184 - Cost-benefit analysis in structural fire safety with Thomas Gernay and Chenzi Ma

Transcript

WEBVTT

00:00:00.261 --> 00:00:02.107
Hello everybody, welcome to the Fire Science Show.

00:00:02.107 --> 00:00:08.727
Today we're talking about money Money in fire safety, or rather how much fire safety costs.

00:00:08.727 --> 00:00:26.381
From my guests today I've learned that the cost of fire safety in a building could be up to 12% of the building costs, and that actually is in line with a small survey that we've done many years ago asking some fellow directors of construction companies who said it was like bulk 10% of their buildings.

00:00:26.381 --> 00:00:28.126
So in fact, fire safety costs a lot.

00:00:28.126 --> 00:00:30.472
Everyone knows that we don't save on fire safety.

00:00:30.472 --> 00:00:34.911
That's what a lot of people tell me but in the end everyone has a finite amount of resources.

00:00:34.911 --> 00:00:38.567
It's not possible or feasible to do everything you want in a building.

00:00:38.567 --> 00:00:44.329
If we don't save on safety, then why, in Poland, is so hard for me to get sprinkles in my buildings?

00:00:44.329 --> 00:00:46.804
In fact, in some cases we are saving on safety.

00:00:46.804 --> 00:00:53.347
We're just not saving on things that are very difficult to go around or require a ridiculously long process to approve.

00:00:53.347 --> 00:01:01.932
Now in today's episode, my guests Professor Thomas Gernay and his third-year PhD student Chenzi Ma from Johns Hopkins University.

00:01:01.932 --> 00:01:05.310
They've tried to answer the question how much fire safety costs.

00:01:05.310 --> 00:01:19.632
They are looking into passive fire safety, into structural fire safety of composite concrete floors, which means they're looking into things like spray motors and intumescent paints and overall savings on your structure, but also looking into structural stability.

00:01:19.632 --> 00:01:30.480
So perhaps you can, instead of applying a sprayars for two hours, maybe you can increase the reinforcement in your concrete floor and provide the same value in terms of fire safety.

00:01:30.480 --> 00:01:43.835
They look into monetary value of those works, including a lot of interesting elements such as maintenance of that structure, such as lifecycle, cost, sustainability, workload, etc.

00:01:43.835 --> 00:02:01.727
You learned about that in the episode, but in the end, what you get is some extent, let's say, unbiased it's always biased, but this one is really clean image of how much do we spend on providing this particular aspect of fire safety, which is structural fire safety, and I love it.

00:02:01.727 --> 00:02:02.831
I love it because it allows you to look on your code.

00:02:02.831 --> 00:02:05.099
If you take their framework, it allows you to look on your code.

00:02:05.099 --> 00:02:12.188
If you take their framework, it allows you to look on your code and say how costly is your country and what can you do better?

00:02:12.188 --> 00:02:21.350
And it's something that they absolutely have to do for my country, because I feel that we've reached a point where we really could spend our money better.

00:02:21.350 --> 00:02:33.312
What Chenzi developed in here is a brilliant simulator available online that allows them to compare hundreds of buildings and have some higher level view on the stuff.

00:02:33.312 --> 00:02:44.026
So, even if you don't try to apply this in your country, I hope this discussion is valuable for you and, in general, it allows us to reflect on how much do we spend on safety.

00:02:44.026 --> 00:02:49.816
So please let me welcome Professor Thomas Gernay and Chenzi Ma from Johns Hopkins University.

00:02:49.816 --> 00:02:52.207
Let's spin the intro and jump into the episode.

00:02:56.841 --> 00:02:58.387
Welcome to the Fire Science Show.

00:02:58.387 --> 00:03:20.435
My name is Wojciech Wegrzynski and I will be your host.

00:03:20.435 --> 00:03:24.997
This podcast is published in partnership with OFR, a multi-award-winning independent consultancy dedicated to addressing fire safety challenges.

00:03:24.997 --> 00:03:36.652
Established in the UK in 2016 as a startup business of two highly experienced fire engineering consultants, the business has grown phenomenally to eight offices across the country, from Edinburgh to Bath.

00:03:36.652 --> 00:03:46.211
Colleagues are on a mission to continually explore the challenges that fire creates for clients and society, applying the best research, experience and diligence for effective, tailored solution.

00:03:46.211 --> 00:03:50.145
In 2025, there will be new opportunities to work with OFR.

00:03:50.145 --> 00:03:58.026
Ofr will grow its team once more and is keen to hear from industry professionals who would like to collaborate fire safety safety features this year.

00:03:58.026 --> 00:03:59.582
Get in touch at ofrconsultants.

00:03:59.582 --> 00:04:01.068
com.

00:04:01.068 --> 00:04:08.616
Hello everybody, I'm here today with two guests from Johns Hopkins University Professor Thomas Gernay hey, thomas, good to have you back in the podcast.

00:04:08.837 --> 00:04:10.725
Hello, wojcic, thank you very much for having me.

00:04:11.701 --> 00:04:16.690
Great to have you, as always, and Chenzi Ma from the same university.

00:04:16.690 --> 00:04:17.884
Hey, chen, good to have you on the show.

00:04:17.884 --> 00:04:20.107
Hi, wojcic, thanks for the invitation.

00:04:27.779 --> 00:04:32.759
It's my great pleasure to be here, and the subject of today's discussion is fire protection costs in general, because you've performed a really outstanding cost-benefit analysis of various fire designs.

00:04:32.759 --> 00:04:36.230
It's something we wouldn't really do that much in the fire safety.

00:04:36.230 --> 00:04:39.889
I have this experience in Poland like no one saves on fire safety.

00:04:39.889 --> 00:04:54.769
We spend whatever money is to be spent on fire safety, but actually when you start thinking about what benefits different aspects of fire safety bring you to the building, no one really knows like how much $1 of spends, how much fire safety that gives you.

00:04:54.769 --> 00:04:56.694
And here comes your paper.

00:04:56.694 --> 00:05:06.745
I assume this is part of a bigger project, so perhaps, thomas, you would like to introduce where does this come from and what inspired you to do this type of really complicated study?

00:05:07.259 --> 00:05:08.406
Yes, absolutely.

