071 - Risk as a tool for thinking with Ruben van Coile

[00:00:00] Wojciech Wegrzynski: Hello, everybody. Welcome to the fire science show. Today, we are going to talk about risk. Once again, in the podcast risk was a topic that we have touched a little bit. I mean, episode 23, I've hosted Jaime Cadena Gomez and David Lange who were talking about. A maximum. Allowable damage approach to risk. And how can that be used in the design of the buildings? Was the. 

[00:00:23] Baseline level of risk in buildings. How using risk as a framework and help you build a better building and understand your building much better. It was a. Very nice episode and I enjoyed it a lot. And then in episode 45, I had Ruben van Coile and Danny Hopkin talking about the foundations of fire safety and how risk can be such a concept that is fundamental to fire safety. And I truly believe this fundamental fire safety. 

[00:00:50] So I thought. It could be nice to dig more into that actually to make it a little more approachable because we've been talking with this experts.[00:01:00] On the risk. Uh, about their sophisticated methodologies or use of the risk at really, really highest level to unrave l, the fundamental fire safety of our buildings, but so far we haven't really covered risk. on its own in the show. So I really wanted to fill that gap. Because I know it is an important concept. It is a fundamental concepts to all of us. But it's not very approachable. One it's, it's a little difficult. It's not very intuitive. So I would really love to talk about the risk as a concept on 

[00:01:35] And for that. I've reached out to Ruben van Coile once again and he has, uh, agreed to star the episode again. Uh, a nice, welcome to Dr Ruben van Coile form Ghent University to the show. And that's exactly what we will be talking about today. Risk as a concept risk as method to understand the. Fire safety of your [00:02:00] building risk as a. 

[00:02:01] Fundamental benchmark off the safety in your building As a tool. To take engineering decisions, a very nice engineering tool. I hope that if you have never used risk as a concept, well, actually you most likely did. If you're a fire engineer, I'm a hundred percent sure each of us has. At some point. 

[00:02:20] Use risk as a concept. Uh, to develop via safety strategy for a building. If you know whether or not you So now, I guess you can use it more as a tool as is, it should be used. And I hope this episode It's something that helps you work with this concept a little more. And I hope you will be more open to using risk as a, as a tool in first, if the engineering. 

[00:02:44] So a very useful episode, it's small to guide on how to risk. It's not a course on how to use risk. But it covers a lot about what risk is and how can be used. As an engineering tool. So yeah, lets spin the the intro and jumping into the [00:03:00] episodes.

[00:03:00] Everybody. Welcome to the Fire Science Show. I'm joined today here by Dr. Ruben van Coile from Ghent University. Hey Ruben, great to have you back in the podcast.

[00:03:31] Ruben van Coile: Hi Wojciech. Thank you very much for the invitation.

[00:03:34] Wojciech Wegrzynski: Oh man, I'm, I'm so happy you took this, uh, invitation and, took this subject and I'm stressed because it's, hard stuff to talk about, uh, simple things in fire. I've, asked you to discuss some concepts about risk in fire and, uh, risk engineering in, in general because I believe for risk engineering is, Fundamentally important for the future of fire science and fire engineering.

[00:03:58] And yet I find [00:04:00] it is difficult to approach. And, I know a lot of people, it is not intuitive. They don't really understand what is being said when you talk in risk to them. I think we may have, communication problem in here. Like how do we engineers communicate? Aspects of, risk and how people perceive them.

[00:04:23] And, I would love to talk to you about some fundamental concepts, in this. So, I'll just drop a bump on you, like, why not, like Ruben, What's, what's risk ? uh,

[00:04:35] Ruben van Coile: Okay, Well, I wouldn't say definition as in be careful when talking. I mean, not going to make an edia or something now, but my definition about risk, very much engineering based, I would say. So, probabilities, consequences, and say a combination of probability and consequences. And I like to actually leave it there cause many people go a step further.

[00:04:59] [00:05:00] And then immediately say about I and multiplication. or, and then if there, different, we're already talking about eventually, for example, different scenarios each with their probability and their consequence, and they say, Oh, you're going to multiply the probability and the consequence, and then you take the sum of all of, of that.

[00:05:18] But, but you are losing information by doing that. So to. Question, what is is risk? Well, a combination of probability of and consequences. So what you actually, if you understand risk, then you understand how in fire, situation your building is going to perform in fire. Both probabilities, likelihoods of different performances and consequences, as in what is done, what does it mean for, people's lives, property, and.

[00:05:46] Wojciech Wegrzynski: I like, I like how you, how you put it, that that's, I knew it is a good idea to invite you in in some, school of engineering that I thought went to. We also talked about the perception of risk as [00:06:00] one of the. Elements of risk itself. I, I guess it, it's maybe more towards the risk acceptance that, uh, risk perceived as unacceptable.

[00:06:10] Even if the mathematical formula would give you, uh, something that that's okay. it could be perceived as, as not acceptable. So, so we were taught that risk perception is inherent aspect of fire risk on its own. Now, I'm, I'm looking, I have my eyes on. ISO 16 7 32. , Fire safety engineering, fire risk assessment, standard with the vocabulary there.

[00:06:36] I hate, uh, encyclopedia, but actually they, they, for fire risk, they say combination of probability of a fire end and quantified measure of its consequence. And, that's, that's so close to your, definition.

[00:06:46] Ruben van Coile: So that's really, again, for me, that's really because we we're talking about engineering, right? So this ISO standard as well, it's about engineering. So risk perception. very interesting. Probably. Say it's, it's very important. [00:07:00] But from an engineering perspective, I would say let's be careful and not immediately jump into that.

[00:07:04] I mean, people like to refer to all these aspects of risk perception. They also like to refer to, concepts of behavior, economics and so on, about how people value, very low, probability events or also how they value consequences. To me, this is more like trying to explain how people. Behave instead of providing an engineering tool, which then can be used to guide decisions as an input for the decision.

[00:07:34] And if then in the end you say, Well, despite the input we gave quite, let's say, objective as an objective as possible, right? An engineering input, despite all of this other human factors come in. I mean human factors in the sense of subjective factors. Okay? That's all we couldn't find, and maybe we need risk perception to sometimes explain stuff. As input, I'll prefer to stay away from

[00:07:55] Wojciech Wegrzynski: Yeah, that's, very nice. Do you think this. Approach to risk as an [00:08:00] engineering tool must be kept like that. So it it is useful in engineering, without this, um, how to say it, even without how people perceive it. I, I mean, it to be objective, it is to be objective.

