WEBVTT Kind: captions Language: en-US 00:00:01.000 --> 00:00:09.900 [Silence] 00:00:09.900 --> 00:00:17.800 [inaudible background conversations] 00:00:17.800 --> 00:00:20.420 Good morning, everyone. 00:00:20.420 --> 00:00:23.720 Hi. Welcome to today’s weekly seminar. 00:00:24.800 --> 00:00:27.360 It’s my pleasure to introduce Anne Hulsey. 00:00:27.360 --> 00:00:31.079 Anne is our speaker today. And Anne is going to be talking about 00:00:31.079 --> 00:00:35.430 tall buildings and cordons. It’s a topic very near and dear to my heart. 00:00:35.430 --> 00:00:38.629 When I was the public information manager, second in command 00:00:38.629 --> 00:00:40.399 at Christchurch, we did a lot of communication around the 00:00:40.399 --> 00:00:45.469 cordons there. And so she’s going to be talking about that today. 00:00:45.469 --> 00:00:48.350 So Anne’s interest in earthquakes really began when she grew up 00:00:48.350 --> 00:00:52.980 in Turkey and lived in Turkey from 4 years old to age 16. 00:00:52.980 --> 00:00:56.420 And she was there during the 1999 earthquakes. 00:00:56.420 --> 00:01:01.720 And that made a huge impression on her and really gathered her interest 00:01:01.720 --> 00:01:05.300 in the phenomenon. And she has a really intriguing background. 00:01:05.300 --> 00:01:09.060 She’s got a bachelor’s in liberal arts from University of Texas. 00:01:09.060 --> 00:01:11.880 And she’s got two master’s degrees – one in engineering and one in 00:01:11.890 --> 00:01:15.850 public affairs, also from the University of Texas. 00:01:15.850 --> 00:01:18.200 And then her Ph.D., which she’s currently doing 00:01:18.200 --> 00:01:21.720 at Stanford, is with the Urban Resilience Group. 00:01:21.720 --> 00:01:25.480 And in between all of this great work at university, she’s also been 00:01:25.490 --> 00:01:28.810 doing internships at NIST. She’s on three internships there. 00:01:28.810 --> 00:01:33.240 And, yes, so, without further ado, Anne. 00:01:33.240 --> 00:01:35.640 - Thank you, Sara. Is this on? 00:01:36.120 --> 00:01:38.120 Thank you, Sara. And thank you for this opportunity to 00:01:38.130 --> 00:01:43.159 talk about my current research into the – quantifying the post-earthquake 00:01:43.160 --> 00:01:47.200 downtime that would be induced by cordons around damaged tall buildings. 00:01:47.900 --> 00:01:49.710 So just to motivate what we’re talking about here, 00:01:49.710 --> 00:01:54.420 I’m going to take you back to Christchurch, New Zealand, in 2011. 00:01:54.420 --> 00:01:56.760 And we had cordons that were restricting access 00:01:56.760 --> 00:01:58.880 to Christchurch’s Central Business District. 00:01:58.880 --> 00:02:03.030 And part of it was restricted for over two years. 00:02:03.700 --> 00:02:06.759 And, as you can see, many of the longest restrictions were – 00:02:06.760 --> 00:02:10.060 which you can see in the darkest red in that central map there – 00:02:10.060 --> 00:02:13.700 were centered around tall buildings. 00:02:13.700 --> 00:02:17.590 And specifically the fact that these buildings were so heavily damaged. 00:02:17.590 --> 00:02:21.040 And the taller the building is, the larger its, quote, fall zone. 00:02:21.040 --> 00:02:24.860 So if the building is damaged, and then, in an aftershock, it were to fall, it has 00:02:24.860 --> 00:02:29.090 a larger range that it could potentially damage the buildings around it. 00:02:29.090 --> 00:02:33.620 And so you can see, in the image on the left, that’s the Clarendon Tower there, 00:02:33.620 --> 00:02:37.360 kind of leaning inwards from the left – from the right side of the picture. 00:02:37.360 --> 00:02:42.420 And its location is there in the map, marked in – it’s clearly right in the 00:02:42.420 --> 00:02:47.780 center of that darker red area where we had the really long restriction. 00:02:47.780 --> 00:02:49.720 And then the same on the right image. 00:02:49.730 --> 00:02:52.170 We have the Victoria Apartments in the center. 00:02:52.170 --> 00:02:55.819 In blue, in front of it, is Craigs House. And you can see that they’re not actually 00:02:55.819 --> 00:02:58.870 very aligned with each other. They actually both are 00:02:58.870 --> 00:03:02.730 leaning in separate directions. And so that’s call residual drift. 00:03:02.730 --> 00:03:06.720 It’s the amount that the building is leaning after the earthquake, and it 00:03:06.720 --> 00:03:11.040 causes concern that we’re going to have issues if there is an aftershock. 00:03:12.740 --> 00:03:17.620 So the motivation for looking at this is really to support policy decisions. 00:03:17.620 --> 00:03:21.020 Community resilience is becoming an increasing priority for different 00:03:21.030 --> 00:03:26.110 policymakers across the nation and across the world as well. 00:03:26.110 --> 00:03:29.850 In San Francisco, the Planning and Urban Research Association 00:03:29.850 --> 00:03:34.980 started looking at this back in 2009. And they’re the first to really take this 00:03:34.980 --> 00:03:39.290 model of a table, which we have on the right, and look at different services that 00:03:39.290 --> 00:03:43.500 the community provides and break it out and say, these are the timeframes 00:03:43.500 --> 00:03:46.960 at which we want each of these services to be back up and running 00:03:46.960 --> 00:03:51.040 again after the earthquake. So, if we look specifically at this 00:03:51.040 --> 00:03:55.450 particular line that says we want 50% of offices and workplaces 00:03:55.450 --> 00:03:59.940 back open, you can see a blue target at four months. 00:03:59.940 --> 00:04:03.830 So that’s what they would like to have. And what they would imagine a resilient 00:04:03.830 --> 00:04:08.370 city would be is to have all of that coming back on at four months. 00:04:08.370 --> 00:04:11.000 The X represents what they currently think would happen 00:04:11.000 --> 00:04:15.650 if the earthquake in question were to happen today or tomorrow. 00:04:15.650 --> 00:04:19.599 And so what becomes interesting, then, is trying to understand how to 00:04:19.599 --> 00:04:23.900 pull that X back in towards the blue area and to reduce 00:04:23.900 --> 00:04:27.140 that resilience gap, is what we call that. 00:04:27.140 --> 00:04:29.500 And so, at the Stanford Urban Resilience Initiative, 00:04:29.500 --> 00:04:34.219 we’re trying to do a few things to help support these kinds of efforts. 00:04:34.219 --> 00:04:37.819 So one is to quantify what the resilience is. 00:04:37.819 --> 00:04:41.830 An example of that would be to assess the recovery through time, 00:04:41.830 --> 00:04:45.520 by which I mean figuring out what the correct location is for that X – 00:04:45.520 --> 00:04:50.400 really getting a good understanding of what our current situation is and 00:04:50.400 --> 00:04:55.300 doing that in a robust, quantifiable way. And then, second, we want to be able to 00:04:55.300 --> 00:04:59.749 support decision-makers as they try to think about how we can pull that X back 00:04:59.749 --> 00:05:03.389 towards the blue – so reduce that gap. And we can do this by comparing 00:05:03.389 --> 00:05:08.300 different policy options and letting them know what – 00:05:08.300 --> 00:05:11.420 how they’re going to get the biggest bang for their buck, essentially. 00:05:11.420 --> 00:05:14.960 But when we look at this table that looks somewhat nice and clean, 00:05:14.979 --> 00:05:17.680 we have to remember that we’re talking about an actual city here. 00:05:17.680 --> 00:05:19.680 This is downtown San Francisco. 00:05:19.680 --> 00:05:23.610 And, if you notice, there’s a lot of tall buildings right next to each other. 00:05:23.610 --> 00:05:26.960 And that is concerning because of the previous images 00:05:26.960 --> 00:05:30.979 that we saw of Christchurch, where just one tall building 00:05:30.979 --> 00:05:34.800 leaning a bit made a big cordon around it. 00:05:34.800 --> 00:05:37.419 And so, when we’re doing this assessment, we need to make sure 00:05:37.420 --> 00:05:42.270 that we’re considering these potential effects in our recovery models. 00:05:42.720 --> 00:05:46.180 Thankfully, San Francisco is already starting to think about these. 00:05:46.189 --> 00:05:49.199 The Office of Resilience and Capital Planning recently released 00:05:49.199 --> 00:05:51.969 its tall building study. And because it does pertain 00:05:51.969 --> 00:05:54.330 so much to what I’m doing in my research, I’m going to 00:05:54.330 --> 00:05:58.020 go through that a little bit and talk about what we found. 00:05:58.020 --> 00:06:00.960 So this was a pioneering study. It’s the first, definitely in the nation, 00:06:00.960 --> 00:06:04.449 and probably in the world, where they looked at tall buildings 00:06:04.449 --> 00:06:09.680 as a special cohort – a unique subset of the buildings within the community 00:06:09.680 --> 00:06:13.809 and thought about, how can we look at this as a whole and figure out solutions 00:06:13.809 --> 00:06:20.680 for this subset because it’s such a big contributor to our community functions. 00:06:20.680 --> 00:06:24.569 And so two reasons for that. One is just the simple fact that they 00:06:24.569 --> 00:06:28.080 host a lot of the proportion of our different community functions 00:06:28.080 --> 00:06:31.080 within the buildings themselves. So if we’re looking at – 00:06:31.080 --> 00:06:34.580 we want 50% office space back open by a certain time, 00:06:34.580 --> 00:06:37.250 a lot of that is going to be in these tall buildings. 00:06:37.250 --> 00:06:41.219 And then also the fact that they are so tall can pose risks much further 00:06:41.220 --> 00:06:45.320 than its own footprint. Again, this is the idea of that cordon. 00:06:46.229 --> 00:06:49.780 So the study was initiated by the city of San Francisco with the intent 00:06:49.789 --> 00:06:54.289 to examine the earthquake performance of San Francisco’s tall buildings. 00:06:54.289 --> 00:06:58.219 And from that, we would then develop recommendations to address policies 00:06:58.219 --> 00:07:02.619 and practices for new design, for assessment and retrofit of 00:07:02.619 --> 00:07:07.020 existing buildings, and then thinking about post-earthquake inspection 00:07:07.020 --> 00:07:11.180 and response. So how are we going to move through the process? 00:07:11.940 --> 00:07:16.240 So the first thing to do was to figure out what is the database of tall buildings 00:07:16.240 --> 00:07:20.879 that we’re looking at. On the left, you can see just a map of 00:07:20.879 --> 00:07:25.919 the northeastern quadrant of the corner – of the city. 00:07:25.920 --> 00:07:29.660 You can see a lot of short buildings below 75 feet in the blue, but then, 00:07:29.660 --> 00:07:33.800 as you approach the downtown, they start to get taller. 00:07:33.800 --> 00:07:38.000 And on the right, you can see the buildings that we ended up collecting. 00:07:38.009 --> 00:07:42.089 It’s 156 buildings that are over 240 feet. 00:07:42.089 --> 00:07:46.040 I’ll discuss in a little bit why we chose 240 feet. 00:07:47.340 --> 00:07:51.000 But what we had to work with when trying to take on this endeavor 00:07:51.000 --> 00:07:55.120 was that we started with San Francisco’s open data portal, 00:07:55.120 --> 00:07:57.249 and we had a few different relevant data sets. 00:07:57.249 --> 00:08:01.789 We had some tax information. We had some building footprints. 00:08:01.789 --> 00:08:06.960 But a lot of this was difficult to compile, especially in a rigorous way where 00:08:06.960 --> 00:08:10.459 we could be certain that we had the right information for each building. 00:08:10.459 --> 00:08:15.819 And so, for this particular project at this level of interest, we needed to 00:08:15.819 --> 00:08:19.560 make sure we went a lot more in-depth. And so, for those buildings over 00:08:19.560 --> 00:08:24.189 240 feet, we went right back to the construction permit documents. 00:08:24.189 --> 00:08:26.499 We went to Department of Building Inspections and looked at 00:08:26.499 --> 00:08:29.509 all their structural drawings. And what we were trying to do 00:08:29.509 --> 00:08:32.870 was extract different attributes about each building that would be 00:08:32.870 --> 00:08:34.979 useful and relevant to us. 00:08:34.979 --> 00:08:38.220 In some cases, the buildings had a Building Occupancy 00:08:38.220 --> 00:08:40.380 and Resumption Program. They were – they were part of 00:08:40.380 --> 00:08:44.099 that program, which meant that they had a BORP report. 