WEBVTT
Kind: captions
Language: en

00:00:01.000 --> 00:00:03.380
[background conversations]

00:00:03.380 --> 00:00:05.140
Good morning.
Hello.

00:00:05.140 --> 00:00:07.420
So welcome to this –
today’s, or this week’s,

00:00:07.420 --> 00:00:11.100
Earthquake Science Center seminar.
And before we introduce the speaker,

00:00:11.100 --> 00:00:12.940
I just want to remind
everyone that next week,

00:00:12.940 --> 00:00:16.730
Diane Moore of Building 3A
will be here to talk about

00:00:16.730 --> 00:00:20.730
the Bartlett Springs Fault and some of
the work she’s been doing on that.

00:00:20.730 --> 00:00:22.840
And then Mehmet,
if you don’t …

00:00:25.260 --> 00:00:29.520
- Well, I’m happy to welcome
David McCallen here today.

00:00:29.520 --> 00:00:33.060
And he’s currently – is it on?
- It’s on. [inaudible]

00:00:33.060 --> 00:00:39.640
- Okay. [laughs]
Associate vice president of UC-Berkeley

00:00:39.640 --> 00:00:45.280
Office of President and associate vice
president for UC National Laboratories.

00:00:45.920 --> 00:00:51.300
He received his Ph.D.
degree from UC-Davis.

00:00:51.300 --> 00:00:54.440
And his research interests
are in the area of

00:00:54.440 --> 00:00:58.940
advanced computational
modeling of structures.

00:01:00.160 --> 00:01:05.500
He participates in the UC
oversight of – this is a major task –

00:01:05.500 --> 00:01:08.840
of the three UC-managed
National Laboratories.

00:01:08.840 --> 00:01:13.600
That is Lawrence Berkeley,
Lawrence Livermore, and Los Alamos.

00:01:14.420 --> 00:01:19.280
Prior to his current positions, he spent
25 years in a variety of technical

00:01:19.280 --> 00:01:27.400
managerial positions at NL – Lawrence
Livermore National Laboratory now.

00:01:27.400 --> 00:01:31.720
He also continues in a technical role
as a visiting researcher at Lawrence

00:01:31.720 --> 00:01:35.960
Berkeley National Laboratory
where he leads two major

00:01:35.960 --> 00:01:40.560
U.S. DOE research projects
developing advanced relation

00:01:40.560 --> 00:01:45.690
capabilities for assessment
of earthquake hazard and risk.

00:01:45.690 --> 00:01:50.050
So I’m really happy to welcome
David here, and I’m sure you’ll

00:01:50.050 --> 00:01:54.420
enjoy what he has to say.
- Well, thank you very much.

00:01:54.420 --> 00:01:56.380
It truly is a pleasure to
be here with you today.

00:01:56.380 --> 00:02:01.920
I was just mentioning to some of my old
acquaintances in college this morning,

00:02:01.920 --> 00:02:06.570
the last time I was down here to sort of
judge the epoch, your doors were open,

00:02:06.570 --> 00:02:10.490
and there was a map center that the
public could walk in and out of.

00:02:10.490 --> 00:02:14.329
So it certainly is a sign of the times
in the – the battened-down mentality.

00:02:14.329 --> 00:02:17.590
And I don’t know how long it’s
been that way, but that sort of

00:02:17.590 --> 00:02:21.720
gauges you about the duration it’s
been since I came down here.

00:02:21.720 --> 00:02:24.409
And I used to come down
quite a bit in my role at Livermore.

00:02:24.409 --> 00:02:28.540
We were working on a number
of earthquake hazard issues

00:02:28.540 --> 00:02:31.430
for Yucca Mountain – and talked with
Tom a little bit about that this morning –

00:02:31.430 --> 00:02:32.901
and other things.
So I had the opportunity

00:02:32.901 --> 00:02:35.799
to come down and visit
with Tom and Paul Spudich

00:02:35.799 --> 00:02:38.720
and others through the years.
And it was – it was always fun

00:02:38.720 --> 00:02:40.760
to come down here and
have those lively discussions.