[00:08:12] You, you cannot put, uh, another metric on top of that because, uh, I mean, it says this, but I don't like it anyway. Right.

[00:08:18] Ruben van Coile: Well that, that last point is also very interesting there. So when you say like, uh, should you keep it simple or basic in engineering sense? I would say yes, definitely. And it, it relates to what you said at the end. You, you don't want you, but you want actually a transparency. So to me, I'm probably, I'm very biased, but to me in the end, okay, probabilities and consequences, I can understand that.

[00:08:46] But I can understand that at least I, you tell me that there is a likelihood of certain events can be fire related, but can just as well be the weather and there are different consequence magnitudes and Okay, I can grasp that and I can make my own [00:09:00] decision if necessary, but as soon as you already as an engineer or somebody who quantifies things, starts to manipulate, This may be a negative term, but, Okay.

[00:09:09] Do conversions or calculations with these numbers. Today I'm going to take into account a factor for risk aversion or for the consequences too high, I'm going to multiply it with or or take a power. You're making things much less transparent. So I very much like the transparency of just probabilities and consequences.

[00:09:29] We, and then we can, We know what we can talk about. Everybody can agree on the values, we can disagree on how they were calculated. Maybe is that something else? But we don't have to talk about factors which hide.

[00:09:40] Wojciech Wegrzynski: Do, do, Do you think we should, uh, in that case, Communicate risk in this dry way as well. Like, this is a number, it's a measure. Nothing else. It doesn't have, feelings, it doesn't have subjective, understanding. I mean, like a temperature ignition. Temperature is this, if temperature is lower, does not ignite.

[00:09:59] If it, it's [00:10:00] higher, there's a good chance he designate. And in the same way, the risk is this, so it is below. Limit whatever limit we've said. this, this decision is, should be formulated based on this value, and other decisions should be formulated on, on, on that value. Because it is, it very dry. It's dry.

[00:10:17] You know, it's, it's, but maybe it should be, maybe if it's a engineering tool, if it's a measure, I guess it should.

[00:10:25] Ruben van Coile: if you, So if the question is, Do I think it should be communicated as objective as possible? And the answer is yes. definitely the old questions about putting in values goes beyond what you should do when you're evaluating a design anyway. mean, in this, in the stage of evaluating the risk of design, How it and should be communicated.

[00:10:47] That's something else. I mean, there, the contents then give the best knowledge. You have the likelihoods, the consequences that draw this picture of performance of the building because that, [00:11:00] that's what it is, right? People say, Oh, risk is something different. No risk is the performance of the building.

[00:11:05] That's what it is. You. Tell me anything else then, Okay, I have a performance for design, fire and so on. Okay. So that you have one little point in your entire space, which defines probabilities and consequences, and you have mapped one or two or 20. So, and you, you are making yourself a picture of the performance of the building and the total performance of the building, the full picture.

[00:11:27] This is the risk profile. Risk performance of the building. So should I be communicated? Really? Yes. How do you do that? Well,

[00:11:35] because there are many dimensions, it become okay, that's something

[00:11:39] Wojciech Wegrzynski: yeah. Yeah. And then, so you, you already started go going in, into the communication space in, drawing the, the graphs or, putting, points somewhere in, in a space, Like every, scenario has its individual like. Probability, individual consequences, I guess.

[00:11:55] Uh, you map them in in some way, whether you, you multiply, whether you do [00:12:00] some other mathematical operation to calculate that. It's, the method, I guess mostly we, we multiply them, but I guess others, other ways could, could be formed then. Um, You end up with a scatter of the points on your metrics, like if you had one axis with probability, the other axis with consequences.

[00:12:20] That's the usual way how we would, present risk in a two dimensional manner. I guess now that difficult choice is how to, where to draw the lines, like which point on this matrix is, okay. Which point on this matrix is, is not okay. and, and I, I know from the talk we, we did with, with Danny Hopkin we were talking a lot about this de minimis uh, level.

[00:12:45] We were talking a lot about ALARP region In the, uh, graph we were talking a lot about, um, unto tolerable risk. And, and I, I guess it's very connected to, this metrics and how we can use that presentation to [00:13:00] understand the, the further choices. So let, let's try to define, this regions on the metrics.

[00:13:05] Maybe let's start with de minimis region, which I assume would be the lowest risk, right? Or.

[00:13:11] Ruben van Coile: Okay.

[00:13:12] Well, that's very fine. We have to maybe, map that or introduce it a little bit because you are, in your questionnaire you said, Oh, we this multiplication. But actually when you are referring to two visualization, like it ended, that's probability and consequences. There is no multiplication involved yet.

[00:13:30] Right? There would only be this multiplication probably of consequences when people want to end up with a single.

[00:13:36] Wojciech Wegrzynski: Yeah. Yeah, you're right. Absolutely.

[00:13:37] Ruben van Coile: So at that point you're just putting your, and we are talking about scenarios because this is the easy way to talk and think. But okay, of course every scenario can be subdivided in further scenario and so on.

[00:13:49] So actually they are just, it's a huge number of possible situations, right? Infinite. So that's one. You have your scenarios, and you can map them probabilities, [00:14:00] consequences. first thing to stay say there is that consequences. Dimensions can be very different. So you have consequence on people's life.

[00:14:10] Lost consequence in dollar value, lost consequence in environmental damage. So this is very much multidimensional. What we generally do, if we do it at all, it's keep it 2d. So we have one vertical access probability, horizontal. Consequence. And then the, what you are referring to, if you're talking about life safety, for example, an FN diagram.

[00:14:32] So frequency, vertical access, the number of lives lost or at risk on the horizontal access. important thing to say, and, maybe I will have to, uh, refer my students to, uh, talk because every year on the exam there are mistakes on this. Is that on the vertical access to this probabilities as a probability of exceed.

[00:14:54] So the curves, they only go down. If you go on the left hand side, the vertical access, you [00:15:00] start there at the, at the point, the curve only goes down cause it's probability of exceeds it's probability of having one fatality or more. And then the next step, probability of two fatalities or more, which necessarily is lowered than having one fatality or more.

