276 - How smoke control fails (and how to find a fix)?

Why Failure Matters In Design

Wojciech 0:00

Hello, everybody. Welcome to the Fire Science Show. My premise for this podcast was to bring interesting conversations that I'm very often holding with a lot of interesting people around me to a broader audience. It was always a pity that, interesting discussions happen in such a closed, uh, group of people, and here we are today. I've been having this conversation with many people in the last months, mostly discussing visibility in smoke and the progression that we're making on, uh, changing the visibility in smoke paradigm, which you will hear about awfully lot once we finalize all the works. but in the meantime, when discussing visibility, it's obviously a tenability criterion that we use to understand whether smoke control works or not. The focus is usually on whether it is maintained or not, and my proposal is that we should rather look on why it's not met when it's not met, you know. To instead of focusing on, on the criterion, focus on the failure as a mechanism. How, how does a system fail? And if we know how does it fail, what can we do to make it not fail, you know? When I was writing the visibility grant with, uh, with Professor Arnold, we opened the, the whole project with a quote from Petroski, uh, which goes by, "I believe that the concept of failure is central to understanding engineering, for engineering design has its first and foremost objective, the obviation of failure." And I found it extremely fitting because that's, that's exactly what we want to do with our systems to, to them not fail we want them to meet the criteria that we want them to, to achieve in practice. And at the same time, you know, failure is a necessary component of good iterative design. If you never failed, how do you know how successful you are and then how far away from that failure you are? It goes somewhere around the optimum design. So in this podcast episode, we're gonna discuss how smoke control systems fail and what can we do about it. Is it always the answer to increase the extraction capacity, huh? That's, that's a, that's a challenging thing, and I hope after this hour with me, we're gonna have a much clearer idea about it. Let's spin the intro and jump into the episode.

Smoke Control Systems And Definitions

squadcaster-e4ag_1_05-27-2026_144803 2:45

The Fire Science Show podcast is brought to you in collaboration with OFR Consultants, a multi-award-winning independent consultancy dedicated to addressing fire safety challenges. OFR is the UK's leading fire risk consultancy that this year celebrates its 10th anniversary. As experts in fire engineering, they are fully committed to delivering preeminent expertise to protect people, property, and the environment. With over 30 chartered engineers and a team of fire researchers at their core, they continually explore the challenges that fire creates for their clients and society so that the best research, experience, and diligence can be applied for effective tailored solutions. In 2026, OFR will grow its team once again and is keen to hear from industry professionals who want to collaborate on fire safety features this year. Get in touch at ofrconsultants.com. And now back to the episode

Wojciech 3:46

I, I think the failure is, is not always a bad thing when we consider it within the iterative design. You know, when, when people, uh, build muscle, they train to failure or like one, two reps to failure that you have to stop training to be most efficient. But, uh, you never know when that is if you've never failed. Like that's very difficult to tell when is the, the failure gonna happen. And the same kind of thing is in the engineering design when we're designing a system and we actually have time, space, and capacity to, to actually do proper iterative design. You have to reach the point where it fails to understand how much margin of safety you have in a building. I mean, of course it's-- I'm not calling that we always should optimize till the, uh, last nail, uh, but that, that's not the point. But understanding when and how the system fails gives beautiful array of tools, new tools that you gain to improve the, this performance. So in today's episode, we're gonna go into the concept of failure in smoke control systems. That's a very broad name that is perhaps likely understood differently in different parts of the world, in this different legislations, in different code So in, in my world, the smoke control systems would be a umbrella term that covers nearly everything that is used to somehow control the air and the smoke in case of a fire in a building. I would define them as, uh, smoke and heat extraction ventilation systems, SHEVs, which is my extraction tool. I will also have a variant of that called the N-SHEVs, which is the natural, natural smoke and heat extraction ventilation systems based on, uh, openings, windows, uh, smoke vents, et cetera. I would subdivide the systems by the kind of mode of operation or how they utilize buoyancy. I would have transversal ventilation systems in which I'm interested in having a smoke layer, a buoyant smoke layer from which I can extract the smoke. I close the smoke within a concept of a smoke reservoir underneath my ceiling, and there I place my extraction vents to extract it from that location. Or I could have, uh, longitudinal systems in which I would be pulling and pushing the smoke, uh, along a pathway, be it tunnel, be it a car park. I've also seen that done in corridors where you have extraction at one end, you have inlet in, in the other end, and there's only one way the smoke should go, uh, from inlet to outlet. And another set of definitions which I find very challenging is that you may be talking about smoke control systems, smoke clearance systems, uh, systems for evacuation, systems for firefighting. I think those definitions all go down to how do we consider the success of the system, what it should achieve. Should it provide clear escape paths? Should it just reduce the temperatures in the, in the compartments? Should it perhaps assist firefighters by continuously extracting heat and smoke from the building and just improving the conditions that they will, they will see? a lot of this-- th-those failures will come down back to the objectives that we set, but we will, we will come back to that, uh, later in the podcast episode. If somehow you're using different definitions than I am for, for those systems, I hope we can figure out some common ground on what I'm talking about that is understandable to you. It, it's k- kind of a mess with all those, uh, names and systems and And that in, for some cases, an FDA and ISO and CN would all use different names for probably the same thing. Anyway, uh, when we talk about failure as well, another important thing to be considered is, mm, a failure of a system in the real world and a failure of a system within an engineering

