Nov. 12, 2025

226 - New Swiss fire safety code with Gianluca De Sanctis and Sofia Kourgiantaki

226 - New Swiss fire safety code with Gianluca De Sanctis and Sofia Kourgiantaki
The player is loading ...
226 - New Swiss fire safety code with Gianluca De Sanctis and Sofia Kourgiantaki
226 - New Swiss fire safety code with Gianluca De Sanctis and Sofia Kourgiantaki
Apple Podcasts podcast player badge
Spotify podcast player badge
Apple Podcasts podcast player icon Spotify podcast player icon

It is a massive effort to rewrite a national fire safety code around measurable risk, explicit targets, and cost-effectiveness. But sometimes, there are great reasons to do so. In this episode, together with Gianluca De Sanctis and Sofia Kourgiantaki we take you inside Switzerland’s sweeping reform, where a new federal law sets a maximum individual risk for life safety, ties property protection to a clear marginal cost rule, and harmonises practice across cantons. Together, we trace how the framework defines acceptance criteria, builds a shared “model code” of probabilistic inputs, and keeps prescriptive pathways for standard projects—only now grounded in risk-optimised measures.

You’ll hear how the system replaces vague equivalence with transparent math. Life safety is anchored at 5×10^-5 fatalities per user per year; if a building exceeds that threshold, measures are required until it doesn’t, regardless of cost. Beyond the threshold, optimisation is driven by the marginal cost principle and a nationally defined social willingness to pay, aligning fire with flood, transport, and earthquake risk policy. For property, the rule is simple and strict: do not spend more than the expected damage you remove.

While the code was being developed, Sofia put the method to the test in a retail centre case study using Bayesian networks and ASET/RSET. The model compared detection, sprinklers, and smoke exhaust while capturing occupancy, fuel loads, growth rates, system reliability, and fire service response. The surprising result: in a seven-meter hall, detection met the life-safety target on its own, and the most cost-effective optimisation paired detection with sprinklers, while smoke exhaust added little benefit in that geometry. The lesson isn’t that one system always wins; it’s that context and data should decide, not habit.

Switzerland didn’t stop at policy. A peer-review approval process, ETH’s advanced training in probability and risk, and a national model code make the approach usable and reviewable. The reform is in technical review ahead of political approval, with mechanisms for minor updates as evidence grows.

Direct links to the document:
- German Version: https://mitwirkung-vkf.ch/de/
- French Version: https://mitwirkung-vkf.ch/fr/

Also, there are 4 short videos in German, French and Italian that describes the new framework of the new codes:
https://www.bsvonline.ch/de/brandschutzvorschriften/projekt-bsv-2026/videos

A part of this shift in culture is also the new MAS in fire at the ETH, which you can learn more about in here:
https://mas-brandschutz.ethz.ch/

----
The Fire Science Show is produced by the Fire Science Media in collaboration with OFR Consultants. Thank you to the podcast sponsor for their continuous support towards our mission.

00:00 - Why Switzerland Is Changing Fire Codes

02:53 - Guests Introduced And Roles In Reform

06:27 - Why Move Beyond Prescriptive Design

10:45 - Law, Harmonization, And Access To The Code

14:06 - How The Framework Blends Prescriptive And Risk

18:16 - Setting Risk Targets And Acceptance Criteria

23:00 - Marginal Cost Principle And Social Willingness To Pay

27:54 - Case Study Setup: Retail Center And Methods

32:10 - Bayesian Networks, Data, And ASET/RSET

36:20 - Findings: Which Measures Actually Pay Off

39:20 - Model Code Parameters And Uncertainty Handling

44:00 - Approval, Competence, And Peer Review

47:12 - Data Sources, Limits, And Robust Decisions

51:40 - Timeline, Revisions, And Education Pipeline

55:30 - Can Other Countries Adopt This Approach

01:00:00 - Host Reflections And Next Steps

WEBVTT

00:00:00.160 --> 00:00:01.679
Hello everybody, welcome to the Fire Science Show.

00:00:01.679 --> 00:00:07.280
Today I am taking you to Switzerland, where there's a big change occurring in the law systems.

00:00:07.280 --> 00:00:12.160
To start with, I my experience with the Swiss fire engineering is not very clear.

00:00:12.160 --> 00:00:16.879
But from people I know, they're very solid and pragmatic foreign engineers.

00:00:16.879 --> 00:00:24.559
I also understood that their code that they were using every day was about prescriptive-based code, so something similar that we would have in Poland.

00:00:24.559 --> 00:00:37.359
And then some time ago, I've learned that there is a major shift that you're ranking in the Swiss fire system, which is driving their whole framework towards risk-based solutions and kind of performance-based design.

00:00:37.359 --> 00:00:39.359
So a very, very interesting shift.

00:00:39.359 --> 00:00:48.640
And I know this was not only interesting to me, but uh to many of the listeners, because at multiple occasions people were asking me to try and cover this change in the podcast.

00:00:48.640 --> 00:01:00.079
So I did my thing, and I've invited two speakers who I've seen giving a keynote on this shift in the Swiss Spark at SFP Enro Corporation earlier this year.

00:01:00.079 --> 00:01:04.560
That is Gianluca De Sanctis from Bassler and Hoffman AG.

00:01:04.560 --> 00:01:08.879
And Sofia Kourgiantaki from Migros Engineering Solutions.

00:01:08.879 --> 00:01:20.239
And together they covered the development of the code and they covered the case studies which were used to validate or verify how the code performs in ReWorld projects.

00:01:20.239 --> 00:01:25.359
So in this talk, we'll dive deeper into the changes made to the Swiss Park code.

00:01:25.359 --> 00:01:29.760
But it's not meant to educate you on how the code looks like.

00:01:29.760 --> 00:01:42.560
This episode is more about a framework or things that you have to do, things you have to consider where you try to shift from a fully prescriptive-based solution to fully risk-based solutions.

00:01:42.560 --> 00:01:43.519
And so on.

00:01:43.519 --> 00:01:48.000
Also discussions when this makes sense, when this does not make sense.

00:01:48.000 --> 00:01:53.040
In fact, our new code is also enforcing prescriptive solutions for a simpler project.

00:01:53.040 --> 00:02:01.439
How to evaluate targets, how to evaluate uncertainties, how to make your whole engineering community ready for such a shift.

00:02:01.439 --> 00:02:04.560
Because you cannot do this overnight very quickly.

00:02:04.560 --> 00:02:08.479
It's a bit a long-lasting process which is still taking place in Switzerland.

00:02:08.479 --> 00:02:15.680
So all those things together, I think, is quite an interesting exercise on risk-based design and performance-based design.

00:02:15.680 --> 00:02:23.120
And how different is this from the world of prescriptive-based design that we still have in most of the places in the urban in the world.

00:02:23.120 --> 00:02:25.919
So given all that, I hope you will enjoy this episode.

00:02:25.919 --> 00:02:28.560
Let's spin intro and jump into the episode.

00:02:28.560 --> 00:02:34.400
Welcome to the Firescience Show.

00:02:34.400 --> 00:02:38.240
My name is Wojciech Wegrzynski, and I will be your host.

00:02:38.240 --> 00:03:07.360
The Firescience Show is into its third year of continued support from its sponsor OFR consultants, who are an independent multi-award-winning fire engineering consultancy with a reputation for delivering innovative safety-driven solutions.

00:03:07.360 --> 00:03:21.120
As the UK leading independent firewalls consultancy, OFR's globally established team have developed a reputation for preeminent fire engineering expertise with colleagues working across the world to help protect people, property, and the planet.

00:03:21.120 --> 00:03:37.360
Established in the UK in 2016 as a startup business by two highly experienced fire engineering consultants, the business continues to grow at a phenomenal rate with offices across the country in eight locations, from Edinburgh to Bath, and plans for future expansions.

00:03:37.360 --> 00:03:46.080
If you're keen to find out more or join OFR Consultants during this exciting period of growth, visit their website at OFRConsultants.com.

00:03:46.080 --> 00:03:47.919
And now back to the episode.

00:03:47.919 --> 00:03:48.800
Hello everybody.

00:03:48.800 --> 00:03:50.240
Welcome to Fire Science Show.

00:03:50.240 --> 00:03:53.520
I am today with uh two quests from Switzerland.

00:03:53.520 --> 00:03:57.360
Uh first, Gianluca De Sanctis from Bassler and Hoffman, AG.

00:03:57.360 --> 00:03:58.240
Hey Gianluca.

00:03:58.240 --> 00:03:58.879
Hey.

00:03:58.879 --> 00:04:04.400
And together with us, Sofia Kourgiantaki from uh Migros Engineering Solutions.

00:04:04.400 --> 00:04:06.400
Hi Sofia, nice to have you in the podcast.

00:04:06.800 --> 00:04:08.080
Hi, uh everyone.

00:04:08.080 --> 00:04:09.599
It's uh such a pleasure to be here.

00:04:09.840 --> 00:04:13.360
Yeah, it's a pleasure for me, and uh, everyone is looking forward to this episode.