[00:15:13] Right? So, okay, we have this graph, maybe a bit difficult for listeners to

[00:15:17] Wojciech Wegrzynski: No, no, it's a challenging, uh, it's a challenging task to do it in the voice show, but also gives you a completely different dimension, understanding it. So I, I'm very happy to do that. Yeah.

[00:15:28] Ruben van Coile: Yeah, well, like this, the thing I just highlighted, this probability of exceedence that I always, I'd say it guess well when we have a lecture on this, but somehow it gets lost. Maybe when it's not visual it will get it's, it'll get lost less either way. So on in this space, we. A curve, a set of lines, again, can discuss how do then do that, but okay, set of line with probability of exceedance consequence.

[00:15:57] And this then defines the risk profile building, [00:16:00] which then can be compared indeed against difference at curves. And one of those curves is what I like to call a de minimis curve. Not everybody agrees on terminology. Uh, not everybody agrees, even if they use the di minimus. Wording, they might mean something else.

[00:16:17] So this is all very tricky. What's very important though, is that you realize what actually means, and I am very clear about this. It is a line which indicates, at which point it doesn't make sense to invest in investigating. So it's, it is like saying, uh, your risk profile is here. It is very low, basically, and this is an economic justification.

[00:16:44] basically any effort you now put in trying to find design changes or safety measures. So anything which would reduce this risk will cost you your time and effort so much relative. The way the risk is [00:17:00] that we say upfront, You know what, This is fine. You don't need to do this anymore. But importantly, if you would know that a certain measure should be done.

[00:17:10] It's cost efficient to be complete, but if you, if you know a measure should be done, then you still have to do it. I mean, it's not because, and this is one point where there is a bit of disagreement sometimes, people say, Oh no, you want, you're there, you're fine. You don't have to do anything. No, it's not the definition that you don't have to do anything.

[00:17:26] You don't have to investigate because your investigation also costs time and

[00:17:29] Wojciech Wegrzynski: so it's a, group of probabilities and consequences, which in essence give you such a low value of risk that it's pointless to even investigate them because, uh, most likely if you investigate it, you would come up with it. It's safe. You don't have to do anything. it would be such a low, uh, low, Can I say low risk?

[00:17:48] Ruben van Coile: Yes. So we talked about immediately about the lowest curve, minimum higher boundary limit will be tolerability limit, we're probably gonna talk about later and maybe a bit more detail. But if the design is not [00:18:00] tolerable, big problem, we have to something. Now, the fact that you are talking about the minimum means that you have already below this.

[00:18:06] So that, and when you're below the tolerability limit, you're saying, Ah, I can live with this design and since it's a risk profile, if this has to be, if doing anything to improve it costs too much is not reasonable, then we can stick with the current design. But if there is something we know we can do, we have to do it.

[00:18:28] And the above the limits you have to investigate. Below the minimum limit. You don't have to investigate because it's so low that you indeed already say like, Look,

[00:18:37] Wojciech Wegrzynski: This is uh

[00:18:38] Ruben van Coile: probably. Yeah. And you can then define a fourth boundary. But I, I don't see the def the point really a lower one where you say, Look, below this even lower level, it's really negligible, but why would something really be negligible?

[00:18:52] Wojciech Wegrzynski: maybe. Maybe you have to increase the risk . So, Yeah.

[00:18:56] Ruben van Coile: Well, yeah. And then, because then you have papers who, who then, [00:19:00] um, or people, people discussing like, ah, you talking about risk to life? Risk to life is never negligible. Yeah. I mean, Oh fine. And I don't think we need to have that discussion. For me, it's a discussion without really. Because if it's, if you say it's not negligible, but if you still say it's economically, you shouldn't investigate anyway.

[00:19:18] The conclusion 

[00:19:19] Wojciech Wegrzynski: what you said about we shouldn't, um, we should invest in safety. Like if I'm sitting in one building and the fire is another building, I, there is no point for them to invest in my safety anymore because I, I'm just not affected by the fire. I, I even sometimes take it further. if I investigate, I dunno, a shopping mall and, let's say I, I have designed a perfect smoke control system that maintains my smoke layer well above the occupants heads I achieve.

[00:19:47] Steady, steady state, solution within my investigation. So with the time, it does not change. I have, I maintain this, this balance. There is never smoke on my evacuation route. It's, designed for a [00:20:00] certain, fairly high size of, of the, of the fire in the building. Then there is no point to.

[00:20:06] Improve this system anymore. Like I've, I've reached all my perfect goals of, of a smoke control system in that building, and if I'd now doubled the cubic capacity of the system, the exhaust. It would not mean the people in the building are more safe. Like you cannot double escape the fire. I mean, they are already in a very unfortunate even having to be evacuated from the building with all the risks or all the challenges related to being evacuated and everything that can happen.

[00:20:36] While that, but my further investments do not, Anything towards the, the threat they're escaping from, which is the smoke. So, and actually from my professional experience, I know that if I increase this, I would most likely worsen the situation because of the complexities in the, in the air supply that would follow increasing the excess rate.

[00:20:57] But in, in the end, like I've ended up with a [00:21:00] system, there is absolutely no point in Improving that unless we start talking about the design fire, which has a probability of fires. Maybe I've designed it for a very probable fire, but maybe I should have designed it for very, very, very improbable fire, but still something that has a probability to it.

[00:21:17] But that, that, that's another, Let's, let's come back to that in a second. maybe we, we should talk about tolerability now. if, if we reach this point.

[00:21:25] Ruben van Coile: Yes. making a think distinction, what you just said. So with the information you just gave possibly, you're, you're saying the risk profile is the mini and you don't need to investigate anything. Possibly you're in an alarm region, In which case you're saying, um, that, Well, I already know that it will not be cost effective to.

[00:21:47] Wojciech Wegrzynski: mm-hmm. . Okay.

[00:21:48] Ruben van Coile: Uh, so, and just to make sure we, we are not on a sidetrack for some listeners, when I say ALARP we return to that after talking about deliverability, but I'm not referring to a [00:22:00] specific legislation. I mean, I, I, I like to think in concepts. So that's also what I hope we get from today's talking to more concepts. And then you can disagree about specific words and definitions.

