WEBVTT Kind: captions Language: en-US 00:00:01.959 --> 00:00:05.890 Good morning, and welcome to today’s Earthquake Science Seminar. 00:00:05.890 --> 00:00:08.290 Before we get started, a couple of quick announcements. 00:00:08.290 --> 00:00:11.630 First, this is Rob and I’s last week as your seminar coordinators. 00:00:11.630 --> 00:00:15.700 So thanks to Rob for co-coordinating with me, and thank you to Jeanne 00:00:15.700 --> 00:00:20.040 and Sara McBride who are taking over from us moving forward. 00:00:20.040 --> 00:00:23.900 Second, next week’s seminar is from Amos Nur, who’s coming down from 00:00:23.900 --> 00:00:27.460 Stanford to talk about the history of scientific thinking about earthquakes. 00:00:27.460 --> 00:00:30.660 It was a seminar he gave, I think, this past winter at Stanford, 00:00:30.660 --> 00:00:33.649 and I’ve heard it’s quite good, so I’m excited to see that. 00:00:33.649 --> 00:00:35.880 This week, we’re joined by Nasser Marafi. 00:00:35.880 --> 00:00:39.020 He’s a postdoc at the University of Washington who’s worked extensively 00:00:39.030 --> 00:00:42.339 with Art and Erin up at the USGS office in Seattle 00:00:42.339 --> 00:00:46.450 on the problem of a Cascadia megathrust earthquake and the 00:00:46.450 --> 00:00:52.739 consequences it could have to structural buildings, in Nasser’s case. 00:00:52.740 --> 00:00:57.360 Nasser did his undergraduate and master’s work at Penn State and then 00:00:57.360 --> 00:01:00.880 worked in industry in New York City and Kuwait for several years before 00:01:00.890 --> 00:01:02.929 joining us at the University of Washington, where he 00:01:02.929 --> 00:01:05.940 completed his Ph.D. in June, this ... - August. 00:01:05.940 --> 00:01:11.660 - ... August [chuckles] – recently and is now a postdoc. So welcome, Nasser. 00:01:14.240 --> 00:01:16.880 - Can you guys hear me okay? Perfect. 00:01:16.880 --> 00:01:20.980 Okay, well, thank you all for being here. I’m Nasser Marafi, and today, 00:01:20.990 --> 00:01:24.450 I’ll be talking to you guys about the impacts of a magnitude 9 Cascadia 00:01:24.450 --> 00:01:28.270 subduction zone earthquake on structures located in deep sedimentary basins. 00:01:28.270 --> 00:01:32.640 The work that I’m going to be showing here was part of my Ph.D. dissertation, 00:01:32.640 --> 00:01:36.320 and I was advised by Dr. Jeffrey Berman and 00:01:36.320 --> 00:01:39.300 Dr. Marc Eberhard at the University of Washington. 00:01:40.560 --> 00:01:43.640 So the seismic hazard in the Puget Sound region includes 00:01:43.640 --> 00:01:46.580 large-magnitude earthquakes due to the Cascadia subduction zone. 00:01:46.580 --> 00:01:48.799 The USGS estimates that this type of earthquake 00:01:48.800 --> 00:01:51.940 has a 10% chance of occurring in the next 50 years. 00:01:51.940 --> 00:01:56.119 Motions during this type of earthquake are expected to be long in duration. 00:01:56.120 --> 00:02:00.320 However, building codes today do not account for the effects of duration. 00:02:01.280 --> 00:02:05.420 Additionally, the Seattle region and much of the Puget Sound region is 00:02:05.420 --> 00:02:10.050 overlying a deep sedimentary basin that is known to amplify ground shaking. 00:02:10.050 --> 00:02:15.060 This effect is also not explicitly considered in current design codes. 00:02:15.060 --> 00:02:18.910 And, as of today, we don’t have recordings for this type of earthquake. 00:02:18.910 --> 00:02:22.860 So its effects on structures are poorly understood. 00:02:23.580 --> 00:02:27.290 To address the paucity of recorded motions, we have a research project 00:02:27.290 --> 00:02:30.190 at the University of Washington with the United States Geological Survey 00:02:30.190 --> 00:02:36.440 that are simulating possible magnitude 9 scenarios of a Cascadia subduction zone. 00:02:36.440 --> 00:02:41.240 This project uses these simulations to study the impacts on tsunamis, 00:02:41.240 --> 00:02:44.790 the impacts of this earthquake on buildings and infrastructure, 00:02:44.790 --> 00:02:46.990 liquefaction, and landslides. 00:02:46.990 --> 00:02:50.000 And the results from these studies are being conveyed 00:02:50.000 --> 00:02:53.680 in terms of probabilistically using risk maps that are 00:02:53.680 --> 00:02:56.820 being presented to stakeholders and the community. 00:02:56.820 --> 00:03:02.680 Part of the project is also looking – sorry, I went back for some reason. 00:03:02.680 --> 00:03:06.660 Part of the project is also looking at earthquake early warning systems. 00:03:06.660 --> 00:03:11.810 My work within this project is focused more on the impacts 00:03:11.810 --> 00:03:14.860 of an M9 on buildings and infrastructure. 00:03:14.860 --> 00:03:18.580 So my seminar today will start off by talking about the characteristics 00:03:18.580 --> 00:03:21.310 of the simulated motions – mainly characteristics that are 00:03:21.310 --> 00:03:24.930 known to affect structural response. The second half of the talk 00:03:24.930 --> 00:03:29.340 will look at individual and regional performance of structures 00:03:29.340 --> 00:03:32.940 in the Puget Sound during a magnitude 9 event. 00:03:33.470 --> 00:03:38.680 So the 3D simulations that I will be using are generated by Art Frankel 00:03:38.680 --> 00:03:42.500 and Erin Wirth from the United States Geological Survey. 00:03:42.500 --> 00:03:46.570 They are varying a series of rupture parameters for each realization to 00:03:46.570 --> 00:03:50.920 address the problem probabilistically. We have about 30 scenarios. 00:03:50.920 --> 00:03:56.400 Each scenario uses a seismic wave velocity model of the Pacific Northwest. 00:03:56.400 --> 00:04:00.150 These velocity models include several deep basins. 00:04:00.150 --> 00:04:03.140 And then high-performance computers are then used to generate deterministic 00:04:03.140 --> 00:04:06.660 motions with period contents above 1 second. 00:04:06.660 --> 00:04:12.840 These motions were then combined with stochastically generated motions 00:04:12.840 --> 00:04:15.810 with frequency content below 1 second to create 00:04:15.810 --> 00:04:21.040 broadbands across a 1-by-1-kilometer grid in the Pacific Northwest. 00:04:21.660 --> 00:04:24.800 Here’s a – here’s an example of two scenarios that have been generated 00:04:24.800 --> 00:04:29.540 by Art and Erin with a hypocenter just off the coast of Washington. 00:04:30.760 --> 00:04:36.800 Realization 1 has a hypocenter that’s just – that’s just further offshore 00:04:36.810 --> 00:04:39.960 of Seattle rupturing toward Seattle. 00:04:39.960 --> 00:04:45.000 And Realization 2 has a hypocenter closer to Seattle rupturing away. 00:04:45.000 --> 00:04:46.880 And what you can start to see is that the ground motions 00:04:46.880 --> 00:04:51.720 produced from these two realizations are quite different. 00:04:54.620 --> 00:04:59.000 Here I’m plotting the velocity history of these two scenarios for Seattle, 00:04:59.000 --> 00:05:02.430 and you can start to see the differences in terms of amplitude 00:05:02.430 --> 00:05:05.880 between the two scenarios and differences in frequency content, 00:05:05.880 --> 00:05:08.220 especially if you start comparing the wiggles. 00:05:08.220 --> 00:05:11.440 And structural engineers have found a way to quantify amplitude 00:05:11.449 --> 00:05:14.530 and frequency content to estimate the engineering demands 00:05:14.530 --> 00:05:22.280 on a structure using single degree of freedom oscillators. 00:05:22.280 --> 00:05:24.750 You can think of those as lollipop structures, 00:05:24.750 --> 00:05:27.770 kind of like the Space Needle in Seattle. 00:05:27.770 --> 00:05:31.879 Each oscillator would have a mass and a – and a stick 00:05:31.880 --> 00:05:37.860 with a specific stiffness and, of course, a period. 00:05:37.860 --> 00:05:41.520 Making the oscillators longer or shorter effectively 00:05:41.520 --> 00:05:45.580 varies the period to represent structures of varying heights. 00:05:45.580 --> 00:05:49.349 Now, these oscillators were subjected to the M9 motions, 00:05:49.349 --> 00:05:52.110 and the maximum acceleration that the oscillator would see 00:05:52.110 --> 00:05:54.360 under this type of earthquake is recorded. 00:05:54.360 --> 00:05:58.360 Now, doing this for multiple oscillators, you can start to plot the 00:05:58.360 --> 00:06:02.800 maximum accelerations that each oscillator would see with period. 00:06:02.800 --> 00:06:07.580 And what this is known as – this is what’s called a response spectra. 00:06:07.580 --> 00:06:12.039 Structural engineers like to use this to figure out what kind of forces 00:06:12.039 --> 00:06:17.759 the structure would see when subjected to this type of motion. 00:06:17.759 --> 00:06:23.650 You would take – you would take a period of a structure, multiply it by – 00:06:23.650 --> 00:06:26.479 take the mass of the structure, multiply it by the spectral acceleration, 00:06:26.479 --> 00:06:29.620 and then you can start to estimate the forces 00:06:29.620 --> 00:06:34.160 that the structure would see, just using F equals m-a. 00:06:34.160 --> 00:06:39.500 Here’s a plot of the response spectra for Realization 1 and Realization 2. 