Real World Failures Versus Simulations

Wojciech 7:45

evaluation framework. And those would be completely two different things, uh, mainly because of how the fault tree would look like. Uh, what I mean by that is that when you are making an e- evaluation, engineering evaluation of the system performance, let's say you run a CFD simulation to understand the smoke flow inside of the building and how well your extraction systems are capable of, of removing that smoke from the building By running that simulations, you immediately assume that everything works. You know, you don't run a simulation for a smoke system that does not work. Well, perhaps you could do that, and that, that's probably a good idea to eventually run such a simulation in some cases. But in general, you are looking into system that has started, that works, that for a fire that has been detected, et cetera. In the real world, that's already a success if you reach that point, and it-- I, I would not say it's guaranteed. It takes a lot of work and effort in designing and maintaining the systems to make sure that the probability of that is close to one. It's never gonna be one, but, uh, we would like it to be as close to one as possible. There are multiple things that can go wrong on a building which can cause a failure of a system, which is some sort of wrong cooperation of the system. It could be caused by a human. Very simply, a lot of systems could fail if you block or open a door that is necessary for a system performance. If you have just one air inlet to your room, and that's one set of doors, and you kind of block them, system's never gonna work, and in your analysis, you would assume they work. Uh, it could be transfers, like there is perhaps some pathway from where you introduce the makeup air to your room, to, to your compartment of interest, and it has some transfer dampers along the way, and maybe they don't open, maybe they're blocked, maybe they're broken or something. That-that's gonna create a tremendous technical challenge for the extraction system, and it's gonna increase the velocities on every other source of air, which is a huge, uh, issue for the operations of the system. You may also have a-- we-- That happened to us many, many times. You may have some technical openings in the ductwork, for example, that are not closed. We had that in a road tunnel, um, a giant road tunnel, massive, massive extraction duct you could walk into. You could drive the car inside the duct. That-that's how big it was. And it was transferring hundreds of cubic meters per second. Enormous, enormous duct. And it had three entrance points to the duct along the length of the tunnel, which was many kilometers. And those entrance points were basically just, you know, uh, holes, manholes in the bottom of the, of the duct, which were obviously supposed to be closed. And one time, uh, after some inspection of the, the, the duct, someone has not closed that, uh, manhole, and we've lost like forty percent of capacity of the system. Can you imagine forty percent was lost because there was one more opening along the way which we have not accounted for? And, uh, no simulation can, can show you that. It's just life, and it takes, as I said, a lot of effort to make sure that everything works. Um, those wrongful operations could be also systemic issues. Like you could have delays in your s- that happened to us many times. You may have delays in the system, either programmed delays, which m- just introduce additional time, for example, to investigate the fire, or you could have delays that come from how the equipment works. If you have 1,000 signals to send out, perhaps the system chooses which signals go first. And if it does that, there's perhaps a delay time between every single signal it sends out, which means, uh, there could be significant delay that builds up, uh, to the systems that you're most interested in. That, that, that is a possibility. We've seen that happen. You could have programming errors. That als- also happens. If you have that, the system's not gonna You could also have some ordinary types of failure of a, of a system, which we of course try to account for on, on the design end, and we try to design around them to make sure that there's enough resilience, that there is enough backup options that they do not pose a hazard. But for example, a power failure, that's a catastrophical one for, for a smoke control system. You can have a device failure. If you have a, a fan that's designed for 400 degrees, you'd expect it to be able to, to extract 400 degrees, uh, smoke. But, you know, my experience with testing those fans is that while they're designed to achieve this performance, they sometimes by the just chance break inside of the test. Like, you, you-- that's very rare that the fan passes the test by-- with the first, uh, try. So even though, uh, you have a rated fan and you cannot get anything better for that, uh, for that location, for the temperatures, it's just the perfect, the best fit device you could have, it's still-- it's not 100% guaranteed that it's gonna work for the whole time. It's, it's just, you know, compliance is not yet, uh, real world performance, and we have to be aware of that. So all of those things can accumulate to cause a fundamental error or fault within the system, which creates a very, um, bad outcome in terms of the performance of the system, potentially leads to a very bad outcome. And this is the real world failure, the real world failure in which, it's very difficult to engineer around. I think what we're gonna go through in this episode, understanding the modes of failure in your building and and where the vulnerabilities are in your building help us create those backups, those, you know, resiliencies in the system that actually also prevent this catastrophical outcomes of, of, uh, random failures within the devices. Uh, but, uh, it, it-- generally, when you engineer, you focus on the performance. So if you engineer a system and it's supposed to achieve a specific performance and you're sure about it, a failure because that performance was insufficient in the building should, should not exist. And yeah, so just remember that in your engineering analysis, when you start your simulation, you've already had a tremendous amount of success with everything in the building. And now we focus on the performance of the system. So I have my, let's say, shopping mall. I have my source of the fire, my design fire. I place it in the location that I choose as, uh, Most representative for the performance that I want to measure. I have my tenability criteria that are agreed with my authority having jurisdiction. I'm looking for ten meters visibility. I'm looking for temperature of 60 degrees along my escape routes. I'm looking for radiation below two and a half kilowatt. Let's say