00:08:44.100 --> 00:08:48.780 And this – BORP means that an owner is going to contract with an engineer, 00:08:48.780 --> 00:08:51.600 and the engineer is then going to look at the building, figure out 00:08:51.610 --> 00:08:55.790 where the potential issue areas are, where the damage would be, 00:08:55.790 --> 00:08:58.300 and then they’re going to make a plan for how to inspect it 00:08:58.300 --> 00:09:01.879 after the earthquake. And then, essentially, 00:09:01.879 --> 00:09:04.290 what this does is gives the owner a bit more agency 00:09:04.290 --> 00:09:07.260 in the process of getting their building back online. 00:09:07.260 --> 00:09:09.420 Without a BORP report, you’re going to be at the mercy 00:09:09.420 --> 00:09:12.170 of the scheduling of the city as they try to send out 00:09:12.170 --> 00:09:15.670 different engineers to inspect all of these buildings. 00:09:15.670 --> 00:09:19.430 If you do have the BORP report, then the engineer that you’ve contracted 00:09:19.430 --> 00:09:24.139 with is authorized by the city to make that assessment and re-open it. 00:09:24.139 --> 00:09:28.730 And so this really gives the owners the ability to take a little bit more initiative, 00:09:28.730 --> 00:09:32.310 have some agency in making sure that they’re getting back online. 00:09:32.310 --> 00:09:35.740 But for my purpose, the reason why this is useful is that that initial study 00:09:35.740 --> 00:09:38.790 of the building pulls out a lot of the same attributes that I was 00:09:38.790 --> 00:09:41.910 doing from the structural drawings. So if there was a BORP report, 00:09:41.910 --> 00:09:45.519 it was much easier to just go grab those attributes for it. 00:09:45.519 --> 00:09:49.660 Additionally, we looked at surveys and interviews with different structural 00:09:49.660 --> 00:09:54.279 designers of these buildings. Also, there’s kind of a small group 00:09:54.279 --> 00:09:57.960 of peer reviewers for these buildings that is part of the code 00:09:57.960 --> 00:10:01.420 and the permitting process. And so there’s a few people 00:10:01.420 --> 00:10:03.709 who are very aware of lots of the buildings out there. 00:10:03.709 --> 00:10:07.580 So we had some conversations with them to get more information. 00:10:07.580 --> 00:10:10.850 And then finally, there are some buildings that it’s really hard to 00:10:10.850 --> 00:10:13.670 get any information about, especially if it’s a much older building. 00:10:13.670 --> 00:10:17.500 There may not be information at the Department of Building Inspections. 00:10:17.500 --> 00:10:22.339 So, in those cases, we went to Emporis, which is an online database of buildings. 00:10:22.339 --> 00:10:25.480 It doesn’t have as much information as what we want, but at least it 00:10:25.480 --> 00:10:28.730 gives us a placeholder, essentially. We say, we know that there is 00:10:28.730 --> 00:10:32.670 this building. It’s about this height. It’s this date. We don’t know more about 00:10:32.670 --> 00:10:37.769 it, but it’s a placeholder to say we know about it and we’re trying to find more. 00:10:37.769 --> 00:10:42.179 - The BORP reports are from the [inaudible] building office? 00:10:42.179 --> 00:10:44.800 - They’re housed there, yes. But they’re … 00:10:44.800 --> 00:10:48.610 - [inaudible] available to the public? - They’re still – we didn’t have 00:10:48.610 --> 00:10:51.269 public access to them. Because it was a San Francisco project, 00:10:51.269 --> 00:10:55.430 that’s how we got them. Part of what’s coming out of this 00:10:55.430 --> 00:10:59.930 project is a closer look at those BORP reports and what the whole system is. 00:10:59.930 --> 00:11:03.000 So things may be changing with how they’re handled. 00:11:06.120 --> 00:11:09.370 So one of the reasons why we’re looking at these tall buildings, again, 00:11:09.370 --> 00:11:14.380 as I said, is that they host a large percentage of the community functions. 00:11:14.380 --> 00:11:16.620 So, when we’re talking about community functions, 00:11:16.620 --> 00:11:20.720 essentially we’re saying, what kind of occupancy does the building host? 00:11:20.720 --> 00:11:24.510 So we have different occupancies plotted here on the map. 00:11:24.510 --> 00:11:27.310 Again, office is one of a lot of importance. 00:11:27.310 --> 00:11:31.300 Another one might be residential. So if we start to think, what proportion 00:11:31.300 --> 00:11:35.370 do these buildings represent, we can do it in one of two ways. 00:11:35.370 --> 00:11:38.970 So I’m going to talk about these percentages based on San Francisco 00:11:38.970 --> 00:11:43.649 Districts 3 and 6, which you can see up in the right-hand corner there. 00:11:43.649 --> 00:11:48.920 This is the two districts that house what we think of as downtown. 00:11:50.400 --> 00:11:52.970 So first of all, we can think of it in terms of, what is the number 00:11:52.970 --> 00:11:57.410 of buildings that are – that contain office space? 00:11:57.410 --> 00:12:05.519 So if there’s 86 buildings that are taller than 240 feet out of 750 buildings, 00:12:05.519 --> 00:12:11.199 that’s about 11% of our office space is contained within these tall buildings. 00:12:11.199 --> 00:12:14.250 But if you think about it instead in terms of square footage, 00:12:14.250 --> 00:12:17.999 and that’s a better representation of how many businesses could fit into 00:12:17.999 --> 00:12:23.290 the single building, we see that we have about 40 million out of 70 million. 00:12:23.290 --> 00:12:26.290 So more like 64%. So, again, this really gives us 00:12:26.290 --> 00:12:30.420 a lot more bang for our buck in terms of addressing a fewer number of buildings 00:12:30.420 --> 00:12:34.180 but getting a lot more office space available. 00:12:34.180 --> 00:12:38.339 If we do the same thing for residential, it’s almost even more striking. 00:12:38.339 --> 00:12:44.040 We have 29 out of about 4,000 buildings, so much less than 1%. 00:12:44.040 --> 00:12:49.740 But we have 33% of our square footage in those buildings. 00:12:50.460 --> 00:12:53.769 At the same time, I don’t want to give the impression that 240 is 00:12:53.769 --> 00:12:57.879 some magic number. This is what we used, but we had 00:12:57.879 --> 00:13:02.430 a few different options for what it would look like to characterize tallness. 00:13:02.430 --> 00:13:08.579 So this list is basically drawn from different structural codes – safety codes, 00:13:08.579 --> 00:13:13.139 things like that. And we looked at three possible options. First is 240 feet. 00:13:13.139 --> 00:13:16.329 That’s coming from seismic and structural codes. 00:13:16.329 --> 00:13:20.949 Essentially, there is a height limit in the code that says, any shear wall 00:13:20.949 --> 00:13:28.480 and braced frame systems, which are two popular kinds of structural systems, 00:13:28.480 --> 00:13:32.930 you can’t go above 240 feet if that’s what you’re using. 00:13:32.930 --> 00:13:38.249 Another possible option is 160 feet. And this is in San Francisco’s 00:13:38.249 --> 00:13:44.449 current Administrative Bulletin 083. It’s the definition of tall buildings 00:13:44.449 --> 00:13:49.540 for that provision. And that provision is a way to use non-prescriptive 00:13:49.540 --> 00:13:52.310 seismic design procedures. So what that means is that, 00:13:52.310 --> 00:13:56.040 if you’re using a prescriptive code, you have a lot of very specific rules 00:13:56.040 --> 00:13:59.160 about how you have to build. One of them is that you can’t 00:13:59.160 --> 00:14:02.269 go above that height limit – 240 feet if you’re using the 00:14:02.269 --> 00:14:05.980 shear wall and braced frame systems. And so what this allows you to do 00:14:05.980 --> 00:14:11.100 is use other ways instead of just those prescriptive rules to design the building 00:14:11.100 --> 00:14:15.589 and then do a lot of analysis to prove that it has the same response – 00:14:15.589 --> 00:14:21.529 the same kind of – yeah, it responds in the same way that we would – the 00:14:21.529 --> 00:14:25.850 code would have intended if you were using those more prescriptive rules. 00:14:25.850 --> 00:14:29.080 So in the definition, again, of that bulletin, 00:14:29.080 --> 00:14:32.840 they’re calling tall buildings anything over 160 feet. 00:14:32.840 --> 00:14:36.880 Another option is 75 feet. That’s coming from fire safety codes. 00:14:36.889 --> 00:14:40.269 And essentially, that is derived from the height that a 00:14:40.269 --> 00:14:43.589 fire truck’s ladder can reach. And so if you’re above that height, 00:14:43.589 --> 00:14:47.569 then you’re going to have to use some different modes of firefighting. 00:14:47.569 --> 00:14:53.120 So if we’re looking at these three options – 240, 160, and 75 feet, 00:14:53.120 --> 00:14:56.600 this is what that looks like in the map. So I’ve now zoomed into the map 00:14:56.600 --> 00:14:59.829 I showed previously so we can get closer into the downtown. 00:14:59.829 --> 00:15:02.439 And specifically, we’re looking at this study area, 00:15:02.439 --> 00:15:05.519 which is outlined by the blue polygon. 00:15:05.519 --> 00:15:09.249 And this represents the densest area of the city. 00:15:09.249 --> 00:15:13.660 So if we plot the buildings that are within that polygon on the right with 00:15:13.660 --> 00:15:20.000 height versus their date in which they were built, it’s clear that it’s hard to 00:15:20.000 --> 00:15:23.740 find a really good place to cut it off. There’s actually somewhat more 00:15:23.749 --> 00:15:26.259 of a distinction between older buildings and newer buildings, 00:15:26.259 --> 00:15:29.829 but not a clear one in terms of what the right height limit is. 00:15:29.829 --> 00:15:35.629 So in the red is all of the buildings that would match the – be above 240 feet. 00:15:35.629 --> 00:15:40.180 That’s about 20%. And then the other bands of color 00:15:40.180 --> 00:15:43.900 show how much fall into those particular height categories. 00:15:43.900 --> 00:15:46.899 But, as you can see, we don’t really have a good way of 00:15:46.899 --> 00:15:50.940 measuring what it should be. And yet, we had to start somewhere, 00:15:50.940 --> 00:15:57.550 so what we did was, we started with 240 feet, made the database that way, 00:15:57.550 --> 00:16:01.250 did the study with that information, and then part of the recommendations 00:16:01.250 --> 00:16:06.999 that came out of it was to make this a lot deeper and go down to 75 feet. 00:16:07.000 --> 00:16:12.580 So that’s actually underway currently, but I’m not part of that study. 00:16:13.930 --> 00:16:17.820 So we now have our 240-foot-and-above buildings 00:16:17.839 --> 00:16:20.899 so we can start to look at what – how we can characterize them. 00:16:20.899 --> 00:16:23.980 And the first is by looking at the structural system. 00:16:23.980 --> 00:16:26.790 So the structural system is the way in which the building 00:16:26.790 --> 00:16:31.749 is going to resist lateral loads. So if you have a building, at all times, 00:16:31.749 --> 00:16:35.309 you’re going to have gravity loads – so it’s pushing down, so you have 00:16:35.309 --> 00:16:39.620 a vertical-resistant system. But then also, during an earthquake 00:16:39.620 --> 00:16:43.529 or during wind, you’re going to start to have lateral forces coming in. 00:16:43.529 --> 00:16:48.550 So the structural system is what resists that lateral forces. 00:16:48.550 --> 00:16:51.920 So we have it categorized in three different colors here. 00:16:51.920 --> 00:16:54.430 One is green for the steel moment frames. 00:16:54.430 --> 00:16:59.379 Essentially, what that means is that you have a steel column and a steel beam. 00:16:59.379 --> 00:17:02.660 And then the connection is where you have the resistance. 00:17:02.660 --> 00:17:06.630 So it’s a really stiff connection. It’s trying to always hold it at 00:17:06.630 --> 00:17:11.100 90 degrees, and that’s what’s keeping things from moving too much. 00:17:11.100 --> 00:17:15.459 Another option is the reinforced concrete shear wall. 00:17:15.459 --> 00:17:17.730 We have a few different options for what that looks like. 00:17:17.730 --> 00:17:20.050 Some of it may have a secondary system. 