00:05:08.406 --> 00:05:20.249
So it's really based on two realizations it's, on the one hand, that as a society we are actually investing a lot in passive fire protection, in protecting our built environment from fire.

00:05:20.249 --> 00:05:26.951
So if you are spending a lot of money, it's good to know if that money is well spent and if we could do things better.

00:05:26.951 --> 00:05:46.389
And the other aspect is that in the community fire engineering community, structural fire engineering community we have been working for a long time to develop the tools to conduct performance-based designs to model the response of structures in fire, and yet these approaches are still rarely used.

00:05:46.389 --> 00:05:46.990
In practice.

00:05:46.990 --> 00:05:48.826
We see that they are not embraced.

00:05:48.920 --> 00:05:52.730
Most people or most buildings are still built using prescriptive designs.

00:05:52.730 --> 00:06:07.281
So the motivation behind this project was to assess the comparative performance of a prescriptive design versus a performance-based design and compare these not only on technical merits but also on cost benefits.

00:06:07.281 --> 00:06:10.290
So the project is supported by the NIST.

00:06:10.290 --> 00:06:31.612
It's called Economic Impact of Performance-Based Fire Design of Composite Steel Frame Structures, because we also leverage and we take advantage of great data that the NIST has been generating over the last few years at the National Fire Research Laboratory, conducting full-scale experiments on composite structures in fire.

00:06:31.612 --> 00:06:36.411
So these experiments provide us the data to build on this economic analysis.

00:06:37.220 --> 00:06:48.793
Can you define composite steel frame structure, because a lot of listeners of Fire Science Show wouldn't be structural engineers, so perhaps we need to be aware of defining stuff for people.

00:06:49.240 --> 00:06:53.552
Yes, here we are looking at buildings that use steel framing.

00:06:53.552 --> 00:07:04.807
So you have steel columns and steel beams with composite steel concrete flooring, so you have a steel deck and then the concrete is poured on top of the steel deck, so the concrete and steel work together.

00:07:04.807 --> 00:07:10.490
This is a very efficient structural system that's commonly used in multi-story buildings.

00:07:10.490 --> 00:07:18.348
For example, if you think about a 10-story office building in a downtown area, that would be a very typical construction method in the US.

00:07:19.161 --> 00:07:20.807
And Chenzi, what brought you to the project?

00:07:20.807 --> 00:07:24.086
How did you start your journey with this cost-benefit analysis in here?

00:07:25.180 --> 00:07:28.331
I did some research on the us uh fire statistic.

00:07:28.331 --> 00:07:31.199
I noted that the fire is very costly.

00:07:31.199 --> 00:07:41.321
So in the us alone the annual cost of fire safety matter is around 57 billion us dollars, which is amazing.

00:07:41.321 --> 00:07:52.690
Besides the construction cost in the fire safety measures and fire is also very frequent in this Every year there is around 575.

00:07:52.690 --> 00:07:55.730
This is my main motivation to this project.

00:07:55.730 --> 00:07:58.850
The fire is very cost, it's very frequent.

00:07:58.850 --> 00:08:04.805
We have to identify how our design performs, how much it can save me.

00:08:05.422 --> 00:08:06.064
Good, good, good.

00:08:06.064 --> 00:08:08.850
One thing that I could start with.

00:08:08.850 --> 00:08:11.500
It's an interesting question when you start thinking about it.

00:08:11.500 --> 00:08:18.514
But if we spend money on passive fire protection, what really is the purpose of that?

00:08:18.514 --> 00:08:23.372
Is this just to prevent the structural collapse of a building?

00:08:23.372 --> 00:08:25.427
Is this the sole and only goal?

00:08:25.427 --> 00:08:30.906
Or are we looking for preventing that in a very specific time, like for an hour, it doesn't collapse?

00:08:30.906 --> 00:08:34.068
Or are there further like have you looked into that?

00:08:34.068 --> 00:08:35.001
Because you have to.

00:08:35.001 --> 00:08:39.166
If you go PBD, performance-based design, you have to have some goal in mind.

00:08:39.166 --> 00:08:44.485
What the goals would be in, let's say, performance-based design and traditional prescriptive design.

00:08:45.048 --> 00:08:47.687
You're right, the goal should go beyond safety.

00:08:47.687 --> 00:08:52.111
So obviously, safety of occupants and firefighters is a prime objective.

00:08:52.111 --> 00:09:08.174
But with these type of approaches because it's in a performance-based design method, as you were saying you lay out objectives for the building, so you can also look at buildings that survive the fire, how long it takes to recover and to be reused.

00:09:08.174 --> 00:09:09.645
So this is something we looked at.

00:09:09.645 --> 00:09:26.212
We developed fragility functions under different natural fire scenarios, which is a way of quantifying the probability of being in different damage states after an event and then from there, we evaluate how long it would need to be repaired and to recover.

00:09:26.212 --> 00:09:31.692
And, chengyi, maybe you can talk more about the work you did on indirect losses and recovery.

00:09:32.179 --> 00:09:32.500
Yeah.

00:09:32.500 --> 00:09:39.774
So right now we use the same safety target on the performance-based design, pre-security design.

00:09:39.774 --> 00:09:43.410
Let's take the NIST test as an example.

00:09:43.410 --> 00:09:47.892
The target is to resist to our standard fire.

00:09:47.892 --> 00:09:51.529
So what's the difference between PPD and prescriptive?

00:09:51.529 --> 00:09:55.171
It's the post-fire recovery.

00:09:55.171 --> 00:10:03.072
For the prescriptive design, when the fire happened the fire may spread to adjacent compartment or even the whole building.

00:10:03.072 --> 00:10:12.087
But for the performance speed design, the probability of fire spread might be much lower and correspondingly the fire cost.

00:10:12.087 --> 00:10:47.135
Building closure time, the duration for PDBD could be much lower than the prescriptive design and accordingly the corresponding extra losses due to the building closure, due to the building shutting down, would be much less for the PVD compared to prescriptive design From my experience in here, we of course have a prescriptive system for structural fire design and all we work with is fire resistance as the vehicle to get there.