00:06:39.500 --> 00:06:44.560 And what you start to see is that the response spectra 00:06:44.560 --> 00:06:47.660 between the two motions is quite different. 00:06:48.240 --> 00:06:53.400 For example, if I had a 2-second structure with a strength that 00:06:53.409 --> 00:06:57.889 corresponded to roughly 50% of its mass, that structure would yield and 00:06:57.889 --> 00:07:03.069 deform inelastically under Realization 1. But it would just respond elastically 00:07:03.069 --> 00:07:09.060 under Realization 2 and return back to its original position after the earthquake. 00:07:10.020 --> 00:07:13.080 Now, as I mentioned previously, the M9 Project wanted to quantify 00:07:13.080 --> 00:07:16.040 the effects of a magnitude 9 earthquake probabilistically. 00:07:16.040 --> 00:07:20.469 So we’re simulating over 30 scenarios of this type of earthquake. 00:07:20.469 --> 00:07:23.449 Here is the response spectra for all 30, and you can start to see that there 00:07:23.449 --> 00:07:28.940 is quite a large variation in spectral acceleration between the 30 scenarios. 00:07:29.420 --> 00:07:32.340 Shown in red is the maximum considered earthquake, 00:07:32.340 --> 00:07:40.430 which typically – is the response spectra typically used in building design. 00:07:40.430 --> 00:07:42.939 And what you can start to see is that many of the realizations 00:07:42.939 --> 00:07:47.569 actually exceed that that is used in design today. 00:07:47.569 --> 00:07:51.779 For example, at 2 seconds, 20 out of the 30 realizations 00:07:51.779 --> 00:07:55.169 exceed the spectral accelerations used in design. 00:07:55.169 --> 00:08:00.240 Fortunately, the building codes that we use today have a capping on 00:08:00.240 --> 00:08:03.330 the minimum design base shear, or minimum design spectral acceleration. 00:08:03.330 --> 00:08:07.219 So, at very long periods, the design minimum 00:08:07.220 --> 00:08:10.000 spectral acceleration would start to govern. 00:08:11.320 --> 00:08:15.260 Now, you can partially attribute this effect to the last Ice Age, 00:08:15.270 --> 00:08:18.660 where a lot of the ice sheets overlying the Puget Sound region 00:08:18.660 --> 00:08:22.259 compacted the top sediment layer, making it ideal for resisting – 00:08:22.259 --> 00:08:25.070 making it stiff and ideal for resisting vertical loads. 00:08:25.070 --> 00:08:28.509 In fact, unlike other cities, many of the tall buildings in Seattle 00:08:28.509 --> 00:08:33.160 are just founded on shallow foundations that are sitting on that till. 00:08:33.160 --> 00:08:38.720 While the till is strong – has a high bearing stiffness, 00:08:38.720 --> 00:08:43.150 it has a much lower shear wave velocity than the surrounding bedrock. 00:08:43.150 --> 00:08:45.320 And what this means is that, if you have seismic waves 00:08:45.320 --> 00:08:49.140 traveling towards the basin, they’re amplified in three ways. 00:08:49.140 --> 00:08:52.250 The first type of amplification is due to the impedance contrast 00:08:52.250 --> 00:08:54.480 between the different sediment layers. 00:08:54.480 --> 00:08:57.590 The second type of amplification that you would get is due to the 00:08:57.590 --> 00:09:00.950 shear-to-surface-wave conversion at the basin edge. 00:09:00.950 --> 00:09:05.020 And then the last type of amplification is the focusing effect of the shear waves 00:09:05.020 --> 00:09:08.640 due to the overall lens-like shape of the basin. 00:09:10.380 --> 00:09:17.140 Now, how do we quantify this? Well, engineers have found – 00:09:17.140 --> 00:09:22.340 typically use the depth to crystalline basement rock to use as a basin proxy. 00:09:22.340 --> 00:09:26.200 And this measure is called Z-2.5, which is basically just the depth 00:09:26.200 --> 00:09:32.240 from the surface to sediment layer with a velocity model – with a velocity – 00:09:32.240 --> 00:09:35.560 a shear wave velocity of 2,500 meters per second. 00:09:35.560 --> 00:09:39.950 So, in the deep parts of the basin, the Z-2.5 value is expected to be large, 00:09:39.950 --> 00:09:43.020 whereas, towards the edge or outside the basin, 00:09:43.020 --> 00:09:47.620 this Z-2.5 value is expected to be a lot smaller. 00:09:47.620 --> 00:09:50.280 Here’s a map that illustrates the shape and size 00:09:50.280 --> 00:09:54.290 of the basin using the Z-2.5 basin proxy. 00:09:54.290 --> 00:09:57.870 And, as you can see, Seattle is sort of in the bull’s-eye, where it’s in the 00:09:57.870 --> 00:10:02.890 deepest part of the basin reaching values of 7 kilometers for Z-2.5. 00:10:02.890 --> 00:10:08.570 So, to quantify, or to understand the effect of the basin, we chose another site 00:10:08.570 --> 00:10:14.860 that is about 80 kilometers south of Seattle that is outside the basin. 00:10:14.860 --> 00:10:18.870 Now, both of these sites are within the same source-to-site distance 00:10:18.870 --> 00:10:22.770 from the fault rupture plane. So ground motions would attenuate 00:10:22.770 --> 00:10:27.400 similarly if the basin wasn’t there between the two sites. 00:10:27.400 --> 00:10:29.780 And so, if you look at the time histories for La Grande, 00:10:29.780 --> 00:10:32.430 and you start to compare it to Seattle, you start to see that there is 00:10:32.430 --> 00:10:35.620 both variations in amplitude and frequency content. 00:10:35.620 --> 00:10:40.180 And going back to the response spectra, you start to see that the spectral 00:10:40.180 --> 00:10:46.640 accelerations in Seattle are much larger than those observed in La Grande. 00:10:47.710 --> 00:10:52.120 This variation in spectral acceleration can also be seen through contour maps. 00:10:52.130 --> 00:10:55.560 Here I’m plotting two contour maps of spectral acceleration – 00:10:55.560 --> 00:10:58.320 one at 0.5 seconds, and one at 2 seconds. 00:10:58.320 --> 00:11:02.060 And, at short periods, the effects of the basins aren’t actually 00:11:02.060 --> 00:11:05.380 modeled deterministically. So we – so we’re not seeing 00:11:05.380 --> 00:11:09.080 any basin amplification or de-amplification at that period. 00:11:09.080 --> 00:11:14.580 But instead, we’re just seeing attenuation of ground shaking with distance, 00:11:14.580 --> 00:11:19.220 as you would expect from predictions of ground motion models. 00:11:19.220 --> 00:11:23.340 However, at long periods, where the simulations are predicting the basin 00:11:23.340 --> 00:11:29.440 amplifications, you start to see that the spectral accelerations are much larger in 00:11:29.440 --> 00:11:35.940 areas with deep – with large Z-2.5, or areas in the deepest part of the basin. 00:11:40.000 --> 00:11:43.880 Now, to quantify the amount of amplification that we’re seeing 00:11:43.880 --> 00:11:47.140 on spectral acceleration between inside and outside basin locations, 00:11:47.140 --> 00:11:50.701 we’ve developed this basin amplification factor using ground motion model – 00:11:50.701 --> 00:11:54.760 or, ground motion prediction equation residuals to account for 00:11:54.760 --> 00:11:57.480 the amount of amplification or de-amplification 00:11:57.480 --> 00:12:01.560 due to local site effects and source-to-site distance. 00:12:01.570 --> 00:12:05.250 And what these factors really mean is that, if you have two sites – 00:12:05.250 --> 00:12:09.770 one in the deepest part – one inside the basin and one outside the basin, 00:12:09.770 --> 00:12:14.820 and they have – all else – site parameters being equal, the ratio in spectral 00:12:14.820 --> 00:12:19.720 acceleration between these two sites is the basin amplification factor. 00:12:20.560 --> 00:12:25.500 Here I’ve calculated all the basin amplification factors 00:12:25.500 --> 00:12:28.080 that you would get from the 30 simulations. 00:12:28.080 --> 00:12:32.860 And I’m plotting it here as – with respect to period. 00:12:32.860 --> 00:12:37.080 And so basically, the inside basin stations, we’re using ones 00:12:37.080 --> 00:12:41.360 that are within the Seattle area with Z-2.5 of about 6 kilometers. 00:12:41.360 --> 00:12:44.450 And the outside basin stations, we’re using ones with Z-2.5 00:12:44.450 --> 00:12:46.420 of about 1 kilometer. 00:12:46.420 --> 00:12:53.180 And the ratio in spectral accelerations between the two locations are quite 00:12:53.180 --> 00:12:57.590 different, and they – the amplification that you get inside the basin is actually 00:12:57.590 --> 00:13:05.080 period-dependent, where it reaches values of about 6 at 3 seconds. 00:13:10.210 --> 00:13:16.980 So we’ve also found similar trends and similar basin amplification factors 00:13:16.980 --> 00:13:20.760 with other basins. Here I’m showing a contour map 00:13:20.760 --> 00:13:26.240 of the Yufutsu Basin in Japan on the left. And we’ve computed basin amplification 00:13:26.240 --> 00:13:32.280 factors for a magnitude 8.3 Nisqually interface earthquake. 