00:17:20.050 --> 00:17:22.960 Again, that has to do with that height limitation. 00:17:23.740 --> 00:17:27.920 But essentially, what a shear wall does is it’s a big reinforced concrete wall, 00:17:27.929 --> 00:17:31.700 and just the stiffness of the wall itself is what keeps it from moving. 00:17:31.700 --> 00:17:35.200 And then finally, we have braced frame systems, which is steel again. 00:17:35.200 --> 00:17:38.720 But, in this case, instead of having a really stiff connection right here, 00:17:38.720 --> 00:17:40.940 what they do is have a brace. 00:17:40.940 --> 00:17:45.860 And so, as it starts to move, that brace will pull everything back in line. 00:17:46.820 --> 00:17:51.520 And we can see here we have older buildings – I’m not showing 00:17:51.520 --> 00:17:54.900 the dates here, but it is fact that the older buildings are the 00:17:54.900 --> 00:17:59.450 ones that are north of Market Street, which would be right here. 00:17:59.450 --> 00:18:02.140 So the older buildings tend to be in green. 00:18:02.140 --> 00:18:04.750 So they’re the steel moment frame systems. 00:18:04.750 --> 00:18:06.840 And the newer buildings, which have been built south 00:18:06.840 --> 00:18:11.130 of Market Street, tend to be in red. So it’s the concrete shear wall systems. 00:18:11.800 --> 00:18:15.640 If we look at this date breakdown a little bit more closely, we can actually 00:18:15.640 --> 00:18:20.590 start to see a more interesting story. So here, just look at the main colors 00:18:20.590 --> 00:18:23.840 again. Again, the blue, red, and green. I’ve broken them out a little bit 00:18:23.840 --> 00:18:26.570 for more specific things, but essentially, it’s just 00:18:26.570 --> 00:18:29.120 those three colors that are of importance. 00:18:29.120 --> 00:18:32.940 And what we can do is start to see how things move through in time. 00:18:35.630 --> 00:18:41.220 So as our – the way that our earthquake engineering profession works, in part, 00:18:41.220 --> 00:18:44.140 is that, after an earthquake, we start to learn lessons. 00:18:44.140 --> 00:18:46.500 We take those lessons, and we put them into the building code. 00:18:46.500 --> 00:18:50.440 And then subsequent buildings have that built into it. 00:18:50.440 --> 00:18:57.360 So one thing that we see is that we have the San Fernando earthquake in 1971. 00:18:57.360 --> 00:19:01.059 And what this showed is that there was a lot of non-ductile behavior of 00:19:01.059 --> 00:19:03.990 reinforced concrete buildings. And so what it means is that it 00:19:03.990 --> 00:19:07.419 wasn’t detailed in such a way – so the reinforcement within the 00:19:07.419 --> 00:19:12.059 concrete wasn’t done in such a way that it can hold together well after it – 00:19:12.059 --> 00:19:14.910 after it starts to get damaged. Which essentially means that 00:19:14.910 --> 00:19:17.960 there isn’t a lot of reserve capacity after the damage begins. 00:19:17.960 --> 00:19:20.789 And that can be a really dangerous situation. 00:19:20.789 --> 00:19:24.400 So what they did in the codes was add a lot of detailing requirements to make 00:19:24.400 --> 00:19:29.299 sure that you do get ductile behavior. But you can also see that, in general, 00:19:29.299 --> 00:19:32.159 there was a lot of confidence lost in these kinds of systems. 00:19:32.159 --> 00:19:37.390 So after 1971, we see – at least at the very beginning, 00:19:37.390 --> 00:19:40.740 we see no reinforced concrete being built. 00:19:40.740 --> 00:19:44.000 Instead, we’re seeing those steel moment frames. 00:19:44.000 --> 00:19:46.200 And you can see a lot of them being built here, especially 00:19:46.200 --> 00:19:49.360 in the ’60s through the ’80s. There was a construction boom at 00:19:49.360 --> 00:19:55.320 the time, and that happened to be the structural system of choice. 00:19:55.320 --> 00:19:58.039 It was – seemed to be behaving well in different earthquakes. 00:19:58.039 --> 00:20:02.500 It was definitely the most economical choice. And so everyone was using it. 00:20:02.500 --> 00:20:08.320 And then, in 1994, there was the Northridge earthquake down near L.A. 00:20:08.320 --> 00:20:11.380 And, almost by accident, they found damage in these connections. 00:20:11.380 --> 00:20:15.400 So, again, in the steel moment frame system, it’s that connection itself 00:20:15.400 --> 00:20:19.630 that is giving us our lateral support. And what they found is that there 00:20:19.630 --> 00:20:25.510 was cracks in the welds between the beam and the columns. 00:20:25.510 --> 00:20:27.830 Which was really concerning. They found it by accident. 00:20:27.830 --> 00:20:29.940 So they said, let’s look at it more systematically. 00:20:29.940 --> 00:20:35.000 Let’s open up all the buildings and, you know, take off the fireproofing, 00:20:35.000 --> 00:20:38.850 take away all the partitions that would be hiding it, and look at it. 00:20:38.850 --> 00:20:42.169 And they found a lot of issues. And so what they did was put 00:20:42.169 --> 00:20:45.620 a moratorium on building with this technique. 00:20:45.620 --> 00:20:49.520 They, again, changed the building code designs to make sure that 00:20:49.520 --> 00:20:54.230 they were addressing this issue. But once again, we essentially saw 00:20:54.230 --> 00:20:59.799 a lot of loss of confidence in the system, and very few were built after that time. 00:20:59.800 --> 00:21:03.760 Note that all of the buildings that were built prior to Loma Prieta 00:21:03.760 --> 00:21:07.540 in 1989 were of this type. They just didn’t know that this 00:21:07.549 --> 00:21:10.279 kind of problem could happen, and they didn’t do that 00:21:10.279 --> 00:21:12.309 systematic opening up and checking. 00:21:12.309 --> 00:21:17.880 So it’s unclear as to what the status of the buildings in San Francisco are. 00:21:17.880 --> 00:21:20.870 And then the last thing that we have is that we start to see a lot more 00:21:20.870 --> 00:21:23.960 of the red buildings – the reinforced concrete buildings. 00:21:23.960 --> 00:21:28.860 And the darkest, which is just solely the reinforced concrete shear wall 00:21:28.860 --> 00:21:33.770 with no secondary system, we can see a big boom in the 2010s. 00:21:33.770 --> 00:21:39.420 And what that is – what that represents is the performance-based 00:21:39.420 --> 00:21:43.320 provisions that I described earlier. So it allows you to not use the 00:21:43.320 --> 00:21:47.130 prescriptive methods of the code but actually design it in another way and 00:21:47.130 --> 00:21:52.240 then prove that it’s having the same kind of building response. 00:21:52.240 --> 00:21:56.059 And so that has led to a lot more building of these 00:21:56.059 --> 00:21:58.440 reinforced concrete shear wall systems. 00:21:58.440 --> 00:22:04.200 It’s a really interesting development, and it seems like it’s going well. 00:22:04.880 --> 00:22:08.100 If you look, instead of at the structural systems, but at the occupancies – 00:22:08.110 --> 00:22:11.799 again, this is what is most of interest when we’re looking at the resilience 00:22:11.800 --> 00:22:16.000 of a city – we can see that we have a lot of office buildings. 00:22:16.000 --> 00:22:20.340 Over half of them are office buildings. About a quarter are residential. 00:22:20.340 --> 00:22:24.660 And then a quarter are split between mixed-use and hotel. 00:22:25.620 --> 00:22:29.280 If you look at this in terms of the breakdown in dates, you can see 00:22:29.289 --> 00:22:35.559 in the heat map on the right, everything above is – everything higher up is older, 00:22:35.559 --> 00:22:39.610 and then below is newer. So you can see that the offices 00:22:39.610 --> 00:22:43.210 were mostly built earlier, and the residential space 00:22:43.210 --> 00:22:47.010 is mostly built more recently. If you happen to remember the 00:22:47.010 --> 00:22:50.419 kind of map that I had given earlier, which showed that everything 00:22:50.419 --> 00:22:54.440 above Market was the steel buildings, and those were the older buildings, 00:22:54.440 --> 00:22:57.610 you can also see that that happens to be the office buildings now. 00:22:57.610 --> 00:23:01.110 And similarly, for the reinforced concrete shear wall systems 00:23:01.110 --> 00:23:05.580 that were built subsequently, down below Market Street, 00:23:05.580 --> 00:23:08.820 those happened to mostly be residential systems. 00:23:10.220 --> 00:23:16.300 So if we think about this in terms of different cohorts, we – so what are 00:23:16.309 --> 00:23:19.320 different groupings of the buildings, and how can we address them as 00:23:19.320 --> 00:23:24.320 distinct groups, we can see a few things. One is that we do have some of these 00:23:24.320 --> 00:23:29.080 pre-1980 non-ductile concrete systems. So, again, those are ones that don’t have 00:23:29.080 --> 00:23:34.020 the updated detailing requirements, and they don’t have the reserve capacity. 00:23:34.020 --> 00:23:37.260 There’s not a lot in this particular data set because we’re looking at 00:23:37.260 --> 00:23:43.330 240 feet, and it’s hard to build in that style above this amount. 00:23:44.520 --> 00:23:47.500 But we don’t want to forget that these are a really big problem. 00:23:47.510 --> 00:23:50.399 It’s, in fact, a much – it’s a small fraction of 00:23:50.399 --> 00:23:54.630 a much larger inventory that we would find if we were to go to lower feet. 00:23:54.630 --> 00:23:57.130 And, again, we are intending to go to lower feet. 00:23:57.130 --> 00:24:01.840 And so we want to make sure that these are getting addressed and flagged. 00:24:01.840 --> 00:24:06.019 There are ordinances in L.A. and Santa Monica and others being 00:24:06.019 --> 00:24:10.490 considered elsewhere for how to address and retrofit these buildings. 00:24:10.490 --> 00:24:13.520 So we definitely don’t want to forget about these ones. 00:24:13.520 --> 00:24:17.750 But the biggest chunk is our pre-Northridge steel-framed buildings – 00:24:17.750 --> 00:24:23.080 the ones that potentially have damage from the Loma Prieta earthquake. 00:24:23.080 --> 00:24:26.660 And then finally, we have the newer buildings that usually 00:24:26.669 --> 00:24:30.539 tend to be residential buildings made up of the concrete – 00:24:30.539 --> 00:24:33.279 reinforced concrete shear walls. 00:24:33.279 --> 00:24:35.880 So because we wanted to, in these recommendations, 00:24:35.880 --> 00:24:38.279 address what should be done for newer buildings, one of the things we 00:24:38.279 --> 00:24:42.809 did was to look at this kind of archetype. 00:24:42.809 --> 00:24:47.990 So we have a reinforced concrete shear wall residential building. 00:24:47.990 --> 00:24:55.110 We also did one other one that’s a buckling restrained braced frame. 00:24:55.110 --> 00:24:59.059 So it’s one of those braced frame systems that I was describing. 00:24:59.059 --> 00:25:03.299 And it’s coming online for being used in office buildings, especially. 00:25:03.299 --> 00:25:06.269 So we’ve designed these two buildings and wanted to see 00:25:06.269 --> 00:25:09.780 what kind of response they would have after an earthquake. 00:25:10.420 --> 00:25:14.169 So we did it in two different site classes in the downtown area, 00:25:14.169 --> 00:25:18.450 which is marked as Site Class D. You have much poorer soil, 00:25:18.450 --> 00:25:21.130 and so the ground motions would be amplified there. 00:25:21.130 --> 00:25:26.720 Whereas, as Site Class B, which is on Rincon Hill, that’s a much stiffer – 00:25:26.720 --> 00:25:30.730 almost rock – not quite rock. And so that’s Site Class B, 00:25:30.730 --> 00:25:35.529 and it’s going to not amplify the ground motions quite as much. 00:25:35.529 --> 00:25:40.500 So what we found is that, for Site Class D, the expected 00:25:40.500 --> 00:25:44.250 repair cost would be about 8% of the replacement value, so that’s just, 00:25:44.250 --> 00:25:46.679 straight up replacing the building, how much would that cost. 00:25:46.679 --> 00:25:50.880 It’s 8% of that for the reinforced concrete shear wall building, 00:25:50.880 --> 00:25:54.520 and then 3% of it for the BRBF building. 00:25:54.520 --> 00:25:57.970 This doesn’t seem too bad if it were just talking about the cost. 00:25:57.970 --> 00:26:00.820 When we’re talking about the recovery time, so how long 00:26:00.820 --> 00:26:04.559 it would take to get back into the building, it’s more concerning. 