When CFD Says The System Failed

Wojciech 14:40

that those are my tenability criteria. In this case, I've designed a smoke extraction system. I start my simulation, and let's say after six minutes I see that in one point of my evacuation route, 1.8 meter above it, I have visibility of nine meters. Obviously my system has failed. So what goes into that failure? What does it even mean that the system has failed in this scenario? What are the elements that go into this system not meeting my requirements? I would divide it into three components. One is the scenario for which the system is chosen because the performance is largely scenario dependent. Two, the tenability criteria themselves, how they are defined. Of course, uh, by defining different tenability criteria, you achieve different, uh, failure, uh, or the system fails at different times. And three, there's the system itself. Can we design the system in a better way to actually provide the performance within that design fire and with those tenability Le-let's try to go through them one by one.

Design Fire Assumptions That Drive Results

Wojciech 15:57

So I would start with the design fires and, uh, the whole array of design assumptions that go into this, um, this scenario. So y-you take 100 fire safety engineers, you give them the same space to design for, you're gonna come up with 100 different design fires in terms of the size of the fire, in terms of the location of the fire, in terms of how the fire is, uh, defined by itself. A lot of things-- A-and that's a part of the engineering judgment really where fire safety engineering is not, uh, you know, uh, a science which tells you what's the most efficient way to place a fire. It's, it's a profession in which you understand that by placing a fire of this type in this location, you're representing such and such hazard. that's what an educated fire safety engineer should be, uh, able to tell. And, uh, there's also a freedom, a liberty in that profession where you design the fire and you, to an extent, take responsibility for that. Um, in the design fire definition, you have things that are kind of obvious, which is the, heat release rate of the fire, the size of the fire, and obviously that will first and foremost mostly influence the outcomes of the, the simulation. Of course, the larger the fire, the worse the conditions inside. But that's not the only variable or not the only thing you can fiddle with, uh, when you're designing, and you should be aware of that. Another one would be soot yield, uh, which is the rate at which the smoke is produced in your simulation. Soot is basically those aerosol particles that fly through the smoke, uh, carbon chains, which are black, dark. They give smoke the smokiness and, uh, basically the smoke that we extract is a fresh air that mixed with those soot particles. They're a very convenient way to trace where the smoke is because they're obviously visible. And, uh, soot yield is already a parameter that can change the outcomes by a factor of even 10. there's a range of soot yield values that you can find in SFP handbook, for example, where different fuels were tested in laboratory conditions, and they give you some values of soot yields. And then you may have real world experiments where this was, uh, measured for real world fires, and they can give you different values. I always found it very ambiguous. Years ago, in 2015-ish, we've written a paper with, uh, Gabriel Evigne on soot yield for engineering design, where we've, uh, put forward that you should use soot yield of 0.1. Not because it's magically, perfect for all kinds of fuels. It's because of, of usefulness. If, if you use soot yield of 0.1, you are not really making a huge error if in reality it's much larger. However, you can f- also find that value