WEBVTT
Kind: captions
Language: en

00:00:01.140 --> 00:00:03.340
So you’re going to
announce the other …

00:00:03.340 --> 00:00:05.980
- Yeah.
Good morning.

00:00:05.980 --> 00:00:08.660
Hello, everyone.
And welcome to the first of our informal

00:00:08.670 --> 00:00:11.889
seminar series on structural
engineering, structural design,

00:00:11.889 --> 00:00:15.820
and ground motion simulation that
I’ve called Shaking the Structures.

00:00:15.820 --> 00:00:20.120
And we’re trying to answer the question
of just how safe these tall buildings are

00:00:20.120 --> 00:00:24.509
and start discussing some of the issues
raised by the New York Times articles

00:00:24.509 --> 00:00:27.330
and others around
seismic design,

00:00:27.330 --> 00:00:29.520
performance-based
earthquake engineering.

00:00:29.520 --> 00:00:32.820
And we’re very lucky to have a –
quite a good series of seminar speakers

00:00:32.820 --> 00:00:36.480
coming up over the month of August
from different perspectives –

00:00:36.480 --> 00:00:41.440
earthquake engineers, geophysicists,
members of both the USGS and

00:00:41.440 --> 00:00:46.840
elsewhere, to try and talk about these
issues from a variety of perspectives.

00:00:46.840 --> 00:00:50.300
So we strongly encourage your
participation in asking questions,

00:00:50.300 --> 00:00:54.230
in being part of the discussions that’ll
happen on the off-hour seminars.

00:00:54.230 --> 00:00:57.320
So next Tuesday morning,
C.B. Crouse is coming down

00:00:57.320 --> 00:01:02.010
from AECOM to discuss
structural engineering.

00:01:02.010 --> 00:01:05.670
As well as, on Thursday afternoon
at 2:30, during the Earthquake Science

00:01:05.670 --> 00:01:09.720
Center coffee hour, we’ll have Nico
Luco from the Golden office talking

00:01:09.720 --> 00:01:14.560
about the national map and the interface
between engineers and USGS science.

00:01:14.560 --> 00:01:16.560
As well as several other
talks throughout the month.

00:01:16.560 --> 00:01:18.900
So we hope
you can join us.

00:01:19.820 --> 00:01:22.680
And before I hand it off
to Sarah to introduce Tom,

00:01:22.680 --> 00:01:26.990
we’re going to do pizza after this
seminar to help kind of promote

00:01:26.990 --> 00:01:32.870
discussion and have people chat
with Tom and others about his talk.

00:01:32.870 --> 00:01:35.470
If you’d like to join us for pizza,
if you could quickly raise your hands,

00:01:35.470 --> 00:01:38.380
we can get a head count
on how much to order.

00:01:39.640 --> 00:01:45.100
[Silence]

00:01:45.980 --> 00:01:48.000
You roughly know, Sarah?
- I don’t – [laughs] I don’t know.

00:01:50.720 --> 00:01:53.500
- We good?
Higher, maybe?

00:01:57.740 --> 00:02:01.500
- Looks like about a dozen.
- Okay. So hopefully you can join us,

00:02:01.510 --> 00:02:05.670
and if you can, we’d love contributions
towards the general pizza fund.

00:02:05.670 --> 00:02:09.700
And now Sarah can introduce Tom.
- Hello.

00:02:09.700 --> 00:02:13.349
So Tom’s instructions were, this is a
long talk, so make the introduction fast.

00:02:13.349 --> 00:02:16.719
This is Tom Heaton.
He is joining us today from Caltech.

00:02:16.720 --> 00:02:20.020
He got his degree at Indiana before
getting his doctorate at Caltech,

00:02:20.020 --> 00:02:22.820
working in private industry,
working for the USGS,

00:02:22.830 --> 00:02:25.340
becoming scientist in charge
of the Pasadena office,

00:02:25.340 --> 00:02:27.279
and then becoming
faculty at Caltech.

00:02:27.280 --> 00:02:29.040
But we still talk to him anyway.
[laughter]

00:02:29.040 --> 00:02:31.320
Please welcome Tom.

00:02:32.100 --> 00:02:36.719
[Applause]

00:02:36.720 --> 00:02:40.320
- Well, thanks.
Is the mic on? Are we …

00:02:40.900 --> 00:02:42.880
Am I getting  the mic?
- Yes.

00:02:42.890 --> 00:02:45.599
- Okay, good.
So it’s a pleasure to be back.

00:02:45.600 --> 00:02:50.880
I used to come here every month when
I was the scientist in charge in Pasadena.

00:02:50.880 --> 00:02:55.260
And then, when I joined Caltech,
I haven’t had the opportunity to be

00:02:55.269 --> 00:03:00.180
up here as much. When I joined Caltech,
a very odd thing happened.

00:03:00.180 --> 00:03:05.219
They gave me a joint position in
engineering and in seismology.

00:03:05.219 --> 00:03:07.049
And through the years,
most of my students

00:03:07.049 --> 00:03:09.200
have actually been
in engineering.

00:03:09.200 --> 00:03:14.090
I was very interested in the problem,
for the last 25 years, really,

00:03:14.090 --> 00:03:16.889
what we really know about
the physics of tall buildings

00:03:16.889 --> 00:03:19.969
and whether or not they’ll
survive big earthquakes.

00:03:19.969 --> 00:03:26.829
So unfortunately, 25 years means this
talk is too long. I’ll try to get through it.

00:03:26.829 --> 00:03:29.919
When we’re done – I’ll warn you
right now, I’ve got a cold.

00:03:29.919 --> 00:03:35.730
So if you know me, I’m happy to
see you, but you don’t want my cold.

00:03:35.730 --> 00:03:39.360
Don’t shake my hand.
So let’s get into this.

00:03:40.510 --> 00:03:44.000
So I put in a bunch of people I’ve
worked with through the years.

00:03:44.000 --> 00:03:47.340
Is Brad Aagaard out there?
There he is.

00:03:47.340 --> 00:03:52.219
So, Brad – I sort of started all this
with Brad Aagaard, who was actually

00:03:52.219 --> 00:03:56.140
a joint student
with John Hall and me.

00:03:56.140 --> 00:04:00.290
And John is probably one of the
best structural engineers in the world.

00:04:00.290 --> 00:04:04.690
Brad started off as a structural
engineer, believe it or not.

00:04:04.690 --> 00:04:08.439
And he got called over into the
Earth science part. Here he is.

00:04:08.439 --> 00:04:11.279
And then there’s been a number
of other students who have worked

00:04:11.279 --> 00:04:15.719
more on the earthquake ground
motion part, early warning part.

00:04:15.719 --> 00:04:20.579
My latest student, Kenny Buyco,
is really a structural engineer,

00:04:20.579 --> 00:04:24.060
and an excellent one.
And I don’t intend to go through all this.

00:04:24.060 --> 00:04:28.080
John Hall at Caltech
has been sort of the –

00:04:28.080 --> 00:04:32.400
behind a lot of the building analysis,
which is really important.

00:04:32.400 --> 00:04:38.620
So let’s see here. So there was this
mention of the New York Times articles.

00:04:38.620 --> 00:04:42.560
And how many people have read
the New York Times articles?

00:04:42.560 --> 00:04:46.550
They’re pretty interesting articles.
They were kind of an aftershock

00:04:46.550 --> 00:04:51.620
to an interesting meeting that
was in Senator Feinstein’s office

00:04:51.620 --> 00:04:53.760
about a year
and a half ago.

00:04:53.770 --> 00:04:58.830
It was – I went to lobby
Feinstein about early warning,

00:04:58.830 --> 00:05:04.370
and she just happened to be in her office.
And she’s a pretty suspicious person.

00:05:04.370 --> 00:05:10.550
I got to say, she’s – you do not –
you do not roll over a Dianne Feinstein.

00:05:10.550 --> 00:05:14.550
And she asked me, she said,
what’s going to happen to San Francisco

00:05:14.550 --> 00:05:18.159
when we get a big earthquake?
And I told her, well,

00:05:18.159 --> 00:05:21.650
there’s a lot of things to worry
about – a lot of problems.

00:05:21.650 --> 00:05:26.759
And she really perked up,
and she basically ended up

00:05:26.759 --> 00:05:30.150
ordering about a half dozen
people into her office

00:05:30.150 --> 00:05:34.560
in another week and a half
to talk about the problem.

00:05:34.560 --> 00:05:37.939
And Arty Gensler, who’s the – kind of
the top architect on the West Coast,

00:05:37.939 --> 00:05:41.789
and a bunch of structural engineers –
Lucy Jones was there,

00:05:41.789 --> 00:05:46.220
and I was there to talk about just
what we’re talking about today.

00:05:46.220 --> 00:05:53.700
And ultimately, Tom Fuller at the
New York Times kind of got wind of

00:05:53.710 --> 00:05:59.060
this meeting, and as a result of that,
he’s written these series of articles

00:05:59.060 --> 00:06:02.689
to kind of discuss whether
we really know what we’re doing.

00:06:02.689 --> 00:06:07.580
So let me get back
to my PowerPoint.

00:06:13.620 --> 00:06:20.380
All right, so here’s a summary of the
talk, is I’ll talk about building physics,

00:06:20.389 --> 00:06:24.940
but I’ll really talk about the
difference between high-frequency

00:06:24.940 --> 00:06:29.020
ground motion – and when I use
the word “high-frequency,” to me,

00:06:29.020 --> 00:06:33.930
that’s kind of – people talk
about peak acceleration,

00:06:33.930 --> 00:06:38.020
which is really mostly controlled
by high frequencies.

00:06:38.020 --> 00:06:43.479
And I’ll say that high-frequency
ground motions, because they saturate

00:06:43.480 --> 00:06:50.460
with magnitude, it turns out that
their statistics are quite standard.

00:06:51.220 --> 00:06:55.960
Basically, normal statistics –
the same kind of statistics you use

00:06:55.960 --> 00:07:02.030
in actuarial things like heart attacks
or murders or auto accidents.

00:07:02.030 --> 00:07:09.430
And then – so high-frequency statistics,
we can do some probabilistic seismic

00:07:09.430 --> 00:07:15.069
hazard analysis in a reliable way
because they seem to be normal.

00:07:15.069 --> 00:07:17.680
But when we talk about
low-frequency ground motions –

00:07:17.680 --> 00:07:21.770
things that are important
to tall buildings – I’ll argue that

00:07:21.770 --> 00:07:28.280
their statistics are not normal statistics.
They’re really power law statistics,

00:07:28.280 --> 00:07:31.520
sometimes called
Pareto distribution statistics.

00:07:31.530 --> 00:07:38.689
And they’re the statistics of epidemics,
wars, the stock market – and they’re

00:07:38.689 --> 00:07:46.500
the kind of statistics that drive people
who set rates completely crazy.

00:07:46.500 --> 00:07:51.120
Because they don’t actually have
things like means and variances.

00:07:51.120 --> 00:07:53.740
They’re very
hard to deal with.

00:07:53.750 --> 00:07:59.900
I’ll then argue, the low-frequency
problem and the high-frequency problem

00:07:59.900 --> 00:08:02.770
are very different kinds
of statistical problems.

00:08:02.770 --> 00:08:04.740
We shouldn’t
confuse them.

00:08:04.740 --> 00:08:10.749
And then I’ll say, in the – in the U.S.,
our current methodologies for

00:08:10.749 --> 00:08:17.350
doing designs based on probabilistic
analyses appear to me to be coming up

00:08:17.350 --> 00:08:22.439
with ground motions that are
systematically smaller than what

00:08:22.439 --> 00:08:28.440
I would choose as an Earth scientist.
So I think there’s a big problem there.

00:08:28.440 --> 00:08:34.300
So in this talk, I’ll concentrate
on near-source ground motion –

00:08:34.300 --> 00:08:39.130
motions close to the actual rupture.
And it’s not that the motions

00:08:39.130 --> 00:08:43.550
at a distance aren’t important.
It’s just it’s so much easier

00:08:43.550 --> 00:08:47.090
to talk about near-source
ground motion in terms of physics.

00:08:47.090 --> 00:08:49.620
And they probably are
the most important ones.

00:08:49.620 --> 00:08:57.070
And – well, here’s the really key issue.
In our cities, the modern tall buildings

00:08:57.070 --> 00:09:02.010
and base-isolated buildings – buildings
that are susceptible to long periods –

00:09:02.010 --> 00:09:06.580
have not really experienced yet
ground motions with displacements

00:09:06.580 --> 00:09:10.880
more than a meter.
And – but they will,

00:09:10.880 --> 00:09:15.110
but lots of these buildings are
being built by people who have

00:09:15.110 --> 00:09:18.010
never seen meter
displacements in a building.