00:10:47.759 --> 00:10:53.013
But I wouldn't say that fire resistance is strongly connected with any sort of a goal.

00:10:53.013 --> 00:10:59.543
I would rather say it's more like you know historical context, like we had one hour for office buildings and it was sufficient.

00:10:59.543 --> 00:11:03.052
Or perhaps two hours in high-rise or four hours or whatever.

00:11:03.052 --> 00:11:11.145
We have four hours in super tall office buildings, which I don't understand really what that gives you, but it's just a requirement.

00:11:11.779 --> 00:11:21.826
And it becomes very interesting when you are not able to meet those goals because of some technological reasons or cost perhaps, and you want to apply for derogation.

00:11:21.826 --> 00:11:30.351
And then you have to prove why you want to derogate and what you're giving as a trade for having a lower fire resistance, for example.

00:11:30.351 --> 00:11:50.392
And when I look at your studies, this is the most perfect thing I could use in those procedures, because I could literally show that investing more and more money in this gives me less and less return on the investment, and that's why I really love this type of analysis.

00:11:50.392 --> 00:12:01.692
Anyway, let's perhaps talk about framework, because it's not that you just took one building and analyzed it, you've literally analyzed more than 100, and you have an entire framework put out for that.

00:12:01.692 --> 00:12:06.403
So perhaps let's discuss the main points of the framework that you've developed for this project.

00:12:06.764 --> 00:12:07.004
Indeed.

00:12:07.004 --> 00:12:11.621
So the objective is to compare the lifetime cost of different designs.

00:12:11.621 --> 00:12:13.927
So we consider a building prototype.

00:12:13.927 --> 00:12:19.568
The engineer has several options in how they would approach the fire design.

00:12:19.568 --> 00:12:27.592
It could go prescriptive or performance-based, and within performance-based, lots of options, and the objective was to assess the lifetime cost.

00:12:27.592 --> 00:12:44.909
So to do that we had to compare or consider the investment cost, the construction, the cost of materials and labor originally, and then also consider avoided losses or fire damage in case the fire occurred over the lifetime of the building.

00:12:44.909 --> 00:12:55.496
And in addition we had a number of co-benefits that we also looked at, such as does one design lead to a faster construction, for example, so that you get the benefit of the building earlier?

00:12:55.496 --> 00:13:05.312
If you can omit some of the fire protection in a large building, it may mean that you get earlier delivery of the building, so you get some rental revenues right.

00:13:05.312 --> 00:13:08.208
We also looked at some of the sustainability aspects.

00:13:08.559 --> 00:13:19.570
Did you look into the lifetime of the protection, because lifetime of a fire protection versus the lifetime of the building, how often it will have to be repaired, replaced.

00:13:20.220 --> 00:13:26.351
Yeah, actually the project, we focus on the lifetime cost of the fire protection.

00:13:26.351 --> 00:13:30.919
So, as you mentioned, regarding the hot oven, it will be replaced.

00:13:30.919 --> 00:13:34.090
We introduced the lifetime maintenance cost.

00:13:34.090 --> 00:13:38.671
So right now we use a simplified method.

00:13:38.671 --> 00:13:57.330
We take the annual maintenance cost as a constant percentage of the construction cost, for example, uh, three percent of the construction cost, this is annual maintenance fee and and did you go into like a whole life cycle, like also demolishing, utilizing afterwards or not yet there?

00:13:57.791 --> 00:14:01.245
so we did that for the fire damage losses part.

00:14:01.245 --> 00:14:15.504
So we did the fine analysis of the damage state after fire and we studied different natural fire scenarios and, based on this damage state, chenji looked at a number of factors on the time to recovery and repair.

00:14:15.504 --> 00:14:22.961
So it's not only the demolition and material cost but also availability of contractors and time to perform the work etc.

00:14:23.484 --> 00:14:29.091
So that is, you tune or you adjust the design method at the beginning.

00:14:29.091 --> 00:14:35.799
You would then get very different outcomes in terms of how long it takes and how costly it is to repair after a fire.

00:14:36.181 --> 00:14:39.269
Chance you may want to add on how the fire damage was established.

00:14:39.269 --> 00:14:40.785
Actually, that's an interesting aspect.

00:14:41.220 --> 00:14:41.561
Yeah.

00:14:41.561 --> 00:14:48.163
So the fire damage estimation is established based on the construction cost database.

00:14:48.163 --> 00:14:52.734
So we build a construction cost database with 130 prototypes.

00:14:52.734 --> 00:14:57.461
So for each prototype we have detailed component costs.

00:14:57.461 --> 00:14:59.283
It's very detailed.

00:14:59.283 --> 00:15:09.270
It's such as one column costs how much, the flow system costs, how much the fan protection, the furniture, the stairwell.

00:15:09.270 --> 00:15:12.253
It contains a lot of construction components.

00:15:12.253 --> 00:15:23.559
So based on this detailed cost we can easily specify when the fire happened what kind of components we need to replace for different damage status.

00:15:24.320 --> 00:15:27.559
So in the damage analysis we use the idea of the Fragility curve.

00:15:27.559 --> 00:15:35.799
Fragility curve provides the relationship between the hazard intensity and the distribution of different damage states.

00:15:35.799 --> 00:15:43.471
So we have four different damage states Damage state one, which represents the least severe damage.

00:15:43.471 --> 00:15:46.234
Damage state four, which is the least severe damage.

00:15:46.234 --> 00:15:47.320
Damage state four, which is most severe damage.

00:15:47.320 --> 00:16:12.827
So for different damage states we define different sets of structure components that is damaged and then we split the cost into construction cost, because we may need to reconstruct this component, and we also consider the demolished cost because, for example, if the floor system is totally destroyed we have to hire some laborer to remove the debris right.

00:16:12.827 --> 00:16:24.667
So in a damage analysis we consider some sort of the demolished cost, the reconstruction cost, as well as some human injuries.

00:16:25.721 --> 00:16:29.250
So maybe we can go through the parts of the framework.

00:16:29.250 --> 00:16:34.673
So where does one start with such an analysis and how does one proceed through the framework?

00:16:35.039 --> 00:16:37.750
Yes, so the framework entails different components.