quite conservative, perhaps too conservative. And it's something that we're thinking about, whether a different value of, of soot yield should be more representative for real world fuels. I, I don't have an im-immediate answer for that. I'm still using soot yield of 0.1 in my engineering design. But at least I'm, I try to be, you know, conscious of the margin of safety and the kind of, uh, you know variety of scenarios that my choice of soot yield covers. And soot yield is not the only one. You also define heat of combustion. So heat of combustion also goes to the equation that tells the software how much smoke you've released. Basically, the software takes your, predefined heat release rate. It divides it by hea- heat of combustion to understand how much grams of fuel are you burning, and then applies soot yield over those grams of fuel to release the smoke. So obviously, the higher, the heat of combustion that you've defined, the less smoke you're gonna produce in your engineering simulation. So if you go with oxygen calorimetry bomb, uh, heats of combustion of, uh, 40-plus megajoules per kilogram, you probably are releasing less smoke than you would in a real-world fire. We, we found a great convergence for the vehicles, uh, from a large database on vehicle fires, which obviously are very mixed, uh, kind of fuel arrays, and that, that, uh, convergence was at around 25 megajoules per kilogram. So that's a nice, um, starting point for me. Finally, uh, the design fire is, is just the design f-- it's in the name. It's a design fire. It's a fire for the purpose of design. It's not a real fire. And, uh, real-world fire can be much different than design fires. O-o-one thing that's important is that the human environment is much more cluttered than we would assume in our simulations. Like, if you take a look around you, you likely have a ton of random combustible materials thrown all around in different places in the compartment. And if you were to model that compartment, in all honesty, you would probably model the desk, the chair, you know, the armchair, the bookshelf, and that's it, and you would not model the curtains or, you know, a small book or, or a piece of cloth hanging on the chair. Uh, real, real-world fires are messy, and they behave a little bit differently. Therefore, the design fire is never able to replicate reality, and that's not the point of it. Now, if I go back to my simulation of a shopping mall, which I have just failed, you know, there's also a possibility that perhaps my design fire was too large. Maybe I should have chosen a smaller soot yield. You know, th-those are things that could actually change the outcome of my simulation. I'm not saying this is the best way to do, and a lot of fire safety engineers, me included, would consider fiddling with the design fire unethical, if you, if you fail a simulation. But in all honesty, you, you have to consider whether the design fire is fit for the purpose. What is it supposed to represent? What kind of information does it give you in the end? What kind of hazard you are sure that your system is ready for? Not as in a particular type of a fire, but just the scale of the fires that you can meet. Does it-- Is it realistic for the, system that you're testing, or is it something, uh, very, very odd? It's different when you simulate a, a shop with plush toys and, and a jewelry sh-store, you know, completely two different cases with very different outcomes. And in standards, they would be all classified underneath a kind of a shopping unit, uh, type of design fire scenario. Something to consider, something, something very important, and, and that one that carries perhaps the biggest, uh, responsibility of fire safety engineer the second part of the consideration would be within the tenability criteria