00:09:18.010 --> 00:09:22.700
And they’re basically designing
for smaller motions.

00:09:25.780 --> 00:09:30.400
And then I’ll ask the question,
if we want to work about the statistics

00:09:30.400 --> 00:09:35.810
of long-period motions
in tall buildings, is it even feasible

00:09:35.810 --> 00:09:39.750
to do statistical
analysis on it?

00:09:39.750 --> 00:09:45.070
Sort of – the insurance company
and actuaries have given up doing

00:09:45.070 --> 00:09:50.220
that kind of analysis for epidemics.
You know, they tell you,

00:09:50.220 --> 00:09:53.720
you can buy life insurance,
but it doesn’t apply if it’s –

00:09:53.720 --> 00:09:57.980
if there’s some sort of pestilence
or a war [inaudible].

00:09:57.980 --> 00:10:03.510
And I’ll say, maybe we can do that
problem, but to be honest with you,

00:10:03.510 --> 00:10:06.090
I don’t know how to do it.
And if we’re going to make the

00:10:06.090 --> 00:10:11.370
claim to the public that
science from the USGS is

00:10:11.370 --> 00:10:15.560
these buildings are good,
we need to do some more work.

00:10:16.460 --> 00:10:20.820
And the real question is,
100 years from now, all the buildings

00:10:20.830 --> 00:10:24.370
we’re putting up now, will we
change the way we’re doing things?

00:10:24.370 --> 00:10:27.550
Or are we making
the right decisions now?

00:10:27.550 --> 00:10:30.930
And we’ll say, well, we went
through all these different earthquakes,

00:10:30.930 --> 00:10:34.480
and Tom Heaton
was just Chicken Little.

00:10:35.640 --> 00:10:40.760
I don’t know.
I’m pretty worried about where we are.

00:10:40.760 --> 00:10:45.680
So current building code –
up until recently, most of the

00:10:45.680 --> 00:10:51.460
building codes have been mainly done
by going out after every earthquake.

00:10:51.470 --> 00:10:55.930
And you look at things that didn’t
work with different kinds of buildings.

00:10:55.930 --> 00:11:01.200
And it’s mainly engineers saying, well,
you didn’t have enough columns,

00:11:01.200 --> 00:11:05.300
you didn’t have enough walls,
and saying, let’s modify the

00:11:05.310 --> 00:11:09.070
current codes so that we
do better in the next earthquake.

00:11:09.070 --> 00:11:15.120
And so most of the building codes
have been developed by looking at

00:11:15.120 --> 00:11:17.640
things that went wrong
in past earthquakes.

00:11:17.640 --> 00:11:23.640
And so if you had enough earthquakes
in a wide variety of buildings,

00:11:23.640 --> 00:11:28.840
sorry, you don’t need the USGS.
I mean, just use all the experience

00:11:28.840 --> 00:11:32.800
that’s out there and just
use the code. It’s good.

00:11:32.800 --> 00:11:37.680
You know, it’s – but the problem is,
of course, we don’t have that much

00:11:37.680 --> 00:11:40.420
experience, especially
for the tall buildings.

00:11:40.430 --> 00:11:46.140
So, in the last 20 years, there’s
been a strong push to go to

00:11:46.140 --> 00:11:53.180
performance-based engineering
and instead to say that, let’s build for

00:11:53.180 --> 00:12:00.440
earthquakes at some recurrence level –
build according to a statistical model.

00:12:00.440 --> 00:12:06.360
And the current practice is to build
so that the expected time

00:12:06.360 --> 00:12:12.400
between collapses of the buildings
are on the order of 2,500 years,

00:12:12.400 --> 00:12:16.160
except in San Francisco,
where it’s 1,500 years.

00:12:16.160 --> 00:12:19.740
Now, you might wonder why
San Francisco is so special,

00:12:19.740 --> 00:12:23.520
and I’ll get back to that.
But it’s not the IQ difference

00:12:23.520 --> 00:12:28.480
between L.A. and San Francisco
that controls this.

00:12:29.330 --> 00:12:36.220
So just kind of a simple building 101,
there’s sort of three end member

00:12:36.220 --> 00:12:41.120
kinds of buildings. On the left are
what we call shear wall buildings.

00:12:41.120 --> 00:12:45.340
And I think a lot of the buildings
here on this campus are shear walls.

00:12:45.340 --> 00:12:50.940
Mainly they’re concrete walls.
Sometimes they make them out of steel.

00:12:50.940 --> 00:12:54.020
Wooden houses
are mainly shear walls.

00:12:54.020 --> 00:12:57.540
And some openings
for doors and windows.

00:12:57.550 --> 00:13:00.340
And they resist the shearing
motion through these walls.

00:13:00.340 --> 00:13:04.200
And if you look from above,
the walls are tied to each other

00:13:04.200 --> 00:13:09.810
at the corners and through the floors.
So these walls are very weak

00:13:09.810 --> 00:13:13.240
out of plane, but in plane,
they’re extremely strong.

00:13:13.240 --> 00:13:16.400
And as long as you connect
the walls together, you’ve got

00:13:16.400 --> 00:13:20.610
a very strong structure.
And those kind of structures,

00:13:20.610 --> 00:13:25.950
they’re very strong, but they – that’s
good, but they tend to be very stiff.

00:13:25.950 --> 00:13:30.420
And I’ll get to this in a second, but it
turns out stiffness is not your friend.

00:13:30.420 --> 00:13:36.990
It’s bad. And wooden houses
are those kind of structures.

00:13:36.990 --> 00:13:41.440
My favorite is 7-Eleven stores.
Anybody by now chasing earthquakes

00:13:41.440 --> 00:13:44.630
around, you know, you go out –
we all go out and look afterwards.

00:13:44.630 --> 00:13:47.500
I’ve never seen a crack
in a 7-Eleven store.

00:13:47.500 --> 00:13:54.000
I mean, the contents are trashed,
but the structures are indestructible.

00:13:54.000 --> 00:13:59.280
And where I work, Caltech,
is built like 7-Eleven stores.

00:14:01.260 --> 00:14:05.680
The other way you can go is a frame
building – columns and beams.

00:14:05.680 --> 00:14:07.640
Looks like
Tinkertoys.

00:14:07.640 --> 00:14:12.600
It’s basically flexible, which actually
is good, believe it or not.

00:14:12.600 --> 00:14:17.000
But it’s weak,
which is bad.

00:14:17.000 --> 00:14:21.860
Weak – I mean, it doesn’t take
a lot of stress to defeat it.

00:14:21.870 --> 00:14:26.670
And it turns out all of our high-rise
buildings are in these flexible things.

00:14:26.670 --> 00:14:30.420
Braced frames are kind of an
in-between thing, and they’re very

00:14:30.420 --> 00:14:34.960
fascinating, but I don’t have time to talk
about them in this – in this presentation.

00:14:34.960 --> 00:14:38.580
So here’s an example of a –
of a shear wall building.

00:14:38.580 --> 00:14:42.400
And the Japanese are – you know,
they’re pretty conservative.

00:14:42.400 --> 00:14:47.200
A lot of bad stuff has happened in
Japan over the last 100 years or so.

00:14:47.200 --> 00:14:50.790
So they tend to build
very conservatively.

00:14:50.790 --> 00:14:55.290
And this were – these were
some concrete shear wall buildings

00:14:55.290 --> 00:14:59.540
that went through the
Niigata earthquake in 1964.

00:14:59.540 --> 00:15:04.800
And you might think that these buildings
were a failure because they fell over.

00:15:04.800 --> 00:15:07.470
But if you look carefully,
you’ll see that what happened

00:15:07.470 --> 00:15:12.790
was the foundation liquified,
and they capsized.

00:15:12.790 --> 00:15:14.790
So the buildings were fine.

00:15:14.790 --> 00:15:18.480
And people continued to live in
these buildings after they fell over.

00:15:18.480 --> 00:15:22.560
They just used the windows for
doors and doors for windows.

00:15:22.560 --> 00:15:26.350
And I’m told they jacked these
buildings back up and continued to

00:15:26.350 --> 00:15:29.360
use them later because
they were just so strong.

00:15:29.360 --> 00:15:33.160
So this is a shear wall
type of construction.

00:15:33.160 --> 00:15:38.250
This is the exact opposite. This is
a frame – steel frame construction.

00:15:38.250 --> 00:15:43.460
This is from Mexico City in the
1985 Michoacán earthquake.

00:15:43.460 --> 00:15:48.160
And these were three supposedly
identical steel towers.

00:15:48.170 --> 00:15:51.640
They were all 300 kilometers
away from the earthquake.

00:15:51.640 --> 00:15:55.590
And you’d think – you know,
they’re 100 meters from each other.

00:15:55.590 --> 00:16:00.670
100 meters and 300 kilometers –
they should, if they’re identical,

00:16:00.670 --> 00:16:04.170
do the same thing.
But this one – this tower,

00:16:04.170 --> 00:16:08.680
when the shaking was over,
there was no discernable damage.

00:16:08.680 --> 00:16:13.560
This central tower actually
had a permanent kink, or tilt.

00:16:13.560 --> 00:16:17.180
The roof was a meter
offset from where it started.

00:16:17.180 --> 00:16:24.920
And this tower, the roof was offset by
100 meters, I think – fell over sideways.

00:16:24.920 --> 00:16:30.870
And it just kind of shows how hard it is
to – if you had to make a model of this,

00:16:30.870 --> 00:16:35.240
they’re all supposedly identical –
same earthquakes, same place –

00:16:35.240 --> 00:16:37.980
very different outcomes.

00:16:37.980 --> 00:16:42.620
So let me talk for a second
about flexible versus strong.

00:16:42.620 --> 00:16:50.340
So stiff buildings tend to be almost rigid.
So you’re accelerating the entire

00:16:50.340 --> 00:16:56.410
mass together, and they tend to
have high stresses because of that –

00:16:56.410 --> 00:17:00.820
everything accelerates
as a rigid unit.

00:17:00.820 --> 00:17:05.439
And so if you make it stiff,
you better make it strong.

00:17:05.439 --> 00:17:09.760
Because you get high stresses in it.
But if you make the building strong,

00:17:09.760 --> 00:17:14.670
you add more walls or more bracing,
and that – what that does is,

00:17:14.670 --> 00:17:18.260
it increases the stresses in the building,
which means you better make it

00:17:18.260 --> 00:17:22.220
even stronger, which means you’re
going to increase the stiffness again.

00:17:22.220 --> 00:17:24.550
It’s kind of
a vicious circle.

00:17:24.550 --> 00:17:27.860
And so, if you go the route of
making a building strong,

00:17:27.860 --> 00:17:32.180
you better make it really strong
because it’ll end up being very stiff.

00:17:32.180 --> 00:17:36.090
You can go the exact opposite,
and for years, in California,

00:17:36.090 --> 00:17:40.650
people thought, well, we could
never really build a tall building to –

00:17:40.650 --> 00:17:43.800
strong enough to go
through big earthquakes.

00:17:43.800 --> 00:17:49.900
And so there were – come some height
limitations on the sizes of buildings.

00:17:49.900 --> 00:17:55.420
And it was only in the ’40s and ’50s when
people did full dynamic analysis

00:17:55.420 --> 00:17:59.800
that they realized, if they went the
route of making things more flexible,

00:17:59.800 --> 00:18:03.040
they’d end up with
much smaller stresses.

00:18:03.040 --> 00:18:06.140
But that’s good –
you have smaller stresses –

00:18:06.140 --> 00:18:10.080
but it ends up decreasing the
strength of the building,

00:18:10.080 --> 00:18:14.780
and now you’re back into another
vicious cycle in the other direction.

00:18:15.740 --> 00:18:20.070
Like I say, in California, all tall
buildings are designed to be flexible.

00:18:20.070 --> 00:18:23.050
Actually, if you go to Chile,
that’s not the case.

00:18:23.050 --> 00:18:26.980
Their tall buildings –
every wall in a Chilean building

00:18:26.980 --> 00:18:30.620
is a concrete shear wall building.
They were fine as long as they were

00:18:30.620 --> 00:18:35.880
going up to about 10 stories, and then
the architects goaded them into pushing

00:18:35.880 --> 00:18:41.340
them up to 50 or 60 stories, and
they’ve got some problems with that.

00:18:41.340 --> 00:18:47.180
Now, we think the buildings that we
have were built according to a code,

00:18:47.180 --> 00:18:51.910
but it’s important to recognize,
when you talk about the U.S. code,

00:18:51.910 --> 00:18:57.880
there are really two codes. There’s a
code for strength versus stiff buildings.