00:26:04.559 --> 00:26:08.750 So for the residential building, we have over five months. 00:26:08.750 --> 00:26:11.399 For the office building, we have over three months. 00:26:11.399 --> 00:26:14.570 And if we think back to the fact that San Francisco is targeting that we have 00:26:14.570 --> 00:26:19.220 50% of office space available after four months, we have an issue here. 00:26:19.220 --> 00:26:23.799 Because these are the best buildings – this is the newest buildings, the most – 00:26:23.799 --> 00:26:26.720 the ones that should be doing the best in terms of kind of 00:26:26.720 --> 00:26:29.970 evening out the problematic buildings that we would have had in the past. 00:26:29.970 --> 00:26:35.440 - To find the damage that you were [inaudible] the repair of, 00:26:35.440 --> 00:26:39.720 what was the ground motion that you used to damage the building, 00:26:39.720 --> 00:26:42.200 and what was the final state of the building? 00:26:42.200 --> 00:26:45.860 Like, for instance, total drift of the building and so on? 00:26:45.860 --> 00:26:49.620 - I didn’t put the drift on my slides here. I didn’t know that there would be 00:26:49.620 --> 00:26:52.370 someone who would ask about that in the audience. 00:26:52.370 --> 00:26:55.480 So these are well within limits. 00:26:56.309 --> 00:27:04.779 So I think that the BRBF was, like, 1% drift. 00:27:04.779 --> 00:27:11.781 And then – so it’s much less than the absolute limit, which would be 3% for 00:27:11.781 --> 00:27:15.700 our maximum considered earthquake, and then for this – this is a designed 00:27:15.700 --> 00:27:19.880 earthquake, so it would be more about – we’d expect – 2% would be the cutoff 00:27:19.880 --> 00:27:23.480 for the drift that would be scaled back at design. 00:27:23.480 --> 00:27:27.679 So it’s well within what we would be targeting 00:27:27.679 --> 00:27:32.299 with a code-based design just looking at safety. 00:27:32.299 --> 00:27:36.899 And yet, we’re seeing thsese functional issues – functional recovery time issues. 00:27:36.899 --> 00:27:41.009 So what that tells us now is that we want to figure out, is there 00:27:41.009 --> 00:27:44.929 a way to design better? Not just think about safety, 00:27:44.929 --> 00:27:47.919 whether or not these buildings could potentially injure the 00:27:47.919 --> 00:27:51.049 people inside of them, and think a little bit farther ahead 00:27:51.049 --> 00:27:55.039 and think about what the recovery would be of these buildings. 00:27:55.039 --> 00:28:00.139 So in terms of our recommendations, they come in four categories. 00:28:00.139 --> 00:28:04.610 The first one is, what actions can be taken prior to an earthquake 00:28:04.610 --> 00:28:07.240 for reducing the risk specific to new buildings? 00:28:07.240 --> 00:28:09.540 And there’s two recommendations that come in this category. 00:28:09.540 --> 00:28:13.679 I’m only looking at one. It’s to establish these recovery-based 00:28:13.679 --> 00:28:17.080 seismic design standards. Again, currently in the code, 00:28:17.080 --> 00:28:20.059 we only look at whether or not the building is safe. 00:28:20.059 --> 00:28:23.820 It would be better, when we’re trying to think about these resilience issues, 00:28:23.820 --> 00:28:26.980 to think more about, how can we get a faster recovery? 00:28:26.980 --> 00:28:31.419 And so the recommendation here is to think about what we want and establish 00:28:31.419 --> 00:28:37.289 those new design standards and figure out what kinds of things in the code 00:28:37.289 --> 00:28:42.140 would drive what we’re wanting in terms of reducing the recovery time. 00:28:42.980 --> 00:28:46.780 Second, we have actions for reducing the risk prior to earthquakes, 00:28:46.790 --> 00:28:48.990 but now thinking about our existing buildings. 00:28:48.990 --> 00:28:52.659 We have four recommendations in this category. 00:28:52.659 --> 00:28:57.440 One of them is to enforce the repair provisions that are already required 00:28:57.440 --> 00:29:01.940 with respect to any damage that may have occurred due to Loma Prieta. 00:29:01.940 --> 00:29:06.529 But they’re not being enforced. And so we want to make sure that 00:29:06.529 --> 00:29:12.769 the city is having the owners go back, look at the damage, see if there is any. 00:29:12.769 --> 00:29:16.600 If there is any, address it. It’s entirely possible that the owners 00:29:16.600 --> 00:29:20.269 have already done this themselves. It just hasn’t been reported to the city, 00:29:20.269 --> 00:29:23.200 and so the city doesn’t know. And so this would be a mechanism 00:29:23.200 --> 00:29:26.160 for getting a bit better insight into what exactly 00:29:26.160 --> 00:29:28.780 the current status of the building stock is. 00:29:28.780 --> 00:29:34.420 - But if they weren’t – [inaudible] fracture the welds in Loma Prieta 00:29:34.420 --> 00:29:39.320 and they’re pre-Northridge welds, if you get a big earthquake, 00:29:39.320 --> 00:29:43.540 they will certainly fracture the welds. Doesn’t matter where the – 00:29:43.540 --> 00:29:47.940 nobody ever found one of those pre-Northridge welds that [inaudible] 00:29:47.940 --> 00:29:52.280 enough to actually [inaudible]. So all you have to know 00:29:52.280 --> 00:29:54.700 is it’s pre-Northridge. - Yeah. 00:29:54.700 --> 00:29:56.280 - And it’ll fracture. 00:29:56.280 --> 00:30:00.420 - Could you repeat the comment or question or … 00:30:01.080 --> 00:30:04.660 - Oh, yeah. The comment is that – so here we’re saying, well, 00:30:04.679 --> 00:30:08.090 we need to at least look and see whether there was damage. 00:30:08.090 --> 00:30:11.559 And if there was damage, to address it. And the point was, Loma Prieta 00:30:11.559 --> 00:30:14.640 was a very small earthquake. If we had a larger earthquake, 00:30:14.640 --> 00:30:18.600 without question, we’re going to have issues in those connections. 00:30:18.600 --> 00:30:22.710 So the comment was, really, we should be looking at it whether or not they were 00:30:22.710 --> 00:30:27.040 damaged in Loma Prieta. Is that … - Well, it’s just that they’re pre-’94. 00:30:27.040 --> 00:30:28.780 We should assume that … - Yeah. 00:30:28.780 --> 00:30:32.340 - … [inaudible] shaking, they will fail. - Yeah. 00:30:36.029 --> 00:30:41.980 So at – yeah, good point. But then the next thing we’re looking at is actions 00:30:41.990 --> 00:30:46.840 to improve the city’s understanding of its tall building seismic risk. 00:30:46.840 --> 00:30:50.590 And this is what we can do further on to continue these studies. 00:30:50.590 --> 00:30:52.929 Two recommendations here. 00:30:52.929 --> 00:30:55.610 But the one that I’m showing is that we want to develop a comprehensive 00:30:55.610 --> 00:31:00.320 recovery plan for the Financial District and the adjacent neighborhoods. 00:31:00.320 --> 00:31:05.039 So what this would look like is to make sure that we are looking at all the actors 00:31:05.039 --> 00:31:10.039 that are in place, all the businesses, all the people, all the owners, 00:31:10.039 --> 00:31:14.940 all of the infrastructure systems, anything that could go on in terms of 00:31:14.940 --> 00:31:19.269 cordons, and figure out ahead of time, what is the best way to move forward 00:31:19.269 --> 00:31:23.440 so that we have taken care of it and we can just kind of start as soon as 00:31:23.440 --> 00:31:27.660 the earthquake occurs. We’ve developed our plan ahead of time. 00:31:29.590 --> 00:31:32.760 So these recommendations were released last October. 00:31:32.769 --> 00:31:36.789 And then, this past month, the mayor signed an executive order directing 00:31:36.789 --> 00:31:39.710 lots of departments to actually enact these recommendations. 00:31:39.710 --> 00:31:42.370 Those departments include Building Inspection, 00:31:42.370 --> 00:31:46.029 Emergency Management, and others. 00:31:46.029 --> 00:31:49.590 But a question that I’m now considering is, what kinds of tools 00:31:49.590 --> 00:31:53.030 can we use to leverage this database when we start to consider 00:31:53.030 --> 00:31:57.260 how the cordons are actually going to affect the city’s recovery? 00:31:58.540 --> 00:32:02.440 So here at USGS, you all are familiar with the HayWired scenario, 00:32:02.440 --> 00:32:04.960 which does a great job of looking at what are the 00:32:04.960 --> 00:32:09.950 aggregate losses that are going to occur during an event. 00:32:09.950 --> 00:32:13.940 But we don’t have a way to start to look at the individual buildings and 00:32:13.940 --> 00:32:17.470 how they would start, you know, causing cordons and interacting with 00:32:17.470 --> 00:32:21.660 the buildings around them when we’re looking at the census tract level. 00:32:21.660 --> 00:32:26.660 There’s a move to start looking at high-resolution regional simulations. 00:32:26.660 --> 00:32:29.960 SimCenter is a center at Berkeley where they’re trying to provide 00:32:29.960 --> 00:32:33.710 a platform for researchers where they can have this end-to-end 00:32:33.710 --> 00:32:38.800 high-resolution regional simulations. And it’s kind of modular. 00:32:38.800 --> 00:32:43.639 So if the researcher has a way that they can start to enhance the module in this 00:32:43.639 --> 00:32:47.049 particular portion – so let’s say they’re looking at the damage to the building 00:32:47.049 --> 00:32:51.059 as opposed to the earthquake shaking – then they can just plug and play and 00:32:51.059 --> 00:32:54.840 put their module right in there, and then they have an end-to-end system. 00:32:54.840 --> 00:32:58.760 So it’s a really exciting project that’s going on. 00:32:58.760 --> 00:33:01.080 And what they’ve done here is taken the same shaking 00:33:01.080 --> 00:33:08.180 that was put into HayWired and looked at it at this building-level – 00:33:08.180 --> 00:33:11.799 you know, building-level resolution. 00:33:11.799 --> 00:33:14.590 And so you can start to see that you can get much more specific 00:33:14.590 --> 00:33:18.769 as to where the damage is occurring. If we actually zoom into San Francisco, 00:33:18.769 --> 00:33:22.299 you can see these really are different buildings, and you could use this to 00:33:22.299 --> 00:33:28.460 tailor different policy recommendations. Maybe it’s a particular area of the city. 00:33:28.460 --> 00:33:33.130 Maybe it’s a particular kind of building. And these become not only a way 00:33:33.130 --> 00:33:37.160 to identify what those issues are, but also a really good way to communicate 00:33:37.160 --> 00:33:41.720 those to the public as they’re trying to work with these policies. 00:33:43.260 --> 00:33:47.960 So then what I want to be able to do is have a framework that’s going to 00:33:47.960 --> 00:33:51.880 follow this schematic, which I’m going to describe, and then I’m going to 00:33:51.880 --> 00:33:56.470 use this to step through each of the steps as we go through. 00:33:56.470 --> 00:34:00.679 So essentially, we want to have urban exposure, what is the building inventory 00:34:00.679 --> 00:34:05.360 that we’re looking at, and then what is the hazard that it’s being exposed to. 00:34:05.360 --> 00:34:08.159 We combine those in the form of vulnerability. 00:34:08.159 --> 00:34:11.780 So we can say, how vulnerable is the particular building? 00:34:11.780 --> 00:34:15.260 What damage is going to happen to it given this hazard? 00:34:15.260 --> 00:34:18.320 Once we have the damage, we can look at impeding factors. 00:34:18.320 --> 00:34:21.780 Impeding factors have also been called irrational components 00:34:21.780 --> 00:34:26.060 of downtime by Mary Comerio. Essentially, it’s anything that 00:34:26.060 --> 00:34:30.590 isn’t repair – the actual time it takes to repair things – that’s also 00:34:30.590 --> 00:34:34.290 contributing to the downtime. So that might be something like getting 00:34:34.290 --> 00:34:38.710 the financing handled, getting the contractor ready, and things like that. 00:34:38.710 --> 00:34:41.680 So together, the damage, the repair from the damage, 00:34:41.680 --> 00:34:45.620 and the impeding factors contribute to the downtime. 00:34:45.620 --> 00:34:49.190 And what I want to do is be able to use the community damage to 00:34:49.190 --> 00:34:54.090 inform cordons, which would then feed back in as impeding factors. 