00:04:13.360 --> 00:04:18.000
Uh I told you it was requested by multiple people on multiple occasions, no pressure here.

00:04:18.000 --> 00:04:31.040
So, what brought us in here is the new risk-based approach in Switzerland that's implemented, and you both have participated at different extents in this implementation and development.

00:04:31.040 --> 00:04:33.600
Uh, let's perhaps first start.

00:04:33.600 --> 00:04:40.639
How did you end up working on a new risk-based you personally and and what what was your role in the process?

00:04:41.040 --> 00:05:01.519
Uh well, I'm a member of SFP Switzerland, and uh I was delegated to uh represent the SFP Switzerland in the stakeholder process, which is used to derive the the safety objectives which we want to address with the code, and also uh derive the acceptance, the risk-based acceptance criteria.

00:05:01.519 --> 00:05:11.360
And afterwards, I was part of uh a working group which uh derived the methods or the process uh for risk-based design.

00:05:11.360 --> 00:05:12.319
And Zofia?

00:05:12.800 --> 00:05:21.279
Uh so um I'm a fire protection expert, originally a civil engineer, and I work for Micro for the last four to five years.

00:05:21.279 --> 00:05:30.079
And Micro is um the largest retail company with um 500 retail stores and logistics centers and industry.

00:05:30.079 --> 00:05:34.240
And so I work in the department of in competent center of fire safety.

00:05:34.240 --> 00:05:40.720
It's an internal department that focuses on fire protection design and developing internal standards.

00:05:41.040 --> 00:05:46.240
And for the um risk code, I I saw that you've implemented case studies.

00:05:46.560 --> 00:05:46.959
Exactly.

00:05:46.959 --> 00:05:58.639
So recently I completed a Master of Advanced Studies at the ETH University in Zurich, and one of the central features was uh risk-based fire safety analysis using probabilistic modeling.

00:05:58.639 --> 00:06:04.160
So my master thesis was actually based on this case study that we conducted together with Gianluca.

00:06:04.480 --> 00:06:05.120
How practical?

00:06:05.120 --> 00:06:07.759
Uh was it good the the master's studies?

00:06:08.079 --> 00:06:13.600
Yes, it was actually great, and it was kind of the perfect opportunity for me to jump into the subject.

00:06:14.160 --> 00:06:18.399
Kind of a spin-off question, but I heard uh a lot of people say a lot of good things about that program.

00:06:18.399 --> 00:06:21.040
So I I just wanted to confirm firsthand.

00:06:21.040 --> 00:06:27.360
Okay, so why does Switzerland need a risk-based code or approach?

00:06:27.680 --> 00:06:33.519
It's a good question because uh everything is fine with the prescriptive code we have now.

00:06:33.519 --> 00:06:46.079
So everyone can use it, uh, everyone is uh happy, not not really everyone, but from the terms of safety, we can say okay, the safety level that we achieve is kind of uh good.

00:06:46.079 --> 00:06:50.800
So there is really the question um why we should change this code.

00:06:50.800 --> 00:06:56.560
Um, and from the perception of the building owners, it is coming always okay.

00:06:56.560 --> 00:07:02.639
We have uh a safe code, but uh our costs for fire safety measures are are very high.

00:07:02.639 --> 00:07:04.720
So are they really efficient?

00:07:04.720 --> 00:07:14.399
And over the last 20 years, there is an additional pressure to the authorities to say, okay, do we have really uh an efficient fire safety design?

00:07:14.399 --> 00:07:24.160
And up to the 2015, they had hard time to answer this question because they didn't have a risk-based uh approach to it.

00:07:24.160 --> 00:07:38.399
And so they started to think about how they they want to improve the general framework, um, how to address safety in fire engineering or fire safety in general, and they came up also with this risk-based approach.

00:07:38.399 --> 00:07:55.759
So it was clear that uh we are moving to a risk-based framework, and it was long time not sure how it looks like because in our current prescriptive code it's stated that the safety has been equivalent to the prescriptive design.

00:07:55.759 --> 00:08:14.079
And there were also research studies conducted um also 10 years before, which showed that this strategy is not really appropriate to demonstrate uh which is not really a good approach to get this efficiency uh consideration into practice.

00:08:14.079 --> 00:08:27.680
So that's why there was also different uh research studies that showed different options, among others, this this uh marginal cost principle approach, which we implemented in the current code now.

00:08:28.079 --> 00:08:35.840
So, yeah, from your perspective as uh well, your company obviously manages a lot of buildings and most likely invests a lot of money into new projects.

00:08:35.840 --> 00:08:43.919
Uh, do you also see the risk-based approach and and this new code as an opportunity to improve efficiency?

00:08:43.919 --> 00:08:46.960
How does it look from the user of the code?

00:08:47.279 --> 00:09:05.039
Well, I think you just asked the right question because uh it's exactly what uh we needed that our company, this new methodology that uh FALK um developed, the Association of Cantonal Fire Insurance Companies, it's actually laid the foundation for risk management in Switzerland.

00:09:05.039 --> 00:09:25.200
And we are using also the same principle, the same methodology, not only for the public safety goods like people and property safety, uh so the building itself, we're using the same methodology also for the private uh safety goods, like our products, our stock of products, our facilities, and also the business continuity.

00:09:25.200 --> 00:09:41.679
We needed a methodology in order to ensure how our uh investments in fire safety, because the final resources are actually limited, how we can um invest them strategically where the greatest risk reduction can be can be achieved.

00:09:41.679 --> 00:09:44.480
It was a perfect opportunity for us.

00:09:44.799 --> 00:09:47.200
Well, opportunity, yes, uh for sure.

00:09:47.200 --> 00:09:49.759
Uh let's dive deeper into the code.

00:09:49.759 --> 00:09:58.720
Uh, we are having a very difficult task in here because we need to discuss what the risk method itself is and how does it work.

00:09:58.720 --> 00:10:14.159
But I'm also extremely interested in how is it incorporated into you know the whole ecosystem in the Switzerland, of which I have absolutely zero knowledge about uh making my job harder.

00:10:14.159 --> 00:10:24.320
But I I need to know how does it how is it implemented and uh how does it tie with with your existing safety approaches?

00:10:24.320 --> 00:10:37.200
Because you know, one thing is to develop a great uh methodology, but uh to implement it and turn it into a tool that actually works, that's a whole different story, and I want both stories in this podcast.

00:10:37.200 --> 00:10:42.240
So maybe first, what level of document is the new risk approach?

00:10:42.240 --> 00:10:45.360
Uh uh how exactly is it brought into the market?

00:10:45.679 --> 00:10:47.759
It's uh a federal law now.

00:10:47.759 --> 00:10:48.799
A federal law, okay.

00:10:48.799 --> 00:10:52.639
A federal law which is um has an impact on every canton.

00:10:52.639 --> 00:10:55.840
So, like a national overall the canton zone.

00:10:55.840 --> 00:10:56.639
Exactly, yeah.

00:10:56.639 --> 00:11:01.759
And the cantons can't deviate from this federal uh law anymore.

00:11:01.759 --> 00:11:14.159
This is the case now that cantons can, in addition to the existing fire safety regulation, define additional measures uh to deviate from this federal code.

00:11:14.159 --> 00:11:29.200
And this cost not harm harmonized uh fire safety management across uh Switzerland, and this was also one motivation uh of the revision to get a harmonized dealing with fire safety across different cantons.

00:11:30.000 --> 00:11:32.080
Is it just one piece of law?

00:11:32.080 --> 00:11:36.240
Is it like there's a law and a second document that explains how to do it?

00:11:36.240 --> 00:11:38.399
Like what's the contents of the package?

00:11:38.720 --> 00:11:42.480
It's one document, it has two parts basically.

00:11:42.480 --> 00:11:54.559
One part is the the legal formulation of what you have to do, and then you have uh also some um let's say um explanation how to deal with it with in in practical.

00:11:54.559 --> 00:12:00.240
So there are some kind of uh explanation how to to read this this legal document.

00:12:00.559 --> 00:12:06.960
Is it accessible to people outside of Switzerland or it costs uh one million francs to buy it?

00:12:07.200 --> 00:12:09.919
Yeah, no, because it's a federal law, it's free.

00:12:09.919 --> 00:12:12.720
Okay um so you can uh download it.

00:12:12.720 --> 00:12:15.039
Now it's under technical revision.

00:12:15.039 --> 00:12:18.080
Okay, so it's also openly available.

00:12:18.080 --> 00:12:24.080
You can uh log in to the uh VKF and then you get the the doc comment in French and German.

00:12:24.080 --> 00:12:24.799
How useful?

00:12:24.799 --> 00:12:26.879
Um just kidding.

00:12:27.840 --> 00:12:29.840
You have to wait for the English version, yes.

00:12:29.840 --> 00:12:31.440
No, no worries, no worries.

00:12:31.440 --> 00:12:34.240
In the age of automatic translators will be good.

00:12:34.240 --> 00:12:38.559
Uh if a link exists, you can find it in the show notes to the listeners.