00:16:37.750 --> 00:16:47.080
You have to evaluate the initial construction cost of your design so that can be facilitated by the database that Changejs built for the different prototypes.

00:16:47.080 --> 00:17:09.988
So either your building is very similar to one of the prototypes, so you will get all the cost, or, if it is a different design, we have actually a software tool that Changejs built that interpolates within this database, so that would provide once again the cost of your fire protection design for your entire building of the composite construction type.

00:17:10.540 --> 00:17:19.527
So in your case you specify I want a 10-story 1,000 square feet floor plan office building, and it just takes market costs.

00:17:19.527 --> 00:17:21.172
I guess some averages.

00:17:21.172 --> 00:17:23.086
Where do you take the numbers from?

00:17:23.086 --> 00:17:24.465
Actually, that's quite interesting.

00:17:24.920 --> 00:17:27.660
We build this database from the IceMins.

00:17:27.660 --> 00:17:34.786
The IceMins are a very popular construction cost database, I think mainly in the US.

00:17:34.786 --> 00:17:37.112
They provide a square footage estimation.

00:17:37.112 --> 00:17:46.515
You can specify the building occupancy type, the aspect ratio, the floor area etc.

00:17:46.515 --> 00:17:48.688
A lot of building parameters.

00:17:49.180 --> 00:17:54.907
Okay, and the cost of fire protection is also specified in those databases as per feet or per square meter.

00:17:55.380 --> 00:17:55.883
Yes, it is.

00:17:55.883 --> 00:18:02.019
The cost of fire protection is also taken from the database and the cost is adjusted based on the fire rating.

00:18:02.019 --> 00:18:13.826
So we also took into account the thickness that would apply based on prescriptive fire rating and adjusted the cost to be refined there in terms of, if you want a two-hour rated building or three-hour rated building.

00:18:13.826 --> 00:18:26.291
The project is, as I mentioned, so funded by the NIST and it's also a collaboration with an industry partner, alia Shafi and Jenny Sideri at Tonton Tomasetti.

00:18:26.291 --> 00:18:31.010
So we worked with industry also to guide us and corroborate some of those cost inputs.

00:18:31.632 --> 00:18:31.952
Brilliant.

00:18:31.952 --> 00:18:36.352
So here I see the first difference between the DPBD and prescriptive.

00:18:36.352 --> 00:18:41.887
You said the cost scales with the fire protection rating, so here performance-based design comes into play.

00:18:41.887 --> 00:18:43.442
Right, because you can optimize.

00:18:43.721 --> 00:18:44.282
That's right.

00:18:44.282 --> 00:18:49.568
So for the same building, the user can run first an estimate based on prescriptive design.

00:18:49.568 --> 00:18:55.836
It would refer to a rating and then modify the design according to a performance-based approach.

00:18:55.836 --> 00:19:09.246
One optimization that would be one trade-off I should say that would be typical in a PBD would be to remove some of the passive fire protection on some of the steel work and instead harden the structure.

00:19:09.246 --> 00:19:18.800
So in the performance-based design you really want to achieve a good fire performance and a high safety level based on the embedded structural fire resistance.

00:19:19.342 --> 00:19:27.128
In practice, for all buildings that means more steel reinforcement in the floors embedded in the concrete of the composite floors.

00:19:27.128 --> 00:19:30.589
That is what the NIST has shown, also through their experiments.

00:19:30.589 --> 00:19:37.130
So now it's a good time to talk just a little bit about those experiments that were conducted at the NIST starting in 2019.

00:19:37.130 --> 00:19:42.845
So one thing that was really unique the NIST they tested at full scale this floor system.

00:19:42.845 --> 00:19:47.964
So this is a bay of six meter by nine meter and they also built the adjacent bay.

00:19:47.964 --> 00:19:49.228
So you even have the restraint.

00:19:49.669 --> 00:19:55.997
And what is really nice is that they repeated the full scale experiments for three variations of design.

00:19:56.037 --> 00:20:05.250
So we have direct comparison of performance from the same lab in the same conditions of the prescriptive design and two variations of performance-based design and they showed experimentally that the prescriptive design and two variations of performance-based design.

00:20:05.250 --> 00:20:29.288
And they showed experimentally that the prescriptive design rated for two hours, actually didn't really achieve two hours of integrity at full scale because the amount of steel reinforcement in the composite floor is insufficient when you look at that large scale and you have the large deflection, so they had crack openings and flame pass-through and in comparison, the performance-based design.

00:20:29.288 --> 00:20:37.333
So where you remove the passive protection on the steel build but you add reinforcement was much more robust and maintained integrity.

00:20:37.333 --> 00:20:42.813
So this is really what we want to show, that it's not just about saving costs, removing insulation.

00:20:42.813 --> 00:20:44.540
Rather, that is a trade-off.

00:20:44.540 --> 00:20:49.742
You remove some of the passive fire protection but you harden the structure and the performance is better.

00:20:49.742 --> 00:20:51.483
And so you can quantify that.

00:20:51.483 --> 00:21:00.708
First you can quantify the initial savings in the construction cost, but then you can also quantify the avoided losses, the better performance.

00:21:00.748 --> 00:21:03.230
But here the NIST experiment 6.9,.

00:21:03.230 --> 00:21:04.890
I assume that must be a natural fire.

00:21:05.391 --> 00:21:06.330
It's actually.

00:21:06.330 --> 00:21:14.994
But actually they have the capability to run a standard fire resistant test, so with a 20 megawatt hood and a huge furnace.

00:21:14.994 --> 00:21:18.836
So they ran the ASTM E109 for two hours on that experiment.

00:21:18.836 --> 00:21:20.897
So it's quite a unique lab there.

00:21:20.978 --> 00:21:23.584
They have a great tool, good, good, good, interesting.

00:21:23.584 --> 00:21:29.547
So investment costs also means like what type of fire protection was applied, how it was applied, the workload costs.

00:21:30.039 --> 00:21:30.685
Yeah, yeah, yeah.

00:21:30.685 --> 00:21:54.113
So we consider the material cost and the labor cost regarding the fire protection, and in our database we also incorporate two different fire protection materials, including both the spread fire SFRM and the Intium Scent, and on average, the Intium Scent costs around five times higher than the SFRM.