Tenability Criteria And Visibility Limits

Wojciech 22:50

themselves. So what are we looking in? And it-- I, I have a big struggle with those tenability criterion because, you know, I'm very into modeling visibility in smoke and understanding visibility in smoke and trying to find new ways, new better ways to understand visibility in smoke. And what I've learned so far is that the model that we're using is not really, um, realistic in the regards of it being able to predict, uh Our ability to see things through smoke. The, the model outcome does not give you visibility, uh, as you would expect it to, to give because it's in the name. It, it's just a clever way of translating smoke concentration into a different value with a different unit that we have agreed on some thresholds on. That- that's it. Convenient, but not really the, the truth. And there's a lot of things that are wrong. Like, if you have a corridor and the corridor is 30 meters long and you're in the middle of it, and you have 10 meters visibility in that corridor or 11. Let's say you have 11, so you've met the functional goal of having more than 10. You cannot see neither of the ends of the corridor from the point you're in. So that, I would say, is tremendously challenging environment to evacuate. And I believe Professor Bryan research from the '80s after MGM fire has shown that this is one of the lead challenges in the evacuation that, uh, people are only willing to go into the smoke where they don't see the other end for a very limited distance or period of time, and often those people perish. So here I have a, a corridor. I'm in it. I'm legal. I've met my tenability goal, but am I safe? I wouldn't say so. I may also have, you know, a, a vestibule, one and a half by one and a half meter, you know, a tiny vestibule in which I have five meters visibility, and there's some smoke in it for some reason. Am I unsafe? I'm way below 10, right? But the room is, like, little bigger than me, so I can see everywhere in the room. So am I really in visibility hazard here or not? That's why I don't like the value of visibility and smoke, uh, being used as threshold because I think the purpose of this condition should be whether I can see what I need to see or, or, or not. Uh, same would go for temperatures. Temperatures are perhaps more straightforward, but I don't think we have a perfectly great, uh, you know, uh, threshold values. And again, it's gonna highly, highly depend on, uh, the moisture content, and it's gonna be extremely dependent on the irritants as well because there's gonna be some combined effects. Um, toxicity would go to the, the same point. Like, th- there's a lot of things that I'm not sure about the tenability criteria. And again, here engineer also has a lot of, uh, freedom to choose, especially in terms of visibility and smoke model. Uh, I'm not sure how your law is structured, but mine only mentions that there should be sufficient visibility on the evacuation route. So I define what me- what it means sufficient. In some legislations, it would say 10 meters, but, uh, does it say For what k value? Does it say at what height you evaluate that? Does it define the soot yield and the design fire which, to which this is tied into? There's a lot of ambiguity in defining this, uh, tenability. And, uh, I, I think th- that's why obsessing, you know, just on the value, just on the number that you get and having this binary type of assessment pass-fail, um, ASID is higher than RSID or is not bec- measuring it up to a second, you know. I, I don't think that's the best engineering. It's the engineering that, that looks good on a report because you can make a table and show to a layman that yes, all the criteria were met, and you're likely expected that your design meets those criteria. in reality, when you're dealing with a design that is somewhere close to the point of failure, which means that you're having an optimal design at some point for, for, for the design fire that you choose, those tenability criteria can be extremely limiting in our ability to engineer. I find it much more, much more valuable I mean, of course, I don't ignore them. I'm not allowed to ignore them. I'm an engineer. I have to, in the end, provide a system in which the tenability criteria are met, and I have to be convinced that they are met, and I have to be convinced that they are correctly chosen for the particular occupancy for the considered design fire for the, for the design that I'm doing. I have to be absolutely convinced that I'm, I'm actually obtaining them. It's not that I don't want to obtain them. I just have to understand very well how does the choice of the tenability criteria tie into the whole strategy of the, of the building and the system that I am designing. This is as much part of the design as the design fire and the design of the system itself So now let's talk about