00:02:40.760 --> 00:02:44.849
So I really appreciate the opportunity
to speak with you today.

00:02:44.849 --> 00:02:47.959
I am not going to talk about my
administrative role for the next hour.

00:02:47.959 --> 00:02:50.879
I think you’d be bored to death
unless you’re interested in, you know,

00:02:50.879 --> 00:02:53.900
the bid for Los Alamos we’re
going through, and might make

00:02:53.900 --> 00:02:56.739
a great conversation,
and we could talk over lunch.

00:02:56.739 --> 00:03:00.230
But I’m really going to
talk about sort of my downtime,

00:03:00.230 --> 00:03:04.120
I would characterize – my fun time.
And I have a supervisor,

00:03:04.120 --> 00:03:07.520
Professor Kim Budil, who is
good enough to let me split my time

00:03:07.520 --> 00:03:10.640
between UC Office of the President and
Lawrence Berkeley National Laboratory,

00:03:10.640 --> 00:03:13.060
where I have a number of
DOE projects going.

00:03:13.060 --> 00:03:17.230
I’m going to talk about one of
our smaller projects, actually –

00:03:17.230 --> 00:03:21.209
something that we’ve kind of done
on a shoestring today, but it’s really

00:03:21.209 --> 00:03:24.950
something that I’m very excited
about and have a lot of interest in.

00:03:24.950 --> 00:03:28.359
We’ve gotten some decent results,
so I want to talk to you about that.

00:03:28.359 --> 00:03:31.290
And so I’m going to begin
by identifying our collaborators

00:03:31.290 --> 00:03:35.620
in this project – always important.
This sort of sensor concept

00:03:35.620 --> 00:03:39.010
we’re working on – and I’ll speak
in a moment to how this idea

00:03:39.010 --> 00:03:42.189
sort of came to me when I was
working at Livermore a few years ago

00:03:42.189 --> 00:03:46.010
in a area totally unrelated to earthquakes
and sort of made a connection.

00:03:46.010 --> 00:03:49.099
But I’ve been involved in
developing this concept.

00:03:49.099 --> 00:03:53.510
Dr. Floriana Petrone from
Sapienza University in Rome

00:03:53.510 --> 00:03:57.469
is a really, really great postdoc –
a postdoctoral scholar we have

00:03:57.469 --> 00:03:59.279
up at Lawrence
Berkeley Laboratory.

00:03:59.279 --> 00:04:02.400
She is heavily involved
in the simulation piece of this.

00:04:02.400 --> 00:04:05.909
Many of you may know Ian Buckle –
a very prominent earthquake engineer

00:04:05.909 --> 00:04:10.200
at the University of Nevada.
Ian has gotten involved in terms of

00:04:10.200 --> 00:04:13.329
the experimentation I’m going to
talk about in part two of this.

00:04:13.329 --> 00:04:16.359
And then, Chico State University,
which was actually my undergraduate

00:04:16.359 --> 00:04:20.639
institution when I first went to college –
a fellow names Jason Coates,

00:04:20.639 --> 00:04:25.650
who is a research engineer in the
Mechatronics Center at Chico State,

00:04:25.650 --> 00:04:28.110
has been developing
and building our hardware.

00:04:28.110 --> 00:04:30.280
I mentioned we’ve done
this on a shoestring.

00:04:30.280 --> 00:04:33.250
We can’t afford to do [chuckles]
a lot of the things we’re doing

00:04:33.250 --> 00:04:35.800
at our National Laboratories.
Our overhead rate is high.

00:04:35.800 --> 00:04:38.370
So we have – we have developed
a program that’s distributed

00:04:38.370 --> 00:04:42.240
across these institutions,
and these are the key players.

00:04:42.240 --> 00:04:45.760
So let me digress for just a
minute and sort of give

00:04:45.760 --> 00:04:50.380
the background and some
notion of how this idea came to be.

00:04:50.380 --> 00:04:54.539
And when I was at Livermore,
I got tasked to being the laser science

00:04:54.539 --> 00:04:57.750
engineering division
leader for a couple of years.