[00:22:11] So tolerability. And so again, we are in the 2D plot. You can also make it, uh, multidimensional if this is something you like, but nobody does that. so we start tod uh, frequency of exceedance, vertical access, horizontal access, number of lives lost. For example. Your risk profile of your building goes there as a curve, which is decreasing, and you have.

[00:22:35] The ous limit, and then another limit, which is a tolerability limit. If you go above the tolerability limit, your design is not okay by definition, and you have to do something which could be just not building the design, or if it is an existing building, you have to do something. And it might be very expensive to do something, but you have to, um, what is the justification for having a tolerability limit? [00:23:00] this inconvenience. I said in the beginning like, Well, let's keep it objective and so on. Well, yes, when you calculate risks and so on, it's objective at tolerability limit deification for having that is that we recognize that. example, high consequence events are not acceptable to society or to individual people, depending what you're talking about.

[00:23:25] And same for high frequency events probably, and different combinations of that. So we just recognize that there is some psychological aspect there, which is put into tolerability limit and. Probability limit actually allows you to benchmark your risk profile against this say aggregated psychological effect. So no justification of economics or so on as far as I am aware of, purely, okay. If you have this building, it's a high rise building. If you have this building and there. Sprinkler control and [00:24:00] this and that, but lots of active systems. But, Okay. Typical example. Maybe there is no staircase because, well, why should we need, We don't need Right. I mean, it's it's 

[00:24:09] a perfect 

[00:24:10] Wojciech Wegrzynski: just, just an elevator which you drive in your, with your Ferrari and you are being driven up to your penthouse staircase is a waste of square feet that you could sell. Yeah.

[00:24:19] Ruben van Coile: Exactly. Exactly. But just actually, which is a very fair point if you would do it purely on, Economic term, I would not be surprised if you can find situations where indeed having no staircases or very low number of staircases will makes economic sense. But we say, Ah, this is still not okay. And why not?

[00:24:38] Well, because in the extremely unlikely event

[00:24:41] Wojciech Wegrzynski: Mm-hmm.

[00:24:42] Ruben van Coile: Extremely unlikely. So unlikely that the monetary value becomes very low, but still in extremely unlikely event that are active control measures the many we have who are great and managed and everything, if everything would fail for whatever reason, we do have a problem and we cannot, justify the day [00:25:00] after that couple of thousand people died. Yeah. There was no

[00:25:04] Wojciech Wegrzynski: so this is something beyond the tolerability that, uh, must be

[00:25:07] Ruben van Coile: Yeah. Yeah. So that would then be, that would be an example of, okay, very short and hypothetical, but it would be an example that's, at least to me, it's not justification. Why would you always need a staircase, right? Because say, uh, whatever happens, you always need a staircase because, This is risk. If you think about risk, because risk is more fundamental than prescriptive guidance, right?

[00:25:26] Different discussion maybe is blowing up discussion here, so risk is the reality. It's the risk profile of the building. Why do we need staircases in every building? Well, apparently because it, it's not tolerable not to have it, and it doesn't matter if you can tell me that it's very expensive to have this staircase.

[00:25:42] That doesn't matter. We need staircase, otherwise not to.

[00:25:46] Wojciech Wegrzynski: I guess the same case could be now brought into a flammable facade with, Cav Super tall building.

[00:25:53] Ruben van Coile: Apparently very interesting as in, cause if you talk about costs and benefits, what we don't [00:26:00] do, as far as I'm aware of. But we, I mean, we are not taking things very holistic, right? And I'm holistic. I mean, even beyond fire safety, some of those, cuttings, they might be actually very good from an environmental perspective.

[00:26:12] mean, not my expertise, so I don't know. But let's say hypothetically it could be a real winner

[00:26:19] and the likelihood of having a fire is very low and so on. So in the end you could say, Well, actually cost effective. we should all go for plannings, but maybe we agree that it's not tolerable to have these kind of consequences.

[00:26:31] And I would then say, Ah, tolerable relates to probabilities and consequences. Yes. So to have these kind of consequences. So if you use this kind of cladding on a low rise building where you say that the consequences are for whatever reason, Very low, then I see no point to be against it.

[00:26:48] Wojciech Wegrzynski: this is a really good discussion because, obviously I could ask you like, what is the, to release or who, who should define that? And we could be talking about ours on, on that, but I, I just like to leave it that you [00:27:00] have this tool. You, you, you have the tool of risk. You have probabilities, you have consequences.

[00:27:05] You have some ways to map them out to see the bigger picture, and you have a tool to define at what point something is ler, at what point it's not. and then you apply this unbiased method to this unbiased collection of your scenarios, and you can really filter them out and see like case by case, what's the case in there?

[00:27:26] And, and take, um, engineering decisions based on that. So wherever this limit lies, which probably is the obligation of, Person responsible for, for setting whoever that is in your country. Politician, architect, uh, firefighter, whoever that may be. it's a tool that you can, that you can work with.

[00:27:47] Ruben van Coile: And to be even stronger and, okay. Not everybody were like this, but I find it very relevant as. Tool for thinking, right? You have tolerability limit and then cause you have to, [00:28:00] you can do the cost effectiveness and so on. Does that mean that you need to specify this limit and give numbers to it?

[00:28:06] Personally, no. There are a number of reasons for that. You could say, okay, it's very difficult, Who should decide and so on that, that's all true. Also, setting a limit, actually. People to go as close as possible to it. I see no reason to have this limit, because you should actually stay away from it.

[00:28:24] It's so, it's think thinking wise. It's just like, Look, in my head there is this limit. I should stay away from it. Right? And then can I tell myself with my own judgment that I am clearly in a tolerable region? If I can do that, then from my perspective, I, I am 

[00:28:41] fine. 

[00:28:42] Wojciech Wegrzynski: what, what, happens when you are just below the deliverability limit? What does the risk tools tell you?

[00:28:48] Ruben van Coile: Well, so Just said is that it doesn't, The tolerability limit is a concept for thinking. It doesn't need to be quantified. I mean, some countries or legislation might want to quantify it. It doesn't need to be quantified. So if [00:29:00] in a situation where it's not quantified you, what you're actually saying is that your risk is quite high and that you feel uncomfortable and that you are like, Well, am I tolerable here?