Two Core Smoke Failure Mechanisms

Wojciech 27:51

the system. How does the system fail and what we can do about it? And here it-- I think there's, one, my experience with smoke control systems is that there's always a way out. It's never, you know, uh, impossible to solve. I've, I've done hundreds of s- buildings. In many of those, the smoke control systems were failing, and we've been facing extremely challenging circumstances, both from the architecture, from the availability of systems, of power. There were very, very few buildings in which, like, you could not do anything, and those were usually buildings in which you just had, like, fundamentally insufficient capacity for power, for example, to create a system. I'm sorry, I'm not able to create a mechanical system with no electricity. That's it. Um, but when you had some availability of, of, of tools, those tools can be put into very good use and figure out. So we have our shopping mall. We have, uh, smoke at one point eight meter above my evacuation route. It's nine meters. That's, that's what the measurement point tells me. The smoke can get there. That, that's the important part. That, that's the-- that's already the troubleshooting. How did the smoke get there? Because I think there are two fundamental mechanisms in which tenability can be broken in such a scenario. One is that the smoke is captured in a smoke layer. It's stratified because of the buoyancy, so it floats above my clean air at some height. And as it accumulates in the smoke layer, the smoke layer kind of descends because there's more and more smoke, uh, in the smoke reservoir. And eventually, the bottom of that smoke layer may become below the point at which I'm measuring, and the smoke layer contains heavy, dense smoke, which obviously means that, the visibility is gonna be broken or temperature's gonna be broken at the, the height at which I assess because n- I suddenly not assessing the bottom layer where there's no smoke. I assess in the smoke layer. That's one way of failing. And another way of failing could be that the, there for some reason the smoke is not buoyant. It's not gathering in a smoke reservoir what I would-- where I would like it to be. It starts to fill out the whole enclosure and the concentration just raises and raises and raises and eventually passes this threshold criterion where I, I have smoke everywhere. And th-tho-those are two completely different outcomes, uh, both in terms of the severity and the, the, the challenges with fixing them and also the mechanism that led to them. If it's a smoke layer kind of failure mode, uh, one, it's not immediately a loss of life kind of situation because if I still have the smoke layer maintained at like 1.7 meter height, there, there's still a possibility of the escape. It's perhaps not at that point, not something that I would be comfortable with and it doesn't meet my tenability criteria. But in reality, people could just hunk down and, and, and escape. It's not a life-threatening situation immediately. It's different when the smoke accumulates because in that case, uh, you may have no visibility at any height and lose the orientation which is the mechanisms for which the visibility is dangerous if you lose it. if the smoke layer descends, this either means that my smoke reservoir is not big enough, perhaps my ceiling is very low and I just don't have enough space to, to maintain smoke there. Maybe my extraction

Fixing Smoke Layer Descent Problems

Wojciech 31:33

capacity is insufficient. th-those would be the things that I would be look into. Of course, uh, the first best-- the, the first simplest thing is to increase the extraction capacity which hopefully will allow you remove more smoke as it accumulates in the smoke reservoir which will delay the descent of the, of the smoke layer which will prevent this type of failure. That's one way, but often very difficult to execute. You have more ways. First, you can look How many extraction points you have and how are they distributed. If you- your shop is two thousand square meters and you have just two extraction points, you're not gonna have a great time. You need to distribute them quite evenly in the space and, uh, we've generally found that the more of, uh, smaller vents is better than just a few large vents. It's not universally true, but in most cases it is. You may want to look into plug-holing phenomena. We're typically installing the vents on the bottom of the ducts and, uh, this can create conditions in which plug-holing phenomenon occurs, which means that the vent is sucking clear air from underneath the smoke layer and does not remove the smoke. And of course, this is something that you don't want to have in your building. We would recommend placing the vents on the sides of the ducts and in some cases even on top of the duct, uh, in, in some spaces. But usually the side orientation of the vents already improves the, the, the situation in the compartment, gives better, better performance than vents installed on the bottom. Also because you extract the smoke higher, um, than compared to the same duct but the du- the vent on the bottom of it. Um, you may want to use different type of s- o- of systems. Maybe for, especially for shops, a very useful system is where you just have a plenum above a sealed ceiling and you extract at the circumference of the room through a, a gap between the, the plenum and the, the compartment. We found that method of vent-- of extracting smoke from shops as extremely efficient, much more efficient than the traditional kinds of systems. you may have too much clutter on the, the ceiling and perhaps it's, there's nothing you can do. That- that's also an outcome and maybe actually removing some of the ducts from the venting, changing the, the, changing the space in such a way that you mindfully create those smoke reservoirs is a way to go. You have tools, not, not only increasing the extraction capacity The other mode of failure where you would have, uh, the smoke accumulating in a room and it's not buoyant anymore, I