00:13:32.290 --> 00:13:35.780 And what we’re seeing is that the trends in terms of basin amplification 00:13:35.780 --> 00:13:40.950 factors with period are similar, however the amplitude or the 00:13:40.950 --> 00:13:44.820 values of basin amplification factors differ between basins. 00:13:46.810 --> 00:13:53.140 Now, another important ground motion characteristic is the frequency content, 00:13:53.150 --> 00:13:57.770 not only at the initial period of the structure, but also at longer periods. 00:13:57.770 --> 00:14:00.470 This is important because, under strong shaking, 00:14:00.470 --> 00:14:03.700 the structure is expected to deform inelastically, 00:14:03.700 --> 00:14:07.620 and it causes the structure to soften and its period to elongate. 00:14:07.630 --> 00:14:11.430 So when this happens, multiple spectral ordinates 00:14:11.430 --> 00:14:15.000 start to matter and affect their response. 00:14:15.000 --> 00:14:19.860 And so, going back to our Realization 1, Realization 2, you know, 00:14:19.860 --> 00:14:22.450 if you look at 0.5 seconds, the spectral accelerations 00:14:22.450 --> 00:14:26.500 at these short periods are quite similar between the two realizations. 00:14:26.500 --> 00:14:30.560 However, if that spectral acceleration was strong enough to cause the structure 00:14:30.560 --> 00:14:35.241 to yield, the amount of energy that the structure would see in Realization 1 00:14:35.241 --> 00:14:38.400 is quite different than Realization 2. And you would expect that 00:14:38.400 --> 00:14:42.350 Realization 1 would be much more damaging than Realization 2. 00:14:42.350 --> 00:14:47.871 And what we’ve done is, we’ve created a ground motion intensity 00:14:47.871 --> 00:14:51.040 measure that sort of quantifies this. And we’ve done this by integrating 00:14:51.040 --> 00:14:56.620 the response spectrum from the elastic period to the expected elongated period, 00:14:56.620 --> 00:15:02.930 and we’ve normalized that value with the – an integral 00:15:02.930 --> 00:15:05.120 that assumes that the shape does not change. 00:15:05.120 --> 00:15:10.780 So what that means is that, if the spectral shape factor, or SSA, is less than 1, 00:15:10.780 --> 00:15:13.940 than the response spectrum, on average, decreases. 00:15:13.950 --> 00:15:18.430 Whereas, if the spectral shape increases, 00:15:18.430 --> 00:15:22.800 then you would get a spectral shape factor that is higher than 1. 00:15:23.300 --> 00:15:26.360 - [inaudible] 00:15:29.480 --> 00:15:31.220 Sorry. 00:15:34.420 --> 00:15:36.800 Okay. Is it working now? 00:15:36.800 --> 00:15:40.320 - It's just - it's jumping back ... - Oh. It’s jumping back and forth? 00:15:40.320 --> 00:15:45.100 Okay. I’ll try to keep an eye on the screen when I talk. 00:15:45.100 --> 00:15:48.910 All right, so here’s a plot that shows the spectral shape factor 00:15:48.910 --> 00:15:53.200 with respect to period for multiple locations along the basin. 00:15:53.200 --> 00:15:56.650 So the solid black line is in the deepest part of the basin in Seattle, 00:15:56.650 --> 00:15:59.870 and the dashed line is towards the edge of the basin. 00:15:59.870 --> 00:16:02.290 And what you start to see is that the spectral shapes 00:16:02.290 --> 00:16:05.720 inside the basin are actually much more damaging. 00:16:09.870 --> 00:16:12.660 In addition, if you were to compare the spectral shapes 00:16:12.660 --> 00:16:19.300 that we’re seeing in the M9 scenarios with ground motions that are – hm. 00:16:19.900 --> 00:16:21.940 That’s weird. 00:16:25.060 --> 00:16:26.020 Okay. 00:16:26.920 --> 00:16:31.500 If you were to compare the [chuckles] spectral shapes for the M9 scenarios 00:16:31.500 --> 00:16:37.090 to spectral shapes expected of motions that are scaled and selected to match 00:16:37.090 --> 00:16:40.310 the maximum considered earthquake used in design, you start to see 00:16:40.310 --> 00:16:42.920 that the spectral shapes in the – in the M9 scenarios are actually 00:16:42.920 --> 00:16:45.070 much more damaging than those used in design. 00:16:45.070 --> 00:16:48.840 So we would expect some more damage due to that alone. 00:16:49.900 --> 00:16:54.000 Okay. In addition, these earthquakes are long. 00:16:54.000 --> 00:16:59.680 And so – and if you were to compare them with crustal earthquake motions 00:16:59.680 --> 00:17:03.130 that are typically used to evaluate structural systems, 00:17:03.130 --> 00:17:08.240 these motions are about 10 times longer. So the Seattle – the M9 motions are 00:17:08.240 --> 00:17:12.089 about 80 to 120 seconds long in terms of significant duration, 00:17:12.089 --> 00:17:16.220 which is a measure of ground motion duration similar to Arias intensity. 00:17:16.220 --> 00:17:21.240 And it’s just much longer. And what that means is that, 00:17:21.240 --> 00:17:24.970 if the strong shaking is longer, that means that the structure 00:17:24.970 --> 00:17:28.760 would see more cycles, and it would cause more damage. 00:17:29.600 --> 00:17:32.450 And here’s just a contour plot showing significant duration 00:17:32.450 --> 00:17:36.289 and how it varies with distance. And what you – another thing 00:17:36.289 --> 00:17:39.320 that we’ve noticed is that, within the basin, 00:17:39.320 --> 00:17:44.120 the significant duration of the motion was not amplified by the basin. 00:17:45.749 --> 00:17:51.240 So thus far, I’ve shown that the M9 motions have high spectral accelerations, 00:17:51.249 --> 00:17:53.470 damaging spectral shapes, and long durations. 00:17:53.470 --> 00:17:56.330 Well, what does this mean for structural response? 00:17:56.330 --> 00:18:03.300 So to answer this question, we’ve – we’re – like, to quantify the 00:18:03.300 --> 00:18:05.860 effects of the M9 motions on structures, we’re working with the city 00:18:05.860 --> 00:18:09.000 and many engineers – engineering firms in Seattle 00:18:09.000 --> 00:18:12.960 to develop representative Seattle building archetypes. 00:18:12.960 --> 00:18:20.450 And here’s a plot that just shows the tall structures in Seattle built after 1994, 00:18:20.450 --> 00:18:23.610 and what we’ve seen, or what we’ve noticed is that most of 00:18:23.610 --> 00:18:28.799 these buildings use concrete core wall systems, as shown here. 00:18:28.799 --> 00:18:33.309 And for this reason, we’ve decided to only concentrate on 00:18:33.309 --> 00:18:37.130 concrete core wall systems as our lateral force resisting system. 00:18:37.130 --> 00:18:40.419 So most of the archetypes that we design in this presentation 00:18:40.420 --> 00:18:43.420 are of concrete core walls. 00:18:45.090 --> 00:18:50.200 So previously, I’ve shown that the basin amplifies a wide range of periods, 00:18:50.200 --> 00:18:55.159 and so to comprehensively understand the effects of the M9 motions, 00:18:55.159 --> 00:18:58.649 we’re not only looking at one structure, but we’re looking – we’re designing 00:18:58.649 --> 00:19:04.549 a range of structures, from four stories to 40 stories tall. 00:19:04.549 --> 00:19:07.850 These buildings are quite similar in nature. 00:19:07.850 --> 00:19:10.320 They have similar floor-to-floor heights. 00:19:10.320 --> 00:19:14.279 They have similar floor plates, so the overall aspect ratio 00:19:14.279 --> 00:19:18.279 of the building is quite similar between the archetypes. 00:19:18.279 --> 00:19:24.240 And they are designed with – they sort of represent tall residential 00:19:24.240 --> 00:19:28.610 condo buildings in Seattle and in other regions in the West Coast. 00:19:28.610 --> 00:19:32.929 They have a central concrete core, which is right here, and that 00:19:32.929 --> 00:19:38.060 concrete core is what is resisting the lateral forces. 00:19:38.060 --> 00:19:42.320 And most of the archetypes have basement levels. 00:19:43.540 --> 00:19:48.580 So we have four variations, so we have – we’re looking at four- to 40-story 00:19:48.590 --> 00:19:52.769 buildings, but we have four variations in each design. 00:19:52.769 --> 00:19:59.080 We have – the first variation is using the current building code – ASCE 7-10. 00:19:59.080 --> 00:20:02.460 And we have another variation that’s looking at the building code 00:20:02.460 --> 00:20:05.160 that will be adopted in Seattle in 2020. 00:20:05.160 --> 00:20:09.649 We’re also looking at design variations to represent sort of 00:20:09.649 --> 00:20:15.409 short structures that are designed to the minimal allowable by code. 00:20:15.409 --> 00:20:18.470 And then, in Seattle, all buildings that are 00:20:18.470 --> 00:20:23.029 taller than 240 feet have to go through this rigorous design process. 00:20:23.029 --> 00:20:28.139 And this is what we’re calling a code-enhanced procedure, where – 00:20:28.139 --> 00:20:31.049 which I’ll talk about in more detail later, but – 00:20:31.049 --> 00:20:34.560 so that’s another design variation that we’re looking at. 00:20:36.140 --> 00:20:39.040 I don’t know why it keeps flicking back – flickering back and forth. 