00:21:54.300 --> 00:21:56.268
Okay, and where do those come from?

00:21:56.268 --> 00:21:58.468
Just the cost of the solution.

00:21:58.468 --> 00:22:00.605
I wonder how universal is that?

00:22:00.605 --> 00:22:05.368
Because the industry is listening, so I guess people will have opinions on that.

00:22:05.940 --> 00:22:16.066
So I think this is indeed, as you say, a specificity of the US market, that the solution that is typically applied is on-site sprayed fire-resistive material sprayed on the steelwork.

00:22:16.066 --> 00:22:23.489
It's rare to have shop painting or even intumescent painting applied, but this is something that the industry is increasingly looking at.

00:22:23.489 --> 00:22:40.262
So you know, the prescriptive spread fire-resistive material solution has been employed for many decades, but as those performance-based design approaches are getting more and more known and people are thinking about how structures are protecting against fire, the industry is also looking overseas and the UK and other countries.

00:22:40.262 --> 00:22:42.390
Shop painting is more common.

00:22:42.390 --> 00:22:45.819
It can speed up the the erection process on site.

00:22:45.819 --> 00:22:49.920
I mean, there are potential benefits, especially if you don't need to protect the entire structure.

00:22:49.920 --> 00:22:58.607
So this is I think this is evolving, but but right now the painting solution is seen as much more expensive here in the us but that's the cost component.

00:22:58.627 --> 00:23:05.089
What about the safety component, like if you have one hour of, let's say, fireboard versus one hour of spray paint?

00:23:05.089 --> 00:23:12.160
Do you find them in your methodology, are they equivalent safety, or you go in-depth into the properties of those materials?

00:23:12.740 --> 00:23:25.058
So to evaluate the safety or the fire performance of the different design solutions, we don't have data, like we do for construction costs, because fire is rare, fortunately do for construction costs, because fire is rare, fortunately.

00:23:25.058 --> 00:23:37.200
So we relied on simulation and modeling, and we used finite element modeling with Saphir and lots of models that we've been developing over the years to simulate the response of the natural fires and on the range of different fire scenarios.

00:23:37.200 --> 00:23:45.405
So there we model each design and each material with their properties and we again simulate the anticipated performance.

00:23:45.539 --> 00:24:03.007
So if an engineer has specific data on a specific insulation product of gypsum boards or others, you would then see the difference that would be captured by the method in terms of different performance, damages for GT functions and, eventually, costs.

00:24:03.740 --> 00:24:04.385
And to close on investment costs.

00:24:04.385 --> 00:24:05.859
I want to move to avoided losses and to close on investment costs.

00:24:05.859 --> 00:24:08.885
I want to move to avoided losses but to close on investment costs.

00:24:08.885 --> 00:24:11.848
You mentioned some sustainability aspects.

00:24:11.848 --> 00:24:13.705
What did you mean by that?

00:24:13.705 --> 00:24:15.366
Did you calculate CO2 emissions?

00:24:15.366 --> 00:24:17.868
What was the sustainability aspect?

00:24:18.060 --> 00:24:19.559
We calculated the CO2 emissions.

00:24:19.559 --> 00:24:28.299
So for a specific fire protection material we can extract the environmental product report from the internet.

00:24:28.299 --> 00:24:40.021
So from that report we can get the relationship between the weight of that material with the weight of CO2 emission and from our cost database we can get the total cost of the fire protection material.

00:24:40.021 --> 00:24:52.512
We can get the total cost of the fire protection material and we can get the total weight of the fire protection material, which provides us a relationship between the cost and the CO2 emission.

00:24:52.512 --> 00:24:57.416
So we captured the CO2 emission based on its total cost.

00:25:05.680 --> 00:25:08.195
One thing, referring to what you just said if you capture the weight, do you also optimize the structure for the weight of passive protection, or this is too far Right now?

00:25:08.195 --> 00:25:11.627
We didn't consider the optimization of the total weight of the construction.

00:25:12.402 --> 00:25:13.365
We had that on one project.

00:25:13.365 --> 00:25:24.540
We had to consider fire protection versus weight because the location of the project above a metro line was so sensitive to the weight of the building and the way how the building was constructed.

00:25:24.540 --> 00:25:27.343
So we don't affect the metro line underneath too much.

00:25:27.343 --> 00:25:28.443
Very interesting discussion.

00:25:28.443 --> 00:25:32.288
Anyway, let's move perhaps to the avoided losses.

00:25:32.288 --> 00:25:37.673
So, as we talked about the costs and everything you said, to me that's quite reasonable.

00:25:37.673 --> 00:25:40.455
You summarize whatever comes in.

00:25:40.455 --> 00:25:47.695
You basically distilled it into atoms, like all the aspects that go in.

00:25:47.695 --> 00:25:50.039
You have hundreds of variants.

00:25:50.039 --> 00:25:54.743
You drop the cost on them and you have the finite number of that.

00:25:54.743 --> 00:25:57.496
But now avoided losses, that's an interesting part.

00:25:57.496 --> 00:26:05.653
First, you mentioned the fragility curves, but how do you put those avoided losses in the lifetime of a building?

00:26:05.653 --> 00:26:08.728
Do you have an expected loss from a fire?

00:26:09.410 --> 00:26:10.112
Yes, you're right.

00:26:10.112 --> 00:26:12.327
So then we come into the world of uncertainty.

00:26:12.327 --> 00:26:21.234
As you very properly said, we can calculate the construction cost, but we can just estimate the probability of an expected loss.

00:26:21.234 --> 00:26:23.145
So we don't have data.

00:26:23.145 --> 00:26:25.769
There are not, fortunately, not many fires.

00:26:25.769 --> 00:26:32.412
So we use simulation, we use our modeling tools that, again, have been developing research committee for many years.

00:26:32.412 --> 00:26:42.910
So we run lots of simulations, finite element simulations of the structural response under a range of fire, and we apply uncertainty in the fire development.

00:26:42.910 --> 00:26:46.891
We use natural fire curves in the thermal response and in the structural response.