00:12:38.559 --> 00:12:51.279
Uh Sophia, the case studies and and the the documents that were a part of the process, is this also somewhat accessible or um well my master's thesis is accessible at the website of the ETH university.

00:12:51.600 --> 00:13:06.720
Um the case study itself was also financed from the Cantonal Fire Safety Company of Canton Vode, so it belongs to Canton Vode, so there is not a link accessible for that, but I have a big overview in my master's thesis.

00:13:07.039 --> 00:13:07.679
Fantastic.

00:13:07.679 --> 00:13:08.480
Good, good.

00:13:08.879 --> 00:13:19.039
But on the FAUKAF website, you find also additional case studies or uh research studies uh which supported the whole development of the risk-based code.

00:13:19.279 --> 00:13:20.720
Okay, another another question.

00:13:20.720 --> 00:13:26.240
Does this approach completely replace the previous approach, or is this an addition?

00:13:26.559 --> 00:13:40.399
Yes, uh, it uh replaces completely the current design approach, but uh I have to say it's not really uh the case that we just have a risk-based approach now, and if for every building you have to perform a risk-based analysis.

00:13:40.399 --> 00:13:42.240
This is absolutely not the case.

00:13:42.240 --> 00:14:09.360
Uh, it's rather that we have uh still we will have uh a proscriptive design where people who are not skilled as a fire safety engineer can apply easily the fire safety code and get a good um safety level with um very uh easy applying uh measures or measures which can be easily um taken from the code and then implemented in the design.

00:14:09.360 --> 00:14:17.840
But you have also the possibility to demonstrate your safety um by deterministic or a risk-based uh performance-based design.

00:14:18.159 --> 00:14:24.720
Is it in power for any uh construction in Switzerland or just from a threshold?

00:14:24.720 --> 00:14:31.200
And does it apply to new builds or refurbishments and uh all other repurposing projects?

00:14:31.600 --> 00:14:42.559
No, it's uh really um an overall law for every building uh in Switzerland, except for buildings which are dealt in a different way on a federal basis.

00:14:42.559 --> 00:15:01.519
So, for example, uh if there are exist tunnel regulations um concerning fire safety from another party, uh then you have to take account of this regulation and then um the the fire safety regulations from the FAUCEF uh will not take that part in it.

00:15:01.519 --> 00:15:03.840
Which I believe already were risk-based.

00:15:03.840 --> 00:15:08.399
They are they are uh risk-based uh strategies uh for tunnel safety, yes.

00:15:08.879 --> 00:15:13.039
I had people praise those regulations in uh in my podcast already.

00:15:13.039 --> 00:15:13.519
Okay.

00:15:13.519 --> 00:15:18.320
So um we know how it is uh set, where is it?

00:15:18.320 --> 00:15:21.039
So let's talk about what's in it.

00:15:21.039 --> 00:15:27.759
Could you give me a quick overview of how the risk method that you applied looks like?

00:15:27.759 --> 00:15:29.840
You have like literally three minutes.

00:15:30.159 --> 00:15:30.480
Okay.

00:15:30.480 --> 00:15:33.840
Um maybe I have to say about the framework.

00:15:33.840 --> 00:15:41.360
From the framework, we have defined in a stakeholder process uh uh the the safety goods we want to address.

00:15:41.360 --> 00:15:48.480
So it's uh we will restrict it on life safety and building safety, uh building property safety, actually.

00:15:48.799 --> 00:15:51.679
As in not collapsing or as in business continuity?

00:15:51.679 --> 00:15:55.279
As a damage prevention in monetary terms.

00:15:56.240 --> 00:16:04.320
And for those two design objectives, we defined risk-based acceptance criteria, which are the basis for all design approach.

00:16:04.320 --> 00:16:08.080
So you can do a prescriptive or performance-based design approach.

00:16:08.080 --> 00:16:08.960
Doesn't matter.

00:16:08.960 --> 00:16:12.080
So you have always the same risk-based basis.

00:16:12.080 --> 00:16:27.039
The idea is that it doesn't matter which design approach you choose, uh, either a prescriptive or uh or a deterministic risk-based performance-based design, you will always fulfill the risk-based acceptance criteria defined in the stakeholder process.

00:16:27.039 --> 00:16:39.519
And either you show explicit compliance uh by a risk-based performance-based design, or you show implicit compliance by applying a prescriptive design or a deterministic performance-based design.

00:16:39.840 --> 00:16:47.360
Applying prescriptive means you already met the risk-based criteria by assumption, or is you have to demonstrate?

00:16:48.240 --> 00:16:58.159
If you think of a building or a general building, you can make a risk-based design, and then you you get the optimal set of measures you have to implement.

00:16:58.159 --> 00:17:06.079
And if you have a lot of those typical buildings, then you can just say, okay, uh, we are dealing in a prescriptive manner.

00:17:06.079 --> 00:17:12.000
For the those types of buildings, we are setting those optimized risk optimized measures.

00:17:12.000 --> 00:17:14.880
And this is basically a prescriptive rule.

00:17:14.880 --> 00:17:26.880
So for the prescriptive design, we derived uh risk-based optimized uh uh measures, and we also know in what terms the prescriptive design is valid for.

00:17:26.880 --> 00:17:35.920
So if we deviate or a building deviates from this typical building setup, then we have to move to a risk-based or performance-based design.

00:17:36.240 --> 00:17:45.279
Sophia, from your perspective, uh it's actually um at Mikro we have retail stores with more of a standardized uh design.

00:17:45.279 --> 00:17:49.279
And here the prescriptive design actually makes sense.

00:17:49.279 --> 00:17:54.960
What do you we also um as a big portfolio owner, we also have uh our own industry and logistics?

00:17:54.960 --> 00:17:57.920
So for these special buildings, we need another approach.

00:17:57.920 --> 00:18:02.720
The prescriptive design uh isn't actually the right approach always for us.

00:18:02.720 --> 00:18:10.559
We need sometimes to precisely calculate the safety level and see which safety measures bring best to a risk reduction.

00:18:10.880 --> 00:18:18.960
Would you in that case perform some sort of risk assessment for the baseline prescriptive solution and compare it with uh with the target?

00:18:18.960 --> 00:18:31.920
I mean, I ask this because you explicitly said at the beginning that uh you wanted to escape the equivalency concept, but it are there buildings in which you would consider both and measure like the outcomes of them?

00:18:31.920 --> 00:18:35.039
Like uh how how do you switch from one mode to another?

00:18:35.440 --> 00:18:37.039
Okay, it it depends.

00:18:37.039 --> 00:18:47.680
Uh sometimes uh we take into consideration what uh the needs of of all stakeholders are, from the owner, for example, what the needs for this usage of the building is.

00:18:47.680 --> 00:18:54.000
Industry buildings, sometimes there are safety measures that in the prescriptive design that actually don't work out.

00:18:54.000 --> 00:18:56.400
You just see that they don't work out well.

00:18:56.400 --> 00:19:02.799
Yeah, like in a freeze storage, maybe a sprinkler system is not the best solution.

00:19:02.799 --> 00:19:06.880
Because it will it will probably react a little bit late.

00:19:06.880 --> 00:19:11.839
So you need to look into other kinds of uh fire safety measures.

00:19:12.880 --> 00:19:22.720
So basically you find the prescriptive code like that requires you certain things for certain spaces, but the variety within makes those things probably an inapplicable.

00:19:23.359 --> 00:19:23.599
Exactly.

00:19:23.599 --> 00:19:31.039
It's a combination of uh experience uh at work and the needs and taking into account all stakeholders.

00:19:31.039 --> 00:19:37.440
Um can we say the operative interests we also have to take into account the operative risk uh interests?

00:19:38.079 --> 00:19:45.759
But do the authorities treat the solutions that they have to treat the solutions in the same way, whether they come from prescriptive or risk-based.

00:19:45.759 --> 00:19:53.279
So this kind of also enforces uh the people who are commissioning their approving the building to to accept it, right?

00:19:53.279 --> 00:19:55.920
Because I that that that's also a potential trap in such a system.

00:19:56.559 --> 00:20:02.400
Actually, um competence and education is a really big part of it, also from the for the new legislation.

00:20:02.400 --> 00:20:11.039
Is uh competence is crucial, both for engineers who perform the risk-based analysis and uh for the authorities who have to review them at the end.

00:20:11.359 --> 00:20:11.519
Yeah.

00:20:11.519 --> 00:20:20.240
I mean, uh yeah, this is such a it's it's not a simple change to change one document in law, it's like literally shaking the whole you know approach.

00:20:20.240 --> 00:20:21.359
I I I love this.

00:20:21.359 --> 00:20:22.000
Uh yeah.

00:20:22.000 --> 00:20:24.559
Jean-Luc, I wanted to ask about equivalency.

00:20:24.559 --> 00:20:33.039
So obviously, if you have two pathways, I understand Sofia's point on when uh a company would choose one or another, that makes complete sense.

00:20:33.039 --> 00:20:43.920
But I like you you said explicitly that the equivalency concept did not lead to efficient design, but there's always you know some sort of need, or maybe not a need.