Fixing Mixing And Lost Buoyancy

Wojciech 34:10

find that much more, um, challenging to solve. Uh, but to challenge it, you have to figure out how did it fail, and what, what's the reason why the smoke is not buoyant? Because it should be. If you have a design fire of m- many megawatts, the smoke should be buoyant. And usually, the reason for that is because of some mixing. So you either have an air inlet high in the smoke layer, which introduces cold air into the smoke layer, which creates conditions in which the, uh, the smoke would lose its buoyancy through mixing with the cold, um, cold air. You may have inlets on the correct height, like on the bottom of your room, but with high velocities. High velocity here is a challenging concept. Uh, I would say one meter per second is the s- the fastest that you could be sure that it's gonna be good. Um, two meters per second is already, uh, quite brave, and anything above two meters per second is gonna mix your smoke and, and air unfortunately. So y- you have to be aware that if you inject the air with high velocities, you, you have to do it very, very far away from the location where you extract from. If you have a very large shop subdivided into multiple smoke reservoirs, injecting air into different smoke reservoirs than the one that you're extracting from is the, the way to go. If you have to inject air to the smoke reservoir where you are extracting to the compartment from which you are extracting, you have to be mindful about the velocities. And you, you also have to... If you utilize something like, uh, doors or gates, make sure that they open. If, for example, you use five doors but only two of them open, you're gonna have much higher velocity on those two doors, which can ruin the system. So you have to be sure that those, uh, utilities that you use to inject the air into the room will actually be working in a fire because if they're not, then something else will have to compensate and this compensation will usually lead to an elevated, uh, velocities, which are a cause of mixing and losing buoyancy by the smoke. You also have to be mindful where the g- air goes because, uh, like it's not gonna stop when it hits the wall. It's gonna, you know, uh, bounce off the wall. It can flow up. It can flow left, right. If you have big obstacles on the way that they can actually redirect the stream of air from the floor to the ceiling, and, uh, this can create tremendous, uh, effects that, that lead to loss of buoyancy and destroy the, the, the performance of the system. These are the things that you have to look into when diagnosing this, uh, this kind of failure and when you understand where the air mixes. You have to move back in time, see where it starts, where it grows. When you, when you investigate that, you then have a possibility to fix it. That's, that's the beautiful part, uh, troubleshooting to, to, to find a fix. If I was at

Troubleshooting With Flow Paths And Data

Wojciech 37:12

the project and I would have to do troubleshooting of it Perhaps the first step would be that I would just look at the flow paths. I would like to understand where the air is coming into my building and when it-- where is it, uh, going out of the building, and what happens in between. What's the exact path that the air flows from the inlet to the outlet to understand what's happening along the way? That would be like number one troubleshooting analysis. And actually, if you have, uh, systems that are based on longitudinal concept, this is als-also fundamental to understand. in longitudinal systems, the failure would be not the loss of buoyancy, but perhaps, uh, back layering of the smoke or smoke moving upstream, uh, smoke not respecting the virtual boundary that you define for it and, uh, moving against the, the direction that you've designed. That, that's the failure in longitudinal system. And understanding the flow path in here is fundamentally important. If you're working with pressurization systems and for some reason they don't work, that's the thing you look into. Like does the air has a way to move from the space with your expected higher pressure to the space with the expected lower pressure? I-Is-- does the flow path exist, you know, from inlet to outlet? Because if, if it does not exist, then you're not gonna have a pressure difference, and the PDS system is not gonna work. So flow path would be my number one thing I would like to understand when troubleshooting a system that has performance that is not matching my expectations. Another one would be looking into the values of velocities and pressures that I have in my building. Important thing, if you have natural openings, uh, pressure and velocities are very much related to each other because the difference of static pressure turns into dynamic pressure, which is velocity in your opening. Therefore, you have to understand that those are kind of the same thing, just measured in a different way. But anyway, you need to, you need to be able to investigate what kind of velocities you have and what were the expected ones. Is it too fast? Too slow? Uh, what's the capacity at my points? Looking into those numeral values un-unravels, you know, a lot of, uh, issues along the way. And that's if I troubleshoot the system on a real building, which for some reason doesn't work, that's probably also a first thing I would look at, like, or the second thing I would do that-- do at the building, measure the capacities at the extraction points. Do they meet the required ones? Because in real building, uh, the vents have tendency to not be, you know, exactly equal in their ability to extract air. It takes a lot of work to actually, you know, make sure that they extract at the same rate. The next thing I would be looking at, obstacles. Uh, what's blocking my air? What's deflecting my air? What, what's along the way? Perhaps on the building, there are things that were not included in the simulation. Fair. We don't know everything when we simulate, but perhaps there are new things that we have to understand and, and assess the impact of. We also would look at the same time at vulnerabilities. Is there like a single opening that if I block it, the whole system will stop working? If yes, then I probably should propose something to remove that vulnerability. I don't want to have a single point failures in my systems because that's, that's dangerous. So those are the third thing I would look into. Finally, if it's a part of ACID asset