00:26:47.520 --> 00:27:06.528
One thing we were very happy with this project is that we had again this unique data from the NIST experiments and the test run by Lisa Cho, and we thank Matt Huller and Matt Bundy and so on for the great data that we had so we could validate our models at full scale, at least under one fire, the standard fire in that case.

00:27:06.528 --> 00:27:09.269
But then, yeah, we run lots of simulations.

00:27:09.269 --> 00:27:14.160
We built fragility functions in order to make this research also more generally applicable.

00:27:14.160 --> 00:27:34.262
So you can compare the performance of prescriptive and performance-based design against any intensity of the fire and with all this probabilistic framework we get to different probabilities of being in different damage states and then the next step is to evaluate the losses and repair costs associated with those different damage states we have.

00:27:34.323 --> 00:27:41.965
Also, there are uncertainty, but we analyze it component by component in terms of demolition, reconstruction, time for unavailability and so on.

00:27:41.965 --> 00:27:43.759
I should also add that youition reconstruction time for unavailability and so on.

00:27:43.759 --> 00:28:04.131
I should also add that you know, again, we built on what we've been working on before, and a very important project was a project funded by the NFPA with a great friend and colleague, ruben Van Coyle from University of Ghent, who was leading that project, and we were working with Shunani and Andrea Luccherini as well.

00:28:04.131 --> 00:28:14.412
And in that project by the NFPA already we were looking at economic impact of fire and we had done some work to quantify also injuries and lives lost etc.

00:28:14.412 --> 00:28:19.231
So there's a very low probability of these, fortunately, but very high cost.

00:28:19.231 --> 00:28:32.006
So again, with a probabilistic approach, we account also for these and I'll just briefly thank Amanda Kimball and Birgit Messerschmidt also for the support a few years ago on that which we are still building on.

00:28:32.920 --> 00:28:35.609
The damage assessment for a specific fire.

00:28:35.609 --> 00:28:37.307
How do you proceed with that?

00:28:37.307 --> 00:28:39.607
You choose a probability of a fire.

00:28:39.607 --> 00:28:42.509
Let's say there will be a one megawatt fire in the building.

00:28:42.509 --> 00:28:45.867
You simulate that fire in the building and and have some outcomes.

00:28:45.867 --> 00:28:47.780
And then you proceed to another one, another one.

00:28:47.780 --> 00:28:52.211
You know probabilities of each of them and then you come up with the final curve.

00:28:53.541 --> 00:28:55.423
No, actually so.

00:28:55.423 --> 00:29:02.412
For the first we we identified the criteria of to define different damage status.

00:29:02.412 --> 00:29:09.570
So here we use residue deflection, so we have three different thresholds of the deflection.

00:29:09.570 --> 00:29:17.140
So once the residue deflection exceeds the threshold we assign a distinct damage state for this simulation.

00:29:17.140 --> 00:29:33.493
So because we consider the uncertainty in the fire, because here we consider the oxygen content, we consider uncertainty in the fire protection material, such as conductivity, the density etc.

00:29:33.493 --> 00:29:39.532
We also consider the uncertainty in the mechanical property of the rebar.

00:29:39.532 --> 00:29:47.708
So we consider five, five randomly in these materials, in these fires, and we run simulations.

00:29:48.342 --> 00:30:00.944
For a distinct simulation we can get a specific outcome, for example, such as the displacement, radio displacement, and we run 10 different simulations.

00:30:00.944 --> 00:30:08.220
We can get 10 different radio deflections and then we compare this response with our threshold.

00:30:08.220 --> 00:30:14.630
If this response is higher than the threshold, we assign a specific damage state to this run.

00:30:14.630 --> 00:30:31.711
So at a specific file load we can get a distribution of different response and then we can convert them to the distribution of different damage states and then we go through from low fire load to the high fire load.

00:30:31.711 --> 00:30:35.791
Then we can create the fertility curve.

00:30:36.359 --> 00:30:39.599
But the damage states are connected to the architecture of the building.

00:30:39.599 --> 00:30:42.048
I know the height, presence of sprinklers, stuff like that.

00:30:42.740 --> 00:30:45.569
So the damage states are structural damage states.

00:30:45.569 --> 00:30:47.667
First of all there is integrity failure.

00:30:47.667 --> 00:31:03.868
So if really we have the simulation shows under a specific fire that the structure, the floor would fail, then usually the simulation stops because nonlinear fire intimate simulation and we assume then the fire propagates and that would be a very severe damage state.

00:31:03.868 --> 00:31:15.727
And then we have intermediate damage where we define them based on the residual deflections in the structure which are correlated to the repair needs and whether you can reuse the structure after work.

00:31:15.727 --> 00:31:22.147
So this depends on the span and the base size, but it doesn't really depend.

00:31:22.147 --> 00:31:33.808
I mean, as we work only with steel concrete, you know steel framing, composite floor structures, we have similar definition of damage states across the spectrum of buildings we are looking at.

00:31:34.884 --> 00:31:36.240
And now those damage states.

00:31:36.240 --> 00:31:43.259
How do you like, is there an expected occurrence of those fires in the lifetime of a building?

00:31:43.259 --> 00:31:50.413
How do you transfer from this into a tangible number of avoided losses?

00:31:50.880 --> 00:31:55.471
Yes, you're right, that's how we do is expected numbers of fire frequency.

00:31:55.471 --> 00:32:00.592
So one component is to build this library of structural fire response.

00:32:00.592 --> 00:32:01.755
That's what we explained.

00:32:01.755 --> 00:32:03.960
This library of structural fire response that's what we explained.

00:32:03.960 --> 00:32:07.990
It's based on running lots of simulations under a range of fires with uncertainties.

00:32:07.990 --> 00:32:14.667
But then we still have to evaluate the probability to have a fire over the lifetime of the building.

00:32:14.667 --> 00:32:17.670
So this we do based on fire statistics in the US.

00:32:17.670 --> 00:32:26.911
So looking at the numbers of actual fire occurrences in similar typologies of buildings commercial, you know, multi-story office buildings, for example.

00:32:26.911 --> 00:32:32.732
So if we have a lifetime of 50 years for a building, we know it's probably that the fire would occur.

00:32:32.732 --> 00:32:41.529
We also need an assumption on the probable or the expected severity of that fire which we correlate to the expected fuel load.