00:18:57.880 --> 00:19:02.270
And then there’s a separate code
that’s written in a different way

00:19:02.270 --> 00:19:05.540
for flexible buildings.
So when you decide to design

00:19:05.540 --> 00:19:08.780
a building, you kind of,
from the beginning, have to decide,

00:19:08.780 --> 00:19:12.920
am I building a stiff building,
or am I building a flexible building?

00:19:12.920 --> 00:19:14.320
They’re different things.

00:19:14.320 --> 00:19:21.380
So here’s an interesting example from
the Cucapah-El Mayor earthquake.

00:19:21.380 --> 00:19:26.100
So here’s the city of
Mexicali just south of our border.

00:19:26.100 --> 00:19:30.700
And the magnitude 7.2
Cucapah-El Mayor earthquake’s

00:19:30.700 --> 00:19:35.500
about 20 kilometers from Mexicali.
It was a pretty hazardous situation.

00:19:35.500 --> 00:19:43.100
And when I first got to Mexicali years
ago, I thought, oh, this city is doomed.

00:19:43.100 --> 00:19:48.620
You know, if Art Frankel was here,
he’d tell us they were doomed.

00:19:48.620 --> 00:19:53.750
So – and then this
earthquake happened, you know,

00:19:53.750 --> 00:19:56.850
on a Sunday in the middle of the day.
And I thought, oh, there’s going to be

00:19:56.850 --> 00:20:01.150
a lot of suffering in this –
from this thing because

00:20:01.150 --> 00:20:05.060
this is not a wealthy city.
A million people.

00:20:05.060 --> 00:20:10.160
Here’s the – here’s the populated area.
It’s pretty close to the earthquake area.

00:20:10.160 --> 00:20:14.780
There were a million people here –
two fatalities. Amazing.

00:20:14.780 --> 00:20:18.360
Incredible success
if you think about it.

00:20:18.360 --> 00:20:23.480
And here’s my interpretation of
what happened was that Cemex,

00:20:23.480 --> 00:20:28.660
the national cement company,
put out a flier to tell people

00:20:28.660 --> 00:20:32.660
how to build concrete block
construction – shear wall construction.

00:20:32.660 --> 00:20:35.740
One-story – most of these
are one-story houses,

00:20:35.740 --> 00:20:38.180
reinforced concrete
wall buildings.

00:20:38.180 --> 00:20:43.970
Basically, they were building
7-Eleven stores, and they did fine.

00:20:43.970 --> 00:20:49.930
So the architects don’t like it, but –
so here’s a – you might have seen this.

00:20:49.930 --> 00:20:55.880
It’s a cool video from Mexicali
during the Cucapah earthquake.

00:20:55.880 --> 00:20:59.390
This is a surveillance
camera on a pole.

00:20:59.390 --> 00:21:05.840
And first comes the high-frequency peak
accelerations from the P and S waves.

00:21:05.840 --> 00:21:09.040
And you can see it shaking
things around pretty quickly.

00:21:09.040 --> 00:21:14.090
And here’s a eight-story very stiff,
strong building in the back

00:21:14.090 --> 00:21:19.730
that’s not affected by it.
So this is – this would get you excited,

00:21:19.730 --> 00:21:22.610
but actually the
building did very well,

00:21:22.610 --> 00:21:27.750
and most buildings did very well, even
though this was pretty exciting motion.

00:21:27.750 --> 00:21:32.780
And now there’s some longer periods
that are coming in as surface waves

00:21:32.780 --> 00:21:37.400
from this earthquake.
Here comes the Love wave.

00:21:37.400 --> 00:21:44.140
And so that’s not dangerous to
those buildings, but if you had

00:21:44.150 --> 00:21:47.760
a flexible tall building there,
you’d be pretty excited now.

00:21:47.760 --> 00:21:51.190
Or, if you were in the pool, you might
be pretty excited. [laughter]

00:21:51.190 --> 00:21:53.800
This is why you need an
early warning system, right?

00:21:53.800 --> 00:22:00.440
So – and let’s go on.
So here’s the real point. [laughter]

00:22:00.440 --> 00:22:03.170
Well, architects
build these things.

00:22:03.170 --> 00:22:07.160
Obviously they must
not have seen that video.

00:22:07.160 --> 00:22:11.220
And by the way, the motions at the
top of a 10-story building are

00:22:11.220 --> 00:22:15.220
going to be 10 times the motions
that you saw at the ground,

00:22:15.220 --> 00:22:18.120
so the people – if you got that
kind of motion, and you were in

00:22:18.120 --> 00:22:24.280
that pool – it’s called an infinity pool –
you’d definitely go infinite there.

00:22:24.280 --> 00:22:27.540
You’d be approaching
the singularity, I think.

00:22:27.540 --> 00:22:32.140
And this is an interesting point,
and the – this is sort of the endpoint

00:22:32.150 --> 00:22:38.390
that I’ll get to is that the architects
are not necessarily thinking about what’s

00:22:38.390 --> 00:22:43.260
going on with the earthquake problem.
So here’s another example of that.

00:22:43.260 --> 00:22:48.130
If you’ve seen this, this is the
Rainier Tower in downtown Seattle.

00:22:48.130 --> 00:22:52.220
And if you can see at the base here,
the base dimensions are about

00:22:52.220 --> 00:22:56.080
half the dimensions of the
upper part of the building.

00:22:56.080 --> 00:22:59.800
And the structural engineer said,
I can do that.

00:22:59.809 --> 00:23:02.190
I mean, won all kinds
of awards for it.

00:23:02.190 --> 00:23:07.770
Of course, this was built in 1977,
before anyone had any idea that

00:23:07.770 --> 00:23:11.730
the Cascadia subduction zone
could have big earthquakes on it.

00:23:11.730 --> 00:23:16.300
It’s still there today, but I wouldn’t
work in that building. I just wouldn’t.

00:23:16.300 --> 00:23:19.160
So now let’s talk about
sort of the physics of

00:23:19.160 --> 00:23:21.520
how these flexible
buildings work.

00:23:21.520 --> 00:23:25.340
This is a frame
type of construction.

00:23:25.340 --> 00:23:28.040
There are columns that
carry the weight of the building.

00:23:28.040 --> 00:23:33.570
And the columns can be
made out of steel or concrete.

00:23:33.570 --> 00:23:35.230
In the code,
it doesn’t really matter.

00:23:35.230 --> 00:23:39.990
It just says they have to have a certain –
go through certain things.

00:23:39.990 --> 00:23:43.300
But people make steel
or concrete frames.

00:23:43.300 --> 00:23:48.350
Most of this talk will be about steel
because it’s much easier to simulate.

00:23:48.350 --> 00:23:51.520
Concrete’s pretty complicated.

00:23:52.060 --> 00:23:56.580
And the lateral stiffness in this system,
as you push it sideways, mainly comes

00:23:56.590 --> 00:24:02.230
from the flexure of the beams.
So you’ve got these beams connected

00:24:02.230 --> 00:24:08.010
to the columns, and if you had –
the connection here was a door hinge,

00:24:08.010 --> 00:24:10.250
then you’d get no
flexure of the beams.

00:24:10.250 --> 00:24:13.550
They’d just – and the
building would just flop over.

00:24:13.550 --> 00:24:17.710
And it would collapse when it got
what’s called a plastic hinge at the base.

00:24:17.710 --> 00:24:21.841
So a plastic hinge – think of
a coat hanger, and you bend it,

00:24:21.841 --> 00:24:26.840
and eventually it gets a kink in it.
That kink is what’s called a plastic hinge.

00:24:26.840 --> 00:24:31.050
And when you get a kink in – a plastic
hinge in your columns, that’s bad.

00:24:31.050 --> 00:24:35.140
Because the column – the main thing of
the column, it has to carry the weight

00:24:35.140 --> 00:24:39.630
of the building. So you’re not going
to change that weight. That’s gravity.

00:24:39.630 --> 00:24:46.980
And once the resisting force of the
column to that bending is less than

00:24:46.980 --> 00:24:49.840
the weight of the building’s
putting it on it,

00:24:49.840 --> 00:24:52.590
there’s nothing you can do
to keep it from coming down.

00:24:52.590 --> 00:24:56.540
And that’s the collapse
mechanism of this kind of building.

00:24:56.540 --> 00:25:02.690
So here’s sort of building physics 101.
Here’s a building that was in

00:25:02.690 --> 00:25:07.080
downtown Los Angeles for Northridge.
And here are records that were –

00:25:07.080 --> 00:25:11.150
oops – recorded at different
floors in the building.

00:25:11.150 --> 00:25:13.740
Here’s the ground motion.
See this pulse?

00:25:13.740 --> 00:25:16.700
You can see the pulse
is traveling up the building.

00:25:16.700 --> 00:25:20.400
And when it gets to the top of the
building, just like in seismology,

00:25:20.400 --> 00:25:24.000
the motions double at the
top because of the free surface.

00:25:24.000 --> 00:25:26.300
And this is a
horizontal motion.

00:25:26.300 --> 00:25:30.740
This is basically an
SH wave going up the building.

00:25:30.740 --> 00:25:35.260
And the wave speed at which it goes up
the building, you could measure it

00:25:35.270 --> 00:25:41.390
from these records, or you could get it
by taking the free period of the building

00:25:41.390 --> 00:25:46.080
and multiplying – taking the building
height – 4 times the building height,

00:25:46.080 --> 00:25:49.500
divide by the fundamental period,
would give you the wave speed

00:25:49.510 --> 00:25:51.690
of this wave
going up the building.

00:25:51.690 --> 00:25:54.460
So that wave speed’s
kind of important.

00:25:54.460 --> 00:26:00.190
High wave speeds mean it’s stiff.
Low wave speeds means it’s flexible.

00:26:00.190 --> 00:26:06.480
So the inter-story shearing –
the how much the roof of the –

00:26:06.480 --> 00:26:11.080
deflects versus the floor is –
as long as everything’s linear,

00:26:11.080 --> 00:26:17.309
it’s the horizontal particle velocity
of the floors divided by the wave speed.

00:26:17.309 --> 00:26:21.800
Which is – for these buildings, it’s on
the order of 100 meters per second.

00:26:21.800 --> 00:26:25.990
So that’s kind of the
wave speed you get for bay mud.

00:26:25.990 --> 00:26:30.150
So these buildings have
shear wave speeds like bay mud,

00:26:30.150 --> 00:26:34.910
but their densities are much less
because they’re mainly air.

00:26:34.910 --> 00:26:42.900
The inter-story shear stress – the forces
that go with that strain – end up being

00:26:42.900 --> 00:26:49.030
the particle velocity of the floor times
the wave speed times the density.

00:26:49.030 --> 00:26:51.700
So if you make it dense,
the stresses go up.

00:26:51.700 --> 00:26:55.610
If you make it stiff, that means
the shear wave velocity goes up.

00:26:55.610 --> 00:26:59.010
And then the stresses go up.
That’s why I’m saying – I said before,

00:26:59.010 --> 00:27:02.690
people didn’t want to make things
too stiff because the stresses go up.

00:27:02.690 --> 00:27:08.320
Now, the key things in these buildings
is that, when the inter-story shearing

00:27:08.320 --> 00:27:15.270
gets bigger than about 8 parts and
10 to the 3rd – a little less than 1% –

00:27:15.270 --> 00:27:20.280
the steel actually starts to yield.
So it starts to permanently bend.

00:27:22.400 --> 00:27:25.860
And so people have gone for
smaller wave speeds because they

00:27:25.860 --> 00:27:31.130
want them more flexible. It decreases
the stresses, but the strains go up.

00:27:31.130 --> 00:27:38.270
And, if you get big enough strains,
if the strain gets to about 6 parts

00:27:38.270 --> 00:27:41.470
and 100, you’re going to
lose the building just because of

00:27:41.470 --> 00:27:44.420
the gravitational load
of the – of the building.

00:27:44.420 --> 00:27:48.850
So here’s a picture of a frame –
steel frame that was being built.

00:27:48.850 --> 00:27:54.880
And, as I said, it’s important for the
columns and the beams to remain

00:27:54.880 --> 00:27:58.360
rigidly connected. If I put door
hinges here, it wouldn’t work.

00:27:58.360 --> 00:28:03.450
So the way these connections are done,
and have been done since the ’50s,

00:28:03.450 --> 00:28:08.980
is they weld the phalanges of the beams
to the phalanges of the columns to keep

00:28:08.980 --> 00:28:14.230
that connection rigid and perpendicular.
They used to do it – you’d see in the

00:28:14.230 --> 00:28:19.330
old movies, they’d throw around rivets
and bolts and things in the movies.

00:28:19.330 --> 00:28:24.870
But starting in the ’50s, they discovered
this was faster and less expensive.