00:20:39.040 --> 00:20:42.299 Okay. If I do that, it sort of gets fixed. 00:20:42.299 --> 00:20:45.690 Okay. So we have four performance groups – one that represents 00:20:45.690 --> 00:20:49.830 code minimum sort of short buildings that are designed in Seattle, one that – 00:20:49.830 --> 00:20:51.960 the code enhanced, which sort of represents 00:20:51.960 --> 00:20:56.070 the rigorous design process that tall buildings in Seattle go through, 00:20:56.070 --> 00:20:57.610 and then we’re looking at two different codes. 00:20:57.610 --> 00:21:00.419 Today, just in the interest of time, I’m only going to be talking about 00:21:00.419 --> 00:21:03.710 the newer code that will be adopted in Seattle in 2020. 00:21:03.710 --> 00:21:08.899 So I’ll look at code minimum designs according to this ASCE 7-16 code, 00:21:08.899 --> 00:21:13.210 and code enhanced designs to this ASCE 7-16. 00:21:13.210 --> 00:21:15.960 So, for all the archetypes that we’ve designed, we’ve built 00:21:15.960 --> 00:21:19.340 numerical structural analysis models that can simulate the response 00:21:19.340 --> 00:21:23.190 and predict the performance of the – of the archetype when subjected to 00:21:23.190 --> 00:21:28.379 different ground motions. These analysis models simulate 00:21:28.380 --> 00:21:31.740 the stress and strains in the lateral force resisting system. 00:21:31.740 --> 00:21:37.499 And more importantly, are calibrated to numerous experimental tests. 00:21:38.860 --> 00:21:43.300 So each archetype is subjected to this suite of M9 motions for Seattle and 00:21:43.309 --> 00:21:46.950 suite of M9 motions for La Grande – so inside and outside basin. 00:21:46.950 --> 00:21:51.499 And then, for each archetype and each ground motion, 00:21:51.500 --> 00:21:56.240 we’re measuring the maximum story drift that the building would see. 00:21:56.240 --> 00:21:59.980 And this is an engineering design parameter that we’re measuring, 00:21:59.980 --> 00:22:06.399 and it can sort of be used later to relate to the amount of damage 00:22:06.400 --> 00:22:08.850 that that building would see. 00:22:10.200 --> 00:22:13.980 Here’s a plot that shows the median of the maximum story drifts 00:22:13.990 --> 00:22:18.980 with respect to number of stories for Seattle and La Grande 00:22:18.980 --> 00:22:23.720 and for the code minimum archetypes. And what we start to see is that 00:22:23.720 --> 00:22:31.509 the Seattle motions are causing 7 times more story drifts than La Grande. 00:22:31.509 --> 00:22:35.010 And this is just – this is because of those three ground motion characteristics 00:22:35.010 --> 00:22:38.369 that I’ve talked about – the large spectral accelerations, the damaging spectral 00:22:38.369 --> 00:22:41.259 shapes, and the durations. The La Grande motions are sort of 00:22:41.259 --> 00:22:48.240 at this – at spectral accelerations that are much lower than what you would 00:22:48.240 --> 00:22:54.320 need to cause a structure to yield, so they’re not seeing large story drifts. 00:22:54.320 --> 00:22:58.380 And so another thing that we wanted to do is we wanted to compare 00:22:58.380 --> 00:23:05.659 these drifts that we’re getting in the M9 simulations to motions 00:23:05.659 --> 00:23:08.100 that are somewhat consistent with the building code. 00:23:08.100 --> 00:23:11.769 And so, to do this, we’ve selected about 100 motions to match the 00:23:11.769 --> 00:23:15.799 conditional mean and variance spectra of a 2,500-year earthquake. 00:23:15.799 --> 00:23:20.879 Again, the same type – the same return period that the code sort of – 00:23:20.879 --> 00:23:26.880 sort of accounts – or, uses for Seattle. 00:23:26.880 --> 00:23:29.059 This is called the maximum considered earthquake. 00:23:29.059 --> 00:23:33.990 And these motions that we’re – these 100 motions are selected from crustal, 00:23:33.990 --> 00:23:40.060 inter-slab, and interface earthquakes so that they’re consistent with the type of 00:23:40.060 --> 00:23:45.460 earthquake sources that would control the 2,500-year hazard for Seattle. 00:23:45.980 --> 00:23:50.840 And here I’m showing the – in orange, I’m showing the story drifts 00:23:50.850 --> 00:23:53.690 that you would see for the maximum considered earthquake. 00:23:53.690 --> 00:24:00.260 And, while they’re sort of less than – less than the M9 motions for Seattle, 00:24:00.260 --> 00:24:06.330 they’re also within this 3% limit for – in terms of story drift for 00:24:06.330 --> 00:24:09.000 tall buildings that are used – that is used. 00:24:10.660 --> 00:24:16.320 So, now, I’m only showing you – so there was 30 ground motions 00:24:16.320 --> 00:24:18.700 in the black dots – at each black dot. 00:24:18.700 --> 00:24:24.060 And there’s about 100 motions in each of the squares. 00:24:24.060 --> 00:24:27.580 And I’m only showing you the median response, but it’s important to sort of 00:24:27.580 --> 00:24:34.740 look at how the story drifts vary within the set of 30 and 100 ground motions. 00:24:34.740 --> 00:24:38.519 And we do this by plotting the probability of exceeding 00:24:38.519 --> 00:24:42.210 a certain story drift with maximum inter-story drift. 00:24:42.210 --> 00:24:46.760 So you can think of this as being kind of like the integral – 00:24:46.760 --> 00:24:53.039 1 minus the integral of the CDF of the distribution of inter-story drifts. 00:24:53.039 --> 00:24:56.800 And, if you go to a 50% probability of exceeding, well, that’s just 00:24:56.800 --> 00:25:00.019 a median response that I was showing previously. 00:25:00.019 --> 00:25:06.639 But if you start looking at the tail ends, sort of at low probabilities of 00:25:06.639 --> 00:25:12.100 exceedance, you start to see that the M9 motions have much larger 00:25:12.100 --> 00:25:15.710 story drifts at these tail ends. And these drifts may be 00:25:15.710 --> 00:25:18.710 large enough to exceed the drift capacity of many structural 00:25:18.710 --> 00:25:23.509 components in the building that would likely lead to collapse. 00:25:23.509 --> 00:25:28.710 And so using all of these points within the ground motion sets that we have – 00:25:28.710 --> 00:25:35.610 the M9 Seattle and the MCE-R, we can start to – we can use those drifts and 00:25:35.610 --> 00:25:41.119 translate them to collapse probabilities using structural component fragilities. 00:25:41.119 --> 00:25:46.639 And here, I’m showing a plot that shows the probability of collapse 00:25:46.639 --> 00:25:52.649 of each archetype under the M9 scenario and under 00:25:52.649 --> 00:25:55.070 a maximum considered earthquake scenario. 00:25:55.070 --> 00:26:00.559 And what you start to see is that the large drifts in the M9 scenarios, 00:26:00.559 --> 00:26:04.779 those tail ends, start to control the collapse probability 00:26:04.779 --> 00:26:07.940 and where the collapse probability, in the M9 scenarios, 00:26:07.940 --> 00:26:11.289 are much larger than that of the maximum considered earthquake. 00:26:11.289 --> 00:26:15.700 Now, previously, the median drifts were the same, but now the collapse – 00:26:15.700 --> 00:26:20.559 due to those high drift – story drifts at the tail ends, the collapse probability 00:26:20.560 --> 00:26:28.020 starts to be much – starts to lead to poorer performance and … 00:26:30.100 --> 00:26:34.920 Now – well, so what is the – you know, what’s a good collapse probability? 00:26:34.920 --> 00:26:41.179 Well, the building code targets a 10% chance of collapse in the MCE-R. 00:26:41.179 --> 00:26:44.730 And so if you were to look at the collapse probabilities 00:26:44.730 --> 00:26:48.830 of the MCE motion, they’re within the code – the building code target. 00:26:48.830 --> 00:26:51.119 So they’re fine according to the building code. 00:26:51.119 --> 00:26:55.070 But these M9 motions are much more damaging that what would we expect. 00:26:55.070 --> 00:26:58.119 And so we’re seeing these larger collapse probabilities – 00:26:58.119 --> 00:27:01.480 much larger than what the code would target. 00:27:01.480 --> 00:27:05.279 So this – all this – the results that I’m showing thus far 00:27:05.280 --> 00:27:09.200 are all for the code minimum archetype. So these are the archetypes that are 00:27:09.200 --> 00:27:12.780 designed to barely meet the code requirements. 00:27:12.780 --> 00:27:15.730 Well, what happens if I look at the code enhanced archetypes – 00:27:15.730 --> 00:27:18.720 the ones that are – that represents tall buildings in Seattle? 00:27:18.720 --> 00:27:21.769 Now, those archetypes – you know, those buildings are designed 00:27:21.769 --> 00:27:26.080 typically using – there’s a design peer review process. 00:27:26.080 --> 00:27:30.799 Those buildings also consider basin effects in the design. 