[00:29:08] so this. Pushes the engineer to say, Well actually I should reduce the risk because I am not comfortable. And this is the, this is the concept. If your country has a strict limit, well okay, then you could say you're in ALARP region now because you're underneath stability limit and you have to look at cost effectiveness.

[00:29:26] But when you're very close and this, this, cause, these discussions come up a lot. I mean, I might be jumping a little bit here, but people. Often bring up this very same comment, maybe with different words. Say, Ah, yeah, but if you're very close to tolerability limit, and then how sure are you about your inputs?

[00:29:41] If your input is a bit different, maybe you're above tolerability limit and so on. Yeah, this, these discussions make sense if your tolerability limit is really fixed. If this is really something you can be above or below. But if it is a concept where you say above that, I'm not comfortable. This is not.

[00:29:58] Below that, [00:30:00] Okay, I'm fine and I just need to check if I'm actually cost effective here. Then having the variation in your parameters doesn't matter so much anymore

[00:30:06] Wojciech Wegrzynski: Mm-hmm.

[00:30:07] Ruben van Coile: cause it's part of your evaluation.

[00:30:09] Wojciech Wegrzynski: Um, now I wanted to jump into, something else. For me, the concept of risk. It such a profound thing because you could accommodate also things like, Fail probabilities, performance of your devices, how does, the risk deal with it? I, I assume they are hidden within the, the scenarios that you investigate but, but may, maybe you can tell me a bigger picture how you would approach measuring that in, in your scenario.

[00:30:36] Like when I said about my smoke control system case in the shopping. But I may have a case where my one fan has failed, or I may have a case where the smoke is not detected and it is delayed by some time. Which of these events has some, uh, probability? There's also a distribution in the fire size, because I just assume a design fire, which is let's say an agreed uh, representation of.

[00:30:59] some sort of [00:31:00] very bad, but still probable fire in, in that case. But I have no idea what it really will be. It will be a scatter of different, amounts of heat, different amounts of smoke, production parameters. it's all distributions. So, how the world of risk works with, with, assumptions that are not just sharp one point thing, but distribu.

[00:31:20] Ruben van Coile: Okay. The, a number of aspects there, and to some extent, it's, a choice for the engineer and the choice, I mean, that you are building a model. Right. You're building a model to model the risk, of the building. And you couldn't do that in different ways. So, one way to go ahead. If you have parameters with a continuous distribution, like, like your, fire size, potentially say, Okay, five MegaBots, 5.16, 6.2.

[00:31:51] So it's a continuous distribution, then you can. Do repeated calculations if you are using calculation model, and just [00:32:00] sample from your input distributions and get a view on your output.

[00:32:04] wojciech_wegrzynski: Mm

[00:32:05] Ruben van Coile: Now, this maybe or maybe not, sounds easy. It's actually relatively difficult. Cause if I say get a view on your output, if your output is a 3D view of how smoke is propagating, then. What this means if you do repeated sampling, it's not so clear anymore. Cause then what is easy? If you say, ah, how many minutes is my, certain door available? Right. Okay. So that you can do and say, ah, it's five minutes. Uh, and this situ simulation 5.2 in that, and you get your distribution of this time that it's doors available.

[00:32:37] But if it's a 3D smoke picture, you, you don't have this. So you, you definitely lose some information. Now I say it's model because you have the other, part as well, you said about the small control system failing, depends what you mean with failure. But if you say it just, it activates or it does not activate in that case, you, for example, have two trees or two, uh, in your eventually, right?

[00:32:58] So then you jump into [00:33:00] an eventually should have put it that way and you say, Okay, one event, the smoke control system activates one event. Smoke control system does not.

[00:33:07] wojciech_wegrzynski: Mm-hmm.

[00:33:08] Ruben van Coile: And you can add granularity to that if you want, but now you have in your CFD or whatever you're using, you have two different models.

[00:33:15] Now you have one where the system is act not active, and one where the system is active. And now if you want, you can now have even do Monte Carlo simulations or anything else inside. Branches of your event. To then, for each of those scenarios, they still take into accounts the uncertainty in your fire size, or you can split it up into five discrete fire sizes.

[00:33:38] is 

[00:33:38] Track 1: also. 

[00:33:38] Wojciech Wegrzynski: and and then what I can do is, is take, if I know the, the combined probability of a certain branch of my Evans tree happening, because I assume the probabilities are either known or. Number can be given to them, so as I venture through my event three, I know this branch, okay, 30% this branch, 70% in the branch of [00:34:00] 30%, another event is 50 50, so I have two times 15 other branch and so on.

[00:34:04] You can work out the individual probability of each of your final branches. The last branch. Then you can do some sort of engineering calculation to DataMine the consequence. And then you have points to feed your plot of probability consequence, and you can start seeing them as they align maybe. And even where nothing in my building, works and the fire is a hundred times bigger than it should be. It will give me a consequence of catastrophic failure, but I will see, okay, the probability was like one to minus, I don't know, 12. So it would go so far into, into the plot that we could never, um, consider that. I guess Are there, are there events that are unto tolerable no matter their probability that, that, that's something I wondered 

[00:34:54] Ruben van Coile: to me conceptually, yes. Right, So you have your tolerability limit and tolerability limit basically says, events which[00:35:00] are there, or designs which have a risk profile, which goes into intolerability region. Those designs are not acceptable. Now, I do think indeed it is reasonable to cut off somewhere at a very, very low probability.

[00:35:12] Anything you represent, just because it becomes a bit ridiculous to really say that you're calculating stuff up to such a precision. but on the other hand, yeah, this is, the case about the staircase we talked about earlier, right? So because your, your sprinkler system and you have a 24 7, 10 people management team and so on, so your buildings can be extremely safe, but still you're saying, Well, we still need staircase, 

[00:35:35] so, 

[00:35:35] Wojciech Wegrzynski: Uh, I, I guess the outcome is also like, if your building has this point, then it probably should not be built then, that's the ultimate thing also to, to figure out, okay, if my building has a scenario that leads to, uh, you know, a death of a thousand people, and, there no, no matter my safety strategy, there is this, this, this chain of events that leads to, to this tragic consequence maybe.

[00:35:57] You either should redesign it completely or, maybe not [00:36:00] build it. I like, again, again, you are showing, risk not as, a mythical concept from the land of unicorns. It is a tool, tool for engineers to use. I, I like that.