00:28:24.870 --> 00:28:31.480
And they thought that this was fine
because they did testing of specimens,

00:28:31.480 --> 00:28:35.340
and they found that steel would
bend before any fractures

00:28:35.350 --> 00:28:39.690
would show up in the welding.
But they weren’t able to test full-scale

00:28:39.690 --> 00:28:43.190
connections because, if you go out
and you look at a real building,

00:28:43.190 --> 00:28:47.990
a piece of steel, it’s enormous.
And to actually bend that steel

00:28:47.990 --> 00:28:51.370
in a laboratory takes a
enormous machine to do it.

00:28:51.370 --> 00:28:55.280
And nobody wanted to spend
the money for a full-scale test.

00:28:55.280 --> 00:29:00.580
And what they wanted to see is,
in large enough loads, that the beams

00:29:00.580 --> 00:29:04.850
would get – this plastic
hinge in the beams.

00:29:04.850 --> 00:29:09.490
And you can see here, it’s got
some permanent bending in it.

00:29:09.490 --> 00:29:13.130
But they don’t want to see plastic
hinges in the columns because

00:29:13.130 --> 00:29:18.000
they have to carry the weight.
So in the Northridge earthquake,

00:29:18.000 --> 00:29:21.059
there were a number of buildings
that went through the earthquake.

00:29:21.059 --> 00:29:25.160
And people went out to inspect them
after the earthquake – steel buildings.

00:29:25.160 --> 00:29:28.570
And they were shocked to discover –
and I’ll show you what it looks like –

00:29:28.570 --> 00:29:30.580
a number of
failed connections.

00:29:30.580 --> 00:29:33.900
But nobody found
any bent steel anywhere.

00:29:33.900 --> 00:29:37.550
And the reason was because those
steel connections all fractured

00:29:37.550 --> 00:29:41.740
before they got large enough forces
to cause bending in the steel.

00:29:41.740 --> 00:29:47.720
So here’s a full-scale test that was set up
after the Northridge earthquake.

00:29:47.720 --> 00:29:52.200
And this was done down at
UC-San Diego. [video playing music]

00:29:52.200 --> 00:29:57.380
Chia-Ming Uang has a fantastic
facility down there to test

00:29:57.380 --> 00:30:02.860
various moment connections, so …
[music playing]

00:30:02.860 --> 00:30:06.940
So here’s the test. Here’s the
column welded at these connections.

00:30:06.940 --> 00:30:12.490
Here’s a big piston out on this end
pushing the beam up and down.

00:30:12.490 --> 00:30:15.610
And it’s meant to simulate
the kinds of motions

00:30:15.610 --> 00:30:17.530
that were in the
Northridge earthquake.

00:30:17.530 --> 00:30:20.090
So this was done right after
the Northridge earthquake.

00:30:20.090 --> 00:30:24.030
[music stops]
And here’s the piston right here.

00:30:24.030 --> 00:30:27.750
You can see it’s increasing
its stroke with time.

00:30:27.750 --> 00:30:33.200
Here’s the moment connection
here and here. [motor sounds]

00:30:33.200 --> 00:30:36.480
[crashing sounds]
And it fractured.

00:30:36.480 --> 00:30:39.360
That’s a weld fracture.

00:30:39.360 --> 00:30:42.240
People were shocked
to discover this.

00:30:43.000 --> 00:30:47.500
[motor sounds]

00:30:47.500 --> 00:30:53.720
So basically, any building built
prior to the Northridge earthquake

00:30:53.720 --> 00:30:57.320
has got these brittle weld connections
in them. [video stops]

00:30:57.320 --> 00:31:02.680
And so John Hall went out and
was very interested in this problem.

00:31:02.680 --> 00:31:08.559
And he designed, on a finite element
computer program, a 20-story –

00:31:08.559 --> 00:31:12.549
a number of 20-story buildings
built according to several codes.

00:31:12.549 --> 00:31:18.070
One, the Japanese code.
One to the U.S. code of 1994 –

00:31:18.070 --> 00:31:22.540
Unified Building Code.
The Japanese code was 1992.

00:31:22.540 --> 00:31:27.740
And these were meant to be
kind of typical of San Francisco.

00:31:27.740 --> 00:31:33.880
This is what they were using in Japan.
And he included in these designs

00:31:33.880 --> 00:31:37.820
either perfect connections that
wouldn’t fracture, or he put in

00:31:37.820 --> 00:31:43.130
connections that would fracture
typical to what they saw in those tests.

00:31:43.130 --> 00:31:47.890
And so if – here’s what’s called a
pushover analysis of those 20 –

00:31:47.890 --> 00:31:55.470
well, there’s actually also –
designed some six-story buildings.

00:31:55.470 --> 00:32:03.200
So there’s four 20-story buildings – four
curves – and four six-story buildings.

00:32:03.200 --> 00:32:07.500
The purple ones are
Japanese-code buildings.

00:32:07.500 --> 00:32:12.980
And blue and red
down here is U.S. code.

00:32:13.860 --> 00:32:20.640
Blue is – I’m sorry – blue is the U.S.
code with perfect connections here.

00:32:20.640 --> 00:32:24.740
And if it’s the solid line, then it’s –
what’s typically out there –

00:32:24.740 --> 00:32:27.200
a building with
brittle connections.

00:32:27.210 --> 00:32:31.790
And so what’s on the vertical axis
is a horizontal load given as

00:32:31.790 --> 00:32:34.350
a percentage of the
weight of the building.

00:32:34.350 --> 00:32:38.440
So it’s kind of like – when you do
these tests, you do them dynamically.

00:32:38.440 --> 00:32:43.340
And you take the base, and you start
accelerating it horizontally at an

00:32:43.340 --> 00:32:49.549
ever-increasing rate – a very slow rate –
that the acceleration is increasing.

00:32:49.549 --> 00:32:52.780
And then you plot the
lateral deflection of the roof.

00:32:52.780 --> 00:32:59.240
So you’re shearing it more and more.
And then you see what the roof is doing.

00:32:59.240 --> 00:33:02.340
In these regions, the buildings
are linearly elastic.

00:33:02.340 --> 00:33:04.799
It goes up,
comes back down, goes up.

00:33:04.799 --> 00:33:09.660
So the slope is the stiffness of
the building for linearly elastic.

00:33:09.660 --> 00:33:17.180
And then, if there’s no weld fracture,
eventually the steel starts to bend.

00:33:17.180 --> 00:33:21.170
And that’s this yielding.
And this is what people intended

00:33:21.170 --> 00:33:24.890
for these buildings to
look like in a pushover curve.

00:33:24.890 --> 00:33:29.480
And then you get out to a large enough
displacement, and eventually gravity

00:33:29.480 --> 00:33:33.270
takes over, and the buildings
just fall down anyway.

00:33:33.270 --> 00:33:38.429
So basically, you can’t take roof
displacements larger than about

00:33:38.429 --> 00:33:44.640
2-1/2 meters in these buildings, even
with good welds, or else they collapse.

00:33:44.640 --> 00:33:48.100
So this is U.S. code.
This is Japanese code.

00:33:48.100 --> 00:33:50.090
It’s higher strength.

00:33:50.090 --> 00:33:54.920
And the brittle welds really
decrease the strength of the building.

00:33:54.920 --> 00:33:57.800
Usually people say,
steel building, incredibly strong.

00:33:57.800 --> 00:34:02.340
But actually, the maximum
horizontal forces in here as a percentage

00:34:02.340 --> 00:34:07.240
of the weight is – for these brittle
welds, is only 6% of their weight.

00:34:07.240 --> 00:34:10.669
So if you put them on a 6-degree
slope, they’d just fall over.

00:34:10.669 --> 00:34:15.210
You couldn’t build the Tower
of Pisa this way. Wouldn’t work.

00:34:15.210 --> 00:34:22.629
So this particular finite element analysis
was specifically designed to go out

00:34:22.629 --> 00:34:29.369
to very large inelastic strains and
to measure collapse of buildings.

00:34:29.369 --> 00:34:38.669
And so here’s a picture that John put
together showing how a building would

00:34:38.669 --> 00:34:45.139
collapse and that. This is the building –
20 stories – before the earthquake.

00:34:45.139 --> 00:34:49.409
And in this simulation,
at 6 seconds, the base starts to

00:34:49.409 --> 00:34:54.470
move over to the left by
just a little less than 2 meters.

00:34:54.470 --> 00:34:56.950
And so it’s
moved to the left.

00:34:56.950 --> 00:35:00.479
The top has not caught up yet,
so it’s lagging behind.

00:35:00.479 --> 00:35:05.009
These little triangles are the fracturing
of welds in this simulation.

00:35:05.009 --> 00:35:10.640
And then, at 7 seconds, the base starts
to move back towards where it started

00:35:10.640 --> 00:35:16.600
from, but the top continues to the left.
So this is just a pulse over and back.

00:35:16.600 --> 00:35:22.359
And the base goes back to where it
started, and the top just keeps going.

00:35:22.359 --> 00:35:26.279
So this is called a side-sway
collapse mechanism.

00:35:26.279 --> 00:35:28.950
And it’s probably the primary
way things would collapse.

00:35:28.950 --> 00:35:32.420
It’s very similar to what we saw
in that Mexico City building.

00:35:32.420 --> 00:35:37.670
So here’s actually – the Japanese
have put a lot more resources into

00:35:37.670 --> 00:35:43.680
their engineering. And this is from
the E-defense shake table tests.

00:35:44.120 --> 00:35:46.080
I hope it comes up.

00:35:46.760 --> 00:35:48.680
All right.
Here we go.

00:35:50.500 --> 00:35:54.660
[video playing music]
So this is – need to drill.

00:35:59.000 --> 00:36:02.080
Skip ad.
Sorry.

00:36:02.300 --> 00:36:05.340
[humming sounds]
So here’s the E-defense shake table.

00:36:05.340 --> 00:36:07.200
They have this
enormous shake table.

00:36:07.200 --> 00:36:12.920
Here’s a 19-story
2/3 scale actual building.

00:36:12.920 --> 00:36:14.800
You can see
how flexible it is.

00:36:14.800 --> 00:36:19.500
It’s going through a Kobe
ground motion simulation here.

00:36:19.500 --> 00:36:22.640
And you can see how
flexible this thing is.

00:36:23.320 --> 00:36:26.420
[humming sounds]

00:36:26.420 --> 00:36:30.840
It’s got this big other
steel frame there because

00:36:30.840 --> 00:36:33.380
they want to catch the
building when it collapses.

00:36:33.380 --> 00:36:37.220
It would destroy the facility
if they actually let it collapse.

00:36:37.220 --> 00:36:41.340
So right there.
It’s just collapsed.

00:36:41.340 --> 00:36:47.080
And you can see that it’s gotten
this collapse in the bottom stories.

00:36:47.080 --> 00:36:51.500
Very similar to that picture
I just showed you from John Hall.

00:36:51.500 --> 00:36:55.500
So that’s the kind of testing you really
want to do to know what’s going on.

00:36:55.500 --> 00:36:59.000
But this is really
expensive stuff.

00:36:59.700 --> 00:37:04.460
[humming sounds]

00:37:05.360 --> 00:37:13.120
[Silence]

00:37:13.900 --> 00:37:19.600
[music playing] All right, so using that
20-story buildings, we ran through

00:37:19.609 --> 00:37:25.549
a number of simulations of what
might happen here in the peninsula

00:37:25.549 --> 00:37:28.760
if there was a repeat
of the 1906 earthquake.

00:37:28.760 --> 00:37:32.100
You might remember back around the
time of the 1906 earthquake [beeps],

00:37:32.109 --> 00:37:37.339
there were a number of special studies
about what 1906 might look like. [beeps]

00:37:37.339 --> 00:37:41.789
And Brad led an effort to make
simulations of the one-period

00:37:41.789 --> 00:37:44.660
ground motions in the
Bay Area [beeping and music]

00:37:44.660 --> 00:37:47.420
that would be compatible
with what we knew about …

00:37:47.420 --> 00:37:49.060
[speaking in background in foreign
language, musical sounds]

00:37:49.060 --> 00:37:50.380
What’s that?

00:37:50.380 --> 00:37:52.020
[musical sounds]

00:37:52.020 --> 00:37:53.700
Oh. YouTube is still going?

00:37:53.700 --> 00:37:55.360
[background voice in foreign language]

00:37:55.360 --> 00:37:56.360
Geez.

00:37:58.000 --> 00:38:00.380
Yeah, yeah, yeah.
Of course.

00:38:01.360 --> 00:38:05.220
[Silence]

00:38:05.920 --> 00:38:09.280
Let’s close it all.
All right.

00:38:10.200 --> 00:38:11.720
Now where am I?

00:38:15.180 --> 00:38:19.240
I guess I got really
out sync here.

00:38:20.020 --> 00:38:25.120
[Slience]

00:38:25.780 --> 00:38:34.500
So this is from a simulation of 1906 –
Brad’s ground motions.