00:27:30.799 --> 00:27:35.799 And then they use – while short buildings may only – 00:27:35.799 --> 00:27:39.710 might only rely on elastic analyses, these tall buildings actually 00:27:39.710 --> 00:27:44.000 use nonlinear analyses to predict their response. 00:27:44.000 --> 00:27:48.600 So if – so I have a group of archetypes – again, 00:27:48.600 --> 00:27:52.119 designed using this code-enhanced procedure. 00:27:52.119 --> 00:27:55.889 And what you start to see is that the code enhanced archetypes 00:27:55.889 --> 00:28:00.779 right here have much lower collapse probabilities in both 00:28:00.780 --> 00:28:08.120 the MCE and in the M9 compared to the code minimum archetypes. 00:28:09.400 --> 00:28:13.640 And so here I’ve also included basin effects in the MCE. 00:28:13.640 --> 00:28:16.359 Now, if you remember from my first slide, 00:28:16.359 --> 00:28:19.519 basin effects aren’t actually included in the current design – in the current 00:28:19.519 --> 00:28:25.140 version of the – of the design code. I think future national seismic hazard 00:28:25.140 --> 00:28:28.049 maps are actually working on including basin effects in them, 00:28:28.049 --> 00:28:31.809 but right – as of today, basin effects aren’t – are not explicitly included 00:28:31.809 --> 00:28:35.049 in the national hazard maps. But if we were to take the 00:28:35.049 --> 00:28:38.830 MCE earthquake, and we were to include basin effects in it, 00:28:38.830 --> 00:28:41.400 we can start to estimate what the collapse risk 00:28:41.400 --> 00:28:46.180 of an MCE earthquake with basin effects. And what you can start to see 00:28:46.180 --> 00:28:50.280 is that the collapse probability is even – are much larger. 00:28:50.280 --> 00:28:54.620 Larger than the M9 motions for the code minimum designs 00:28:54.620 --> 00:29:00.519 and similar in terms of the collapse probability to the M9 motions. 00:29:00.519 --> 00:29:10.730 So the – so, again, I’m comparing the M9 collapse risk to this 10% target, 00:29:10.730 --> 00:29:14.350 but is that correct? Because, when the code says 00:29:14.350 --> 00:29:20.679 10% collapse in the MCE, well, that MCE is a 2,500-year-return earthquake. 00:29:20.679 --> 00:29:26.090 The M9, for the Pacific Northwest, has about a 500-year return period – 00:29:26.090 --> 00:29:31.419 a much shorter return period. So 10% target in the M9 will not 00:29:31.419 --> 00:29:36.370 translate to 10% collapse in the MCE. So if I were to hash out my math 00:29:36.370 --> 00:29:42.360 and try to figure out what the collapse probability, given a 2,500-year return 00:29:42.360 --> 00:29:45.409 period under an M9, it actually would be much larger. 00:29:45.409 --> 00:29:50.109 And so these probabilities of collapse that you’re seeing under an M9 00:29:50.109 --> 00:29:57.559 will actually start to exceed that target that the code is trying to achieve. 00:29:57.560 --> 00:30:02.000 So this is important when looking at these results as well. 00:30:02.820 --> 00:30:07.800 So most of the – so all the results I’ve shown thus far is for individual 00:30:07.800 --> 00:30:13.870 archetypes and for two – for one location – so, yeah, downtown Seattle. 00:30:13.870 --> 00:30:22.340 How would the collapse – how would the collapse risk vary regionally? 00:30:22.340 --> 00:30:28.239 And so I’ve shown you that multiple ground motion characteristics matter. 00:30:28.239 --> 00:30:31.159 Spectral accelerations matters. Significant duration or duration 00:30:31.159 --> 00:30:34.399 of the motion matters. And the spectral shape matters. 00:30:34.400 --> 00:30:37.940 And so what we’ve done is we’ve formulated a ground motion 00:30:37.940 --> 00:30:41.480 intensity measure that sort of combines these three effects, 00:30:41.480 --> 00:30:44.740 and we call it the effective spectral acceleration. 00:30:44.740 --> 00:30:47.720 Now, I haven’t – I’m not going to get into the math of how we calculate this, 00:30:47.720 --> 00:30:51.220 but I do have a slide for anyone who’s interested later. 00:30:51.820 --> 00:30:59.200 So here’s – so why does this work? Well, so here’s two collapse fragilities 00:30:59.200 --> 00:31:05.160 derived just using – derived using the results for the code minimum archetypes. 00:31:05.160 --> 00:31:09.060 And the collapse fragility, which basically just predicts 00:31:09.060 --> 00:31:14.680 the probability of a building collapsing given earthquake intensity, 00:31:14.680 --> 00:31:17.040 and the earthquake intensity measure that I’ve used here 00:31:17.049 --> 00:31:20.269 is spectral acceleration. The earthquake intensity measure 00:31:20.269 --> 00:31:24.600 that I’ve used here is effective spectral acceleration. 00:31:24.600 --> 00:31:29.159 And what you start to see is, if I derive the results from the analysis 00:31:29.159 --> 00:31:33.970 that I’ve run using the MCE motions and the M9 motions, you start to see 00:31:33.970 --> 00:31:38.081 that are variations in collapse prediction if I use the M9 motions 00:31:38.081 --> 00:31:41.429 versus if I use the maximum considered earthquake motions. 00:31:41.429 --> 00:31:46.840 Now, this variation is really due to ground motion characteristics, 00:31:46.840 --> 00:31:51.779 like duration and shape, that aren’t captured by spectral acceleration alone. 00:31:51.780 --> 00:31:56.500 And so if I jump towards this effective spectral acceleration 00:31:56.500 --> 00:32:00.420 intensity measure that accounts for spectral acceleration, shape, 00:32:00.429 --> 00:32:04.989 and duration, you can start to see that the collapse predictions 00:32:04.989 --> 00:32:09.259 start to be similar regardless of the ground motion sets. 00:32:09.259 --> 00:32:15.799 So if – so if I use this effective spectral acceleration regionally, 00:32:15.799 --> 00:32:20.769 and I can start to – start to see areas of high effective spectral acceleration 00:32:20.769 --> 00:32:26.759 intensity and try to sort of predict where most of my building damage will be. 00:32:27.980 --> 00:32:33.520 And we can do this by evaluating this integral. And I’ll break it down for you. 00:32:33.529 --> 00:32:37.859 The integral is basically trying to predict the collapse probability 00:32:37.859 --> 00:32:43.479 for a given point on a map by using the collapse fragility functions 00:32:43.480 --> 00:32:47.360 using effective spectral acceleration that I’ve defined previously – 00:32:47.360 --> 00:32:49.190 or, that I’ve showed previously. 00:32:49.190 --> 00:32:52.400 And then, since I have – I don’t have only one scenario. 00:32:52.400 --> 00:32:55.859 I have 30 scenarios, and so I want to capture the uncertainty 00:32:55.859 --> 00:33:01.019 in effective spectral acceleration. So if I take that – integrate that 00:33:01.019 --> 00:33:05.359 with the collapse fragility, and I integrate that with 00:33:05.359 --> 00:33:10.990 some uncertainty with the design strength of the building and uncertainty 00:33:10.990 --> 00:33:16.080 due to some modeling variation – or, modeling uncertainty, 00:33:16.080 --> 00:33:20.600 and if I evaluate this integral for each location around the Puget Sound, 00:33:20.600 --> 00:33:26.889 I can sort of predict the probability of collapse for different period buildings. 00:33:26.889 --> 00:33:31.300 And so here I’ve shown a 0.5-second collapse – 00:33:31.300 --> 00:33:34.960 regional collapse probability, 1 second, and 2 seconds. 00:33:34.960 --> 00:33:41.139 And it’s sort of – can pinpoint areas where I’m expecting to see 00:33:41.139 --> 00:33:44.539 large damage, not only because the spectral accelerations are high, 00:33:44.540 --> 00:33:49.300 but because I have long durations and damaging spectral shapes. 00:33:51.440 --> 00:33:55.780 And so, to conclude, the simulating motions of a magnitude 9 Cascadia 00:33:55.780 --> 00:33:58.740 subduction zone earthquake are damaging because I’ve shown you 00:33:58.740 --> 00:34:01.830 that they have larger spectral accelerations – more so than what 00:34:01.830 --> 00:34:07.740 is considered in design, damaging spectral shapes, long durations. 00:34:07.740 --> 00:34:12.799 The building performance that we’re seeing is quite different 00:34:12.799 --> 00:34:19.950 because of the large drifts that are – the large drifts for part of 00:34:19.950 --> 00:34:24.889 the simulations that are sort of driving the collapse susceptibility. 00:34:24.889 --> 00:34:31.970 Excluding basins in the MCE sort of led to that 10% target to be met. 00:34:31.970 --> 00:34:35.379 So we know that, if we are designing buildings according to 00:34:35.379 --> 00:34:41.720 what the hazard is today, you’re fine. But if you were to use a hazard such 00:34:41.720 --> 00:34:45.210 as the M9, or the hazard that we’re predicting from the M9 simulations, 00:34:45.210 --> 00:34:49.770 we start to see these larger collapse probabilities than what we would expect. 00:34:49.770 --> 00:34:52.770 Code minimum collapse probabilities greatly exceeded 00:34:52.770 --> 00:34:56.990 the 10% target in the M9. 