Timing Strategies That Avoid More Fans

Wojciech 40:30

analysis and I see there, there's obviously a time component to that. And, uh, not always more means better. Sometimes delaying activation of some things. If, if it's the makeup air that's breaking your system operation and if it's, few minutes that you're struggling to get from the system, perhaps placing a ramp function so that system capacity grows with the time and accelerates from zero to full power in like five minutes, maybe that's a solution. Perhaps you don't need the whole capacity in the earliest phase of the fire, um, and you don't want to have any mixing by chance. And, uh, reducing this capacity in the early stage of the fire and then growing it up is perhaps better than starting the system with the full capacity straight on. That's something that happened to us many times when designing car park systems. Um, perhaps you want to start a specific point before other points. We also had that multiple times, especially in tunnels where you have stations in between them. The site of the station at which you enable your longitudinal systems matters which one starts first. So we have to account for, you know, temporal shift in, uh, in starting some of the systems so that the final sequence of operations is the one that, that, you know, allows us for the best outcome in the considered building. Uh, you can play a lot with those time strategies and introducing delays or reducing delays in specific points of time, having overlap with the systems, starting them in a specific, sequence of operations, utilizing ramp functions. Th-those are all tools that really, really help, uh, in the building. And, and it is actually amazing that many times we were on a building where The system would just fail. It was obvious it didn't achieve the performance, and playing with those time components or, or fine-tuning the sequence of operations actually solved the problem. People were quite amazed that, wow, you could do it without installing more fans. That's, that's amazing. And, uh and yeah, that, that's, that's what we try to achieve.

Key Takeaways And Listener Prompt

Wojciech 42:43

So, um, to summarize, if you run your CFD simulation, the moment you start your simulation, you've already had a lot of success with your system because it did not fail in anything that would prevent it from operating. You should be aware of that because in the real building, it, it could be a little bit different. And then when you have a challenging, uh, simulations where the system fails, there's multiple potential reasons for that and multiple things to investigate. From the assumptions, the design file, the tenability criteria, understanding how fit they are for the project, what is the sensitivity of your simulation to those assumptions, through system performance, understanding what exactly happens, in what order of events does it happen, where does it happen, how quickly does this particular tenability criteria crosses their threshold value, how narrow it is. Is it in the whole space, in the whole volume? Is it in a layer? understanding and looking into those things, understanding the flow path into your building, understanding the velocities, the pressure gradients in your building, understanding the architectural ob-obstacles, and, uh, looking into potential vulnerabilities. This is the pathway from something that does not work in my CFD to a fixed system that, uh, in the end is something that operates at the level at which I am comfortable with and at which I am happy with. I hope this was useful in some way, uh, with you. I, I always found it kind of amazing how much you can do with a failed system. People would think that the system has failed. "Oh no, we have to design a new one," or, "Oh no, we have to now introduce twice more fans to, to get it working." But in reality, there's usually a lot of things you can do with a system When you approach it with this mentality of trying to find what has failed, how it has failed, and then applying the specific fix for the exact failure mechanism that you've identified. I found it very useful in my engineering practice, and I wonder how you handle problems like that in your engineering practice. Let me know if you can. I hope this was an interesting food for thought, and I hope you've enjoyed this podcast episode. I'm today at, uh, SFP in Venice. Uh, tomorrow we start the conference on heritage buildings. You know, heritage buildings are really difficult when we design smoke control system and have to retrofit smoke control systems into a very rigid, uh, you know, building with a lot of, uh, fundamental limitations, uh, related to that. That's one of the reasons I recorded this episode. You know, when I was preparing a presentation to here, I, I had a lot of failed systems in front of me that I had to fix. So the pathway is not just, you know, four failed systems, it's four failed systems in historical buildings as well. I hope this, uh, seminar was gonna be-- will be good. If, if you're here in Venice, uh, make sure to, to have a chat with me. Let's meet at the coffee breaks or somewhere. And, uh, if not, well then I hope you at least enjoyed the podcast episodes. And if you do that, I'm sure that next week you're gonna enjoy again because, uh, next Wednesday more fire science is coming your way. Thanks for being here with me. Cheers. Bye.