00:04:57.750 --> 00:05:03.759
And really, at that time, that project –
that focus of that large division was

00:05:03.760 --> 00:05:06.640
building the National Ignition Facility –
you may have heard about it –

00:05:06.640 --> 00:05:10.319
at Livermore.
And this just shows a schematic

00:05:10.319 --> 00:05:12.599
of the NIF –
the National Ignition Facility.

00:05:12.599 --> 00:05:17.360
To give you some context of scales,
If you laid football fields across here,

00:05:17.360 --> 00:05:19.910
this is just about a
football field of width.

00:05:19.910 --> 00:05:22.909
You can line up three football
fields of length down here.

00:05:22.909 --> 00:05:26.259
And then you have a
big, giant target chamber here.

00:05:26.259 --> 00:05:30.840
And this National Ignition Facility
was built for fundamental research

00:05:30.840 --> 00:05:33.580
and applied research and
high-energy density physics.

00:05:33.580 --> 00:05:36.419
And I think people know what
the product – principle product

00:05:36.420 --> 00:05:40.260
of Livermore is. They are a
stockpile stewardship laboratory.

00:05:40.260 --> 00:05:43.610
Post-1992, they could
no longer test nuclear weapons.

00:05:43.610 --> 00:05:48.000
And there’s been – was a big effort
to develop simulation capabilities

00:05:48.000 --> 00:05:53.720
and experimental capabilities that would
allow them to assure the vitality and

00:05:53.720 --> 00:05:57.220
the reliability of the nation’s
deterrent in the absence of underground

00:05:57.220 --> 00:06:01.169
nuclear testing in Nevada. This facility
was an important part of that.

00:06:01.169 --> 00:06:07.970
It has 192 laser beams. These laser beams
start out as one single low-power beam.

00:06:07.970 --> 00:06:11.730
Goes into a beam splitter.
Splits out into 192 beams.

00:06:11.730 --> 00:06:14.139
Those beams rattle
back and forth through here

00:06:14.139 --> 00:06:18.289
very quickly and what are
called flash lamps go off.

00:06:18.289 --> 00:06:22.900
Those flash lamps pump up that laser in
terms of its energy – those 192 beams.

00:06:22.900 --> 00:06:26.680
Those 192 beams come into turning
mirrors, and they all focus down,

00:06:26.680 --> 00:06:31.349
in theory, to a single point at
the center of that target chamber.

00:06:31.349 --> 00:06:36.370
And they want to focus very accurately.
You have a very, very small target that

00:06:36.370 --> 00:06:43.040
has low-atomic-number materials inside.
And the idea is to compress that target

00:06:43.040 --> 00:06:47.540
and achieve fusion, for the
first time in the laboratory,

00:06:47.540 --> 00:06:50.150
or conditions close to fusion.
And that why it’s called

00:06:50.150 --> 00:06:52.800
the National Ignition Facility.
You want to get ignition.

00:06:52.800 --> 00:06:55.500
You want to create
fusion in a laboratory.

00:06:55.500 --> 00:06:58.470
So to give you some scale –
idea of scale, these little

00:06:58.470 --> 00:07:04.060
red boxes are surrounding
humans in these various locations.

00:07:04.060 --> 00:07:07.840
So this is a giant facility.
A $4 billion facility.

00:07:07.849 --> 00:07:13.479
And when we built this, our principle
threat was not earthquakes.

00:07:13.479 --> 00:07:18.069
Our principle threat was vibrations.
This whole facility is an optical bench.

00:07:18.069 --> 00:07:22.750
And if you have too much vibration in
that system, due to vehicles going by,

00:07:22.750 --> 00:07:26.370
due to air conditioners running,
due to people walking down the hall,

00:07:26.370 --> 00:07:29.910
you cannot achieve the
pointing accuracy you need,

00:07:29.910 --> 00:07:31.969
which is – which is
very, very stringent.

00:07:31.969 --> 00:07:34.810
So one of my engineers –
we had to determine, as we built

00:07:34.810 --> 00:07:39.569
this $4 billion facility, if we were
ultimately going to be able to achieve

00:07:39.569 --> 00:07:43.090
our pointing accuracy. Because,
if we’re not, we’ve got a big problem.