00:34:56.990 --> 00:35:01.650 The code enhanced archetypes had M9 collapse probabilities 00:35:01.650 --> 00:35:06.320 that were similar to the MCE target, but it’s important to – ah – 00:35:06.320 --> 00:35:10.500 it’s important to note that the M9 has 00:35:10.500 --> 00:35:14.060 a much shorter return period than the MCE. 00:35:14.060 --> 00:35:18.460 And then the regional collapse prediction can be predicted using 00:35:18.460 --> 00:35:22.120 this effective spectral acceleration measure that accounts for the 00:35:22.120 --> 00:35:26.250 spectral acceleration, variation, shape, and duration regional variation. 00:35:26.250 --> 00:35:31.040 And sort of – you can isolate areas of high collapse probability. 00:35:32.100 --> 00:35:35.900 Before I leave, I’d like to acknowledge multiple funding agencies for supporting 00:35:35.900 --> 00:35:41.280 this project, and of course the M9 team for supporting this research. 00:35:41.280 --> 00:35:43.140 Thank you. 00:35:43.140 --> 00:35:49.540 [Applause] 00:35:49.540 --> 00:35:51.100 - [inaudible] - Yeah. It’s – I don’t know why it 00:35:51.100 --> 00:35:53.160 was flickering between slides. 00:35:53.160 --> 00:35:55.720 - I realized in your introduction – or, in my introduction of you, 00:35:55.720 --> 00:35:59.030 I sold Nasser a bit short. Recently, the City of Seattle has 00:35:59.030 --> 00:36:02.550 changed its building code in response to the work that Nasser has done with 00:36:02.550 --> 00:36:07.790 Art and Erin to try and increase the safety of buildings built after December. 00:36:07.790 --> 00:36:10.140 - Right, so … - Of this year. 00:36:10.140 --> 00:36:15.260 - Yeah. So there’s, like, two things going on right now where the national 00:36:15.260 --> 00:36:20.060 seismic hazard maps are going to include basin effects in the next cycle. 00:36:20.060 --> 00:36:24.180 So the spectral accelerations that you’ll see in the – in the building codes 00:36:24.180 --> 00:36:28.440 are going to go up because now they include basin amplifications. 00:36:28.440 --> 00:36:36.920 And then, for tall buildings in Seattle, they were including basin amplifications 00:36:36.920 --> 00:36:42.120 in the design process, but as of, I think, end of November, 00:36:42.120 --> 00:36:45.780 those basin amplification factors are actually going up. 00:36:45.789 --> 00:36:49.880 Because what we found is that the basin amplification factors that 00:36:49.880 --> 00:36:54.260 we’re seeing in the M9 simulations are actually quite different to those 00:36:54.260 --> 00:36:59.869 that are being used today. Because the ones that are used 00:36:59.869 --> 00:37:05.720 today are derived from crustal earthquakes in California. 00:37:06.820 --> 00:37:09.400 - I just didn’t want to leave without saying that you – 00:37:09.400 --> 00:37:13.980 already contributing to the City of Seattle’s safety. Questions? 00:37:17.100 --> 00:37:24.660 - Yeah, that set of records that, you know, are for the maximum 00:37:24.670 --> 00:37:29.109 considered earthquake, one, what is the maximum considered earthquake? 00:37:29.109 --> 00:37:33.660 And, two, give us a little bit more information on what are the magnitude 00:37:33.660 --> 00:37:41.000 and distances of those records that were evaluated to compare to M9 – 00:37:41.000 --> 00:37:43.660 well, that are part of the building code right now, I guess. 00:37:43.660 --> 00:37:45.400 - Yeah. Yeah. 00:37:45.400 --> 00:37:53.320 So the maximum considered earthquake represents a ground motion with shaking 00:37:53.320 --> 00:38:01.200 that corresponds to a 2,500-year earthquake for that particular location. 00:38:01.200 --> 00:38:06.700 Now, that 2,500-year shaking may be controlled 00:38:06.700 --> 00:38:10.630 by different types of earthquake mechanisms. 00:38:10.630 --> 00:38:14.260 So we have some – so, in Seattle, we have the Seattle Fault. 00:38:14.260 --> 00:38:19.060 So there’s some motions that I’m using that are from crustal earthquakes. 00:38:19.060 --> 00:38:22.700 And so those motions are actually not as recorded motions. 00:38:22.700 --> 00:38:28.020 I sort of look at recorded ground motions, and then I would 00:38:28.020 --> 00:38:35.000 scale them up to match the response spectra of a maximum – 00:38:35.000 --> 00:38:38.200 of a 2,500-year-return earthquake. 00:38:38.210 --> 00:38:41.079 And so I have some that represent the Seattle Fault. 00:38:41.079 --> 00:38:44.220 I have some that represent deep inter-slab earthquakes. 00:38:44.220 --> 00:38:47.630 And I have some that represent interface earthquakes. 00:38:47.630 --> 00:38:51.390 And then I also have some earthquakes with some pulse-like characteristics 00:38:51.390 --> 00:38:56.609 because the Seattle Fault is, I think, like, maybe 2 miles, 3 miles 00:38:56.609 --> 00:39:00.750 from downtown Seattle. A lot of the motions that we would 00:39:00.750 --> 00:39:05.049 expect from the Seattle Fault – a 2,500-year-level motion 00:39:05.049 --> 00:39:08.500 from the Seattle Fault would have some pulse-like characteristics. 00:39:08.500 --> 00:39:12.099 So we’ve selected some motions to have some of those pulse-like characteristics. 00:39:12.900 --> 00:39:15.140 - Yeah. Okay. 00:39:15.140 --> 00:39:20.540 None of which sound much like M9. - No. [laughs] 00:39:20.540 --> 00:39:25.700 - And then the next question is, were those same set of earthquakes 00:39:25.700 --> 00:39:33.220 used to set the design specification of, 00:39:33.220 --> 00:39:38.580 what, the minimum or the present-day design considerations? 00:39:39.280 --> 00:39:43.710 - Um, I – hm. 00:39:43.710 --> 00:39:49.270 So the – I guess the design-level earthquake – 00:39:49.270 --> 00:39:56.620 the 2,500-year-level earthquake was sort of picked out to – 00:39:56.620 --> 00:40:03.840 so buildings are designed to have a 1% chance of collapse in 50 years. 00:40:03.840 --> 00:40:06.990 That’s what the building code targets, okay? 00:40:06.990 --> 00:40:13.200 And so what engineers have done is they’ve sort of back-calculated, well, 00:40:13.200 --> 00:40:18.671 to get that target, what do I need – what kind of earthquake return period 00:40:18.671 --> 00:40:25.640 do I need to design for so that I achieve that 1% chance of collapse in 50 years? 00:40:25.640 --> 00:40:30.960 So that’s how – that’s how we get to the 2,500-year earthquake. 00:40:32.240 --> 00:40:36.029 - Yeah, but sooner or later, it’s – yeah, you – sooner or later, you’d need 00:40:36.029 --> 00:40:38.240 a ground motion to calculate that … - Yeah. 00:40:38.240 --> 00:40:47.940 - … 1% chance [inaudible]. And I’m just wondering, to what extent 00:40:47.940 --> 00:40:55.200 those ground motions have already been taken into account 00:40:55.200 --> 00:40:59.300 in the present design. 00:40:59.300 --> 00:41:04.140 Did they know, for example, what you showed is, when the 00:41:04.140 --> 00:41:08.200 probability gets pretty close to 10%, you know, they’re all less than that, 00:41:08.210 --> 00:41:11.019 and then it kind of kisses that at the end. - Right. 00:41:11.019 --> 00:41:15.920 - And whether – that’s just reproducing what went into 00:41:15.920 --> 00:41:19.840 the design considerations in the first place. 00:41:19.840 --> 00:41:27.799 - So – yeah, so the – all the recorded motions that we have today are sort of 00:41:27.799 --> 00:41:35.400 fit to these ground motion models that can predict the response spectra, 00:41:35.400 --> 00:41:41.090 given distance and magnitude and other site parameters. 00:41:41.090 --> 00:41:45.270 And so there are ground motion models that represent earthquakes 00:41:45.270 --> 00:41:47.610 from different source mechanisms. 00:41:47.610 --> 00:41:52.730 And what engineers do is they sort of use these ground motion models. 00:41:52.730 --> 00:41:59.520 They take earthquake recurrence models, and they try to figure out what level of 00:41:59.520 --> 00:42:05.270 shaking, or what spectral accelerations I need to design for, given the 00:42:05.270 --> 00:42:09.690 earthquake rupture models and given these ground motion models that are 00:42:09.690 --> 00:42:13.039 sort of derived empirically using the ground motions that are recorded today. 00:42:13.039 --> 00:42:16.849 So what – so whenever we get a new earthquake, 00:42:16.849 --> 00:42:19.630 that goes into the next round of ground motion models. 00:42:19.630 --> 00:42:23.470 And then, there are folks that sort of estimate, well, what is 00:42:23.470 --> 00:42:26.110 the return periods of all these different types of earthquakes? 00:42:26.110 --> 00:42:32.580 We take those predictions, and we take our predictions for ground shaking 00:42:32.580 --> 00:42:40.380 intensity, and then we use those two things to calculate what the – what the – 00:42:40.380 --> 00:42:45.340 what spectral accelerations we would expect to see in the 100 years, 200 years, 00:42:45.340 --> 00:42:48.720 300 years, 2,000 years, and 5,000 years, and so on. 00:42:48.720 --> 00:42:53.350 And then use – so that is what is called a hazard curve. 