and everything. Right. I think what is key is to really move forward in providing the tools and the framework to. Post approach these questions in a systematic way.
[00:49:29] Once we make the decision, we want to go that way such as performance based, and then the tools to enable the assessments to be rigorous and rooted in the physics. And that's really what computational modeling does that. It gives you an understanding of the physics of the problem in terms of how the fire is going to develop and then how the structure is going , to respond, right, as opposed to taking a simplified model, which is.
[00:49:51] Useful for benchmarking, but which does not give you these insights and the sensitivity and so on, but it's true that then you move to the higher level and it becomes a [00:50:00] human decision problem. And so we cannot formalize it the same way as engineers, because. You need to speak with the stakeholders and it'll depend on the economics and on the social and so on.
[00:50:10] So, uh, it's, it's true that it can seem frustrating, but to some extent, I think it's, once you move the needle from the, the engineering problem to, to all of the technical social problems that, is just different
[00:50:23] different considerations.
[00:50:24] Wojciech Wegrzynski: I wonder if in 50 years we'll still be thinking Mindset of fire resistance and, and how, how we'll develop. I'm happy to view as this develops Thomas. that was a great talk to you. I'm very thankful for taking me into the world , of structural fire engineering and I'm, I'm sure it was interesting, probably more interesting to non-structural fire engineers, because we finally got an insight to how you guys do your.
[00:50:52] so, yeah, huge. Thanks. and, I hope to see you here again with another interesting subject
[00:50:59] Thomas Gernay: Well, thank you. It [00:51:00] was just really my, my pleasure. I had lots of fun discussing and I hope to see you soon. They may be at the
[00:51:04] summer school.
[00:51:05] Wojciech Wegrzynski: that I would love that. Thank you very much.
[00:51:08] And that's end of this episode, I thought about making a fancy out ultra as usual. But in fact, the end of this discussion has pretty much covered it all. We're talking about fancy tools, , AI, bam integration, and other stuff. While we don't have objectives and design scenarios fixed, and that's kind of devastating, really, I feel really sad about the fact that we still do not have this seemingly simple, but as Thomas said, quite complex stuff figured out. And if you're a researcher in the field and you want to find a place where you can make great impact, well, there it is this is a much needed place for research. And while we are developing that structural fire engineering is there to help us understand the behavior of the whole structures. We hope we've pretty well highlighted. The difference between the performance of [00:52:00] a member performance of a single element in your building and the performance of the whole building itself at The example of Cardington and a membrane action in the slabs , is a perfect example of that. How, how different it is. Like if we just focus on performance of single members, as we usually do, how do you understand the performance of the whole structure? How it will behave in the fire? Will it behave better, will behave worse.
[00:52:23] Yeah. Luckily the current experiences that they usually behave better. And Thomas mentioned that these actions that were found made the buildings more robust. So even some requirements could have been softened down on the secondary members. It's really interesting how understanding fire science can significantly improve the way, how we design structures.
[00:52:45] Some has also mentioned. New Euro codes coming up. And, for those who don't know, Euro codes is a series of standards in European union. They're free to access, publicly available and they have, Plenty of information [00:53:00] on structural engineering and structural fire engineering with all modern materials.
[00:53:05] I really found interesting that he mentioned, uh, Luxemburg annex to eurocode, which covers glass. And I must really jump into that because , it's super interesting topic to me at this point. I hope it was not too heavy for summer break, but, , I guess, , fire science , is just great.
[00:53:20] Even if it's even if it's hard, it's still great. I hope you've enjoyed this episode and I hope to see you and I hope to see you here again next Wednesday. Thank you. Cheers.