00:07:43.090 --> 00:07:47.330
So we did a lot of design things and
very specialized optical design to make

00:07:47.330 --> 00:07:50.990
that platform and that bench stable.
But we really didn’t know until we

00:07:50.990 --> 00:07:53.110
built the damn thing if we
were going to be able to

00:07:53.110 --> 00:07:55.169
meet our
performance requirements.

00:07:55.169 --> 00:07:58.320
Even after we built it,
we had to scratch our head and say,

00:07:58.320 --> 00:08:01.420
okay, well, how do we know if
we’ve got that pointing accuracy?

00:08:01.420 --> 00:08:05.180
So a couple of our engineers came
up with a fairly clever thing.

00:08:05.180 --> 00:08:07.789
At target chamber center,
they went out and found

00:08:07.789 --> 00:08:12.349
one of these silicon photodiodes
positioning sensitive devices.

00:08:12.349 --> 00:08:15.789
And I’m not a solid-state physicist,
so bear with me a minute.

00:08:15.789 --> 00:08:18.969
I’ll tell you the fundamentals
of how I understand this works.

00:08:18.969 --> 00:08:23.231
So the laser impinges on this device –
this silicon device, as you see

00:08:23.231 --> 00:08:26.530
over on the right-hand side.
And this is an actual picture of one.

00:08:26.530 --> 00:08:30.699
When the laser goes into the substrate,
there’s a quantum effect,

00:08:30.699 --> 00:08:35.130
and free – electrons are set free.
Free electrons.

00:08:35.130 --> 00:08:39.139
If there’s a voltage across this thing,
those electrons can then migrate

00:08:39.139 --> 00:08:41.779
through a current
out to the edges.

00:08:41.779 --> 00:08:46.490
Turns out the current is directly
proportional to the resistance in the

00:08:46.490 --> 00:08:50.950
material, which is directly proportional
to the distance from the edge.

00:08:50.950 --> 00:08:54.769
So the bottom line is, if you can
measure those currents very quickly,

00:08:54.769 --> 00:08:58.130
you can very, very accurately
determine where that laser hits –

00:08:58.130 --> 00:09:00.310
I mean,
really, really accurately.

00:09:00.310 --> 00:09:04.730
And so I would – I’ve always said,
this is the money shot right here.

00:09:04.730 --> 00:09:08.170
Built the $4 billion laser.
We wanted to be able to hit that target

00:09:08.170 --> 00:09:12.959
with the laser beams within 50 microns,
which is pretty damn small.

00:09:12.959 --> 00:09:15.019
So we put one of these chips
at the center, and we just

00:09:15.019 --> 00:09:18.420
fired the laser a whole bunch of times.
And we found out where those lasers

00:09:18.420 --> 00:09:21.730
hit this thing, and here’s the cloud.
You can see there’s a bias.

00:09:21.730 --> 00:09:25.510
But lo and behold, the 50 microns
we wanted to achieve is right here,

00:09:25.510 --> 00:09:29.850
so there was a big sigh of relief
at Livermore when we did this.

00:09:29.850 --> 00:09:32.610
Because we didn’t have
a huge vibrational problem.

00:09:32.610 --> 00:09:35.649
And if you think of this, you know,
we really met the criteria without

00:09:35.649 --> 00:09:39.180
a lot of room to spare,
so that was – that’s the money shot.

00:09:39.180 --> 00:09:42.329
So, you know, we’re going to
get around to earthquakes today.

00:09:42.329 --> 00:09:45.750
We’ll jump there next.
But the point of this is,

00:09:45.750 --> 00:09:48.520
I had a notion that, god,
you know, one of the –

00:09:48.520 --> 00:09:53.380
some of the attributes of this device
is that it’s extremely broadband.

00:09:53.380 --> 00:09:55.600
From an engineering
standpoint, the physics

00:09:55.610 --> 00:09:58.090
of this thing
occur instantaneously.