00:42:53.350 --> 00:42:55.799 And then we take that hazard curve, and we integrate it with 00:42:55.799 --> 00:43:01.029 a building fragility curve that sort of corresponds to collapse. 00:43:01.029 --> 00:43:07.740 And then sort of the process then would give us a collapse – 00:43:07.740 --> 00:43:11.599 a collapse probability in 50 years. 00:43:11.599 --> 00:43:18.619 So then – so then we’re trying to target 1%, so how do we – how much – 00:43:18.619 --> 00:43:24.440 how do we vary our design intensity to get that 1% in 50 years? 00:43:24.440 --> 00:43:28.039 - I think, for all of this [chuckles] – and I’m sorry to take up 00:43:28.040 --> 00:43:30.520 so much time [inaudible]. - No, that’s fine. 00:43:31.300 --> 00:43:35.700 Present-day design ground motions are based on what we know now. 00:43:35.700 --> 00:43:38.820 - Yes. - Or what we knew five or 10 years ago, 00:43:38.820 --> 00:43:43.340 probably more correctly. The real value of M9 is to give you 00:43:43.341 --> 00:43:50.050 a look at something that I am confident will happen, even though 00:43:50.050 --> 00:43:52.610 we’re not quite sure when. - Right. Right. 00:43:52.610 --> 00:43:58.980 - And it’s this – it’s this forward nature of M9 that makes it so important. 00:43:58.980 --> 00:44:01.300 - Right. - And so different from current practice. 00:44:01.300 --> 00:44:04.340 - Right, right. Yeah, and there’s a lot of people at the USGS 00:44:04.349 --> 00:44:08.589 that are actually trying to figure out how to take what we’ve – what we’ve 00:44:08.589 --> 00:44:14.410 found in physics-based simulations and integrate it to – with the 00:44:14.410 --> 00:44:17.059 national seismic hazard map. So the national seismic hazard maps 00:44:17.059 --> 00:44:21.950 are not only derived from recorded earthquakes, but from earthquakes 00:44:21.950 --> 00:44:26.420 that we can sort of predict using these physics-based simulations. 00:44:27.460 --> 00:44:30.020 - I’m going to go because I’m holding a mic. 00:44:30.020 --> 00:44:33.470 Nasser, in one of your figures on basin amplification, you showed data – 00:44:33.470 --> 00:44:36.440 you didn’t speak to it, but you showed data from the Nisqually earthquake that 00:44:36.440 --> 00:44:38.000 was recorded in the Puget Lowlands. 00:44:38.000 --> 00:44:42.319 Do you have thoughts, or can you explain the difference between the shape 00:44:42.320 --> 00:44:47.020 and peak response of the Nisqually motions versus the synthetic M9? 00:44:47.020 --> 00:44:53.040 - Yeah. So we’ve – I’ve found that the trends with period in terms of 00:44:53.040 --> 00:44:58.840 basin amplifications are similar between Japan, Nisqually, and the M9. 00:44:58.840 --> 00:45:04.200 But what we’ve – what we’re seeing is that there’s variations in 00:45:04.200 --> 00:45:08.080 basin amplification factors. So the magnitudes aren’t matching up. 00:45:08.080 --> 00:45:14.880 And it’s kind of difficult to sort of pinpoint why that is, but you could – 00:45:14.880 --> 00:45:17.849 some suggestions I have is, maybe it’s just path effects. 00:45:17.849 --> 00:45:25.930 Nisqually was a deep earthquake. So the seismic waves were sort of 00:45:25.930 --> 00:45:29.180 attacking the basin or attacking [inaudible] from a different angle, 00:45:29.180 --> 00:45:34.619 which might change the amount of basin amplifications that you’re seeing. 00:45:34.619 --> 00:45:38.430 The M9 earthquakes are much shallower, and they’re further away. 00:45:38.430 --> 00:45:42.520 And so the amount of amplifications that we’re seeing from that could be 00:45:42.520 --> 00:45:46.740 different just because of how the motions got to the basin. 00:45:46.740 --> 00:45:51.200 - [inaudible] shape of the peak response of Nisqually at 5 or 6 seconds. 00:45:51.200 --> 00:45:53.740 - At 5 or 6 seconds, yeah. 00:45:53.740 --> 00:45:58.660 Yeah. So it’s a different source mechanism. 00:45:58.660 --> 00:46:03.970 And so one of the more recent questions that we’re having is, well, 00:46:03.970 --> 00:46:08.570 we know that the M9 motions are – the basin amplification factors – 00:46:08.570 --> 00:46:11.270 or, the basin is amplifying the M9 motions this way. 00:46:11.270 --> 00:46:16.870 Well, is this consistent with the kind of amplifications that you would 00:46:16.870 --> 00:46:21.540 see in different – in the Seattle Fault? Is it consistent – is it similar to 00:46:21.540 --> 00:46:23.349 what you would see for deep inter-slab earthquakes? 00:46:23.349 --> 00:46:30.059 So there’s some – so there are some efforts that are ongoing looking at 00:46:30.060 --> 00:46:34.440 amplifications for different source mechanism earthquakes. 00:46:38.280 --> 00:46:42.200 - I have two short questions. - Mm-hmm. 00:46:42.200 --> 00:46:44.760 - Maybe the answer will be wrong, but I don’t know. 00:46:44.760 --> 00:46:50.480 The first one is the concept [inaudible] last slide. 00:46:50.480 --> 00:46:56.700 Damaging spectral … - Shape? 00:46:56.700 --> 00:46:58.280 - Shapes, yeah. - Mm-hmm. 00:46:58.280 --> 00:47:01.500 - Would you elaborate on that a bit? And then the second one is, 00:47:01.500 --> 00:47:07.220 what is your definition of collapse? I mean, do you mean pancake? 00:47:07.220 --> 00:47:14.380 Or do you mean it is damaged to the point where the building cannot 00:47:14.380 --> 00:47:21.260 be occupied anymore and probably [inaudible] to be torn down? 00:47:21.260 --> 00:47:23.540 What is your collapse … - What brings my collapse, yeah. 00:47:23.540 --> 00:47:26.860 No, I – that’s all information that I left out of the slide – 00:47:26.860 --> 00:47:32.020 or, left out of this presentation. So, to answer your first question, 00:47:32.020 --> 00:47:34.760 what I mean by damaging spectral shapes is that, 00:47:34.760 --> 00:47:39.220 if you have – if you – going back to this slide. 00:47:40.320 --> 00:47:46.780 So there’s quite a few realizations where you sort of 00:47:46.780 --> 00:47:50.260 get this hump in the response spectra. 00:47:50.260 --> 00:47:55.470 And so, if you’re at a period that’s around 1 second, half a second, 00:47:55.470 --> 00:48:00.339 you know, and that spectral intensity is strong enough to cause initial yielding, 00:48:00.339 --> 00:48:04.710 and your structure starts to yield, and it’s – it deforms inelastically, 00:48:04.710 --> 00:48:08.369 and its period elongates, what that does is that 00:48:08.369 --> 00:48:16.069 long-period energy further away start to affect the response. 00:48:16.069 --> 00:48:22.410 And so, if you have a ground motion with high long-period energy, 00:48:22.410 --> 00:48:27.570 that kind of motion will be more damaging than motions 00:48:27.570 --> 00:48:31.730 that have low-period energy at long periods. 00:48:31.730 --> 00:48:36.750 And so that’s what I mean by more damaging spectral shapes. 00:48:36.750 --> 00:48:43.490 And so I tried to prove this by saying, well, I’ve computed this spectral shape 00:48:43.490 --> 00:48:47.700 factor, which is basically just the integral of the response spectrum 00:48:47.700 --> 00:48:52.210 from periods of interest. And then I plot this spectral 00:48:52.210 --> 00:48:55.529 shape factor with period. And what we start to see is that 00:48:55.529 --> 00:49:03.280 the intensity of spectral – of spectral shape in Seattle is way more damaging 00:49:03.280 --> 00:49:08.450 than motions outside the basin and more damaging than motions that are sort of 00:49:08.450 --> 00:49:12.180 selected and scaled to match the maximum considered earthquake today. 00:49:12.180 --> 00:49:15.580 So that’s what I meant by damaging spectral shapes. 00:49:16.460 --> 00:49:19.740 To answer your second question – does that answer your first question? 00:49:20.420 --> 00:49:24.740 - [inaudible] the motions are large enough to … 00:49:24.740 --> 00:49:27.900 - Cause – yeah, cause yielding. - Right. Okay, then you 00:49:27.900 --> 00:49:30.140 have [inaudible]. That’s fine. - Yeah. 00:49:30.140 --> 00:49:33.060 - So the second one [inaudible] … - Yeah. So the second question is – 00:49:33.060 --> 00:49:37.040 so I – so the second question I actually left out of the slide – 00:49:37.040 --> 00:49:43.460 left out of the slideshow just to shorten it. But the way that I define collapse is – 00:49:43.460 --> 00:49:56.260 so my models simulate the deformations in the lateral force-resisting system. 00:49:56.270 --> 00:49:59.579 But – and the lateral force-resisting system can 00:49:59.580 --> 00:50:08.000 take these stresses and strains. But what we’re finding is that the 00:50:08.000 --> 00:50:14.540 components associated with the gravity system – so the column-to-slab 00:50:14.540 --> 00:50:19.640 connections, the connection of the slab to the – to the concrete core, 00:50:19.650 --> 00:50:24.480 those connections may not be able to go to such large drifts. 