00:32:41.640 --> 00:32:43.909
We know it's not perfect, but it's a way of doing that.

00:32:43.909 --> 00:32:53.005
So if it's an office building, and either the authority or the engineer assumes a fuel load distribution in that type of building.

00:32:53.005 --> 00:32:55.666
We correlate that to the expected fire severity.

00:32:55.666 --> 00:33:02.230
So priority of fire occurrence, expected fire severity and then expected damage states based on all our simulations.

00:33:02.230 --> 00:33:14.828
And from those expected damage states, which is a distribution, we have then the economic loss assessment with repairs, demolition labor and also time of unavailability of the building.

00:33:15.440 --> 00:33:18.589
But that's a loss and you use avoided loss.

00:33:18.589 --> 00:33:21.387
So where does avoided come from?

00:33:21.808 --> 00:33:22.049
I see.

00:33:22.049 --> 00:33:26.799
So when I talk about avoided loss, you should understand that it's comparative.

00:33:27.141 --> 00:33:32.073
So here, the whole framework is aimed at comparing design options.

00:33:32.073 --> 00:33:42.087
So we would evaluate losses for the prescriptive design and we would also evaluate losses for the alternative, performance-based design, and the difference would be the avoided loss.

00:33:42.087 --> 00:34:08.889
So what we are showing for this the case study specifically is that by trading fire protection on secondary beams and instead adding more reinforcement in the composite floor, not only there is a possibility to save in initial cost a little bit, but also you saved in those avoided losses, because the losses of the performance based design are expected to be much smaller because the system is more robust.

00:34:10.253 --> 00:34:17.574
Very interesting and in your analysis you must eventually clash those numbers like the investment costs and avoided losses.

00:34:17.574 --> 00:34:19.847
Is this like a dollar per dollar basis?

00:34:19.847 --> 00:34:20.985
Are there weights on that?

00:34:20.985 --> 00:34:21.922
How do you compare?

00:34:21.922 --> 00:34:29.431
Like the investment costs, which is obvious, right, it's money spent upfront versus a loss.

00:34:29.431 --> 00:34:31.668
It could happen in the lifetime of a building.

00:34:31.668 --> 00:34:33.360
It could not happen in the lifetime of the building.

00:34:33.360 --> 00:34:37.543
You cannot say it's not the same value In this methodology.

00:34:37.543 --> 00:34:40.150
Is it dollar by dollar or there's a method to that?

00:34:40.619 --> 00:34:42.266
Yeah, very good question.

00:34:42.266 --> 00:34:43.826
So there are two aspects to it.

00:34:43.826 --> 00:34:46.309
I would say first, it's a present net value approach.

00:34:46.309 --> 00:34:51.833
So obviously dollar in the future is weighted by the inflation rate and so on.

00:34:51.833 --> 00:34:55.530
So that's a way of comparing things that do not happen at the same time.

00:34:55.530 --> 00:34:58.909
But what you refer to is more of an uncertainty component, right?

00:34:58.909 --> 00:35:00.025
So I know how much I'm going to.

00:35:00.025 --> 00:35:01.422
So it's a very good point.

00:35:01.422 --> 00:35:03.969
We have lots of discussion also in the industry on that.

00:35:03.969 --> 00:35:12.579
I would even add one more thing is that with the performance-based design, a big hurdle is the uncertainty that the design will be accepted.

00:35:12.579 --> 00:35:16.972
So the owner wants the building made and the engineer has to deliver.

00:35:16.972 --> 00:35:21.344
And if they go the prescriptive route, for sure it's going to be signed off.

00:35:21.344 --> 00:35:26.670
But if they want to do an alternative design, there is uncertainty, you know, will it be accepted or not?

00:35:26.670 --> 00:35:30.016
That's really hard to weigh in in that approach.

00:35:31.940 --> 00:35:33.061
It's very difficult.

00:35:33.061 --> 00:35:45.206
Perhaps you've been in those discussions with investors, but you know, saving $1 today to avoid the loss of $1 in the future is not such a great deal, to be honest.

00:35:45.347 --> 00:35:49.849
Yeah, so the thing is that here we are building on.

00:35:49.849 --> 00:35:56.612
So for several years I think the committee has shown the technical benefits of the performance-based design right.

00:35:56.612 --> 00:36:04.655
You have more freedom, you can optimize, you can have maybe better resilience, et cetera, and here we want to show the economic benefits.

00:36:04.655 --> 00:36:08.657
But in some cases what we are showing is that you can have both.

00:36:08.657 --> 00:36:30.929
Not always, but in some cases you can save now because the construction might be cheaper, because again you remove some of the fire protection, you have less on-site work, you may have completed the building a little bit earlier because it goes faster to increase the diameter of pre-bars in slab than to spray material on all the beams.

00:36:32.061 --> 00:36:37.306
You may have a saving now plus more avoided losses, so plus a better performance.

00:36:37.306 --> 00:36:43.525
But of course it's not always the case and also there are these uncertain components that I was referring to.

00:36:43.525 --> 00:36:44.509
Will it be accepted?

00:36:44.509 --> 00:36:48.045
There is also additional cost of engineering, which may be very variable.

00:36:48.045 --> 00:37:06.193
So it's not all clear cut, but we think that we are providing here data to show that there is an economic case in addition to the technical case and in many markets because we also look at different costs of labor depending on the location and so on In many markets you might save now plus save in the future.

00:37:07.021 --> 00:37:18.829
I think, from my perspective, I'm doing this podcast selfishly to teach myself and apply stuff and, by the way, there's hundreds of people listening along me, but I appreciate learning from the world's best.

00:37:18.829 --> 00:37:27.985
You know, in my case I see an immediate benefit of having some benchmark losses that would be associated with the standardized solution.

00:37:27.985 --> 00:37:43.507
And then, technically, I could run an optimization study where I would just I would like this number to not grow, you know, and see like what kind of elements I can remove or switch, like to get the lowest investment costs while maintaining the same loss.

00:37:43.507 --> 00:37:46.025
That would be a good optimization front.

00:37:46.025 --> 00:37:48.070
I think that could actually work.