00:38:34.500 --> 00:38:40.180
And those different buildings –
the 20-story buildings.

00:38:41.000 --> 00:38:45.860
And up on the top are
buildings with brittle welds.

00:38:45.860 --> 00:38:49.440
Actually, the Japanese had
brittle welds as well as the U.S.

00:38:50.160 --> 00:38:56.180
And on the bottom are perfect welds,
and these maps put those

00:38:56.190 --> 00:39:01.430
20-story buildings everywhere on the
finite element grid so that you could say,

00:39:01.430 --> 00:39:04.359
if there was a 20-story
building everywhere,

00:39:04.359 --> 00:39:07.319
which is what you’re
headed for right now,

00:39:07.320 --> 00:39:11.100
this is what it would look like
if you had another 1906 earthquake.

00:39:11.100 --> 00:39:14.720
And so brittle
welds on the top.

00:39:14.720 --> 00:39:19.460
Anything that is red
is basically collapsed.

00:39:19.460 --> 00:39:24.520
And anything that’s green is
basically damaged beyond repair.

00:39:24.529 --> 00:39:27.569
And in these simulations,
there are no bay muds.

00:39:27.569 --> 00:39:31.579
There’s no big problems with –
foundation problems,

00:39:31.579 --> 00:39:33.559
which would
make things worse.

00:39:33.559 --> 00:39:36.520
And the Japanese buildings
do better than the U.S. buildings,

00:39:36.520 --> 00:39:39.349
and especially
the brittle welds.

00:39:39.349 --> 00:39:43.099
It could be pretty ugly if
we had a repeat of one of those.

00:39:43.100 --> 00:39:46.760
We don’t really know, but if we’re
going to go out and tell people all these

00:39:46.760 --> 00:39:51.380
buildings are good for 2,500 years,
but this is our best understanding

00:39:51.380 --> 00:39:57.690
of what a 1906 would look like,
it’s saying there’s probably a problem.

00:39:57.690 --> 00:40:01.700
This epicenter was in the Golden Gate.
It’s even worse if you take the

00:40:01.700 --> 00:40:07.819
epicenter and move it up to
Bodega Bay or someplace like that.

00:40:07.819 --> 00:40:12.559
And then we did a number
of simulations based on what

00:40:12.559 --> 00:40:16.749
buildings would do based on
simulations that was done at SCEC.

00:40:16.749 --> 00:40:22.049
And here was a simulation that was done
by Rob Graves – another USGS person.

00:40:22.049 --> 00:40:26.710
And we did that same trick, and
anything in red in here is a collapse.

00:40:26.710 --> 00:40:33.599
This would be a magnitude 7.3 on the
Puente Hills thrust beneath Los Angeles.

00:40:33.599 --> 00:40:37.130
And we really – you don’t want to
be there if we get that earthquake.

00:40:37.130 --> 00:40:41.369
For those of us in southern California,
this is kind of our worst nightmare.

00:40:41.369 --> 00:40:49.340
And so we ran lots of these things.
Whoops. Where did I go? Sorry.

00:40:53.180 --> 00:40:54.700
Where’s my …

00:40:56.940 --> 00:40:59.600
All right.
I lost my cursor. There it is.

00:40:59.609 --> 00:41:05.309
All right, so we went through an
incredible suite of simulated

00:41:05.309 --> 00:41:08.900
ground motions –
64,000 of them.

00:41:08.900 --> 00:41:13.549
And for every one of those ground
motions, we decided what happened.

00:41:13.549 --> 00:41:20.140
And if the buildings didn’t have much
permanent bending, we made it gray.

00:41:20.140 --> 00:41:24.390
If they collapsed and ended up
on the ground, we made the point red.

00:41:24.390 --> 00:41:28.640
And if they were bent, when it was
done so much that you couldn’t repair

00:41:28.640 --> 00:41:34.390
the building, which is strains of
greater than about 1%, we turned it

00:41:34.390 --> 00:41:38.170
these black colors – not repairable.
And then we plotted all those

00:41:38.170 --> 00:41:42.360
simulations versus the peak
displacements of the record

00:41:42.360 --> 00:41:46.559
and the peak velocities of each record.
And I think you can see that all

00:41:46.559 --> 00:41:51.779
the collapses are mainly upper right.
And there are kind of some areas

00:41:51.779 --> 00:41:56.730
where – separating where things are
okay and where things are not okay.

00:41:56.730 --> 00:42:01.900
And basically, the bottom line is,
to get collapse, you need

00:42:01.900 --> 00:42:07.619
enough velocity in the ground
motion to make the steel yield.

00:42:07.619 --> 00:42:10.950
And then you need
enough displacement in the

00:42:10.950 --> 00:42:15.739
ground motion to get the roof
offset so much that it’s unstable.

00:42:15.739 --> 00:42:20.720
So you can look at these things
in terms of just what’s the velocity of

00:42:20.720 --> 00:42:23.670
the ground motion and the displacement.
And it gives you a pretty good

00:42:23.670 --> 00:42:28.239
idea of what this building
will take and won’t take.

00:42:28.239 --> 00:42:31.660
And so we ran those through all
these different kinds of buildings.

00:42:31.660 --> 00:42:35.930
Here the Japanese code
20-story building with perfect welds.

00:42:35.930 --> 00:42:42.979
The U.S. with perfect welds.
Again, versus velocity on the

00:42:42.979 --> 00:42:45.660
vertical axis and displacement
on the horizontal axis.

00:42:45.660 --> 00:42:47.530
And here are the
brittle-weld buildings.

00:42:47.530 --> 00:42:52.740
You can see the brittle-weld buildings
are very much in a class of their own.

00:42:52.740 --> 00:42:59.060
And so peak velocities for the
U.S. brittle-weld buildings

00:42:59.069 --> 00:43:01.749
greater than 60 centimeters
of displacement

00:43:01.749 --> 00:43:07.819
or 60 centimeters per second of velocity
are bad news for those buildings.

00:43:07.819 --> 00:43:12.479
And personally, I really like this
kind of plot because it makes it

00:43:12.479 --> 00:43:16.529
very clear what the buildings
actually can and can’t take.

00:43:16.529 --> 00:43:20.230
So right now, people are telling me,
oh, that building is built for

00:43:20.230 --> 00:43:24.579
2,500-year ground motion. I have
no idea what they’re talking about.

00:43:24.579 --> 00:43:28.430
I want to know what that
building is on a plot like this.

00:43:28.430 --> 00:43:30.940
Is it better than these?
Worse?

00:43:30.940 --> 00:43:34.530
This is what we really need
to know about these things.

00:43:34.530 --> 00:43:40.269
So Kenny Buyco, my latest student,
went through and looked at a bunch

00:43:40.269 --> 00:43:45.019
of buildings built with
different codes through the years.

00:43:45.019 --> 00:43:49.910
The ’94 building code.
The ’85 building code.

00:43:49.910 --> 00:43:51.769
And the 1973 building code.

00:43:51.769 --> 00:43:54.819
And he has some even
more modern ones in there.

00:43:54.819 --> 00:43:59.160
And here are the pushover curves
for those different building codes.

00:43:59.160 --> 00:44:01.299
Up to ’94,
they do brittle welds.

00:44:01.299 --> 00:44:04.250
They’re all controlled
by the brittle welds.

00:44:04.250 --> 00:44:08.020
But even if they had good welds,
the designs are pretty – they haven’t

00:44:08.020 --> 00:44:14.300
changed much over time from 1983 –
'73 all the way up to present.

00:44:14.309 --> 00:44:18.310
So people say, well, we’re working
like crazy to figure out what the

00:44:18.310 --> 00:44:21.349
ground motions are so we
can give them to the engineers.

00:44:21.349 --> 00:44:24.979
Well, they’re not going to change
anything. They don’t change anything.

00:44:24.979 --> 00:44:26.519
Because everything’s
been working.

00:44:26.519 --> 00:44:32.180
And a lot of this is – actually, they’ll
tell you, is controlled by wind drift.

00:44:32.180 --> 00:44:36.640
And through the years –
I remember when I was just young,

00:44:36.640 --> 00:44:39.029
people said, well,
wind completely controls.

00:44:39.029 --> 00:44:42.019
I don’t have to worry about
earthquakes for these buildings.

00:44:42.019 --> 00:44:44.569
And through the years, we’ve
gotten bigger and bigger records.

00:44:44.569 --> 00:44:46.680
And now they’re worried
about the earthquakes,

00:44:46.680 --> 00:44:50.329
but they say, well,
the wind saved us.

00:44:50.329 --> 00:44:53.109
So there’s a lot of
wind in this problem.

00:44:53.109 --> 00:44:59.269
So this is – this is a simulation of
a 60-story building built according

00:44:59.269 --> 00:45:06.450
to the ’94 code – a steel-frame building.
Running the Denali record through it –

00:45:06.450 --> 00:45:10.780
this is Abel Dizon, who was
one of John Hall’s students.

00:45:10.780 --> 00:45:15.740
And with brittle welds, that record
collapses a 60-story building.

00:45:15.740 --> 00:45:20.340
And even with good welds,
it’s quite a challenge for it.

00:45:21.900 --> 00:45:27.780
Right now, most of the discussion
about how to characterize ground motion

00:45:27.800 --> 00:45:32.120
is around 5% damped
response spectral acceleration.

00:45:32.120 --> 00:45:34.960
Now, most people in here are
not earthquake engineers.

00:45:34.960 --> 00:45:37.880
You probably don’t
think about response spectra.

00:45:37.880 --> 00:45:41.390
But response spectra, again,
is just a linear oscillator with

00:45:41.390 --> 00:45:46.690
different damping and period,
and the maximum response of it.

00:45:46.690 --> 00:45:51.339
And people typically – all of our
things are linear analyses, and we

00:45:51.339 --> 00:45:58.210
use 5% damping, which is
a long way from critically damped.

00:45:58.210 --> 00:46:03.569
And if you did a – kind of a cyclic
picture of force of your oscillator

00:46:03.569 --> 00:46:08.719
versus the displacement of your
oscillator, it would go in an ellipse.

00:46:08.719 --> 00:46:12.140
And the area in the
ellipse would be controlled

00:46:12.140 --> 00:46:16.320
by the damping in your –
in your oscillator.

00:46:16.320 --> 00:46:19.320
And right now,
we’re using 5% damping,

00:46:19.320 --> 00:46:22.789
which is kind of this green area.
But when you do the full finite element

00:46:22.789 --> 00:46:28.020
analysis, you discover that the finite
element analysis, it comes up linear,

00:46:28.020 --> 00:46:31.719
yields along here, and then it
comes back and yields along here.

00:46:31.720 --> 00:46:35.360
So it has this huge
hysteretic area.

00:46:35.360 --> 00:46:38.520
And all the area in
there is inelastic work.

00:46:38.520 --> 00:46:43.220
And if you were to try to simulate
that with viscous damping,

00:46:43.220 --> 00:46:45.630
you’d need much
bigger viscous damping.

00:46:45.630 --> 00:46:49.760
And furthermore, if you look at it,
the stiffness that controls the

00:46:49.760 --> 00:46:53.839
period of the building in
a linear system is just this slope.

00:46:53.839 --> 00:46:58.619
But if it’s really out in this
yielding range, the stiffness is

00:46:58.620 --> 00:47:04.260
what’s called the tangent stiffness.
This line across here – the slope on that.

00:47:04.260 --> 00:47:09.380
And it turns out that the
bigger the plastic yielding,

00:47:09.390 --> 00:47:12.799
the longer the
period of the structure.

00:47:12.799 --> 00:47:19.979
So instead of using this green curve, we
ought to be using – we’re going to use

00:47:19.979 --> 00:47:25.880
response spectra. We ought to use much
bigger damping and longer periods.

00:47:25.880 --> 00:47:28.160
So here’s a very interesting
plot that was done

00:47:28.160 --> 00:47:31.289
by Shiyan Song,
one of my students.

00:47:31.289 --> 00:47:35.759
We took the – this is actually a
nine-story building – steel building.

00:47:35.759 --> 00:47:39.319
And we took a
ground acceleration – this one –

00:47:39.319 --> 00:47:44.890
and we started increasing the size
of the acceleration, just scaling it up.

00:47:44.890 --> 00:47:47.589
And just before it collapsed,
this is what it looked like

00:47:47.589 --> 00:47:52.789
in terms of acceleration
of the ground – 68% g.

00:47:52.789 --> 00:47:57.720
And if you take this acceleration,
and you ran it through a response

00:47:57.720 --> 00:48:01.999
spectral acceleration of 5% damping,
and you got the acceleration out of that,

00:48:02.000 --> 00:48:06.140
you’d get 91% g
just when it was collapsing.