00:50:24.480 --> 00:50:30.200 And so there are various researchers that have sort of predicted what the 00:50:30.200 --> 00:50:36.150 collapse – or, what the drift capacity of all these connections are. 00:50:36.150 --> 00:50:39.599 And so we have a component – a structural component model that 00:50:39.599 --> 00:50:47.619 predicts the probability of failure of a slab-column connection, given drift. 00:50:47.619 --> 00:50:57.020 And so, if I take my variation of drift demand within a ground motion set, 00:50:57.020 --> 00:51:02.260 and I integrate it with a probability of failing a slab-column connection 00:51:02.260 --> 00:51:10.100 given drift capacity, I can estimate what the collapse probability 00:51:10.100 --> 00:51:14.250 of that building is due to failure of the gravity system. 00:51:14.250 --> 00:51:18.310 So I’m not seeing a lot of pancaking like you would see in sort of 00:51:18.310 --> 00:51:22.559 moment-frame structures. The collapse risk that I’m seeing 00:51:22.560 --> 00:51:29.160 is mainly driven by failure of the slab-column connections. 00:51:32.220 --> 00:51:34.520 Does that … 00:51:34.520 --> 00:51:37.100 - [inaudible] I mean, that’s one explanation. 00:51:37.100 --> 00:51:38.300 - Yeah. 00:51:41.500 --> 00:51:43.480 - I think you have to think about this a little bit. 00:51:43.480 --> 00:51:47.220 - Yeah. Well, we – I’m happy to chat with you after. 00:51:47.220 --> 00:51:50.220 - [inaudible] - There’s a bunch of slides that I – 00:51:50.220 --> 00:51:53.620 I can show them to you when we sit down. Yep? 00:51:53.620 --> 00:51:57.460 - Yeah, you pointed out the difference between your MCE-R scale 00:51:57.460 --> 00:52:03.089 ground motions, but you didn’t really get into, are they – I mean, they match 00:52:03.089 --> 00:52:06.020 the spectral shape that you scale them to. - Yeah. 00:52:06.020 --> 00:52:09.299 - But, you know, you pointed out that, oh, there’s all these difference in the 00:52:09.299 --> 00:52:13.780 collapse probability, but you didn’t say anything about, are the durations similar 00:52:13.780 --> 00:52:18.359 when you have subduction zone events? Is that the key parameter that’s 00:52:18.359 --> 00:52:22.079 missing in your scaled ground motions that creates differences? 00:52:22.079 --> 00:52:27.609 What is it in the M9 ground motions that you have these metrics for 00:52:27.609 --> 00:52:31.290 that you don’t have in your scaled ground motions? 00:52:31.290 --> 00:52:34.559 - So that’s a good question. I don’t have, actually, 00:52:34.559 --> 00:52:38.369 a slide that illustrates the differences in durations 00:52:38.369 --> 00:52:42.160 and shapes between the M9 and the MCE. 00:52:43.120 --> 00:52:50.200 But – so when we’re selecting the MCE – when we’re selecting 00:52:50.210 --> 00:52:57.930 motions that represent the MCE, we sort of capture parameter – 00:52:57.930 --> 00:52:59.970 ground motion intensities such as duration by 00:52:59.970 --> 00:53:03.829 selecting motions from other interface earthquakes. 00:53:03.829 --> 00:53:07.700 So here you see the response spectra of crustal earthquakes, inter-slab 00:53:07.700 --> 00:53:14.000 earthquakes, interface earthquakes that were selected to be at MCE intensity. 00:53:14.000 --> 00:53:18.400 Now, the interface – the crustal earthquakes would have much shorter 00:53:18.400 --> 00:53:23.049 durations. So their durations are probably between 5 and 10 seconds. 00:53:23.049 --> 00:53:26.609 And then – but if you go to the interface earthquakes, those durations 00:53:26.609 --> 00:53:32.230 are going to be around 60, 70 seconds. So these earthquake motions are 00:53:32.230 --> 00:53:34.460 interface earthquake motions that are recording in Japan. 00:53:34.460 --> 00:53:40.590 So they’re mainly Tohoku, Tokachi-Oki, some of the Tohoku aftershocks. 00:53:40.590 --> 00:53:46.920 And so duration is sort of not explicitly considered, but implicitly considered 00:53:46.920 --> 00:53:51.160 by selecting motions that have the same source mechanism. 00:53:52.540 --> 00:53:57.890 The other thing is, the M9 motions have a lot of basin amplification. 00:53:57.890 --> 00:54:05.369 And so – which will increase the spectral acceleration – the response spectra. 00:54:05.369 --> 00:54:10.760 And so, when I select the MCE motions, I’m not selecting it – I’m selecting it 00:54:10.760 --> 00:54:16.740 to what is predicted today by the ground motion models that are used 00:54:16.740 --> 00:54:21.060 in the national seismic hazard maps. So those models that are used 00:54:21.060 --> 00:54:26.069 don’t have basin effects in them. So I’m not considering damaging 00:54:26.069 --> 00:54:31.380 spectral shapes or damaging spectral accelerations there. 00:54:32.280 --> 00:54:35.200 So the differences that we’re seeing between the M9 motions and the 00:54:35.200 --> 00:54:41.960 MCE-R is mainly due to shape and acceleration because 00:54:41.960 --> 00:54:46.460 I’m not accounting for basin amplifications. 00:54:46.460 --> 00:54:50.260 But the duration is sort of implicitly considered. 00:54:50.260 --> 00:54:54.510 But, again, I’m only comparing an – so my M9 motions are all – 00:54:54.510 --> 00:55:00.250 there’s 30 long-duration motions. The MCE – so for 2 – 00:55:00.250 --> 00:55:04.200 for a 2-second period, which is what these MCE motions are for, 00:55:04.200 --> 00:55:07.560 only 50% of them are actually interface earthquakes. 00:55:08.180 --> 00:55:10.259 - The other is – sort of two other comments. 00:55:10.259 --> 00:55:14.450 One is that my understanding from Nico Luco is this 10% in 50 years 00:55:14.450 --> 00:55:21.200 collapse probability, he has very little confidence that it’s actually 10%. 00:55:21.200 --> 00:55:23.520 It’s some percent. - It’s some percent. Yeah. 00:55:23.520 --> 00:55:27.660 - And it’s – and so to actually start to compare numbers, it seems like 00:55:27.660 --> 00:55:33.319 we’re not quite there yet in terms of the engineering design and our sort of – 00:55:33.319 --> 00:55:36.869 the uncertainty of that number is the methodology is set up to get there. 00:55:36.869 --> 00:55:38.820 - Right. - But we’re not there yet, and then … 00:55:38.820 --> 00:55:44.220 - So, yeah, no, that’s a good point. I mean, it’s not – yeah, a 1% chance 00:55:44.220 --> 00:55:49.180 in 50 does not necessarily guarantee you a – or, a 10% chance of MCE – 00:55:49.180 --> 00:55:51.759 a 10% chance of collapse in the MCE does not necessarily 00:55:51.760 --> 00:55:55.880 guarantee you a 1% chance of collapse in 50 years. 00:55:55.880 --> 00:56:03.460 To answer what the 50-year collapse risk would be, I would probably need to do – 00:56:03.460 --> 00:56:07.520 well, I would have to consider all the earthquakes – not just the M9, 00:56:07.529 --> 00:56:11.480 but all the earthquakes that would be observed 00:56:11.480 --> 00:56:15.140 by this building in a 50-year period. 00:56:15.140 --> 00:56:19.820 Or a 2,500-year period to calculate the accurate collapse risk. 00:56:19.839 --> 00:56:23.650 - And the other point I want to make is that you sort of – you seem to be 00:56:23.650 --> 00:56:29.809 mixing probabilities and recurrence intervals that the 2,500-year 00:56:29.809 --> 00:56:34.599 ground motion is not a 25-year recurrence interval of an earthquake. 00:56:34.599 --> 00:56:37.490 It’s the recurrence interval of a ground motion. 00:56:37.490 --> 00:56:41.549 So you may have a 500-year event and be at sort of, you know, 00:56:41.549 --> 00:56:45.039 one sigma above that to get your 2,500-year. 00:56:45.039 --> 00:56:48.430 And so, when you’re comparing the M9s where you have a 00:56:48.430 --> 00:56:51.859 500-year recurrence interval, but 30 scenarios, and you’re going – 00:56:51.859 --> 00:56:55.920 you’re taking account that you have some out on the tail above the median. 00:56:55.920 --> 00:56:57.740 - Right. - You’re starting to mix 00:56:57.740 --> 00:57:01.130 sort of probabilities, and you need to know sort of the probabilities of the 00:57:01.130 --> 00:57:04.009 M9 events in each of those variations … - Right. 00:57:04.009 --> 00:57:06.549 - … when you’re comparing the probabilities of ground motion. 00:57:06.549 --> 00:57:07.859 - Right, right. 00:57:07.859 --> 00:57:18.370 No, it – yeah, I – you’re perfectly right. It’s – so a 500-year return period, there’s, 00:57:18.370 --> 00:57:22.880 you know, an average four or five that could happen in the 2,500-year return. 00:57:22.880 --> 00:57:28.009 So you’d have to figure out what the collapse probability is for – 00:57:28.009 --> 00:57:33.420 in a 2,500-year period under multiple M9s. 00:57:33.420 --> 00:57:39.640 So, yeah. That is a valid point. 00:57:43.620 --> 00:57:48.680 - Okay, final question? Great. 00:57:48.690 --> 00:57:51.500 If you’d like to meet with Nasser this afternoon, I think he has 00:57:51.500 --> 00:57:55.180 some more time. Just come find me or give me a call in the office. 00:57:55.180 --> 00:57:57.540 And then, thanks again. - Thank you all for coming. 00:57:57.540 --> 00:58:00.860 [Applause] 00:58:00.860 --> 00:58:05.100 [Silence]