00:20:43.920 --> 00:20:52.319
Uh someone would like to compare the solutions coming out from prescriptive system, a risk-based system, and and see which one's better, worse.

00:20:52.319 --> 00:20:53.599
How do you treat that?

00:20:53.839 --> 00:21:00.160
The level of safety now in the current code, in the current prescriptive code is defined by the prescriptive measures.

00:21:00.160 --> 00:21:07.680
And we can we have cases where we clearly see that the prescriptive setup is clearly not efficient.

00:21:07.680 --> 00:21:21.839
So if you are deviating from a non-optimal case and try to show equivalency by a risk-based or performance-based design, you end up with equal safety level, but you know that it is not really efficient anyway.

00:21:21.839 --> 00:21:25.039
So this equivalence approach is kind of flawed.

00:21:25.039 --> 00:21:34.400
So that's why we want to explicitly dealing the efficiency of the measures uh in our risk-based decision making.

00:21:34.400 --> 00:21:38.960
So in this way, we we ask for measures to be efficient.

00:21:38.960 --> 00:21:44.640
And in this way, we we set everything up from the prescriptive design to the performance-based design.

00:21:44.960 --> 00:21:57.119
But to know that the prescriptive measures are inefficient, okay, you can do it by your gut feeling and intuition and extreme experience that you you have in your work, and I respect that.

00:21:57.119 --> 00:21:59.440
But uh objectively you have to measure that.

00:21:59.440 --> 00:22:05.920
Where was there any you know a project to actually measure how much safety does the prescriptive give you?

00:22:06.079 --> 00:22:06.319
Yes.

00:22:06.319 --> 00:22:23.440
Well, we we applied in different case studies before the revision and looked at the efficiency of um of measures like sprinkler system, uh, how much they cost, how much they reduce the risk, and then we judged are they efficient for the setup uh of the prescriptive code?

00:22:23.440 --> 00:22:28.880
And we saw that in some parts they are efficient and in some parts they are not efficient.

00:22:29.200 --> 00:22:33.680
Was there any specific projects at which the previous approach spectacularly failed?

00:22:33.680 --> 00:22:35.519
Like it was unsustainable.

00:22:35.519 --> 00:22:43.119
We have some massive airports and you know, extremely large shopping malls, for example, are extreme stress for me in my system.

00:22:43.119 --> 00:22:45.279
So I wonder if you had the same for you.

00:22:45.599 --> 00:22:54.240
We saw that in some specific cases, um, some measure it's always that uh we have a very conservative approach.

00:22:54.240 --> 00:22:56.720
Or in general, it was too safe.

00:22:56.720 --> 00:23:07.759
That supported the perception of the the building owners that uh we have too high level of uh safety and we can reduce it to get a more efficient fire safety design.

00:23:07.759 --> 00:23:17.920
But we saw also that in some parts the the prescriptive design doesn't provide the optimal dissolution um as you would have in a risk-based design.

00:23:17.920 --> 00:23:22.000
So uh it's it's underperforming in in terms of safety.

00:23:22.000 --> 00:23:37.039
And in in this revision, we have a lot of uh deregulation, but we have also some parts where we have uh where we increase the need for measures and we um we get a higher safety level, but it's efficient to do that.

00:23:37.359 --> 00:23:49.359
And I would uh just uh I would like to add that it's uh sometimes for uh specific fire safety measures, it's clear that they are effective or they make sense, and sometimes it's just not clear.

00:23:49.359 --> 00:23:58.160
For example, what we did in this case study is we had a detection system, a sprinkler system, and a smoke exhaust system in the building, and we were not sure we need all of them.

00:23:58.160 --> 00:24:16.720
So in this quantitative analysis, we calculated the risk reduction for each of them and the cost effectiveness in the end, all these three systems, and uh, in the end, we decided for this specific building, the smoke exhaust system was not necessary, uh, definitely according to the new regulations.

00:24:16.720 --> 00:24:20.720
We wouldn't have to um install this as well.

00:24:20.720 --> 00:24:34.000
So it's sometimes it's not clear because uh the investment is too big, and you do a more precise analysis, a quantitative analysis to decide do I need this or do I need to do I don't need uh this one.

00:24:34.880 --> 00:24:39.279
To close on on the philosophy of the system, we'll move to case study in a second.

00:24:39.279 --> 00:24:45.279
Sophia, you mentioned that it came from canton insurance uh group or committee or uh of some sort.

00:24:45.279 --> 00:24:48.000
What what what's the role of insurance in all of this?

00:24:48.000 --> 00:24:51.359
Was like insurers supportive of this development?

00:24:52.319 --> 00:25:01.359
The framework in Switzerland, how they deal with fire safety is that uh the cantonal building insurance have um a kind of governmental task.

00:25:01.359 --> 00:25:12.880
They have to enforce fire safety, but they have also to provide insurance for all the buildings, so they can't choose uh which building they uh they want to insure, they have to ensure all the buildings.

00:25:12.880 --> 00:25:19.200
But in this way, they can also say what kind of uh safety they they want.

00:25:19.200 --> 00:25:22.079
So it's basically uh a hand in hand.

00:25:22.079 --> 00:25:30.720
Um they have uh kind of uh um a task to ensure the the buildings, but they can also say uh what kind of level uh safety level they want.

00:25:30.720 --> 00:25:33.359
That's broken down into very simple pieces.

00:25:33.759 --> 00:25:37.759
Yeah, I mean I'm asking this because you said uh it was potentially too safe.

00:25:37.759 --> 00:25:38.559
I I get that.

00:25:38.559 --> 00:25:41.119
I I get this all the time in my tunnels, for example.

00:25:41.119 --> 00:25:49.200
Some things are over-designed, and they I always say uh you just need a specific level of safety in your building.

00:25:49.200 --> 00:25:53.519
It's not supposed that you escape the building, return to it, and escape again.

00:25:53.519 --> 00:25:55.359
It's like you have to be safe once.

00:25:55.359 --> 00:26:02.799
And and anything above it, it just provides you safety layers for bigger hazards, which are more rare by assumption.

00:26:02.799 --> 00:26:08.319
So I guess from an insurer's perspective, it's not their money they spend on on the investments.

00:26:08.559 --> 00:26:09.920
So exactly.

00:26:09.920 --> 00:26:20.480
And and in this in this this framework, they have also the to prove because they are mandated from the federal federal body, let's put it like this.

00:26:20.480 --> 00:26:31.119
They are mandated from a federal body to ensure safety, and they have also to provide uh kind of uh a proof that they don't demand too safe solutions.

00:26:31.359 --> 00:26:31.920
Ah, okay.

00:26:31.920 --> 00:26:34.559
Wow, Switzerland is such an interesting country.

00:26:34.559 --> 00:26:36.160
This makes sense.

00:26:36.160 --> 00:26:40.319
Uh I would like to follow up also on the risk-based targets.

00:26:40.319 --> 00:26:47.519
So, because that's something I actually struggle with implementing risk-based approaches in Poland.

00:26:47.519 --> 00:27:06.319
I don't have many instances where I can do risk-based analysis in in my ecosystem, but actually in the world of tunneling, there are some spaces where through risk analysis I can demonstrate something or I can change a solution based on the outcomes of the risk analysis.

00:27:06.319 --> 00:27:12.240
However, I do not have risk targets in like explicitly mentioned in my code.

00:27:12.240 --> 00:27:15.359
I just I just it just says do risk analysis.

00:27:15.359 --> 00:27:16.000
Yes.

00:27:16.000 --> 00:27:20.480
You mentioned you have risk-based targets, what they are and how you got to them.

00:27:20.720 --> 00:27:26.480
Um, we have two design objectives: life safety and property um building property safety.

00:27:26.480 --> 00:27:33.039
And for those two design objectives, we defined risk acceptance criteria.

00:27:33.039 --> 00:27:35.680
For life safety, we defined two criteria.

00:27:35.680 --> 00:27:46.559
One is that we have to ensure for every building in our portfolio maximal risk that uh is deemed to be acceptable.

00:27:46.559 --> 00:27:55.279
And beyond that risk, if you take this hurdle, then you can perform a kind of economic optimization of your fire safety measures.

00:27:55.279 --> 00:28:04.559
So that in this way we ensure that every building in our portfolio meets a kind of uh uh a safety level which is deemed to be acceptable.

00:28:04.559 --> 00:28:10.319
Is it one number for all buildings, or it's it's one uh number for each user.

00:28:10.319 --> 00:28:20.319
So every user has has a kind of uh um fatality rate, which we have to prove in design that uh that uh this this discretion is met.

00:28:20.319 --> 00:28:25.039
And beyond that, we can perform monetary optimization of the measures.

00:28:25.359 --> 00:28:27.359
If you meet the maximum risk, okay.

00:28:27.599 --> 00:28:28.079
Yes, exactly.

00:28:28.079 --> 00:28:31.440
It's it's like uh a boundary condition that we have to take.

00:28:31.440 --> 00:28:35.200
And if we take this, then we can make economic optimization.

00:28:35.200 --> 00:28:40.640
And if we don't take this, we have to make measures not depending on the cost of the measures.