00:34:54.090 --> 00:34:56.620 So I’m going to start stepping through this. 00:34:56.620 --> 00:34:59.460 And first we’re going to look at urban exposure. 00:35:03.000 --> 00:35:05.650 So this is the same map that I had shown before. 00:35:05.650 --> 00:35:11.360 We have the red buildings – the tallest buildings with a lot of detail, but we 00:35:11.360 --> 00:35:13.860 don’t have all the buildings around that. And we’re – when we’re thinking about 00:35:13.860 --> 00:35:18.270 the cordons, we need to think about, what are the buildings around them. 00:35:18.270 --> 00:35:21.890 So I mentioned that we have this Open Data Portal information. 00:35:21.890 --> 00:35:25.770 It wasn’t good for the level of resolution we were looking at for that building 00:35:25.770 --> 00:35:29.490 study, but we can use it for this level of information. 00:35:29.490 --> 00:35:35.300 And so I combined three different data sets from there. 00:35:35.300 --> 00:35:38.290 One is property tax rolls. Another is land use. 00:35:38.290 --> 00:35:41.640 And then building footprints. It was difficult because they 00:35:41.640 --> 00:35:44.760 all have different basic units. Some are the parcel level, which 00:35:44.760 --> 00:35:50.780 means it depends on who the owner is. If you have – if you have a 00:35:50.780 --> 00:35:53.500 condominium building, there’s a lot of owners in that building, 00:35:53.500 --> 00:35:57.570 and so you have to figure out how to flatten these things 00:35:57.570 --> 00:36:02.530 and make them work together. But essentially, I was able to do it, 00:36:02.530 --> 00:36:09.670 making a lot of assumptions and decisions, but what it looks like is this. 00:36:09.670 --> 00:36:13.620 So we can see that we start to know where different occupancies are, 00:36:13.620 --> 00:36:17.390 what kinds of materials are being used throughout the city. 00:36:17.390 --> 00:36:20.590 And we ended up with a data set that aggregated the information 00:36:20.590 --> 00:36:25.520 to get a date for each building, a height, occupancy, the square footage, 00:36:25.520 --> 00:36:28.410 and then the construction material. 00:36:28.410 --> 00:36:33.220 Now that we have the urban exposure, we can start to think about our hazard. 00:36:33.220 --> 00:36:36.290 So what we’re doing is using earthquake rupture scenarios. 00:36:36.290 --> 00:36:39.650 So taking a rupture and then seeing how the ground motion is going to 00:36:39.650 --> 00:36:41.460 be propagated through the ground. 00:36:41.460 --> 00:36:45.180 And the purpose of that is that we can reflect the variation of the shaking 00:36:45.180 --> 00:36:48.630 intensities that would be experienced across the community. 00:36:48.630 --> 00:36:50.160 So one obvious way in which there would be 00:36:50.160 --> 00:36:53.240 variation is the location within the city. 00:36:53.240 --> 00:36:56.730 Depending on where you are, you’re going to have different ground motion. 00:36:56.730 --> 00:37:00.510 But also we have different soil stiffnesses. 00:37:00.510 --> 00:37:03.870 So we might have one building that is on stiff rock and 00:37:03.870 --> 00:37:07.640 one that’s on a really soft soil. So one way that we can handle that 00:37:07.640 --> 00:37:13.450 is to take – do our simulation of our ground motions at 00:37:13.450 --> 00:37:16.570 different locations and make sure that we are, in fact, 00:37:16.570 --> 00:37:21.300 getting a representation of each of those different soil conditions. 00:37:21.300 --> 00:37:27.440 So let’s say, for example, that we have a building right here on this rock site. 00:37:27.440 --> 00:37:32.660 What we would do is sample our ground motions from this point here. 00:37:32.660 --> 00:37:36.500 Whereas, if we were over here, even though we’re closest to this rock site, 00:37:36.500 --> 00:37:41.090 we would make sure that we’re sampling from the similar soil type. 00:37:42.720 --> 00:37:48.000 What I’m most interested in terms of the variation across the community 00:37:48.000 --> 00:37:52.380 is the individual building properties. So because we have so many different 00:37:52.380 --> 00:37:55.040 kinds of buildings, so many different heights, so many different 00:37:55.040 --> 00:37:59.500 construction materials, we’re going to have a lot of different response. 00:37:59.500 --> 00:38:02.760 And in order to characterize the dynamic building response, 00:38:02.760 --> 00:38:06.110 what we really need to be doing is having a range of intensity measures. 00:38:06.110 --> 00:38:09.380 We can’t just use peak ground acceleration only. 00:38:09.380 --> 00:38:13.590 So to describe what I do, I’m going to just look at one particular site, 00:38:13.590 --> 00:38:17.620 remembering that we actually have it for all of these sites here. 00:38:17.620 --> 00:38:22.680 And we’re going to look at a San Andreas rupture and a Hayward rupture. 00:38:22.680 --> 00:38:26.180 And we’re going to characterize it by spectral acceleration. 00:38:26.190 --> 00:38:29.570 So this is a response spectrum. And what it says is that, 00:38:29.570 --> 00:38:33.520 for a building with a 1-second period, we’re going to experience 00:38:33.520 --> 00:38:38.150 a certain level of spectral acceleration. Whereas, if we have a different 00:38:38.150 --> 00:38:41.120 building – maybe it’s a taller building with a 3-second period – 00:38:41.120 --> 00:38:43.320 it’s actually going to be lower. 00:38:43.320 --> 00:38:46.670 And so this allows us to get a lot more nuance into how the 00:38:46.670 --> 00:38:51.540 different buildings are going to behave due to the ground motion. 00:38:51.540 --> 00:38:54.920 So the way that we’re doing that is to use OpenSHA to get 00:38:54.920 --> 00:38:58.480 rupture medians and standard deviations at each location. 00:38:58.480 --> 00:39:01.710 Again, recognizing what the soil type is there. 00:39:01.710 --> 00:39:05.130 And then we’re using spatial and multi-period correlation models 00:39:05.130 --> 00:39:08.400 on top of those medians and standard deviations in order to 00:39:08.400 --> 00:39:13.710 generate rupture-consistent response spectra at all of our locations. 00:39:13.710 --> 00:39:16.760 One thing to note is that, because we do know that there is a standard 00:39:16.760 --> 00:39:21.080 deviation around the median, we can actually, for each rupture, 00:39:21.080 --> 00:39:24.510 make a lot of different simulations and catch all of that potential 00:39:24.510 --> 00:39:28.600 variation in the ground motion due to a single rupture. 00:39:29.910 --> 00:39:32.320 So we have our urban exposure and now our hazard. 00:39:32.320 --> 00:39:36.950 How can we connect them to get our vulnerability and therefore our damage? 00:39:36.950 --> 00:39:39.630 So I’m now switching to a hypothetical community because 00:39:39.630 --> 00:39:43.820 it’s a little bit more manageable for the purpose of this conversation. 00:39:43.820 --> 00:39:48.000 But if we have this community where we know what our date is, 00:39:48.000 --> 00:39:52.770 we know what our structural types is, how tall the buildings are, we can 00:39:52.770 --> 00:39:58.140 use FEMA P-58 to start to get some damage and repair time simulations. 00:39:58.140 --> 00:40:03.200 So what we do here is take the spectral acceleration for each building – again, 00:40:03.200 --> 00:40:09.980 for each rupture simulation – and we insert that into the FEMA P-58 model. 00:40:09.980 --> 00:40:13.410 That gives us a structural response. How much is the building, for example, 00:40:13.410 --> 00:40:17.330 going to sway during the earthquake? That information about how the 00:40:17.330 --> 00:40:20.400 building responds then informs component damage. 00:40:20.400 --> 00:40:24.030 So if we have, maybe, some structural components, 00:40:24.030 --> 00:40:26.309 some partition walls, some light fixtures, 00:40:26.309 --> 00:40:31.150 how is that swaying going to affect the components? 00:40:31.150 --> 00:40:34.140 And then, given the damage that is accrued, 00:40:34.140 --> 00:40:36.820 how long is it going to take to repair? 00:40:36.820 --> 00:40:42.120 So, by using this for all of the buildings, we can start to get a map of the time to 00:40:42.130 --> 00:40:48.020 functional recovery for our scenario – for our entire community. 00:40:48.020 --> 00:40:52.190 One thing to note is that the way that FEMA P-58 handles the uncertainty 00:40:52.190 --> 00:40:56.540 that’s going on here is that it does a lot of different realizations. 00:40:56.540 --> 00:41:00.040 So it’s a Monte Carlo simulation. You might have, let’s say, 00:41:00.040 --> 00:41:03.600 1,000 different realizations of what could happen to the building. 00:41:03.600 --> 00:41:08.040 And so we’re carrying this through to the community analysis by, similarly, 00:41:08.040 --> 00:41:11.280 having a lot of different realizations of the community damage. 00:41:11.280 --> 00:41:15.520 So anytime I show a map, that’s one of many possible maps. 00:41:16.700 --> 00:41:20.460 Once we have our community damage, we can think about what cordons 00:41:20.460 --> 00:41:26.050 are going to be set up. So if we have a damaged and 00:41:26.050 --> 00:41:29.740 leaning tall building, denoted here by the dark red, we might have 00:41:29.740 --> 00:41:31.500 a safety cordon that’s placed around it. 00:41:31.500 --> 00:41:36.240 And everything within that cordon is going to be inaccessible. 00:41:36.240 --> 00:41:39.870 So we could think, first of all, about what different kinds of cordons 00:41:39.870 --> 00:41:44.070 have happened in the past. Two examples, both from New Zealand, 00:41:44.070 --> 00:41:47.580 are Christchurch in 2011 and Wellington, after the 00:41:47.580 --> 00:41:52.140 Kaikoura earthquake in 2016. So in Christchurch, what they did 00:41:52.140 --> 00:41:57.720 is they just put up a huge cordon around the Central Business District. 00:41:57.720 --> 00:42:00.820 And then eventually, over time, they kind of shrunk that. 00:42:00.820 --> 00:42:03.520 In Wellington, they took a really different approach. 00:42:03.520 --> 00:42:07.080 Partly because it was less shaking. It wasn’t directly underneath. 00:42:07.080 --> 00:42:10.530 So it wasn’t as bad. But also because they were really concerned. 00:42:10.530 --> 00:42:13.180 They didn’t want to get into the situation that Christchurch had. 00:42:13.180 --> 00:42:15.990 They had seen how bad it was for the economy, 00:42:15.990 --> 00:42:19.150 and they just wanted to avoid that if possible. 00:42:19.150 --> 00:42:22.430 So instead, what they did was took the individual buildings 00:42:22.430 --> 00:42:26.310 and marked which buildings around it would be inaccessible. 00:42:26.310 --> 00:42:27.770 You can see a blue building there. 00:42:27.770 --> 00:42:30.260 That’s actually accessible from one side but not the other. 00:42:30.260 --> 00:42:34.260 So it’s a very different approach to how to address these cordons. 00:42:34.260 --> 00:42:38.540 So therefore, as we’re trying to give decision-makers tools for knowing 00:42:38.540 --> 00:42:43.490 what best to do, it would be really useful to understand the ramifications of these 00:42:43.490 --> 00:42:46.870 different potentials – these different cordoning choices. 00:42:46.870 --> 00:42:50.860 As opposed to telling them, well, you have a 30% chance of needing 00:42:50.860 --> 00:42:55.330 this kind of cordon and a 90 – or, 70% chance of having this kind of cordon, 00:42:55.330 --> 00:42:59.860 we can actually say, if you choose one or the other, this is the ramifications. 00:42:59.860 --> 00:43:04.400 So the different options here are, what is going to trigger a cordon, how 00:43:04.400 --> 00:43:09.580 big the cordon is going to be, and then how long that cordon is going to last for. 00:43:12.780 --> 00:43:17.060 Because we had that much better detailed understanding of what these 00:43:17.060 --> 00:43:21.150 buildings are, we can do high-resolution modeling of these tall buildings. 