00:37:48.070 --> 00:37:50.748
And then it's interesting that your method actually allows for that.

00:37:50.748 --> 00:37:55.030
Did you try something like that, chenzi, or wasn't that part?

00:37:55.480 --> 00:37:58.985
Actually no, we didn't try this aspect.

00:37:58.985 --> 00:38:08.423
But I would like to argue a little bit about what you said saving $1 and avoid $1 loss in the future.

00:38:08.423 --> 00:38:12.974
I mean, with time goes by, everything is becoming more and more expensive.

00:38:12.974 --> 00:38:20.481
Today we repair the floor system we can't cost $100, but in the future probably we need to cost $200.

00:38:20.481 --> 00:38:27.273
So if we convert all the future value in current value perhaps it is always $1 loss.

00:38:27.273 --> 00:38:34.512
So the saving, the avoid losses, I don't think it will ever change significantly with the time.

00:38:35.173 --> 00:38:58.050
Interesting and perhaps let's go to the outcomes of the research, because so far we've spent like 40 minutes discussing this framework and it's really a big framework, like we're talking about it, like it's an easy task, but it's years of research, from what I see in the paper, and it's like most impressive, to be honest.

00:38:58.050 --> 00:39:09.987
But you've created this database of 130 buildings, You've pre-designed them, you've chosen some sort of fire protection measures for them and you've run this analysis.

00:39:09.987 --> 00:39:19.760
What are the main outcomes that listeners can take home from this analysis and perhaps then we can go to more detailed outcomes.

00:39:20.260 --> 00:39:22.822
So we built the construction cost database.

00:39:22.822 --> 00:39:26.884
We analyzed 130 different building prototypes.

00:39:26.884 --> 00:39:30.746
So from this aspect we do find something.

00:39:30.746 --> 00:39:39.650
We found that for the fire safety measure cost, it accounts for 4% to 12% of the total construction cost.

00:39:39.650 --> 00:39:49.155
All fire safety measures yes, all fire, including the sprinkler, the fire pump, all the detectors, all the fire safety measures.

00:39:49.155 --> 00:39:56.670
If we zoom into the passive fire protection on the steel work, it ranges from 0% to 1.2%.

00:39:57.422 --> 00:40:02.324
And for context, what would be the cost of steel and concrete in a building?

00:40:02.324 --> 00:40:03.106
Like a rough number.

00:40:04.001 --> 00:40:07.210
It is around 20% of the total construction cost.

00:40:07.460 --> 00:40:10.989
So it would be like 5% of the cost of the structure itself.

00:40:10.989 --> 00:40:11.911
Right Interesting.

00:40:11.911 --> 00:40:15.570
And how did it vary between the solutions that you've looked into?

00:40:16.199 --> 00:40:17.746
So yes, we find in.

00:40:17.746 --> 00:40:24.833
So the numbers that Chenji just gave are for composite multi-story building in the US with prescriptive fire design.

00:40:24.833 --> 00:40:34.474
Okay, so with the performance-based fire design there is an opportunity to shave off some of that cost, because we can omit some of the fire protection.

00:40:34.474 --> 00:40:41.742
So maybe you get from 1.2% of the cost of the building to maybe 0.6% of the cost of the building to maybe 0.6%.

00:40:41.742 --> 00:40:50.831
But one thing that's important to mention is that the prescriptive design in the US uses very little steel reinforcement in the composite floors.

00:40:50.831 --> 00:40:54.570
It uses 60 square millimeters per meter.

00:40:55.161 --> 00:41:01.248
So maybe listeners from Europe who know about structural design of composite buildings would find that to be very, very small.

00:41:02.119 --> 00:41:09.889
It's a welded wire mesh that's there for construction reasons but it's not really aimed to carrying forces because they rely on the steel deck.

00:41:09.889 --> 00:41:23.409
So the reason I'm mentioning that is that in order to activate tensile membrane action and robust behavior in our performance-based design in fire, we need to increase quite a bit this amount of steel in the floors.

00:41:23.409 --> 00:41:31.250
So in the US you have to pay more if you want tensile membrane action, whereas in Europe, for example, they already start from a higher value.

00:41:31.250 --> 00:41:37.978
So they often find that with PBD they remove fire protection, they don't necessarily need to pay much more elsewhere.

00:41:37.978 --> 00:41:43.313
Right, they would need to pay attention to continuity and overlap and so on, but not that much more material.

00:41:43.313 --> 00:41:52.414
We found that we can save maybe 0.6% of the cost of the building value, but we almost put the same amount back in the steel in the floor.

00:41:52.414 --> 00:42:00.333
The details vary, but it will not be very, very different in one way or another in terms of initial construction costs.

00:42:00.800 --> 00:42:03.590
But there will be differences in avoided losses.

00:42:04.541 --> 00:42:05.264
Very much so.

00:42:05.264 --> 00:42:08.670
Yes, and that's backed up by the data from the NIST.

00:42:08.670 --> 00:42:18.326
So it's not just our simulations, but already the NIST showed in their experiments that the minimum amount of steel in the floor it was not sufficient.

00:42:18.326 --> 00:42:19.786
It led to integrity failure.

00:42:19.786 --> 00:42:22.065
So there is really a question there in whether we should.

00:42:22.065 --> 00:42:30.210
You know that it would be recommended to put more steel anywhere in the composite floors Because once you look at full scale, you know beyond.

00:42:30.210 --> 00:42:33.663
You know the just small standard fire furnace.

00:42:33.663 --> 00:42:39.534
You really need more reinforcement to maintain integrity in that case for the two-hour rating.

00:42:39.534 --> 00:42:42.989
So we showed that also with the simulations on the natural fire.

00:42:42.989 --> 00:42:54.300
The performance enhanced, the avoided losses much better with the performance-based design in terms of priority of failure, of maintaining integrity for the entire fire.

00:42:54.943 --> 00:42:55.967
And for those losses.

00:42:55.967 --> 00:43:04.047
If you compared a solution with one, two, three hours of fire resistance, was the effect really as like?

00:43:04.047 --> 00:43:07.887

184 - Cost-benefit analysis in structural fire safety with Thomas Gernay and Chenzi Ma

Listen On

Recent Episodes