00:48:06.960 --> 00:48:09.820
Where am I –
here, 91% g.

00:48:09.820 --> 00:48:13.080
And then, if you took the full
finite element analysis and

00:48:13.089 --> 00:48:19.240
kept everything linear in it, and you ran
through the – at the base of the building,

00:48:19.240 --> 00:48:22.349
the shear at the base of the
building would be this green curve.

00:48:22.349 --> 00:48:27.670
And that would be 65% of the
weight of the building – this one.

00:48:27.670 --> 00:48:30.180
And then, if you actually
go to the yielding building

00:48:30.180 --> 00:48:36.339
that collapsed,
it’s this red curve here.

00:48:36.339 --> 00:48:42.440
And that’s – that turns out to be 23%
of the weight of the building.

00:48:42.440 --> 00:48:45.620
Because it’s yielding.
It can’t exceed that stress.

00:48:45.620 --> 00:48:51.079
That turns out to be the building yield
stress. So this is what really matters.

00:48:51.079 --> 00:48:54.319
This is what collapses the building –
this red curve at the bottom.

00:48:54.319 --> 00:48:58.339
And currently, we’re using
this orange curve up here.

00:48:58.339 --> 00:49:02.380
But if you took and you increased the
damping on the – on the response

00:49:02.380 --> 00:49:08.400
spectrum up to 70% damping, and you
took the period out to be longer period,

00:49:08.400 --> 00:49:14.140
you’ll get this purple curve,
which looks a lot like the red curve.

00:49:14.380 --> 00:49:22.340
And so it turns out that, if you did
response spectra at 70% damping,

00:49:22.340 --> 00:49:27.160
and you increased the period
by about a factor of 1.5,

00:49:27.170 --> 00:49:31.579
you get very close to what controls
the strength of the building.

00:49:31.579 --> 00:49:35.460
So – let’s skip that one.

00:49:35.460 --> 00:49:39.869
So – this is too much stuff,
but here, this is good.

00:49:39.869 --> 00:49:44.460
So we took 50 different records,
and we did this same trick with them.

00:49:44.460 --> 00:49:49.960
And here’s the accelerations that were
necessary to collapse the building.

00:49:49.960 --> 00:49:55.720
After you filter those records,
make them 70% damped records,

00:49:55.720 --> 00:49:57.940
you end up with
these blue plots.

00:49:57.950 --> 00:50:01.589
And the acceleration at
that filtered record is

00:50:01.589 --> 00:50:04.700
very close to the pushover
yield strength of the building.

00:50:04.700 --> 00:50:10.329
And there’s a lot less dispersion
in the curve if you do that.

00:50:10.329 --> 00:50:17.140
So here’s the dispersion using this
filtered acceleration up on the top.

00:50:17.140 --> 00:50:21.519
And then the next-best
one is just peak velocity.

00:50:21.520 --> 00:50:26.480
And then, below that is what we’re
currently using – spectral acceleration.

00:50:26.480 --> 00:50:30.320
It’s got a lot more dispersion,
but it’s not a very good measure.

00:50:30.320 --> 00:50:34.080
And peak acceleration
is useless.

00:50:34.089 --> 00:50:39.089
Peak displacement is not very good
unless you connect it with peak velocity.

00:50:39.089 --> 00:50:44.959
So if you took all those different
records, and you used the

00:50:44.959 --> 00:50:48.319
low damping – that’s these red records –
if you take the median,

00:50:48.319 --> 00:50:51.769
you get this red curve.
And then people take that curve,

00:50:51.769 --> 00:50:55.009
and they decrease it by something
they call an R factor,

00:50:55.009 --> 00:50:59.410
which is another one-hour talk,
and come to what’s called the

00:50:59.410 --> 00:51:05.170
design spectrum at the –
the design spectrum is this blue thing.

00:51:05.170 --> 00:51:09.920
But if we use 70% damping,
we get rid of all those peaks,

00:51:09.920 --> 00:51:12.680
and it’d be a much more
meaningful measure.

00:51:12.680 --> 00:51:15.660
It’s a much broader-band
measure of the ground motion.

00:51:15.660 --> 00:51:19.219
And so right now, when we go to
the earthquake engineers, and we say,

00:51:19.219 --> 00:51:23.940
we ran all these forward simulations of
big earthquakes, they come back to us,

00:51:23.940 --> 00:51:26.660
and they say,
I can’t use that.

00:51:26.660 --> 00:51:29.229
You don’t have enough
high frequencies in there.

00:51:29.229 --> 00:51:32.840
And you’ve got to verify that you’re
getting the right high frequencies.

00:51:32.840 --> 00:51:37.360
And I say, it’s just
all a red herring.

00:51:37.360 --> 00:51:44.020
Because if you look back at
that plot, what really matters is …

00:51:47.849 --> 00:51:51.360
… is this filtered low frequency.
And what’s that doing is,

00:51:51.369 --> 00:51:56.310
it’s keeping track of the total momentum
of the above part of the building.

00:51:56.310 --> 00:51:58.890
And that’s what’s shearing
the building off at the bottom.

00:51:58.890 --> 00:52:04.049
So the first thing you do, if somebody
gives you a broadband ground motion,

00:52:04.049 --> 00:52:07.779
filter out the high frequency so you
can figure out what’s really going on.

00:52:07.779 --> 00:52:12.440
But now the engineering community is
saying, I can’t use it because you haven’t

00:52:12.440 --> 00:52:16.690
put in the right high frequencies, which
we’re just going to remove anyway.

00:52:16.690 --> 00:52:22.180
So that’s kind of a big
misunderstanding there.

00:52:22.880 --> 00:52:24.960
Oh, my god.
Let’s see.

00:52:24.960 --> 00:52:31.959
So now, just a few comments
about how ground motions change

00:52:31.959 --> 00:52:33.729
in different-sized
earthquakes.

00:52:33.729 --> 00:52:36.800
So, again,
PGA is high frequencies.

00:52:36.800 --> 00:52:40.940
And I’m going to use PGD –
peak ground displacement –

00:52:40.950 --> 00:52:44.369
as a characteristic of low-frequency
earthquakes – or, low-frequency

00:52:44.369 --> 00:52:49.140
ground motion and what makes the
building unstable in the first place.

00:52:49.140 --> 00:52:53.869
And PGV is important and
somewhere in between those two.

00:52:53.869 --> 00:52:59.339
So these are ground motion
prediction equations that were

00:52:59.339 --> 00:53:05.219
designed to work for early warning.
And a lot of emphasis was using

00:53:05.219 --> 00:53:10.079
a very wide band of magnitudes,
both the ground motions in the

00:53:10.080 --> 00:53:15.500
southern California network and
all the PEER NGA motions as well.

00:53:15.500 --> 00:53:19.839
And this plot is peak acceleration.

00:53:19.839 --> 00:53:25.650
This upper-right is peak velocity,
and the lower-right is peak displacement

00:53:25.650 --> 00:53:31.670
as a function of magnitude
for different distance ranges.

00:53:31.670 --> 00:53:35.100
So one is zero kilometers –
very near-source.

00:53:35.100 --> 00:53:39.880
One is 30 kilometers – fairly close.
And one’s distance.

00:53:39.890 --> 00:53:46.119
And so at distant place, the acceleration
continues to go up with magnitude

00:53:46.119 --> 00:53:51.009
all the way up into the 7s. It doesn’t
really saturate at large distances.

00:53:51.009 --> 00:53:54.459
But when you get in close,
it completely saturates.

00:53:54.460 --> 00:53:58.560
And it saturates pretty fast –
about magnitude 4-1/2.

00:53:58.560 --> 00:54:03.480
And you don’t get any bigger
accelerations if you’re in really close.

00:54:03.480 --> 00:54:08.299
Displacements – the long periods, there’s
a kink here, but they never saturate.

00:54:08.300 --> 00:54:11.320
And that makes sense.

00:54:11.320 --> 00:54:13.500
That’s why we call
them big earthquakes.

00:54:13.519 --> 00:54:16.930
Now, this is kind of my
favorite plot of all times.

00:54:16.930 --> 00:54:26.599
It’s just a plot – take all the records in
the NGA database and plot them all –

00:54:26.599 --> 00:54:31.999
for each record, the peak acceleration of
the record versus the peak displacement.

00:54:31.999 --> 00:54:34.950
So kind of low-frequency
and high-frequency.

00:54:34.950 --> 00:54:40.109
And you can see there’s a lot of scatter
between, if you get a big displacement,

00:54:40.109 --> 00:54:42.859
does it mean big acceleration.
But generally, there’s a

00:54:42.859 --> 00:54:47.630
correlation in here until you
get to the less than 10 kilometers.

00:54:47.630 --> 00:54:52.560
The near-source ground motions –
those are these heavy circles – there’s –

00:54:52.560 --> 00:54:56.400
you’ll see there’s no correlation
between displacement and

00:54:56.400 --> 00:55:02.190
acceleration once you’re in close
for the magnitudes bigger than 6.

00:55:02.190 --> 00:55:04.930
These are all magnitudes
bigger than 6.

00:55:04.930 --> 00:55:08.199
And if you plot just the high
frequencies for those near-source

00:55:08.199 --> 00:55:13.960
ground accelerations, this is just
a histogram of number of records as a

00:55:13.960 --> 00:55:18.390
function of the log of the amplitude.
You get a plot that looks like this.

00:55:18.390 --> 00:55:22.720
Actually, this gray curve was a plot
that was made using data prior to

00:55:22.720 --> 00:55:32.170
the Chi-Chi earthquake. And it had
a median of 400 and – about 46% g.

00:55:32.170 --> 00:55:35.579
And then 48% g.
And then, if you added in the

00:55:35.579 --> 00:55:41.289
Chi-Chi records, then you get this new
distribution – lots of close-in records.

00:55:41.289 --> 00:55:46.049
The distribution didn’t change at all.
So this is a normal distribution.

00:55:46.049 --> 00:55:49.849
And the mean is known.
The standard deviation didn’t go.

00:55:49.849 --> 00:55:53.309
And that’s why I say we know
how to do high frequencies in close.

00:55:53.309 --> 00:55:57.140
I mean, this is a
standard normal distribution.

00:55:57.500 --> 00:56:02.299
So, by the way, if you take those
high frequencies, and you plot

00:56:02.299 --> 00:56:07.350
peak acceleration as a – in near-source
as a function of stress drop of the

00:56:07.350 --> 00:56:11.760
earthquake where stress drop is
measured as the change in stress

00:56:11.760 --> 00:56:14.210
from before the earthquake
to after the earthquake,

00:56:14.210 --> 00:56:18.319
there’s no correlation between
peak acceleration and stress drop.

00:56:18.319 --> 00:56:24.420
But in the – in the current NGA
relationships, it’s assumed there is.

00:56:24.420 --> 00:56:28.180
But I think we’re kind of
going in the wrong direction there.

00:56:28.180 --> 00:56:33.680
So these short-period ground motions,
they’re described by Gaussian statistics,

00:56:33.680 --> 00:56:37.130
and we know how to do that stuff.
I don’t want to say any more about it.

00:56:37.130 --> 00:56:40.270
Here’s the same thing
for the low frequencies.

00:56:40.270 --> 00:56:44.920
So now here’s the number
versus the log of the amplitude.

00:56:44.920 --> 00:56:47.820
And it doesn’t look like
a normal distribution.

00:56:47.820 --> 00:56:50.000
I don’t know what
this distribution is.

00:56:50.019 --> 00:56:53.880
But the gray one was made
with data prior to Chi-Chi.

00:56:53.880 --> 00:56:58.279
And then when you add the Chi-Chi
records in here, you get this black one.

00:56:58.280 --> 00:57:02.940
And the distribution changed.
Its median changed.

00:57:02.940 --> 00:57:07.200
And its variance changes.
Everything changes because

00:57:07.209 --> 00:57:10.650
there were lots of big
long-period motions in Chi-Chi.

00:57:10.650 --> 00:57:15.170
And if you put in the simulation
that Brad did for a 1906,

00:57:15.170 --> 00:57:18.569
and you put where we think we
have all the strong motion records,

00:57:18.569 --> 00:57:22.019
this is – the distribution
might look something like

00:57:22.019 --> 00:57:26.479
this dotted line if we
had a 1906 earthquake.

00:57:26.479 --> 00:57:29.940
And our understanding of what
long periods would look like would

00:57:29.940 --> 00:57:36.529
be pretty different if that happened.
Again, the medians change from

00:57:36.529 --> 00:57:41.890
prior to the – to this Chi-Chi
earthquake, it was 18 centimeters.

00:57:41.890 --> 00:57:45.799
After you put in Chi-Chi, it was 30.
And if you put in Brad’s, it was 58.