00:28:40.640 --> 00:28:43.119
So we have to to to get the safety level.

00:28:43.119 --> 00:28:45.519
Is this number written in the in the law?

00:28:45.519 --> 00:28:47.599
This number is written in the law, yes.

00:28:47.599 --> 00:28:48.559
What is the number?

00:28:48.559 --> 00:28:52.880
It's uh five times ten to the minus five per user per year.

00:28:53.200 --> 00:28:58.079
Okay, we we would be using one to ten to minus four, so you're having half of that.

00:28:58.079 --> 00:28:58.559
Yeah.

00:28:58.559 --> 00:29:01.119
Yeah, that that sounds uh that sounds uh reasonable.

00:29:01.359 --> 00:29:17.039
And um the way we get there, uh to the economic optimization I will come maybe later, but the way to get there, we had experience in other fields that they are also dealing with uh such criteria, um such absolute risk criteria.

00:29:17.039 --> 00:29:27.519
And if you set them too high, so if your hurdle for safety is too high for this minimal uh safety level, then you you don't allow economic optimization.

00:29:27.519 --> 00:29:42.559
And one criteria or one qualitative understanding how to set this value was that we don't want to limit the economic optimization and it should be really uh stated in in terms of yeah, it's not too conservative.

00:29:42.559 --> 00:29:44.720
It's it's a high target but achievable.

00:29:44.720 --> 00:29:45.759
Yes, exactly.

00:29:46.000 --> 00:29:53.759
Uh and uh was there any consideration about the value of life, quality of life to establish this number?

00:29:54.079 --> 00:29:56.640
For this number, not specifically.

00:29:56.640 --> 00:30:03.680
Um, it is basically we are in it to the On the safety level in other fields of safety.

00:30:03.680 --> 00:30:14.880
For the economic optimization, there the value of life or the social win is to commit social resources for safety is an essential part of the concept.

00:30:14.880 --> 00:30:16.240
Okay, that's one.

00:30:16.240 --> 00:30:17.200
You said there are two.

00:30:17.200 --> 00:30:18.240
So what's the other one?

00:30:18.240 --> 00:30:21.039
The other one is the marginal cost principle.

00:30:21.039 --> 00:30:24.319
It's in a framework with the marginal cost principle.

00:30:24.319 --> 00:30:30.000
So it's basically just cost and uh comparison between cost and uh risk reduction.

00:30:30.000 --> 00:30:40.160
So for building property safety, we defined uh for every monetary unit we invest into safety measures, we need one monetary unit of risk reduction.

00:30:40.160 --> 00:30:40.720
Uh-huh.

00:30:40.880 --> 00:30:42.720
So at least it evens out.

00:30:43.039 --> 00:30:47.200
Yeah, that's the that was the easy part uh for building property safety.

00:30:47.200 --> 00:31:05.279
And when it comes to life safety, uh the thing is a little bit complicated because if you perform a risk-based uh design or uh risk assessment, then you get the expected consequences as definition um of risk uh in terms of fatality per year, for example.

00:31:05.279 --> 00:31:09.920
And then you have to compare this number to a monetary number of the costs.

00:31:09.920 --> 00:31:20.240
And for this, we apply a concept which is uh known as the social willingness to pay or to commit social resources.

00:31:20.240 --> 00:31:28.160
Uh, and interestingly, this number is also set by a federal organization for spatial development.

00:31:28.160 --> 00:31:36.799
So it's not this organization is not dealing only with fire safety, but it is also dealing with earthquake safety, with uh flood safety, and so on.

00:31:36.799 --> 00:31:49.680
And they set this number uh in a in a general term to uh as an efficiency criteria to derive efficient measures in each level uh of safety, uh in each safety field.

00:31:49.680 --> 00:32:04.880
And the idea is is quite nice because uh if you think about it, why should we spend one Swiss franc in for fire safety if this uh Swiss franc can be more efficiently used in in earthquake safety or in flood safety?

00:32:04.880 --> 00:32:10.640
So it will provide in this field uh a higher benefit for society.

00:32:10.880 --> 00:32:22.640
Yeah, I mean I'm I'm not exactly sure if this is the exact concept I have in mind, but Ruben uh once taught me a concept that it's very hard to establish the value of life in your country.

00:32:22.640 --> 00:32:28.960
It's an extremely difficult discussion to have in your country if you want to uh establish what is the value of a human life.

00:32:28.960 --> 00:32:34.640
But at the same time, yeah, society is willing to invest a specific amount of money to save lives.

00:32:34.640 --> 00:32:39.519
Like it costs you uh 100 million euros to build a hospital, you can build it.

00:32:39.519 --> 00:32:41.519
If it costs 200, then perhaps we don't.

00:32:41.519 --> 00:32:51.039
So through decisions like that, society already made its mind on how much it's willing to invest to provide this uh life safety and and this social willingness.

00:32:51.039 --> 00:32:52.079
I love the name.

00:32:52.079 --> 00:33:01.599
So so if you have someone that actually gives you the number to guide you and it's based across different disciplines, then it's awesome that you have that.

00:33:01.839 --> 00:33:04.400
And uh, this is a whole research field actually.

00:33:04.400 --> 00:33:11.839
Um over the the 60 years, uh they have different approaches how to derive this number.

00:33:11.839 --> 00:33:21.279
Uh so if you go deeper in in this uh this topic, you find a lot of references um also in the national context, not only in Switzerland.

00:33:21.279 --> 00:33:26.480
So and also the OECD, so it's uh it's really an international level.

00:33:26.480 --> 00:33:31.920
They also uh recommend some specific values for different countries.

00:33:31.920 --> 00:33:34.000
And the numbers are there.

00:33:34.000 --> 00:33:39.839
Uh you come back of those um organizations and those methods.

00:33:39.839 --> 00:33:44.720
It's really the case what you want as a stakeholder, what do you want to achieve.

00:33:44.720 --> 00:33:49.200
And it took us two years in the stakeholder process.

00:33:49.200 --> 00:34:01.920
In this process, uh we we had uh over 20 stakeholders participating in discussion, what level of safety and what numbers we want to set for the new code.

00:34:02.240 --> 00:34:02.880
Perfect.

00:34:02.880 --> 00:34:09.280
Um Sophia, so in your case studies, you probably had to calculate all of those.

00:34:09.280 --> 00:34:11.199
I I assume that was the outcome.

00:34:11.199 --> 00:34:19.920
So maybe let's introduce the listeners to the case studies that you've run and talk about how you have applied this methodology and what were your findings.

00:34:19.920 --> 00:34:21.760
Let's start with with what they were.

00:34:22.159 --> 00:34:24.880
Okay, uh, well, it was a research project.

00:34:24.880 --> 00:34:32.960
We decided we our project team was uh Gianluca, me because I had the interest for my master's thesis as well.

00:34:32.960 --> 00:34:52.000
It was initiated by my colleague and mentor Thomas Laschett and under the technical supervision and guidance of Gianluca, and uh it was also financed from the insurance fire insurance company of Canad uh VOD, as I said before, and uh represented by DDA Gignard.

00:34:52.000 --> 00:34:57.199
So, Migro, as a large portfolio owner, was willing to initiate also this research project.

00:34:57.199 --> 00:35:00.000
It was one of our shopping centers.

00:35:00.000 --> 00:35:13.360
So the aim was to apply this new methodology and answer key questions that are um important to both planners from our point of view, uh, or the owner, and to the authorities as well.

00:35:13.360 --> 00:35:23.280
So, from a planner's perspective, uh it was critical to identify the data availability, what data do we need, and how accessible it is for us.

00:35:23.280 --> 00:35:31.280
Um, also the competencies, which skills and tools are required to conduct a quantitative analysis, a risk-based analysis design.

00:35:31.280 --> 00:35:38.079
How robust this concept is, how stable the the approach is when the conditions change, for example.

00:35:38.079 --> 00:35:45.119
And uh finally, how well it integrates into the planning process, how does it influence the project development overall?

00:35:45.119 --> 00:35:53.920
So, from the authority side, uh the focus was on the feasibility of the proposed design and to evaluate the plausibility of the overall approach.

00:35:53.920 --> 00:36:03.119
So, as I said before, it was kind of the perfect opportunity for me for my uh master thesis, and these were the key questions of the thesis as well.

00:36:03.119 --> 00:36:08.880
And uh we conducted this risk-based performance-based design for the risk modeling.

00:36:08.880 --> 00:36:10.719
We use Bayesian networks.

00:36:10.719 --> 00:36:15.840
This is a graphical model that represents the probabilistic relationships between variables.

00:36:15.840 --> 00:36:26.079
We it's um like a simple fault tree, but unlike one, we have multi we can reflect, we can portray uh multiple dependencies at once.

00:36:26.079 --> 00:36:36.559
So it's a more complicated model, but we can include a lot of information like uh the fire protection fire safety measures we want to analyze.

00:36:36.559 --> 00:36:58.320
The ones I mentioned before, the detection system, the sprinkler, and the smoke exhaust system, the effectiveness of these measures, like the probability of failure or the probabilities that they are inactive, uh the fire brigading's response, also the fire development, the fire characteristics, such as the fire growth rate.