[00:36:10] Ruben van Coile: That's why like the 2D visualization, because this gives a lot of information, right? It's not easy to interpret always, but it gives you how often things happen and how severe they could get, and the combinations of the two. If somebody gives you just a single number, which sometimes you can see that they just combine different scenarios in a single number.

[00:36:30] This tells me personally, not much really. If say, this is a metro station and this metro station, the annual risk of, death by fire is this number like, okay, I don't know what means now. Cause did you just lose 200 people in a very low event? Or are you talking about what is this? So, the two is very nice.

[00:36:49] One thing to maybe add and what we just said without, Closing the topic because I don't think it is closed, but it, when we are talking about these very high consequence events and things which are not, [00:37:00] acceptable no matter what, and on, we're getting very close. So that's my feeling with, something you had in one of the previous podcasts some time ago with Jaime for example, allowable damage.

[00:37:10] So with these things, I mean, I see, or I feel, I feel is a better word, that these things actually might match a lot.

[00:37:17] So it's not one or the other. That's what I'm trying to

[00:37:19] say.

[00:37:19] Wojciech Wegrzynski: yeah. Cool. Cool. Okay. We said that, risk would be some sort of collection of, probabilities and consequences. probabilities I feel, are not necessary very much to explain. I, think every engineer should understand the, the concept of probability, but I wonder.

[00:37:37] About the measures of consequences so far. You've mentioned, um, fatalities when we discussed the curves and you've mentioned there can be fn curve where, or, or people, um, affected by the fire also, not necessarily fatalities, but, but people affected by the fire. I guess there could be a monetary, analysis where you would just put, uh, value in whatever currency you [00:38:00] prefer on the damage.

[00:38:01] Ruben van Coile: Definitely,

[00:38:02] Wojciech Wegrzynski: How else would you measure the consequences? Or, or these are like the ultimate ones.

[00:38:09] Ruben van Coile: Anything you want, and this is the thing we don't do enough. 

[00:38:12] Wojciech Wegrzynski: I don't, you don't see that every day. Like, I always work with FN, to be honest, I, honestly don't think I have ever worked with anything else. But yet here opens a whole world of possibilities where you could map stuff in such a different weight to gain a real understanding of a building.

[00:38:29] So, yeah,

[00:38:29] Ruben van Coile: So I should actually open up my own, uh, definition there somewhere. I actually don't have it fresh, so I always like to say the design should be. Socioeconomically acceptable. and this is where we should start, right? Because people always jump into the traditional fire safety objectives, life safety, property protection and so on, but miss things.

[00:38:50] And then in end, some people add cultural heritage because they feel that indeed there some cases which don't match the traditional ones and so on. And, [00:39:00] but the risk of. Doing it that way. I mean, choosing the traditional ones and then adding something which you think about is that you don't take a broad approach and actually think what is important here.

[00:39:10] If you're having a tunnel, then it's very likely, I would say, even extremely likely that life safety is

[00:39:18] wojciech_wegrzynski: mm-hmm.

[00:39:19] Ruben van Coile: Yes. But defining for your design is your business continu. So, yeah, but if you did, you never thought about it or if it didn didn't come, come up, then well, yeah, it was not on your list. So if you think about first, Okay.

[00:39:33] What, why is this tunnel important? Okay. It's connecting to important parts of the city. Whatever. Hopefully it comes up if you have a museum and say, Ah, yes, the evacuation is fine. Um,

[00:39:44] Wojciech Wegrzynski: We, we, we've 

[00:39:45] Ruben van Coile: like, Yeah, okay. 

[00:39:45] But 

[00:39:45] wojciech_wegrzynski: antiques in it. 

[00:39:46] Ruben van Coile: oh yeah, everything is gone. Okay, well probably.

[00:39:50] Wojciech Wegrzynski: not a great outcome.

[00:39:51] Ruben van Coile: Yes. 

[00:39:51] Wojciech Wegrzynski: I have, ISO in front of me, like I said, and con for consequence. oh. These guys are good. Actually. , you know, these guys, these guys are good. They say outcome or [00:40:00] outcomes of an even expressed positively or negatively, quantitatively or qualitatively. Well, that's not fair, broad definition of a consequence, but, like you said, it can be whatever you like it to be.

[00:40:12] Again, 42.

[00:40:13] Ruben van Coile: Well, what I'm, Not a hundred percent sure of, but within the, the ISO framework, Well it's, it is a framework iso, right? So you have the fire engineering and then the stock parts, including structural engineering. But the, fire engineering also refers to higher, more general standards. So this risk definition is, uh, probably not 

[00:40:35] Wojciech Wegrzynski: Could come from, From above. 

[00:40:37] Track 1: Yeah. 

[00:40:37] Wojciech Wegrzynski: above. Above. Okay. 

[00:40:38] And so, you know, preparing for this discussion, Danny has sent me a very comprehensive list of questions to ask you, which I don't, We're not even in one third. And, it seems we'll run out of time, but, let's say one item of, from that list, One that, that really is interesting to me, individual versus societal risk.

[00:40:58] And, it also needs to [00:41:00] be explained. I, I think it's, it's also a fundamental concept. so how would you explain the difference between a risk from the point of view of an individual and risk from the point of view of society, whatever society is?

[00:41:14] Ruben van Coile: Okay, now, now we need to make a very clear distinction. So in the last question you said, uh, you said, the. View of society or point of view 

[00:41:22] Wojciech Wegrzynski: Okay. Maybe point of view 

[00:41:23] wojciech_wegrzynski: is 

[00:41:23] Ruben van Coile: an Well, but I mean, can close that one relatively easy in the sense you can say societal is a public perspective. what does society except individual then is a private perspective.

[00:41:39] What was the private person? except so very straightforward in a way, society. Has, for example, per captive guidance. So society has somehow maybe implicit risk, which accepts maybe there were some individuals who would've been happy to spend less money on fire safety in their building, and maybe some people want to [00:42:00] go above, but this is their private choice and society sets the minimum.

[00:42:04] So that's. Clear. Well me, . So then if you say, Okay, but that's the perspective. If you say like, Okay, societal risk, individual risk, this is something then different. So when we talked about FN diagrams, we were talking about what we are as often refer to societal risk. So the possibility of having multiple fatalities in a single event, and then with a additional statement that multiple fatalities is not very much accepted.