00:57:45.799 --> 00:57:50.729
So when – for the structural engineer,
he thinks he knows what long periods

00:57:50.729 --> 00:57:54.210
look like in an earthquake,
but he’s not seen 1906 yet.

00:57:54.210 --> 00:57:57.410
And it’s going to
change things for him.

00:57:57.410 --> 00:58:01.119
And the question is, here’s –
can we predict what this distribution

00:58:01.120 --> 00:58:03.989
is going to look like
100 years from now?

00:58:04.760 --> 00:58:07.619
I’m not so sure we can.
And I don’t have time

00:58:07.619 --> 00:58:10.849
to go through it, but I think
that’s a pretty hard thing to do.

00:58:10.849 --> 00:58:14.880
But, by the way, if you want to use
stress drop, it’s really important for

00:58:14.880 --> 00:58:20.029
the long periods because the stress drop
is related to the slip on the fault.

00:58:20.029 --> 00:58:24.820
And the slip is what controls
the near-source displacements.

00:58:25.840 --> 00:58:28.720
And these things are a –
that’s a power law, by the way.

00:58:28.720 --> 00:58:31.140
It’s what we call
a Pareto distribution.

00:58:31.140 --> 00:58:36.080
And, again, in actuarial statistics,
it’s like saying, how many people

00:58:36.080 --> 00:58:38.420
are going to die
in a war next year?

00:58:39.060 --> 00:58:42.520
Well, [chuckles] probably we
have a better chance this year than

00:58:42.520 --> 00:58:47.300
most years, but we still don’t know.
I mean, we don’t have any idea.

00:58:47.300 --> 00:58:51.089
A pandemic – I mean, bird flu.
Come on.

00:58:51.089 --> 00:58:54.599
Stock market. I like that.
The stock guys always used to tell me,

00:58:54.599 --> 00:58:58.309
well, your retirement’s going
to be this amount in 20 years.

00:58:58.309 --> 00:59:02.080
And then along came 2008.

00:59:04.880 --> 00:59:12.680
So here’s a really interesting example.
This was the PEER tall-building project,

00:59:12.680 --> 00:59:17.170
which was to do performance-based
design for a 60-story building

00:59:17.170 --> 00:59:21.049
in downtown Los Angeles.
And actually, they – even though

00:59:21.049 --> 00:59:27.940
Caltech is a member of the PEER
consortium, they never told us

00:59:27.940 --> 00:59:33.079
they were running this project,
and they never asked us to be in it.

00:59:33.079 --> 00:59:35.140
And I guess you could
understand why that would be

00:59:35.140 --> 00:59:37.880
from the things
I’ve been saying.

00:59:37.880 --> 00:59:42.540
And it took me a heck of a long time to
get to these results I’m showing here.

00:59:42.540 --> 00:59:49.709
I had to – I had to pitch a fit saying, you
know, how come we can’t see the results

00:59:49.709 --> 00:59:55.839
of an academic study? So 40-story
buildings, 5-1/2-to-6 second period.

00:59:55.839 --> 01:00:01.579
And they said, working with engineering
consultants and experts at SCEC –

01:00:01.579 --> 01:00:07.079
or USGS, really – they did
ground motion simulations for 25 years

01:00:07.079 --> 01:00:15.709
out to 5,000-year repeat periods.
And it’s – these motions are

01:00:15.709 --> 01:00:19.900
similar to what are used here
for Rincon Hill in San Francisco.

01:00:19.900 --> 01:00:27.650
And here are the – a bunch of records
with their response spectra that have

01:00:27.650 --> 01:00:32.369
response spectra scaled so they’re
similar to what comes out of a national

01:00:32.369 --> 01:00:38.950
probabilistic seismic hazard analysis map.
So it’s out at this 5-1/2-second period.

01:00:38.950 --> 01:00:42.480
And then they modified the
records so that they’d be compatible

01:00:42.480 --> 01:00:47.479
with that number. And they
used a dozen different records.

01:00:47.479 --> 01:00:52.239
Here’s the velocities
for those dozen different records,

01:00:52.239 --> 01:00:53.730
all plotted on
top of each other.

01:00:53.730 --> 01:00:56.339
They’re big records.
They’re up near a meter per second.

01:00:56.339 --> 01:00:58.430
They’re quite
challenging for buildings.

01:00:58.430 --> 01:01:03.380
But here’s the one that’s really important
is that maximum displacement, they say,

01:01:03.380 --> 01:01:06.530
for downtown Los Angeles –
we know in 2,500 years,

01:01:06.530 --> 01:01:12.539
it’s not going to be greater than a meter.
And I really question that.

01:01:12.539 --> 01:01:17.499
I mean, do we really mean to,
as Earth scientists, tell the world

01:01:17.499 --> 01:01:21.239
we know we’re not going to have
greater than a meter displacement

01:01:21.240 --> 01:01:24.320
in downtown Los Angeles?
I don’t think so.

01:01:24.320 --> 01:01:27.299
I mean, in that same
2,500-year period of time,

01:01:27.299 --> 01:01:30.990
the L.A. Basin is going to
shorten by 30 meters.

01:01:30.990 --> 01:01:37.270
And they say, but you’ll never see a
ground motion bigger than a meter.

01:01:37.270 --> 01:01:40.900
So here’s an example of a
big motion from a flat

01:01:40.900 --> 01:01:43.940
thrust fault under a city.
This was Kathmandu.

01:01:43.940 --> 01:01:48.299
This was really the record that
kind of got this whole thing

01:01:48.299 --> 01:01:51.799
started in the first place.
Lucy Jones kind of told Tom Fuller,

01:01:51.799 --> 01:01:54.009
well, you ought to see
this record from Kathmandu.

01:01:54.009 --> 01:01:58.049
I mean, it’s this record – I think
some people have probably seen it.

01:01:58.049 --> 01:02:03.520
It’s got – doesn’t have much
acceleration. It’s just 16% g.

01:02:03.520 --> 01:02:06.000
But it’s very long-period
acceleration in here.

01:02:06.000 --> 01:02:09.380
When I said the acceleration
characterizes the short period,

01:02:09.380 --> 01:02:13.239
not for this record.
And the maximum displacements

01:02:13.239 --> 01:02:16.930
in here are about
a meter and a half.

01:02:16.930 --> 01:02:22.599
Peak velocity is up near
a meter and a half per second.

01:02:22.599 --> 01:02:24.349
And when you put them
on a response spectra,

01:02:24.349 --> 01:02:27.519
they’re much bigger than the
design spectrum out at the

01:02:27.519 --> 01:02:32.959
size of 5-1/2 seconds, which is in the
sweet spot for these tall buildings.

01:02:32.959 --> 01:02:36.589
And if you’ve – people who
have seen this, really cool.

01:02:36.589 --> 01:02:38.789
If you don’t believe
that that happens,

01:02:38.789 --> 01:02:44.579
this is a surveillance cam in
Kathmandu during the earthquake.

01:02:44.579 --> 01:02:48.430
And so people are just – this is before,
and the P wave is about to hit.

01:02:48.430 --> 01:02:52.750
And people come out into
the square here in a second.

01:02:52.750 --> 01:02:55.740
So everybody’s feeling
the P wave, running outside.

01:02:55.740 --> 01:02:58.660
And here comes the S wave.

01:03:02.440 --> 01:03:04.780
Long periods.

01:03:04.780 --> 01:03:08.940
So fortunately, Kathmandu
didn’t have any tall buildings.

01:03:08.940 --> 01:03:14.920
Well, so if you really want to figure out
what’s going to happen in these long

01:03:14.920 --> 01:03:18.980
periods, currently, we do everything
through the earthquake magnitude.

01:03:18.980 --> 01:03:22.320
And we say we’re going to
run magnitudes up to 8 or 8.2.

01:03:22.329 --> 01:03:26.199
And we take
statistics on those 8 or 8.2.

01:03:26.199 --> 01:03:28.400
But if you really want to know
what’s going on, you really

01:03:28.400 --> 01:03:33.500
need to know what’s the slip in the
nearby sections of the faults.

01:03:35.060 --> 01:03:39.959
And so it turns out, if you double –
you took the 1906 earthquake and

01:03:39.959 --> 01:03:44.309
kept everything the same, but you
doubled all the displacements in 1906,

01:03:44.309 --> 01:03:50.829
it would jump from a 7.8 to
an 8 in terms of moment magnitude.

01:03:50.829 --> 01:03:54.880
So if you did that, you’d double
all the long-period motions.

01:03:54.880 --> 01:03:59.069
But if you looked on the GMPEs
for long-period motions,

01:03:59.069 --> 01:04:02.640
it would say that the ground
motion is going from a 7.8 to an 8,

01:04:02.640 --> 01:04:05.449
in close, would
only increase by 20%.

01:04:05.449 --> 01:04:09.039
So what’s the big deal?
I mean, to the earthquake engineering

01:04:09.039 --> 01:04:13.549
community, they use the word “eight”
all the time because it doesn’t

01:04:13.549 --> 01:04:16.099
make that much difference to them.
I mean, for an Earth scientist

01:04:16.100 --> 01:04:21.740
to say “eight,” what we mean
is we’ve gone up to 15-meter slips.

01:04:21.740 --> 01:04:24.080
And to us,
that’s an enormous number.

01:04:24.080 --> 01:04:26.800
There’s a big
misunderstanding about that.

01:04:27.340 --> 01:04:30.080
Well, let’s forget that.

01:04:30.089 --> 01:04:35.059
So right now, we’re in this thing where
we claim we know how to design

01:04:35.059 --> 01:04:38.759
for these future earthquakes,
sort of designing for the known.

01:04:38.759 --> 01:04:41.719
Some architect wants to
win an award, so he comes up with

01:04:41.719 --> 01:04:46.039
some design that
looks like it’s implausible.

01:04:46.040 --> 01:04:49.220
And if somebody can build it,
he’ll get an award for it.

01:04:49.220 --> 01:04:53.640
And so he chooses the
geometry of the design.

01:04:53.640 --> 01:04:59.490
And then the structural engineering
and geotechnical engineering community

01:04:59.490 --> 01:05:04.739
goes out and figures out the forces
that would be required for that design.

01:05:04.739 --> 01:05:08.910
And then they figure out the size of –
all the limits on the elements.

01:05:08.910 --> 01:05:11.839
And that’s kind of
performance-based design.

01:05:11.840 --> 01:05:16.100
And then, when they’re done, they say,
it’s built for the 2,500-year earthquake.

01:05:16.100 --> 01:05:20.670
And, you know, to me, as a scientist,
it’s important to acknowledge the

01:05:20.670 --> 01:05:25.819
things we don’t know. And I don’t
think I know this problem very well.

01:05:25.819 --> 01:05:30.309
So there’s – it’s still possible,
in the engineering world,

01:05:30.309 --> 01:05:33.229
to design for things
you don’t know very well.

01:05:33.229 --> 01:05:36.660
And if you don’t know things very well,
you can do the old-fashioned thing.

01:05:36.660 --> 01:05:40.360
You say, well, I got to have a building.
I’m not going to live in the street.

01:05:40.360 --> 01:05:44.500
I got to have a building,
and what do the buildings have to do?

01:05:44.500 --> 01:05:47.140
And I could – I could
use different architectures.

01:05:47.140 --> 01:05:50.140
I could use a 7-Eleven
store if I wanted to.

01:05:50.140 --> 01:05:52.920
Probably not.
It wouldn’t win any awards.

01:05:52.920 --> 01:05:57.099
And then, look at the different designs –
what they cost and how well they work.

01:05:57.099 --> 01:06:00.239
And then go back and now
ask the Earth scientist and say,

01:06:00.239 --> 01:06:04.719
well, I chose this design. I think it
would collapse for this ground motion.

01:06:04.719 --> 01:06:08.859
What do you think? Does that look
like a plausible ground motion?

01:06:08.859 --> 01:06:13.930
But nobody does that.
And, in fact, when I try to find out

01:06:13.930 --> 01:06:17.779
what these buildings are that are being
built down in the SoMa district,

01:06:17.779 --> 01:06:21.719
are actually built to, you can’t find out.
It’s in the code.

01:06:21.719 --> 01:06:26.750
It’s a secret to know what
the actual building things are.

01:06:26.750 --> 01:06:32.000
It’s not allowed to tell anybody.
So there’s no review in the situation.

01:06:32.860 --> 01:06:37.180
Well, so I have a couple
of recommendations.

01:06:37.190 --> 01:06:41.390
One is, it’s not just the
Earth scientists and the engineers.

01:06:41.390 --> 01:06:44.160
The architects are in this,
and they’re – and developers too

01:06:44.160 --> 01:06:46.630
are an important part
of this whole problem.

01:06:46.630 --> 01:06:51.180
And this thing about telling the
public that we’re designing

01:06:51.180 --> 01:06:54.680
for 2,500 years I think
is extremely misleading.