00:43:21.150 --> 00:43:25.210 And that’s going to support the more detailed analysis that we 00:43:25.210 --> 00:43:28.800 really need for thinking about how these cordons would play out. 00:43:28.800 --> 00:43:31.850 So, again, we’re still using FEMA P-58, but now we’re 00:43:31.850 --> 00:43:35.760 not using spectral accelerations. We’re actually using acceleration time 00:43:35.760 --> 00:43:41.330 histories, putting that into a nonlinear model of the building itself to figure out 00:43:41.330 --> 00:43:44.260 what the response – the structural response would be 00:43:44.260 --> 00:43:47.530 in a much more nuanced way. That, then, feeds, again, 00:43:47.530 --> 00:43:50.850 into the component damage and then the repair time. 00:43:50.850 --> 00:43:53.800 And what we get out of this is a residual drift – so that 00:43:53.800 --> 00:43:57.640 comes from the structural response. And this, again, is essentially 00:43:57.640 --> 00:44:00.680 how much the building is leaning after the earthquake. 00:44:00.680 --> 00:44:04.310 Additionally, we can get a repair time for the repairs 00:44:04.310 --> 00:44:08.890 that would be required for stabilization. So what I mean by this is what it 00:44:08.890 --> 00:44:12.050 would take to make the building not dangerous to any other building. 00:44:12.050 --> 00:44:16.090 That’s only going to be thing like structural repairs and exterior clotting. 00:44:16.090 --> 00:44:19.280 We don’t care about whether or not the building itself is usable. 00:44:19.280 --> 00:44:23.250 Just whether it’s dangerous to the buildings around it. 00:44:23.250 --> 00:44:25.870 So with the residual drift, what we can do is figure out 00:44:25.870 --> 00:44:29.730 a threshold that we would use for defining that cordon trigger. 00:44:29.730 --> 00:44:33.260 And the reason why we’re using the residual drift is that it’s 00:44:33.260 --> 00:44:36.900 a good way to characterize the probability of collapse in an earthquake. 00:44:36.900 --> 00:44:40.420 The more damaged it is, the more leaning it would be, 00:44:40.420 --> 00:44:45.050 and the more likely that it would collapse in an aftershock. 00:44:45.050 --> 00:44:47.770 If we have, in fact, triggered a cordon, we have to figure out what the 00:44:47.770 --> 00:44:51.910 duration of the cordon is going to be, and so that depends on this 00:44:51.910 --> 00:44:55.300 repair time to how long it would take to stabilize the building. 00:44:57.480 --> 00:45:00.940 Once we figured out, based on the damage, where the cordons are, 00:45:00.950 --> 00:45:04.030 we can take those cordons and apply them as impeding factors 00:45:04.030 --> 00:45:06.830 to all of the buildings around them. 00:45:06.830 --> 00:45:10.560 So to describe this, I first have to explain what impeding factors are. 00:45:10.560 --> 00:45:13.970 I’m using the model from the REDi Rating System, 00:45:13.970 --> 00:45:17.230 which was developed by ARUP. And essentially what it does 00:45:17.230 --> 00:45:22.820 is has three different paths of what could delay the repair. 00:45:22.820 --> 00:45:26.080 So we have financing or we have engineering mobilization 00:45:26.080 --> 00:45:29.370 and permitting or we have contractor mobilization. 00:45:29.370 --> 00:45:33.700 And essentially, these all are – these logistics are happening in parallel, 00:45:33.700 --> 00:45:36.440 and whichever one takes the longest is going to be the controlling 00:45:36.440 --> 00:45:39.020 factor on when we can start the repair. 00:45:39.020 --> 00:45:43.030 Once the repair is done, that entire time is what we call the downtime. 00:45:43.030 --> 00:45:48.600 So what this looks like is that we would have these three parallel paths. 00:45:48.600 --> 00:45:53.490 In this case, contractor mobilization is controlling at 23 weeks, 00:45:53.490 --> 00:45:55.860 which means repair can start at that time. 00:45:55.860 --> 00:46:01.320 Repair takes 24 weeks, so we have a total of 47 weeks of downtime. 00:46:01.320 --> 00:46:05.580 If, however, we know that there is a cordon that is thrown across this 00:46:05.590 --> 00:46:08.360 building, and we know what the duration of that cordon is from the 00:46:08.360 --> 00:46:11.360 previous slide, we can add that in here. 00:46:11.360 --> 00:46:16.560 If that cordon happens to be longer than the longest impeding factor, then we’re 00:46:16.560 --> 00:46:21.060 going to have an additional margin before we can start the repairs. 00:46:21.060 --> 00:46:25.120 So here on the bottom, you can see the time that is due to its own impeding 00:46:25.120 --> 00:46:30.200 factors, a little bit of time for the cordon, and then the time for repair. 00:46:30.200 --> 00:46:33.030 If we think about – this is just for one building, but if we want to think about it 00:46:33.030 --> 00:46:38.080 for all of the buildings, we can take that little bar and turn it up on its end. 00:46:38.080 --> 00:46:41.540 And I’ve marked here 36%. That’s the percentage of the 00:46:41.540 --> 00:46:46.110 downtime for this building that was due to the cordon itself. 00:46:46.110 --> 00:46:49.500 If we look at another building next to it, we might have a different percentage. 00:46:49.500 --> 00:46:52.720 And then another building. But what we start to see is that the 00:46:52.720 --> 00:46:55.671 cordon all gets lifted at the same time. So that means that these were all in 00:46:55.671 --> 00:46:59.270 the same cordon, happened to get lifted at that time, 00:46:59.270 --> 00:47:02.500 but each building is affected to a different degree in terms of 00:47:02.500 --> 00:47:06.300 the proportion of its downtime that is being caused by that cordon. 00:47:06.300 --> 00:47:10.910 So if we look at this in the map of our little hypothetical community, 00:47:10.910 --> 00:47:15.470 we can see that we have four buildings that needed cordons, and each one 00:47:15.470 --> 00:47:20.240 has a different time before it’s released. And, based on the percentages that 00:47:20.240 --> 00:47:24.860 I described on the left, we have – the buildings within those cordons 00:47:24.860 --> 00:47:29.980 are mapped with the darkness of the green denoting how much 00:47:29.980 --> 00:47:33.640 of the percentage of downtime was due to the cordon. 00:47:33.640 --> 00:47:38.180 If we then make the left plot for this actual case, we can 00:47:38.180 --> 00:47:42.610 start to see this interesting stepping behavior where 00:47:42.610 --> 00:47:45.540 our cordons are getting lifted at different times. 00:47:45.540 --> 00:47:51.320 So our first cordon lifted is for this building, the leftern-most one. 00:47:51.320 --> 00:47:55.510 And it’s released, and we see a lot of buildings that can then 00:47:55.510 --> 00:47:59.840 start to be repaired right after that because it just so happens that there’s 00:47:59.840 --> 00:48:06.250 not a lot of overlap in this circle. The final one to be released is this one here. 00:48:06.250 --> 00:48:10.130 And so, again, we see a lot of buildings that are released at that time. 00:48:10.130 --> 00:48:12.770 But it’s interesting to see that, even though these cordons 00:48:12.770 --> 00:48:17.140 all have the same size, those two intermediate steps are much shorter – 00:48:17.140 --> 00:48:22.060 or, many fewer buildings are getting released when the cordon is let up. 00:48:22.060 --> 00:48:26.180 And the reason for that is that – so this is the second one to be let up. 00:48:26.180 --> 00:48:32.400 And we can see that, of the circle, there’s – not all of it going to be 00:48:32.400 --> 00:48:35.550 released because we have this overlapping circle from the other one. 00:48:35.550 --> 00:48:39.730 So only a portion of the buildings under this cordon are going to 00:48:39.730 --> 00:48:42.940 be accessible again. 00:48:42.940 --> 00:48:46.290 For the second – for the third building, we actually see that there’s 00:48:46.290 --> 00:48:49.840 a very, very small amount of its buildings – just those in this 00:48:49.840 --> 00:48:54.310 part of the circle here – that are going to be affected 00:48:54.310 --> 00:48:57.750 by that release at that moment. And otherwise, all the buildings 00:48:57.750 --> 00:49:02.690 that are near it are additionally affected by that longest cordon. 00:49:02.690 --> 00:49:05.030 So what we can start to see is that there’s this really interesting 00:49:05.030 --> 00:49:07.350 interplay of which cordons are next to 00:49:07.350 --> 00:49:10.800 which cordons and how each building is affected. 00:49:10.800 --> 00:49:14.820 But this isn’t super useful for looking at things holistically. 00:49:14.820 --> 00:49:19.120 It’s interesting to look at conceptually, but in terms of quantifying the 00:49:19.120 --> 00:49:23.300 downtime, we’re going to have to use a different kind of metric. 00:49:23.300 --> 00:49:28.990 So for looking at the downtime, we’re using what’s called recovery curves. 00:49:28.990 --> 00:49:33.420 So what we have here is the community’s recovery plotted against 00:49:33.420 --> 00:49:36.830 time on the – on the X axis. And then we have some kind of 00:49:36.830 --> 00:49:41.320 a recovery metric. In this case, we’re using the available office space. 00:49:41.320 --> 00:49:45.580 That’s a percentage of the pre-event office space 00:49:45.580 --> 00:49:47.280 is the way that I’m doing it here. 00:49:47.280 --> 00:49:51.170 So, at the time of the earthquake, suddenly we lose a lot of space. 00:49:51.170 --> 00:49:54.400 But then, over time, we’re getting access to it again, 00:49:54.400 --> 00:49:58.620 and so our curve is moving back up to 100%. 00:49:58.620 --> 00:50:04.640 And the way that we characterize loss in this is it’s the area above that curve 00:50:04.640 --> 00:50:08.860 is the lost function to the community due to this earthquake. 00:50:08.860 --> 00:50:11.560 And we can start to compare these different curves. 00:50:11.560 --> 00:50:17.841 So if we take the blue curve as the one that would be – the downtime 00:50:17.841 --> 00:50:21.890 that considers the fact that we’re also going to have cordons – and compare it 00:50:21.890 --> 00:50:27.180 to the red curve, which is ignoring these cordon access restrictions, we can see 00:50:27.180 --> 00:50:32.160 that the area between those two curves is going to be the loss to the community 00:50:32.160 --> 00:50:37.500 that’s based specifically on the cordon-induced access restrictions. 00:50:38.020 --> 00:50:41.780 And so this gives us the first step of what we’re trying to do in the story. 00:50:41.790 --> 00:50:45.670 Again, it’s plotting where the current X is given current conditions. 00:50:45.670 --> 00:50:48.650 And the horizontal lines, those thresholds of interest 00:50:48.650 --> 00:50:56.540 at 30, 60, and 90%, are useful because we can say, at this time, 00:50:56.540 --> 00:50:59.720 we got back 60% of our office space. 00:50:59.720 --> 00:51:05.410 Or, at this time, we got back 90%. But, again, we’re not just interested 00:51:05.410 --> 00:51:09.250 in figuring out what is the current conditions, where is that X. 00:51:09.250 --> 00:51:12.510 We also want to figure out how we can help policymakers to push 00:51:12.510 --> 00:51:16.440 these curves upward and essentially reduce the amount of area that’s above 00:51:16.440 --> 00:51:21.140 the curve, which is – again, that denotes the loss to the community. 00:51:22.520 --> 00:51:25.620 So specifically in terms of the cordon-related downtime, 00:51:25.620 --> 00:51:28.270 there’s a couple things that we could do. 00:51:28.270 --> 00:51:32.290 One of it is to reduce the duration of the cordon. So not change which 00:51:32.290 --> 00:51:37.200 cordons are put up, but make sure that we can get them down sooner. 00:51:37.200 --> 00:51:42.220 So we can do this with financial or logistical mitigation measures. 00:51:42.220 --> 00:51:47.290 For example, I had already described how these impeding factors work. 00:51:47.290 --> 00:51:50.960 And I described them in the context of one of the nearby buildings 00:51:50.960 --> 00:51:54.620 and how we would add in the cordon as a different impeding factor. 00:51:54.620 --> 00:51:57.090 But also, the buildings that are getting repaired – the buildings 00:51:57.090 --> 00:52:01.650 that are causing these cordons – also have these same impeding factors. 00:52:01.650 --> 00:52:04.050 So here we have our repair is a shorter time because 00:52:04.050 --> 00:52:06.890 we’re looking just at the stabilization repairs, 00:52:06.890 --> 00:52:13.100 but we still have to manage getting all the logistics taken care of. 00:52:13.100 --> 00:52:16.990 So if we were to find a way to reduce the contractor mobilization, suddenly, 00:52:16.990 --> 00:52:22.150 it would be the engineering mobilization track that would be slowing us down. 00:52:22.150 --> 00:52:24.670 And we have recovered – I can’t quite see that. 00:52:24.670 --> 00:52:28.730 I think it’s three weeks. If we additionally help out with 00:52:28.730 --> 00:52:32.390 the engineering mobilization, we can reduce it again, and now it’ll be the 00:52:32.390 --> 00:52:37.640 financing that is what’s stopping us – or, what’s the controlling factor. 