[00:42:30] And so if you would have. a hundred people dying in a single event. This would be an outcry. We would change our legislation quite sure about that. If you would have a hundred events with one person dying, probably you don't change the

[00:42:43] Wojciech Wegrzynski: The, the media would be brought after the fourth one and then would never, not even mention them anymore. like 

[00:42:50] we like 

[00:42:50] Ruben van Coile: Yes, happening. 

[00:42:51] Wojciech Wegrzynski: with car accidents, right? 

[00:42:53] Ruben van Coile: Yeah. So. That societal is there and it links them to, to reliability limits and to reliability Limit is a limit to societal [00:43:00] risk. What is important though, and it's very difficult, is that if you make an fn Iica, the fn iica depends on the number of exposed people. So this is extremely inconvenient.

[00:43:12] So if you now say are in the legislation or something, I'm going to put a limit on my tolerability. So Right. I'm going to actually have a tolerability limit in legislation. Yeah. Okay. How are you going to do this? Because if, if there are only a hundred people working in the building, then this is a natural limit on the number of people, involved in the fire.

[00:43:31] If you now have a, much bigger site with thousands of people, Is it the same deliverability limit which supplies, I mean, can't be because otherwise the other building is not safe. So either way it's, it's not obvious at all. but if you, Cause again, said it, let me, I like to think about as, as concepts really, If you say for big building, thousands of people working there, the design is toler.

[00:43:53] Okay, so you're now saying that the frequency of having, one person dying or exposed to fire or a [00:44:00] hundred percent so on, that's a tolerable, you do an alarm evaluation or which can be just by expert judgment, but implicitly you come to the conclusion that it's not cost effective to invest more. Okay, fine.

[00:44:11] You have now shown that on average

[00:44:14] the average person there, the risk is.

[00:44:16] but you have not shown that it is fine for everybody. And this is crucial to not forget, and this is done the individual risk. Cause it is not unthinkable that in your big building when you do all your evacuation simulations, if this is what you're doing, that the people who die in your simulations are always the same

[00:44:33] ones. 

[00:44:33] wojciech_wegrzynski: Okay. 

[00:44:33] Ruben van Coile: So you're now having this fend diagrams like, oh 

[00:44:35] there's 

[00:44:36] Wojciech Wegrzynski: one. 

[00:44:36] That diet okay at this

[00:44:38] Ruben van Coile: Yeah, it's a. 5,000 people there. One person dies occasionally, yes. But it's always the same room where people start, where the people are, where they die, as in they start there and they never make make it to the exit. So the individual risk for the people in that.

[00:44:52] Room, for example, mobility impaired people could also be, related statement, their individual risk will be too [00:45:00] high. So societal risk, overall performance of the building. individual risk saying like, Okay, and now looking at all the individuals, is it actually tolerable for everybody 

[00:45:09] Wojciech Wegrzynski: Okay. So, now, now, now I understand this from the perspective of, risk engineering, that, you may have an outcome. That is acceptable from the bigger picture point of view, but, potentially unacceptable from the individual point of view. And if I have found a room where this individual risk is that, hi, I would put my it in there.

[00:45:29] I dunno about, about your university, but I would have preferences there and that's engineering as well. 

[00:45:36] Okay. Good, good, Now I would like to, , take you into some, private business . SFPE working on the new performance based engineering guide, and I had the pleasure, privilege, to, be responsible for the Hazards chapter in.

[00:45:53] And definition of hazards. So I would love to, to hear you, now. Now let's go into the process [00:46:00] of risk engineering. How does one, define their hazards and, and what hazards could be?

[00:46:04] Ruben van Coile: Okay, this is really 

[00:46:06] Wojciech Wegrzynski: Yeah, and I'm gonna, whatever you say, I'm gonna write it down in the standard now. . 

[00:46:09] No, just kidding. Don't worry.

[00:46:12] Ruben van Coile: because what I am, regularly working. With somehow is quantification of. So you said earlier, before we, for example, that my design, fire and uncertainty, my design, fire uncertainty and file ties. Okay. This, and we propagate this and we get a picture of, of the building. When you talk about hazards, you are now in an area which to me is very much about risk management, really, where you're gonna, going to identify which, hazards there are, what can be done to maybe eliminate some of those hazards.

[00:46:44] So it's. a very different skill. I mean, it's, it is very important. So we talk, okay, what is hazard? We have the, the is document here somewhere. So something with a potential, to have, create negative, uh,

[00:46:58] consequences. 

[00:46:59] Wojciech Wegrzynski: it in [00:47:00] It's, it's, even our chain of events with the potential to create harm, something like that. It's not a precise definition, but more or less like that. Yeah.

[00:47:08] Ruben van Coile: So it might be possible to just eliminate some, right? So you have a, a process and an industrial plant, and you say there's a certain hazard there because they are using an open flame and say, Okay, we can eliminate this hazard by changing our process and removing the open flame. within the, the risk evaluation, but I'm talking about the evaluating the risk of a building going all the way there. Is, Well, something I have never done. It's a, it's just, it's an, it's, that's, that's really big, big

[00:47:37] Wojciech Wegrzynski: so again, as long as we stay to the simplest definition of risk, it, it's really accommodating. And now we went outside of it and, and is now risk management, which is a process which is, a job to be done based on the fact that you have this, unbiased measure of, of risk, which allows you to do that because then for hazards, uh, hazards are.

[00:47:57] Interesting because we have [00:48:00] defined them that you need to identify all of them. But it's not the point where you judge whether this hazard makes sense or not. Whether is at high risk or low risk. No, no. That's the, like a set of inputs to your risk analysis. And then through your risk analysis, you will work out what they mean.

[00:48:17] At the, at the hazard definition, you, you should not really evaluate because actually you can, you can miss something and. You, you need to apply the whole methodology to understand the consequence of an event that includes a particular hazard. So, for us it was a challenging thing, to work around like, and also you have to accommodate the fact that someone may miss a hazard it's, um, very difficult to understand all the hazards in your building. 

[00:48:40] Ruben van Coile: Yeah, so using the word all is maybe, uh, 

[00:48:43] Wojciech Wegrzynski: Yeah. But there's always something that, could have been missed, right?