01:06:54.680 --> 01:07:00.670
It says, basically, we know the answer,
and we’re not going to change our mind.

01:07:00.670 --> 01:07:04.309
How can we give somebody a 25-year
answer and say, oh, by the way,

01:07:04.309 --> 01:07:09.119
I might change my mind in 10 years.
That doesn’t make any sense.

01:07:09.119 --> 01:07:12.380
And so it says,
who needs a USGS?

01:07:12.380 --> 01:07:15.770
I mean, who needs a
Caltech to study this problem?

01:07:15.770 --> 01:07:20.459
And at Caltech, I can tell you,
earthquake engineering is disappearing.

01:07:20.459 --> 01:07:24.609
Because the message is coming up,
this is a solved field,

01:07:24.609 --> 01:07:26.720
even though it clearly isn’t.

01:07:26.720 --> 01:07:30.880
But I think we’re really making
a mistake by doing that.

01:07:30.880 --> 01:07:35.280
I don’t think this 2,500-year
design is a scientific conclusion.

01:07:35.280 --> 01:07:40.800
And frankly, my recommendation
is that the USGS should just say,

01:07:40.809 --> 01:07:45.579
we can make the short-period ground
motion maps, but the long-period maps,

01:07:45.579 --> 01:07:48.839
we don’t know how to do that,
so we’re not going to publish any.

01:07:48.839 --> 01:07:51.859
I mean, because that’s
the honest answer is,

01:07:51.859 --> 01:07:54.559
we don’t know
what those numbers are.

01:07:54.559 --> 01:07:57.319
So thanks for
being so patient.

01:07:57.320 --> 01:08:03.420
[Applause]

01:08:05.120 --> 01:08:08.180
- So a really great way to start this
kind of month-long discussion about –

01:08:08.180 --> 01:08:10.420
and thinking critically about
our assumptions about

01:08:10.420 --> 01:08:12.920
how buildings are designed
and trying to learn more.

01:08:12.920 --> 01:08:15.680
So if people have questions for Tom,
we can take those, and then we’ll

01:08:15.680 --> 01:08:20.120
roll into kind of a more
informal discussion over pizza.

01:08:24.200 --> 01:08:30.440
[Silence]

01:08:30.860 --> 01:08:37.840
- Tom, as usual, thank you.
And that was a very challenging lecture.

01:08:39.480 --> 01:08:41.260
I made a lot of notes.

01:08:41.260 --> 01:08:46.480
It will take two days to ask you, but I
will ask two of them, if you don’t mind.

01:08:46.480 --> 01:08:51.480
One of them is, I think I discussed
this with you years ago on a plane,

01:08:51.480 --> 01:08:54.200
where I accidentally
sat next to you,

01:08:54.200 --> 01:08:58.620
and we were discussing
about the effect of damping.

01:08:58.620 --> 01:09:03.660
And I suggested that, for example,
when you are talking about the

01:09:03.660 --> 01:09:11.059
70% damped response spectra, et cetera,
so up until it comes to that 70%,

01:09:11.060 --> 01:09:15.020
that increasing damping
decreases the response.

01:09:15.020 --> 01:09:19.860
And that’s something that I believe that,
in all your considerations,

01:09:19.860 --> 01:09:24.300
and your colleagues’ considerations,
that is not being taken care of.

01:09:24.320 --> 01:09:27.360
Okay, let me ask
my second question.

01:09:27.720 --> 01:09:31.560
I’m – as I said, there are a number of
questions, and I hope that there’s

01:09:31.560 --> 01:09:39.940
enough pizza at lunchtime for all
of us to enter into nice discussions.

01:09:39.940 --> 01:09:46.040
But in all your comments, and I think
that there’s a lot of validity in

01:09:46.049 --> 01:09:54.320
most of what you say, but on that end,
I also looked at the whole world is

01:09:54.320 --> 01:09:59.429
looking green from your point of view.
So my general question is, is there

01:09:59.429 --> 01:10:03.680
anything that the structural
design engineers do that you like?

01:10:03.680 --> 01:10:09.599
- Oh. Don’t – let’s see.
Which one will I take first?

01:10:09.599 --> 01:10:15.860
About the damping, yes, the – increasing
the damping decreases the response.

01:10:15.860 --> 01:10:21.380
But, for years, people decreased
the response because of ductility

01:10:21.380 --> 01:10:24.730
with something called the R factor.
And you know what that is.

01:10:24.730 --> 01:10:28.909
Nobody else here knows what it is,
but if you – if you put in a

01:10:28.909 --> 01:10:33.389
more realistic damping to
put in that plastic damping,

01:10:33.389 --> 01:10:36.520
the R factor just
disappears out of the problem.

01:10:36.520 --> 01:10:39.849
And so you don’t
need to decrease anything.

01:10:39.849 --> 01:10:44.880
People had been decreasing things
in kind of a fake way for years.

01:10:44.880 --> 01:10:47.630
With respect – are there
things about the earthquake

01:10:47.630 --> 01:10:50.200
engineering profession that I like?
Absolutely.

01:10:50.200 --> 01:10:56.639
I mean, all my Earth science colleagues
accuse me of being an engineer.

01:10:56.639 --> 01:11:00.410
And my engineering colleagues
accuse me of being a scientist.

01:11:00.410 --> 01:11:05.040
And all my students have
been engineers, and I like the

01:11:05.040 --> 01:11:10.680
fact that the engineers are quite
rigorous about mechanics and really

01:11:10.680 --> 01:11:16.160
have a very deep understanding
of how things mechanically work.

01:11:16.160 --> 01:11:22.760
And so – and for us, as Earth scientists,
you know, when we – when we put

01:11:22.760 --> 01:11:27.719
out some sort of a paper in the AGU or
whatever, we get to change our mind.

01:11:27.719 --> 01:11:32.740
In five years, we get to say,
five years from now, I had this

01:11:32.740 --> 01:11:36.750
really cool idea, but when
I checked it out, it just didn’t work.

01:11:36.750 --> 01:11:42.650
And nobody will hold that against you.
If you’re a structural engineer, and you

01:11:42.650 --> 01:11:46.829
said, I designed the Millennium Tower,
and five years later, somebody comes

01:11:46.829 --> 01:11:51.400
to you and says, you screwed up,
I’m suing you for every penny

01:11:51.400 --> 01:11:56.320
you’ve got, I mean, you’re
exposed in a way an Earth scientist

01:11:56.320 --> 01:12:01.449
would never be exposed.
And so the engineers

01:12:01.449 --> 01:12:06.100
live a different world.
I mean – and they – I know why

01:12:06.100 --> 01:12:11.659
they’re conservative, and I have
some colleagues – I’m not sure

01:12:11.659 --> 01:12:17.340
whether they’ll call me a friend or not,
but I just want to have a more

01:12:17.340 --> 01:12:23.270
rational discussion with them. And I’m
not really blaming the engineer in here.

01:12:23.270 --> 01:12:27.349
It’s kind of the entire system,
really, that’s the problem.

01:12:27.349 --> 01:12:32.329
I think it’s got a lot in common
with the mortgage-backed securities

01:12:32.329 --> 01:12:36.909
of the big collapse in 2008,
where people were saying,

01:12:36.909 --> 01:12:42.900
all these bonds are rated triple-A, and
of course, it was all kind of nonsense.

01:12:42.900 --> 01:12:46.550
That was – nobody really
had a good understanding.

01:12:46.550 --> 01:12:49.290
But I’m not sure anybody
was really evil there.

01:12:49.290 --> 01:12:53.760
They just – they just couldn’t stop it.
There’s so much money involved

01:12:53.760 --> 01:12:57.761
in the development today.
For the structural engineer, if they

01:12:57.761 --> 01:13:04.639
tried to say what I’m saying here, they’ll
get crushed by it. I mean, it’s tough.

01:13:04.640 --> 01:13:08.079
But I’m so happy you guys
are talking about it, actually.

01:13:08.720 --> 01:13:12.600
- Yeah. I got the microphone.
Quickly, about the stock market.

01:13:12.610 --> 01:13:18.171
Yes, it took quite a beating in 2008
and 2009, but for years in the

01:13:18.180 --> 01:13:27.540
21st century, it’s gone up rather nicely.
And, you know, the short – anyone

01:13:27.540 --> 01:13:30.120
will tell you this,
[chuckles] you know,

01:13:30.120 --> 01:13:36.640
measuring things on a year-to-year basis
in the stock market is not a good idea.

01:13:37.680 --> 01:13:40.880
[chuckles]
That’s a general statement.

01:13:40.880 --> 01:13:47.620
The one thing that I do know something
about in your talk is the ground motion.

01:13:47.620 --> 01:13:52.240
And it seemed to me the
ground motion curves you showed –

01:13:52.240 --> 01:13:55.780
I didn’t get a good look,
but for PGA and PGV,

01:13:55.790 --> 01:14:00.020
they seemed to be considerably
higher than the NGA equation.

01:14:00.020 --> 01:14:08.460
- I think you’re right.
- And that is something that is testable.

01:14:08.460 --> 01:14:11.530
And I guess the question is,
if you know that –

01:14:11.530 --> 01:14:14.760
if we agree on that,
why is that the case?

01:14:14.760 --> 01:14:19.170
- Well, so I’m especially worried
about near-source ground motions

01:14:19.170 --> 01:14:21.489
in big earthquakes, so …
- And so are we.

01:14:21.489 --> 01:14:25.150
- … whenever there’s a record
like the Kathmandu record,

01:14:25.150 --> 01:14:31.020
or we’ve gotten bunch of them
from Japan lately or New Zealand,

01:14:31.020 --> 01:14:35.840
those records are precious to me.
I mean, they’re the real indicator

01:14:35.840 --> 01:14:38.800
of what’s going to happen
in these big earthquakes.

01:14:38.800 --> 01:14:43.309
And when we – when I process
those records, I never put a –

01:14:43.309 --> 01:14:48.139
just a high-pass filter on them.
That just kind of removes

01:14:48.139 --> 01:14:52.010
a lot of the important information.
And in fact, I didn’t have time to talk

01:14:52.010 --> 01:14:56.840
about it here, but part of
Kenny Buyco’s thesis is about

01:14:56.840 --> 01:15:03.210
how using the current processing
techniques often removes important

01:15:03.210 --> 01:15:09.080
parts of the – of the signal, so …
- And people, you know,

01:15:09.080 --> 01:15:12.520
still worry about doing that for the –
for the ground displacements,

01:15:12.520 --> 01:15:17.239
but for the peak accelerations and
the peak ground velocities, you know,

01:15:17.240 --> 01:15:19.740
that’s pretty well-behaved stuff.
- It is.

01:15:19.740 --> 01:15:25.760
- And the relations you showed are still
significantly higher for PGA and PGV.

01:15:25.769 --> 01:15:28.380
- I don’t really think they are.
When we compare what was

01:15:28.380 --> 01:15:32.380
coming out of PGA and PGV,
they’re right in there …

01:15:32.380 --> 01:15:36.809
- Okay.
- … exactly with the – with the NGA.

01:15:36.809 --> 01:15:41.199
So peak acceleration
in close is a half a g –

01:15:41.199 --> 01:15:45.079
it was a half a g when you and I
were young, and it’s still half a g.

01:15:45.079 --> 01:15:49.929
And it will be for the next 100 years.
I mean – or 300 years.

01:15:49.929 --> 01:15:52.239
So that’s pretty stable.

01:15:52.239 --> 01:15:56.619
But the long-period, it really depends
on what’s the slip going to be

01:15:56.619 --> 01:16:01.380
in the next big earthquake, so …
- Well, maybe I didn’t see enough

01:16:01.380 --> 01:16:04.519
of them to justify my comment.
- Well, I’ll tell you later, but …

01:16:04.519 --> 01:16:07.429
- But we compare that stuff after,
okay, okay.

01:16:07.429 --> 01:16:10.780
- We don’t disagree at
the high frequencies.

01:16:12.140 --> 01:16:15.660
- Okay. Well, thank you all for your –
and thanks again, Tom, for your talk.

01:16:15.670 --> 01:16:19.280
Our next speaker will be
Tuesday morning at 9:30.

01:16:19.280 --> 01:16:23.000
So hopefully everyone here will
be there as well, bright and early.

01:16:23.000 --> 01:16:24.690
And thanks
for coming.

01:16:24.690 --> 01:16:29.540
[Applause]

01:16:30.520 --> 01:16:32.860
[Silence]

01:16:32.860 --> 01:16:34.800
[static sounds]

01:16:36.340 --> 01:16:39.120
- Are we still on the air, Tom?
- Yeah.

01:16:39.120 --> 01:16:43.820
[Silence]