00:36:58.320 --> 00:37:26.719
We took into account multiple fire growth rates with their relevant probability of occur occurrency in retail buildings specifically, and also building usage characteristics, like the occupancy density is different at night and in the middle of the day, or maybe in a very rare occasion, such as Christmas, we took all this different information in this we took into account in the model in the modeling all these characteristics.

00:37:26.719 --> 00:37:34.639
So, in the end, we calculate the risk for a life safety and property safety for the extent of damage.

00:37:34.639 --> 00:37:40.639
We also used uh modeling and we did we performed A set airset uh comparisons.

00:37:40.639 --> 00:37:53.599
Um we I calculated the available and required safe aggressive time multiple times for different scenarios, it was spread, and for the building, we did we took into account the extent of damage in the building.

00:37:53.599 --> 00:38:16.719
So we conducted the whole probability analysis and consequence analysis, and we used the risk acceptance criteria, the way Gianluca described them before, and to realize at the end that to achieve this acceptability line, this minimum acceptability line for the people's safety, we just needed one of these three technical measures.

00:38:16.719 --> 00:38:21.920
For example, the most the cheapest one, uh the detection system in this specific case.

00:38:21.920 --> 00:38:27.440
But uh with the marginal cost principle, we also need to apply more.

00:38:27.440 --> 00:38:39.440
And in the end, uh a combination of a detection system and a sprinkler system for this building was uh the perfect combination, we could achieve the the greatest risk reduction, and it was cost effective.

00:38:39.840 --> 00:38:49.280
So basically, with your first approach, any of the three would give you the cut threshold of five to ten minus one five.

00:38:49.280 --> 00:39:11.199
Uh so you would get the baseline, and uh here you are in the economic optimization, and then using the marginal cost principle, you basically established that it is efficient to add another measure in this case, uh, the sprinkler system, but the gain is not enough to justify installing the third one, which was the smoke control in your case.

00:39:11.440 --> 00:39:18.719
Yes, and that was because uh in the specific building, it was an old industrial building with a ceiling height of seven meters.

00:39:18.719 --> 00:39:18.880
Yeah.

00:39:18.880 --> 00:39:26.079
And it was kind of obvious that the smoke exhaust system in this specific case was not very effective.

00:39:26.079 --> 00:39:37.360
But in another building, uh like the standard shopping centers that we have with four meters height, it definitely is something we have to consider, so it's not very obvious.

00:39:37.360 --> 00:39:49.760
So it was a specific situation that gave us these interesting results, and we could consider reducing the um fire compartments, so we could maybe uh don't have a sprinkler system in the building.

00:39:49.760 --> 00:39:57.360
You can also see uh where you can apply different safety measures and which factors uh um affect the results.

00:39:57.679 --> 00:39:58.960
Fantastic, really, really good.

00:39:58.960 --> 00:39:59.920
This is very interesting.

00:39:59.920 --> 00:40:06.480
Gianluca, from the risk model perspective, Sophia mentioned she used Bayesian networks.

00:40:06.480 --> 00:40:11.039
Is the tool also defined in your law, or it just says you have to do risk analysis?

00:40:11.039 --> 00:40:14.880
How deep does it go in defining how the risk analysis should be performed?

00:40:15.199 --> 00:40:19.599
Uh we set rules for the risk-based approach that you have to follow.

00:40:19.599 --> 00:40:23.280
For example, you have to follow the rules of probability theory.

00:40:23.280 --> 00:40:30.000
So you can't invent, for example, or you can't use, for example, the greater method, which is our index uh-based uh method.

00:40:30.000 --> 00:40:37.360
Anymore, so you have to define uh your system in the means of uh probabilities and quantify all the uncertainties.

00:40:37.360 --> 00:40:42.480
And this was one is one example among different uh rules.

00:40:42.480 --> 00:40:49.119
And we also defined in the in the document some basic parameters for the risk-based design.

00:40:49.119 --> 00:40:50.800
It's called the model code.

00:40:50.800 --> 00:40:59.599
The model code defines basically the parameters you have to consider in in your risk-based design, uh, but you can also deviate if you know it better.

00:41:00.079 --> 00:41:06.800
Do you define the trustworthiness of the sources that tell you you know better?

00:41:06.800 --> 00:41:10.880
Like, can you do you have to base it on peer-reviewed uh research?

00:41:10.880 --> 00:41:13.360
Do you can you base it on a podcast episode?

00:41:13.360 --> 00:41:14.239
I don't know.

00:41:16.880 --> 00:41:18.960
For the podcast, I'm not so so sure.

00:41:18.960 --> 00:41:20.639
I I would not recommend that.

00:41:20.639 --> 00:41:23.920
Like well, you can deviate in uh in a specific project.

00:41:23.920 --> 00:41:37.679
So if you if the model code is not giving you additional information or basic information on on some specific point and specific issues we which you have in in the project, then you can also deviate.

00:41:37.679 --> 00:41:40.079
And then you can uh go to the literature.

00:41:40.079 --> 00:41:52.159
Um, I mean then you you enter our daily engineering life, how you do you combine different information, put them together and um assess or perform the whole design.

00:41:52.159 --> 00:42:02.400
But if you are doing so, if you deviate, then you always have to um also to model the all the uncertainties in your deviation appropriately.

00:42:02.400 --> 00:42:08.079
So that means that you have to perform sensitivity analysis and also parametric analysis and so on.

00:42:08.320 --> 00:42:14.800
Is there in the end some uh responsible person, an officer of some sort that approves or disapproves that?

00:42:15.280 --> 00:42:15.519
Yes.

00:42:15.519 --> 00:42:18.159
In Switzerland it's quite a good system.

00:42:18.159 --> 00:42:20.239
Uh we established over the years.

00:42:20.239 --> 00:42:23.119
So we have a kind of a peer review system.

00:42:23.119 --> 00:42:32.239
So every design, prescriptive or also performance-based design, is getting through the authorities and they have to uh approve the design.

00:42:32.239 --> 00:42:44.880
And this was a large discussion how the authorities can approve uh risk-based design, because the knowledge how to perform risk-based analysis is not that broad in Switzerland yet.

00:42:44.880 --> 00:42:50.559
So um this is uh one of the major tasks we have also to face in the future.

00:42:50.559 --> 00:42:53.920
But there are also existing some specific solutions to it.

00:42:54.159 --> 00:43:04.800
Yes, if you already mentioned that the competencies is critical if you want to establish system like that, not only within the designers but also within the authorities who have to go through that.

00:43:04.800 --> 00:43:09.039
Uh in the modeling itself, those have to be deterministic.

00:43:09.039 --> 00:43:12.239
I assume you said the no indexing methods, uh etc.

00:43:12.239 --> 00:43:16.719
But do you specify approaches like ACR set or okay?

00:43:16.960 --> 00:43:23.840
Um ACR set is one way to uh assess the consequences uh which you have to consider in the risk assessment.

00:43:23.840 --> 00:43:34.719
We don't specify how you you have to do um AZR set analysis because this is um we don't we didn't want to write a textbook how to perform your design.

00:43:34.719 --> 00:43:44.880
This is uh built in another part in a different code, but we wanted to just set the main driver which which drive the risks.

00:43:44.880 --> 00:44:06.960
So, for example, we specify occurrence rate in the in the code, um, in order that everyone is doing the same to specific occupation or it is uh tied to the specific occupation classes, um and then you have the possibility also to adjust in um in a way to your project-specific object.

00:44:06.960 --> 00:44:22.320
And then we specify, for example, uh also the um fire spread rates or the probabilities which are associated with fire spreading, uh like fire growth time, a completely probabilistic distribution of the fire growth time.

00:44:22.320 --> 00:44:31.360
Um we specify, for example, uh the burning behavior of your fire load, um, the fire load itself as a probabilistic distribution, and so on.

00:44:31.360 --> 00:44:39.599
So if you look in there, you find a lot of information about how to describe the uncertainties in your system.

00:44:39.599 --> 00:44:52.000
And starting from the the fire occurrence over the efficiency of the measures, but also about model uncertainties and also fibergate intervention, um, is also handled there.

00:44:52.480 --> 00:44:55.840
So if you are when applying this in practice, uh, how useful was there?

00:44:55.840 --> 00:44:59.679
Did you have found some of them which were problematic to you?

00:44:59.679 --> 00:45:02.079
Did you have to seek further?

00:45:02.079 --> 00:45:04.480
What's your assessment of those?

00:45:04.960 --> 00:45:19.119
Uh well, actually, it was very interesting because at the time when um I was conducting the the uh the case study or the probabilistic analysis, John Luca was also parallel developing this model code, these guidelines.

00:45:19.280 --> 00:45:19.679
Uh-huh.

00:45:20.559 --> 00:45:30.000
So it was a direct apply, and it it really helped me to understand why we need this approach in every single step of the way of the analysis.

00:45:30.000 --> 00:45:31.920
Why do we need these spreads?

00:45:31.920 --> 00:45:38.800
Why do we need these distributions in order we had to uh model these um uncertainties?