[00:48:47] Ruben van Coile: Exactly the by definition, I think it's, it's impossible to identify

[00:48:51] all hazards. and then what is the value of looking into that, what the value is indeed that by thinking about hazards that you will, [00:49:00] and identifying them, that you will find ways to eliminate some or to have mitigation measures and so on.

[00:49:06] So it, it's much more. The process, at least for me. then it is about determining is my building now?

[00:49:13] wojciech_wegrzynski: Hmm.

[00:49:13] Ruben van Coile: Is my event curve of my building acceptable? And let's start with looking at the hazard of having my washing machine having an electrical fault. Okay? If you, if you do that, then uh, dec calculation is gonna take a long time.

[00:49:26] Wojciech Wegrzynski: Yeah, we, we finally reached a point where, you've touched the, the part of risk management that I, I think I like the most. And it's something that also when I, you, you've mentioned the episode Jaime and David Lang about the maximal oval damage and. It was very like process related.

[00:49:44] Like there is this framework, this, maybe you could even call it philosophy. Like if you do these things when defining risk for your building, you will eventually come up with a pretty decent design of your building because, Going through the [00:50:00] process and understanding these things through this type of measures makes you think about the solutions in a completely different way, eventually leading to, uh, a safer structure.

[00:50:11] So it's, uh, I think it's a value of risk engineering that is very difficult to quantify. Like how much percent safer my building will be if I do risk analysis for it. I don't know, but it'll be safer it's something, You cannot calculate that. But, this subconscious feeling I have about this, this process that I, I really like.

[00:50:30] And that's also why I, I wanted to do an episode like this. So the whole concept is more approachable by more people and maybe more people will. Be familiar enough or less intimidated, to, to actually use it because the benefits are way, way beyond just putting a number to the consequences and probably to a scenario, right?

[00:50:49] Ruben van Coile: Yeah, definitely drawn on the process side there. Fully agree. And then, The, like the risk assessment where you actually calculate a risk profile. [00:51:00] I'm definitely not saying this should be done for every type of structure. It will probably be all very rarely that that the effort makes sense.

[00:51:07] But I hope that most fire engineers or people involved in in fire safety realize or agree. With me that there is this underlying risk profile for a building, which is its performance. And when you do one or five design fires and you calculate that, you are just testing it a bit, a bit and then making a judgment in the end that, yeah, based on those evaluations, the risk profile will be fine. And so, but if you go the step further, Which will not be necessary in, in many cases, but if you go a step further, you get a better picture. So that's really what it is. It's not an alternative, it's just getting a better picture of the same.

[00:51:47] Wojciech Wegrzynski: Okay, Ruben, thank you.

[00:51:49] Thank you so much for taking me into the, the approachable, uh, risk. And I, I really appreciate taking your time to, to have this difficult discussion. I mean, talking about [00:52:00] fundamental concepts is sometimes the most difficult discussions there are. for, for the end. You can do whatever you want for your closing statement, but I would love, How would you advertise?

[00:52:10] Like why you, Did I ask you that again? Why do you tell your students to, to learn risk? 

[00:52:15] Ruben van Coile: Maybe, and I can say something very similar as probably I said last time, is that it's, it's fundamental,

[00:52:20] Wojciech Wegrzynski: you did. You did. Now I remember. No, let's, let's drop the students. Students are cool. let's talk engineers. Why engineers who have never used risk concept in their life should take a while and, learn a bit about it and appreciate it. Will, will it be useful to them?

[00:52:35] Ruben van Coile: Because they are in the business of designing for risk. So no. No matter how you do it, if it is just by. Your opinion or by doing calculations, number of set, set of them, or many of them, you are in the business of evaluating the risk of a building and making sure it is acceptable. So this is your job? No.

[00:52:56] No matter how you do it.

[00:52:58] Wojciech Wegrzynski: If you like it, of or not, [00:53:00] if you know about it or not, , it's, it's your job. 

[00:53:02] should Okay it Ruben, thank you so much, for coming and, I guess, we will have many more occasions to talk about risk in this podcast. I, I like this direction and I think we're doing something good here, going into this journey and taking the, the listeners with us.

[00:53:19] So thanks a lot for coming and yes, you run.

[00:53:21] Ruben van Coile: Cheer. Thanks lot. Bye.

[00:53:23] Wojciech Wegrzynski: And that's it. Thank you Ruben for Uh, fantastic discussion. It's so difficult to talk about the fundamental concepts and. And Ruben, this is very brutal. To the interviewer. 

[00:53:35] He always takes my questions and smashes it to Atoms and answers one way. It's kind of stressful To review like that, but in the end, , the point of doing this is to learn, and these are the moments where I learned the most. So I guess for my audience, you guys As much. Uh, with me, that's, that's a great thing for If I had to pick one takeaway from this episode for myself. 

[00:53:57] I guess it would be the difference between the societal risk and [00:54:00] individual risk. But, from this perspective of maybe someone is. At, um, higher risk 

[00:54:08] At disproportionate risk to the rest of the population of the building. That, that's a very interesting concept because. I see here, a potential to use this as a, as a way to address the problems of disabled people, the problems of, Lack of accessibility for some the problems with, 

[00:54:25] delivering a straight safety strategy that is inclusive for all in the building. That's a, that's a beautiful concept. And I think. Uh, the tool risk. Could be even better in this the night then I thought so. So that is my takeaway. I wonder what your takeaway is. And if you enjoyed this episode, if you liked it, please let me know. I would love to. 

[00:54:44] Learn if episodes like this are, are useful to you because I can make more of them, you know? And yeah, that would be it. Uh, For today. Thank you all for the five-star ratings that are coming. I've heard that. Some [00:55:00] are confused because you cannot find the five-star rating on the webpage and there's. 

[00:55:05] Approximately 13% of the audiences listens to the podcast through my web page, which I appreciate. That's the point of having webpage. Really? If you want to leave me a rating, you need to go to an app like a. Apple podcasts or Spotify and only diary again. Again, leave the reviews that are counted by the algorithms, the links to, to. 

[00:55:23] Uh, apple podcasts and Spotify are always under the episode descriptions. So there's like one click away due to get into the app. And then you can give me the five-star rating if you really want to do that. And yeah, you actually do. Thank you very much for them so thank you very much for joining me here this wednesday and uh, looking forward to having you here next wednesday See by