00:52:37.640 --> 00:52:41.490 But what it’s done is reduced the amount of downtime, 00:52:41.490 --> 00:52:44.690 reduced the length of the cordon, which will then affect all of 00:52:44.690 --> 00:52:48.060 the buildings around it and get them back online that much quicker. 00:52:48.060 --> 00:52:52.870 So we can plot this as the yellow line in between them. 00:52:52.870 --> 00:52:57.220 What’s happened now is that we haven’t quite gotten to the curve where 00:52:57.220 --> 00:53:00.910 we’re just ignoring the cordon access restrictions because we do still have 00:53:00.910 --> 00:53:04.560 the same cordons. It’s just that we have reduced the amount of time. 00:53:04.560 --> 00:53:07.340 So we’re better than what we had before with the blue, but we’re 00:53:07.340 --> 00:53:10.960 not quite to what it would be like without any cordons. 00:53:11.970 --> 00:53:17.300 The other option is not to just reduce the duration of the cordons 00:53:17.300 --> 00:53:20.450 but actually prevent the need for them in the first place. 00:53:20.450 --> 00:53:24.750 So this looks like retrofits to the structures to make sure that they 00:53:24.750 --> 00:53:28.820 don’t get as damaged and therefore don’t need the cordon. 00:53:28.820 --> 00:53:33.640 An additional benefit of this is that the building itself would have less damage, 00:53:33.640 --> 00:53:36.030 and therefore, it could be brought online sooner. 00:53:36.030 --> 00:53:40.070 So when we plot that curve, it actually does go above the 00:53:40.070 --> 00:53:43.070 red line which ignores the cordon restrictions because 00:53:43.070 --> 00:53:46.600 those buildings themselves are now back online sooner. 00:53:47.300 --> 00:53:50.560 As you can see, see this is idealized data. It’s not simulated. 00:53:50.560 --> 00:53:52.070 It’s not going to be this clean. 00:53:52.070 --> 00:53:56.600 You’re going to have lines crossing and have to decide which is the best option. 00:53:56.600 --> 00:54:00.220 There’s going to be different costs. But those are all policy decisions. 00:54:00.220 --> 00:54:04.110 Those are not for the engineers to make. Our role is to provide to the 00:54:04.110 --> 00:54:08.390 policymakers what these different options are, what the ramifications 00:54:08.390 --> 00:54:12.880 of them are, how much benefit to the community we provide, 00:54:12.880 --> 00:54:16.400 and then how long – or, how much it costs. 00:54:17.800 --> 00:54:21.100 So just to summarize what I’ve discussed here, San Francisco 00:54:21.100 --> 00:54:24.790 is quite actively engaged in its seismic resilience policy. 00:54:24.790 --> 00:54:29.160 And even specifically in considering the effects of its tall buildings. 00:54:29.860 --> 00:54:34.260 The tall buildings study has offered a data set that we can start to leverage 00:54:34.260 --> 00:54:39.010 to address and assess how these cordons are inducing additional loss 00:54:39.010 --> 00:54:42.450 of function to the community. But to do that, we need to use 00:54:42.450 --> 00:54:47.800 high-resolution models that are feasible at a regional scale. 00:54:47.800 --> 00:54:50.840 And so we can combine several different state-of-the-art tools. 00:54:50.840 --> 00:54:55.110 For example, I talked about how we can take OpenSHA and then use spatial 00:54:55.110 --> 00:54:59.680 and multi-period correlation models in order to get spatially distributed 00:54:59.680 --> 00:55:04.540 response spectra, all of which are consistent with a single rupture. 00:55:05.220 --> 00:55:08.980 Additionally, we can take FEMA P-58 and use the open source data that we 00:55:08.980 --> 00:55:13.980 got for our exposure model and use that to get a much more 00:55:13.980 --> 00:55:17.790 comprehensive picture of the building damage, and more specifically, 00:55:17.790 --> 00:55:20.610 the repair and the repair times. 00:55:20.610 --> 00:55:24.720 And then we can use the REDi impeding factors and add onto it our geospatial 00:55:24.720 --> 00:55:28.130 tools to understand the physical relationship between the buildings. 00:55:28.130 --> 00:55:31.660 And that lets us use this combined analysis of 00:55:31.660 --> 00:55:36.660 both the logistical delays and the cordon-related delays. 00:55:36.660 --> 00:55:40.300 And this kind of high-resolution regional analysis can inform 00:55:40.310 --> 00:55:44.390 much more nuanced, more focused resilience policies 00:55:44.390 --> 00:55:47.900 that we can provide to the policymakers. 00:55:47.900 --> 00:55:51.700 So then, in conclusion, resilience-minded policymakers 00:55:51.700 --> 00:55:54.990 are going to need quantitative assessments of the anticipated 00:55:54.990 --> 00:55:57.190 recovery for their community. 00:55:57.190 --> 00:56:02.410 Again, that looks like figuring out where that X is given the current status. 00:56:02.410 --> 00:56:07.020 But if we ignore these safety cordons, which can cause significant delay 00:56:07.020 --> 00:56:10.460 to neighborhoods and to community functions, 00:56:10.460 --> 00:56:12.460 then we’re really missing the point. 00:56:12.460 --> 00:56:17.320 We’re not correctly assessing where that X currently stands. 00:56:17.320 --> 00:56:21.540 But in addition to correctly figuring out what the appropriate location for the 00:56:21.540 --> 00:56:25.510 X is, identifying these buildings and their cordons can give us 00:56:25.510 --> 00:56:30.680 a lot of information as to how to make targeted policies that would 00:56:30.680 --> 00:56:36.560 help to reduce these cordon-related downtimes and mitigate those delays. 00:56:36.560 --> 00:56:39.940 So with that, thank you for your time, and I’m happy to take any questions. 00:56:39.940 --> 00:56:45.400 [Applause] 00:56:46.520 --> 00:56:49.120 - All right. We already have a question from Anne. 00:56:52.860 --> 00:56:56.420 - Anne, that’s a great talk. It was really interesting. Thank you. 00:56:56.420 --> 00:56:59.220 I have some – I really like the way you’re looking at the adjacency 00:56:59.220 --> 00:57:03.260 problem. So, around the cordon, were you able to look at some 00:57:03.260 --> 00:57:10.730 things like having the cordon actually can also make the recovery faster? 00:57:10.730 --> 00:57:12.920 This is one reason they had in the Christchurch like that, 00:57:12.920 --> 00:57:15.710 so they could go in and demo the heck out of everything and 00:57:15.710 --> 00:57:19.730 not worry about trying to work around some buildings. 00:57:19.730 --> 00:57:22.020 So that’s one thing. The other thing is that, 00:57:22.020 --> 00:57:25.020 buildings that are sort of on the edge of cordons in Christchurch, 00:57:25.030 --> 00:57:30.500 their businesses didn’t do well. So there’s these other effects, 00:57:30.500 --> 00:57:35.060 the sense of what’s, you know, safe – the sense of safety. 00:57:35.060 --> 00:57:37.970 So it’s not just about how much space is available, but it’s about 00:57:37.970 --> 00:57:41.180 people’s perceptions of where it’s safe to do business. 00:57:41.180 --> 00:57:43.620 I just wondered if you had thought about any of that. 00:57:43.620 --> 00:57:45.990 - Thank you. I’ve definitely thought about the first one. 00:57:45.990 --> 00:57:56.040 So the question of whether or not there are what Ken Elwood calls perverse 00:57:56.040 --> 00:58:01.030 incentives. So whether or not the cordon actually helps with the recovery. 00:58:01.030 --> 00:58:04.320 So one way to handle that is that, when we’re figuring out what 00:58:04.320 --> 00:58:09.520 the impeding factors are, we can actually make adjustments for that. 00:58:09.530 --> 00:58:12.280 I didn’t go into how we figure out, for each building, what it is. 00:58:12.280 --> 00:58:17.320 It’s kind of a simulation of what we expect based on a different 00:58:17.320 --> 00:58:23.710 median and standard deviations. So one option is to say, when we’re 00:58:23.710 --> 00:58:27.640 trying to figure out, for example, the contractor mobilization time, 00:58:27.640 --> 00:58:31.340 because the whole area is already blocked off, we don’t need to take 00:58:31.340 --> 00:58:36.660 as much time to put up blockades and things like that. 00:58:36.660 --> 00:58:40.490 And so that’s going to be a shorter time. So what we can do is just model that 00:58:40.490 --> 00:58:45.560 by saying our median value for our contractor mobilization is different. 00:58:45.560 --> 00:58:50.470 I haven’t actually implemented that, but that’s how I would go ahead and do that. 00:58:50.470 --> 00:58:55.860 The second question of the public perception of safety is an interesting 00:58:55.860 --> 00:59:00.710 one. So what I’ve done here is just looked at which ones are 00:59:00.710 --> 00:59:06.050 kind of legally accessible as opposed to what people would enter into. 00:59:06.050 --> 00:59:11.080 That starts to, in my mind, get at a question of broader social issues 00:59:11.080 --> 00:59:16.000 and how much the downtime would be causing economic issues. 00:59:16.000 --> 00:59:19.960 So for example, the building next to the cordon, the little retail shop that is 00:59:19.960 --> 00:59:24.150 selling something, and people aren’t going to come to it, that’s starting to 00:59:24.150 --> 00:59:27.220 get into an economic question as opposed to just, 00:59:27.220 --> 00:59:30.920 is this space physically open. I’m not going that far in my particular 00:59:30.920 --> 00:59:35.280 modeling, but I think it’s a really, really interesting area to think about. 00:59:37.500 --> 00:59:40.780 [Silence] 00:59:41.320 --> 00:59:46.280 - Tom Heaton. This issue of, if the building has actually 00:59:46.290 --> 00:59:51.550 got some residual drifts, and the people who were in the building know that the 00:59:51.550 --> 00:59:59.000 building had some permanent damage, even if it’s repaired, did you consider 00:59:59.000 --> 01:00:04.050 that possibly the building won’t have much value economically anymore? 01:00:04.050 --> 01:00:08.100 Because who would want to go back into the same building that’s 01:00:08.100 --> 01:00:11.940 put back the way it was? - Yeah. So that’s a similar question 01:00:11.940 --> 01:00:16.160 of the public perception. No, I have not thought about it. 01:00:16.160 --> 01:00:21.720 Again, really interesting additional layer to add, but not one which I have done. 01:00:21.720 --> 01:00:24.080 I see you have a microphone back there. 01:00:24.940 --> 01:00:27.480 - Thank you for a very nice talk. 01:00:29.900 --> 01:00:33.200 Okay. See, you guys are more used to it. 01:00:34.080 --> 01:00:39.000 Thank you for a very nice talk, and I have a question about your comments 01:00:39.000 --> 01:00:49.090 related to residual drift and leaning. What is your criteria for it? 01:00:49.090 --> 01:00:55.540 For example, if you have a 40-story building, are you talking about only on 01:00:55.540 --> 01:01:02.740 the ground level that is damage occurs, and then therefore the leaning will – 01:01:02.740 --> 01:01:06.300 throughout the whole height of the building? 01:01:06.300 --> 01:01:17.640 And so is the leaning definition – what – where do you stop, I mean? 01:01:17.640 --> 01:01:19.840 What is the limit? - Good question. 01:01:19.840 --> 01:01:26.340 - I mean, it’s a very difficult situation because, if you have, let’s say, 01:01:26.340 --> 01:01:31.820 cracks and [inaudible] in the shear wall [inaudible], that doesn’t mean 01:01:31.820 --> 01:01:41.760 that the whole building is in danger. But what is – what is – what is the – 01:01:41.760 --> 01:01:47.340 in your computations, what was the acceptance criteria for such a situation? 01:01:47.340 --> 01:01:52.450 - Yeah. So there’s a couple different ways that I could answer this. 01:01:52.450 --> 01:01:55.590 So, first of all, in FEMA P-58, which is what I’m using here 01:01:55.590 --> 01:02:01.870 for the residual drift, it’s taking the maximum inter-story drift at any level 01:02:01.870 --> 01:02:04.050 and taking that as the residual drift. 