00:45:39.280 --> 00:45:47.679
I mean, you you've mentioned some of the most difficult to find, like occur occurrence rates, efficiency factors for uh systems.

00:45:47.679 --> 00:45:55.280
Wow, this is like this is a very difficult statistic to have a very solid, like science-based grasp on.

00:45:55.280 --> 00:46:00.880
Almost like I would risk this is almost impossible to get at the level we would like to have it.

00:46:00.880 --> 00:46:13.360
So I I but I I see a value if this is defined in the code, you may not be closest to the truth, but at least everyone is lying at the same rate.

00:46:13.360 --> 00:46:14.320
Exactly.

00:46:14.800 --> 00:46:15.280
Exactly.

00:46:15.519 --> 00:46:40.159
And we are, I mean, to to today, uh our today's approach is that we in a performance-based design, we just set a specific scenario, and then we are performing our uh performance design design based on this scenario, and uh selecting this this uh design scenario is is complicated as well, and you can't also uh state in the in the in the codes.

00:46:40.159 --> 00:46:53.599
There are codes that state you have to consider this specific uh scenario, but uh in some times um it's okay, and in some times for some buildings, um you can't apply this scenario because it doesn't make sense.

00:46:53.599 --> 00:46:57.679
So um and in this way you have the same problems actually.

00:46:57.679 --> 00:47:22.559
So uh and what we we did also in the in the model code is that we relying on the on on data, so for example, the occurrence rate in Switzerland, we have uh a pretty good understanding how often a fire occurs, because we had data from the insurer and they uh track every fire because every fire with uh financial damage, they they uh assess because they insure it.

00:47:22.559 --> 00:47:29.039
So uh we have very good databases for this specific topic, but you you're right.

00:47:29.039 --> 00:47:56.559
In in other parts, for example, uh in the efficiency or the effectiveness of some kind of fire safety measures, we had less good data, and then we had to make some estimates also, but always in the context that we are not interested in having a very exact or the precise risk assessment, because the the aim is having a robust decision about what kind of fire safety measures we have to intake.

00:47:56.559 --> 00:48:04.320
So in in this way, we can say, okay, if you are using the those uh numbers, everyone is doing the same stuff.

00:48:04.320 --> 00:48:10.400
So everyone doing is uh is wrong, but uh everyone doing is the same stuff.

00:48:10.400 --> 00:48:13.119
Everyone is wrong at the same extent, exactly.

00:48:13.119 --> 00:48:21.840
But it's not really worse than uh compared to uh a deterministic performance-based code where you specify some design criteria.

00:48:21.840 --> 00:48:24.400
So uh yeah, that that was a trade-off.

00:48:24.559 --> 00:48:28.480
Yeah, I'd also like to add it's exactly that what Gianluca said.

00:48:28.480 --> 00:48:32.480
We have these uh factors where we are uncertain in the model.

00:48:32.480 --> 00:48:34.000
So what can we do?

00:48:34.000 --> 00:48:38.480
We can take a more conservative approach on them, we can play with the model.

00:48:38.480 --> 00:48:43.599
I had the opportunity to play with this modeling and see how they affect the outcome.

00:48:43.599 --> 00:48:46.800
Also, the the conservative approach, how does it affect the outcome?

00:48:46.800 --> 00:48:52.880
And in the end, I have a robust design which is not so much affected from its of its characteristics.

00:48:53.280 --> 00:48:54.559
I found it very important.

00:48:54.559 --> 00:49:04.239
Like if I compare it to the Polish context, you also don't want engineers to compete on design fires, you don't want you know engineers to compete on suit yield.

00:49:04.239 --> 00:49:14.719
If I have a suit yield of 0.1 and someone else has 0.05, and they have half of the system I do, well, that's kind of obvious, isn't it?

00:49:14.719 --> 00:49:15.280
Right?

00:49:15.280 --> 00:49:22.559
Uh but it's kind of nuanced, and the end client may not recognize the difference, you know.

00:49:22.559 --> 00:49:26.880
And I my system is perceived as the worst because it's so much bigger.

00:49:26.880 --> 00:49:27.840
What's the reason?

00:49:27.840 --> 00:49:33.280
Well, there is a reason, but they perhaps never had a chance to actually fully comprehend the difference.

00:49:33.280 --> 00:49:34.880
So I I appreciate that, yeah.

00:49:34.880 --> 00:49:35.199
Yeah.

00:49:35.199 --> 00:49:42.800
Okay, so uh so uh the final question the is this all you said it's in the comments, so I assume it's not yet in power, or is it in power?

00:49:43.119 --> 00:49:45.760
Now it's uh under technical uh revision.

00:49:45.760 --> 00:49:57.599
Uh so everyone in Switzerland can um say, okay, yeah, this part of the of the new fire safety code is uh is wrong or it has to be be improved, or um yes.

00:49:57.599 --> 00:50:18.960
So that's the the technical revision is going up to the January 15th, and then it's going to the to the Faukaf, and then we have to revise the code, and then it's uh there is a political uh review, and then from 2027 it is in no just in time for a Geneva conference.

00:50:18.960 --> 00:50:22.719
It's uh right before the Geneva Conference, the SFP.

00:50:22.719 --> 00:50:24.400
Uh Europe Conference.

00:50:24.400 --> 00:50:26.960
So that will be a very good celebration.

00:50:26.960 --> 00:50:32.480
Does it have to go through some public vote or something?

00:50:32.559 --> 00:50:34.400
What was the political pathway for the project?

00:50:34.639 --> 00:50:37.840
There is a political um instance that uh revised the code.

00:50:37.840 --> 00:50:43.519
So there is a political approval needed for getting through the whole process, yes.

00:50:43.840 --> 00:50:55.599
While doing that, are you still working on uh some sort of improvements, or you're still running case studies to see how to implement it, or maybe you can apply some of it through derogations or something?

00:50:55.920 --> 00:51:00.880
In the current code, we are not allowed to apply the risk-based design.

00:51:00.880 --> 00:51:02.559
Hence, you need the new one.

00:51:02.559 --> 00:51:07.360
Yes, that's but for only for uh public safety goods.

00:51:07.360 --> 00:51:36.480
But for private safety goods, uh, for example, like Sophia is dealing with um business continuity and so on, uh, there we can also uh deal with uh this risk, and then it helps also to look in the model code, see um what are the the effectiveness or the raility of the uh measures, and then we can um incorporate this um these numbers in the in in the risk assessment for private safety goods.

00:51:36.960 --> 00:51:38.559
It's um uh thank you, John Luca.

00:51:38.559 --> 00:51:41.360
It's uh it's uh what I also mentioned in the beginning.

00:51:41.360 --> 00:51:48.639
Uh we have now the foundation for its risk management, and we are trying to implement it also for the private safety goods.

00:51:49.519 --> 00:51:55.840
Have you already considered some sort of revision cycle or like a living document for a pathway for this?

00:51:56.159 --> 00:51:59.599
Well, fire safety codes are revised every 10 years.

00:51:59.599 --> 00:52:05.440
This is quite long, but minor revisions can be taken um also in a short shorten period.

00:52:05.440 --> 00:52:10.079
There is uh a committee who is dealing with um minor changes in the codes.

00:52:10.079 --> 00:52:10.320
Yeah.

00:52:10.639 --> 00:52:14.239
Yeah, so just for SFP Tzurich in 2037, good.

00:52:14.239 --> 00:52:14.880
Exactly.

00:52:14.880 --> 00:52:18.639
And uh well, one final question to Sophia.

00:52:18.639 --> 00:52:23.440
Did this experience change the way how you approach your projects in your everyday job?

00:52:23.440 --> 00:52:28.079
I think exposure to risk uh changes your perspective even on the prescriptive code.

00:52:28.400 --> 00:52:30.960
Definitely, it was such a valuable experience.

00:52:30.960 --> 00:52:33.440
Also, uh, I learned new methodologies.

00:52:33.440 --> 00:52:45.679
This uh actually, this Master of Advanced Studies at the ETH uh was a big um opportunity for me to gain knowledge in all engineering methods in fire safety as well as probabilistic modeling.

00:52:45.679 --> 00:52:54.000
And on my everyday job, definitely, I definitely implement the things we I learned, and we are also um on this uh methodology.

00:52:54.000 --> 00:53:00.000
We as I said, we work on it on the for the private safety goods, and the results are very, very interesting.

00:53:00.400 --> 00:53:17.119
This was one of the strategic strategic components, this MIS at ETH, uh, where we we in 2015 we we saw that the direction is going to risk-based codes, and we thought, okay, um, how we we can ensure the competency of people.

00:53:17.119 --> 00:53:22.800
And one way to to get the basis is uh by education, and this is a very important step.

00:53:22.800 --> 00:53:45.039
And the the the whole uh Master of Advanced Studies, it's only it's not a fully master, it's a master of advanced studies, has uh five models, and one model is specifically uh treating uh probability and statistics and risk-based design and um performance-based design in a in a general context.