01:02:04.050 --> 01:02:09.000 So it’s not saying the – from the bottom of the building to the top of the building 01:02:09.000 --> 01:02:14.020 and what that difference is. That’s not it. It’s every individual piece. 01:02:14.020 --> 01:02:18.400 Or, every individual story, and then it takes the worst one. 01:02:18.410 --> 01:02:22.230 So that could be addressed in a different way and extracted 01:02:22.230 --> 01:02:25.020 in a different way if we were interested in it, but I’m currently 01:02:25.020 --> 01:02:29.220 just taking that FEMA P-58 model. 01:02:29.220 --> 01:02:34.100 The second question is, what is the threshold of what would trigger it. 01:02:34.100 --> 01:02:39.000 And for that, what we’re doing is – haven’t done it yet, but the intent is to – 01:02:39.000 --> 01:02:44.480 when we create these models – these nonlinear models that we’re 01:02:44.480 --> 01:02:50.000 subjecting to our time histories, what we can do is subject it to 01:02:50.000 --> 01:02:53.610 earthquakes that result in different residual drifts and then use that 01:02:53.610 --> 01:02:58.820 to condition the reduction in – or, the increase in the probability 01:02:58.820 --> 01:03:04.230 of collapse during that – during a subsequent aftershock. 01:03:04.230 --> 01:03:09.420 And so we can start to see, what is the – is there any kind of – 01:03:09.420 --> 01:03:14.380 a nonlinearity in the response, given different residual drifts? 01:03:14.380 --> 01:03:18.140 And then say, okay, that’s going to be the cutoff that we’re using. 01:03:20.120 --> 01:03:25.560 - Hi. That was a really interesting talk. And towards the end of it, you were 01:03:25.560 --> 01:03:30.330 saying how now these results need to be incorporated into policy decisions. 01:03:30.330 --> 01:03:35.581 It seems like there’s a step in between that goes from the results that you 01:03:35.581 --> 01:03:40.480 showed to a cost-benefit analysis that says, you know, okay, if you’re going to 01:03:40.480 --> 01:03:44.910 retrofit this building, what’s a crossover where that outweighs – 01:03:44.910 --> 01:03:48.920 the savings outweigh the cost? And the – also the economic ripple 01:03:48.920 --> 01:03:53.650 effects of the places just outside the cordon that you were mentioning before. 01:03:53.650 --> 01:03:57.880 So I guess one question is, is that being done? 01:03:57.880 --> 01:04:03.200 What’s going on to address that aspect that will actually get it to policy? 01:04:03.200 --> 01:04:06.880 And then, so specifically, maybe with San Francisco, what’s the 01:04:06.880 --> 01:04:11.200 level of uptake you’re seeing with these results and going forward? 01:04:11.200 --> 01:04:16.600 - Yeah. So first of all, in the recommendations that were provided 01:04:16.600 --> 01:04:19.550 from that tall building study, they’re clearly already thinking about these 01:04:19.550 --> 01:04:23.540 cordon issues, and they’re trying to figure out how to get a handle on it. 01:04:23.540 --> 01:04:26.580 And so we know that they are interested in these kind of topics. 01:04:26.580 --> 01:04:30.320 So that’s step number one, and we’re excited about that. 01:04:30.320 --> 01:04:35.610 The next thing is that, as you said, we need to figure out what the 01:04:35.610 --> 01:04:40.150 cost-benefit is. So you can see one of the sponsors here is NIST. 01:04:40.150 --> 01:04:44.310 That project award is for a project specifically about the welded steel 01:04:44.310 --> 01:04:48.390 moment frame – so the pre-Northridge moment frames. 01:04:48.390 --> 01:04:52.240 And it kind of has two parts. One is my part, which is looking at 01:04:52.240 --> 01:04:54.540 how they would affect the community around them. 01:04:54.540 --> 01:04:58.330 The other part is looking at what different retrofit options there might be. 01:04:58.330 --> 01:05:02.350 It’s a little bit hard to look at it broadly because each building is incredibly 01:05:02.350 --> 01:05:05.520 unique, and figuring out what the retrofit would be for that is really 01:05:05.520 --> 01:05:10.450 difficult. But, so we are starting to try and think about that. 01:05:10.450 --> 01:05:12.330 Wen-Yi, mentioned at the bottom there, 01:05:12.330 --> 01:05:15.560 is the one who will be looking more at that. 01:05:16.180 --> 01:05:18.000 Does that answer your question sufficiently? 01:05:18.000 --> 01:05:19.960 Is there another part to it? 01:05:19.960 --> 01:05:21.980 - No. Just seems like a very complex [inaudible]. 01:05:21.980 --> 01:05:23.840 - Yeah. - … analysis that needs to be done. 01:05:23.840 --> 01:05:26.440 - Yeah. For sure. 01:05:30.380 --> 01:05:33.020 - This is getting a little bit down in the weeds, but you glossed 01:05:33.020 --> 01:05:38.060 very quickly over – what building models were sort of just generic 01:05:38.070 --> 01:05:44.290 based on the type of structural system being used versus the actual design? 01:05:44.290 --> 01:05:48.900 Which ones were you actually doing nonlinear time history analysis versus 01:05:48.900 --> 01:05:53.030 just spectral analysis time histories? Can you just give us a quick summary 01:05:53.030 --> 01:05:57.160 of how much of this is – you sort of jumped back and forth … 01:05:57.160 --> 01:05:58.590 - Yeah. - … between generic and very specific. 01:05:58.590 --> 01:06:00.450 And how much of it is specific? 01:06:00.450 --> 01:06:03.290 And which areas were sort of much more generic? 01:06:03.290 --> 01:06:10.100 - Yeah. So for the broad – the broader community, we’re using SP3, which is 01:06:10.100 --> 01:06:16.180 a tool that allows you to implement FEMA P-58 much more easily. 01:06:16.180 --> 01:06:20.470 And they have a batch input. So you can put in certain characteristics 01:06:20.470 --> 01:06:23.960 of your buildings for the entire inventory, and they have 01:06:23.960 --> 01:06:28.340 done a lot of analysis looking at different code periods. 01:06:28.340 --> 01:06:30.660 And so, given the date, how is it going to be built? 01:06:30.660 --> 01:06:32.050 And a lot of things like that. 01:06:32.140 --> 01:06:38.680 And so they then give us back the models, and that’s how we do that. 01:06:38.680 --> 01:06:42.650 I’ve set it up so that the outputs that I’m getting from that are the same as the 01:06:42.650 --> 01:06:45.920 outputs that I would be getting from the other model – 01:06:45.920 --> 01:06:53.480 the more high-resolution model. And so what we do for that is that we 01:06:53.480 --> 01:06:59.420 do the nonlinear analysis, put those EDPs into FEMA P-58, 01:06:59.420 --> 01:07:03.480 and then what we extract on the backside is in the exact same format – 01:07:03.480 --> 01:07:07.940 the exact same output as for the other. So we can use them – when we’re 01:07:07.940 --> 01:07:11.210 starting to interface with the actual information, it’s the exact same format. 01:07:11.210 --> 01:07:12.660 It doesn’t know that it’s different. 01:07:12.660 --> 01:07:15.060 But the inputs that we put in were different. 01:07:15.560 --> 01:07:17.460 Does that help? 01:07:17.460 --> 01:07:18.400 Yeah? 01:07:18.400 --> 01:07:20.260 - [inaudible] - Please. Please. 01:07:20.960 --> 01:07:24.440 - Hi. Anne, again. I like the way you paired your work 01:07:24.440 --> 01:07:28.720 with the San Francisco study. The HayWired scenario is below 01:07:28.720 --> 01:07:31.980 designed earthquake level shaking in San Francisco. 01:07:31.980 --> 01:07:36.950 But our study of tall buildings there had similar downtimes that you 01:07:36.950 --> 01:07:40.140 came up with. And they were due to non-structural damage. 01:07:40.140 --> 01:07:42.630 So I’m just wondering if that’s what you’re also seeing, 01:07:42.630 --> 01:07:47.860 that non-structural damage is where the code – the work in the code is? 01:07:48.520 --> 01:07:55.660 - Yeah. So when we did those two archetypes for – in the San Francisco 01:07:55.660 --> 01:07:58.520 project, we did those two archetypes of new buildings and looked at the 01:07:58.520 --> 01:08:03.520 downtime. It was definitely mostly driven by non-structural components. 01:08:04.680 --> 01:08:09.380 - So when you did the time history analysis, were the time histories from 01:08:09.390 --> 01:08:15.630 the standard kind of analysis done these days from GMPEs and from 01:08:15.630 --> 01:08:20.710 spectrum-compatible ground motions? So you didn’t actually use any 01:08:20.710 --> 01:08:25.960 synthesized ground motions, which are probably far more realistic? 01:08:25.960 --> 01:08:30.440 - Yeah. So that’s – so what I – the examples that I was showing was 01:08:30.440 --> 01:08:35.120 all with the hypothetical community. So that’s its own story. 01:08:35.120 --> 01:08:40.639 But where I’m moving forward through, yes, it would all be done – we haven’t 01:08:40.639 --> 01:08:43.529 yet actually made that decision. I expect that it will all be done using 01:08:43.529 --> 01:08:50.339 recorded ground motions. Yeah. - Have you followed – in Christchurch, 01:08:50.339 --> 01:08:54.699 is there any effort to rescue some of those tilting buildings? 01:08:54.699 --> 01:08:59.210 And what kind of decision-making is going on around that? 01:08:59.210 --> 01:09:03.719 - So initially what happened is that they – it would be hard to imagine 01:09:03.719 --> 01:09:07.699 this happening here, but basically, the government said, okay, we need to 01:09:07.699 --> 01:09:10.819 address this and address it quickly. So we’re going to make the decisions on 01:09:10.819 --> 01:09:14.130 what happens to each of these buildings. And then they would say, this one needs 01:09:14.130 --> 01:09:19.799 to be torn down, or this one needs to be made safe, and things like that. 01:09:19.799 --> 01:09:24.159 And then there are still some buildings that are still unclear as to 01:09:24.159 --> 01:09:28.009 what’s going to happen to them. But because, by now, it’s a much 01:09:28.009 --> 01:09:30.819 smaller perimeter around it, it’s almost like they’re just 01:09:30.820 --> 01:09:35.860 letting the building owners make those decisions at this point. 01:09:35.860 --> 01:09:38.690 - It just seems like – you know, there would be huge gray areas. 01:09:38.690 --> 01:09:42.559 - Yes. Yes. Which is why they’re trying to figure out ahead of time 01:09:42.559 --> 01:09:45.819 what this looks like. So I mentioned that the city is trying to come up with 01:09:45.820 --> 01:09:48.949 the protocols beforehand, and that’s the exact issue 01:09:48.980 --> 01:09:51.380 as to why they need to do that. 01:09:53.600 --> 01:09:58.260 [Silence] 01:09:58.780 --> 01:10:01.840 - Yeah. Interesting talk. My question concerns that, 01:10:01.840 --> 01:10:04.800 when you calculate the probability of collapse, it sounded like you 01:10:04.800 --> 01:10:08.030 assumed the aftershock. But, in fact, you could use 01:10:08.030 --> 01:10:12.239 a probability of that aftershock actually occurring to come up with 01:10:12.240 --> 01:10:14.880 a more realistic probability of collapse. 01:10:14.880 --> 01:10:19.620 - Yeah. Those are all things that are currently under development 01:10:19.630 --> 01:10:23.699 as to how we’re doing it. I painted a picture – I partly showed 01:10:23.699 --> 01:10:26.650 work that I have already done and partly incorporated what is coming. 01:10:26.650 --> 01:10:29.280 So I wasn’t always clear as to which is which. 01:10:29.280 --> 01:10:32.550 The question of what that is going to look like is one that’s currently 01:10:32.550 --> 01:10:36.460 under discussion. So if you have any thoughts, let me know. 01:10:37.080 --> 01:10:40.440 - All right. I think we’re going to – unless anyone has any burning 01:10:40.449 --> 01:10:44.760 questions, I think we’re going to wrap it up for today. 01:10:44.760 --> 01:10:48.239 So we’re going to now take Anne out to lunch at the cafeteria 01:10:48.239 --> 01:10:50.090 just across the way here on campus. 01:10:50.090 --> 01:10:54.000 Anyone is invited to come and join us and continue the discussion with Anne. 01:10:54.000 --> 01:10:58.080 And just so you know, next week, we have Joel Edwards from UC-Santa Cruz. 01:10:58.080 --> 01:11:01.140 He’s going to be speaking about Costa Rica subduction zone. 01:11:01.150 --> 01:11:04.409 So we’ll hope to see you there, and let’s give Anne another round of applause. 01:11:04.409 --> 01:11:05.409 Thanks, Anne. - Thank you. 01:11:05.409 --> 01:11:10.080 [Applause] 01:11:11.500 --> 01:11:20.480 [Silence]