00:53:45.039 --> 00:53:59.440
And one of five models we dedicated to just risk-based design because we we we realized that we need this uh the knowledge for the upcoming fire safety engineers in Switzerland uh of this risk-based design.

00:53:59.840 --> 00:54:05.199
And now I fake being surprised you've strategically done this like as expected.

00:54:05.199 --> 00:54:06.960
No, but this is a great decision.

00:54:06.960 --> 00:54:15.599
Like I think it was Professor Arnold Dix who told me he loves the Swiss approach to risk the most because it's very, you know, like efficient.

00:54:15.599 --> 00:54:20.559
I um he perhaps used the word efficient, but I think efficient is what he had in mind.

00:54:20.559 --> 00:54:30.239
And and I agree, this is it's not just about writing a code, it's about changing the system and acting on all of the fillers.

00:54:30.239 --> 00:54:38.719
If you if you did change the system and you did forget about educating new generation of engineers who can use it, not good.

00:54:38.719 --> 00:54:44.480
If you change the system and you have not said what kind of tools you want, not gonna work.

00:54:44.480 --> 00:54:48.000
No risk-based targets, you're gonna have a horrible time.

00:54:48.000 --> 00:54:52.800
Yeah, uh allow people to choose their own occurrence rates, it's not gonna work.

00:54:52.800 --> 00:55:00.960
So I think uh there will be things that work better and worse in the system for sure when you start applying it everywhere.

00:55:00.960 --> 00:55:09.760
And I guess this this revision cycle gives you opportunity to improve, but I think your starting point is is really, really sound and good.

00:55:09.760 --> 00:55:12.960
And and congratulations to everyone involved in this.

00:55:12.960 --> 00:55:32.639
Uh, I remember at the SFP Edinburgh conference there was uh a quite lengthy and and quite challenging questions and answers session, which uh you have defended yourself quite well back then, but I'm also pretty sure we have answered a lot of those questions in a more elaborate way in here.

00:55:32.639 --> 00:55:34.719
So I'm I'm extremely pleased with that.

00:55:34.719 --> 00:55:36.639
So I leave the last words to you.

00:55:37.119 --> 00:55:39.280
So thank you, Wojciech, for this opportunity.

00:55:39.280 --> 00:56:01.840
I think um this this uh also from my point of view, uh when I could reach how fire safety should look like, I think the Swiss approach is going in the is a very uh a close point to my perfect world of risk-based design and also how uh in the future we have to deal with uh fire safety.

00:56:01.840 --> 00:56:05.599
So it's uh it's a very nice development in Switzerland.

00:56:05.599 --> 00:56:13.039
And um, I I loved the that we we get the opportunity to to show the audience now to um how can it be done?

00:56:13.199 --> 00:56:26.719
And there are a lot of other opportunities to get into in this uh the only thing that worries me, which is not a worry for you, but the thing that worries me is could this be implemented in a place which is not Switzerland?

00:56:26.719 --> 00:56:29.679
No, I I I mean you you know you know what I mean.

00:56:29.679 --> 00:56:34.800
Like yeah, all the puzzles that are in place that make complete sense for you.

00:56:34.800 --> 00:56:45.280
I see you know traps, uh holes with spikes, uh, you know, waiting for me if I wanted to do like it's it's way beyond implementing a risk code.

00:56:45.280 --> 00:56:47.760
So but you already did actually.

00:56:48.000 --> 00:56:48.639
Yeah.

00:56:48.639 --> 00:56:57.199
Uh if you look at the structural fire safety code, uh Eurocode AN 909112.

00:56:57.199 --> 00:57:09.440
This approach, um, which we want to uh set in fire safety now for in general, this has been already done in structural fire safety engineering.

00:57:09.440 --> 00:57:15.119
They derived a deterministic design from probabilistic criteria.

00:57:15.119 --> 00:57:27.199
The deterministic design is used to derive tabulated data uh for architects, which can um for standard buildings, they can just pick the numbers from the tablet data.

00:57:27.199 --> 00:57:36.559
But and what a lot of people don't know, those performance criteria stated in the Eurocodes, they are derived for by the marginal cost principle.

00:57:36.559 --> 00:57:47.280
So it's exactly this this whole setup, and uh in the background you have also a model code which specified all the uncertainties in the structural engineering.

00:57:47.280 --> 00:57:57.440
So, in this this kind of way, this this whole approach is not really new, it is really inspired by the structural engine and engineering uh approach.

00:57:58.000 --> 00:58:26.880
I agree with you, but still when you zoom out from one branch of fire safety, like structural, into the entire fire safety, you suddenly can replace the passive fire protection through sprinklers or perhaps by combination of smoke alarm and vents, if you can prove through your analysis that that's the efficient way, and suddenly you are touching economical incentives of so many players around.

00:58:26.880 --> 00:58:30.480
That's uh that that that that's a completely different anyway.

00:58:30.480 --> 00:58:36.079
Um we've deviated so Sophia, uh what's uh what's your final message to the audience?

00:58:36.480 --> 00:58:46.159
I think um to your point before I I think it cannot be implemented one-to-one, but I think it definitely has potential, this uh risk management approach.

00:58:46.159 --> 00:58:58.880
And it enables tailored solutions for our buildings be beyond prescriptive regulations and as I said, targeted investments in fire protection with uh truly cost-optimized solutions.

00:58:58.880 --> 00:59:00.639
That's actually my point.

00:59:00.639 --> 00:59:08.079
It was um such a pleasure to talk about it, about um my work on this, and it was just a small part in this big project.

00:59:08.079 --> 00:59:10.400
So, yeah, it was such a pleasure to be here today.

00:59:10.400 --> 00:59:11.039
Thank you.

00:59:11.440 --> 00:59:12.400
No part is small.

00:59:12.400 --> 00:59:15.519
Thank you, thank you guys for for coming to the Far Science Show.

00:59:15.519 --> 00:59:15.920
Thank you.

00:59:16.159 --> 00:59:16.480
Thank you.

00:59:16.719 --> 00:59:17.360
Yeah, that's it.

00:59:17.360 --> 00:59:18.079
Thank you for listening.

00:59:18.079 --> 00:59:20.719
So, what do you think about the changes to the Swiss Parco?

00:59:20.719 --> 00:59:25.679
Do you think this could be implemented somewhere that is not Switzerland?

00:59:25.679 --> 00:59:38.800
I really, really wondered uh at the end of the episode whether this is something that he could just translate to a different country and apply there successfully, because it's not just about picking a risk methodology.

00:59:38.800 --> 00:59:46.480
It's not about, you know, setting your law requirements that force you to do a specific pathway in your project.

00:59:46.480 --> 00:59:50.719
It's also not about just finding equivalency between solutions.

00:59:50.719 --> 00:59:58.159
Uh it's a lot, a lot more, as you have heard in this episode, it's about finding prescriptive solutions that work based on the risk principles.

00:59:58.159 --> 01:00:03.599
So uh following the Prescriptive pathway actually fulfills the risk-based goals.

01:00:03.599 --> 01:00:16.320
It's about setting the targets, setting the goals in such a way that they provide sufficient level of safety that they are happy with, but at the same time do not kill the economic incentives to optimize.

01:00:16.320 --> 01:00:19.280
Look for better solutions.

01:00:19.280 --> 01:00:26.239
It's about promoting fire solutions that uh actually make sense economically.

01:00:26.239 --> 01:00:29.199
So you reduce risk when when it makes sense.

01:00:29.199 --> 01:00:30.880
I really, really like it.

01:00:30.880 --> 01:00:36.159
But so many things that you have to have in place to enable this.

01:00:36.159 --> 01:00:41.440
And I think uh in in in Switzerland, the culture is actually very good to apply this.

01:00:41.440 --> 01:00:45.599
I I wonder if you could translate the culture of the risk-mease method.

01:00:45.599 --> 01:00:54.960
I'm keen to observe how will they apply this method in practice, how will it evolve over the years, what will be the changes, what will be the feedback of the real users.

01:00:54.960 --> 01:01:06.880
I'm I'm I'm also quite happy that uh while developing the model, they already do case studies like the ones that Sophia presented, because this allows them to uh double check everything while it's being built.

01:01:06.880 --> 01:01:12.559
But uh the real world will bring so many more case studies that they will learn a lot more about the implementation of the method.

01:01:12.559 --> 01:01:16.400
I'm looking forward to learn what they learn from from that aspect.

01:01:16.400 --> 01:01:25.280
But for now, I'm really amazed about uh how this new approach was developed, and I am looking forward to to see it in practice.

01:01:25.280 --> 01:01:40.639
If you'd like to check it out, uh there's a link in the show notes and it's in German and French, but in the age of AI, I don't think that will stop you if you are curious enough to see uh what's inside, and let's see how where we get with that.

01:01:40.639 --> 01:01:51.039
If there are more news from Switzerland, if there are some experiences from implementation of the materials, you can be sure that you will find them in the fire science show as well.

01:01:51.039 --> 01:01:56.639
Thank you very much for being here with me today, and I'm looking forward to see you here next Wednesday.

01:01:56.639 --> 01:01:57.760
Thank you, bye.