00:09:58.090 --> 00:10:00.980
So you can measure stuff very,
very fast with this type of thing.

00:10:00.980 --> 00:10:03.290
If that – if that laser
is vibrating, you know it.

00:10:03.290 --> 00:10:06.970
You can track that thing to extremely
high frequency with extreme accuracy.

00:10:06.970 --> 00:10:09.670
You can see we measured
within 50 microns.

00:10:09.670 --> 00:10:12.970
And moreover, maybe just
as important, you can measure

00:10:12.970 --> 00:10:16.740
a permanent offset just really easily.
I mean, it’s trivial.

00:10:16.740 --> 00:10:20.500
If you move that laser over
in a static displacement, you know,

00:10:20.500 --> 00:10:23.690
you can measure that.
So, in my mind, I thought, ah,

00:10:23.690 --> 00:10:26.720
you know, this would be wonderful
for measuring things like structural

00:10:26.720 --> 00:10:30.160
vibrations in an earthquake
because it’s incredibly broadband

00:10:30.160 --> 00:10:33.709
and incredibly accurate.
Okay, so the only problem

00:10:33.709 --> 00:10:38.380
with that theory is that these
chips are extremely expensive.

00:10:38.380 --> 00:10:43.220
You know, they’re precision
manufactured, and they’re limited in size.

00:10:43.220 --> 00:10:48.120
So I thought you could scale this
thing up to do – measure building

00:10:48.120 --> 00:10:52.160
response and large displacements.
Turns out, you can’t do that very well.

00:10:52.160 --> 00:10:56.550
Just the engineering does not permit that.
So I was kind of stuck, and sort of

00:10:56.550 --> 00:10:58.700
went away and didn’t
think about it much more.

00:10:58.700 --> 00:11:02.340
But then I came back to it
just before I left Livermore.

00:11:02.350 --> 00:11:05.139
And to motivate discussion,
I want to talk a little bit about

00:11:05.139 --> 00:11:08.220
what engineers care about in a
vibrating and deforming building.

00:11:08.220 --> 00:11:10.279
And I want to introduce the
concept of interstory drift that

00:11:10.279 --> 00:11:13.050
many of you have probably heard about.
And if you think about solid

00:11:13.050 --> 00:11:17.890
mechanisms, and you think about a
steel bar – just think of a simple steel bar.

00:11:17.890 --> 00:11:20.899
You can think of characterizing
the force deflection behavior

00:11:20.899 --> 00:11:24.110
of that bar in terms of
forces and displacements.

00:11:24.110 --> 00:11:27.360
But engineers and continuum mechanics
people don’t like to do that.

00:11:27.360 --> 00:11:29.720
Why? Because you have
to know what the bar is.

00:11:29.720 --> 00:11:33.089
It doesn’t tell you anything to
just say, I put a 1,000-pound force

00:11:33.089 --> 00:11:36.600
on this bar, and it deformed
a tenth of an inch.

00:11:36.600 --> 00:11:39.209
That doesn’t tell you a lot unless
you know the dimensions of the bar.

00:11:39.209 --> 00:11:42.579
So engineers normalize,
and solid mechanics people normalize,

00:11:42.580 --> 00:11:44.200
by thinking of
stress and strain.

00:11:44.200 --> 00:11:49.840
So they think of force per unit area
or change in length per unit length.

00:11:49.850 --> 00:11:53.230
And so they normalize things that way.
I would – I would assert that interstory

00:11:53.230 --> 00:11:56.860
drift is the structural engineer’s
normalization of building deformation.

00:11:56.860 --> 00:12:00.350
If you have a building, and you said,
during an earthquake, the roof of that

00:12:00.350 --> 00:12:04.380
building displaced 10 inches,
that wouldn’t tell you a hell of a lot

00:12:04.380 --> 00:12:09.350
unless you knew whether it was a
two-story building or a 40-story building.

00:12:09.350 --> 00:12:12.269
But if you characterize things
in terms of interstory drift

00:12:12.269 --> 00:12:15.730
and normalized it that way, then it’s
common language between those

00:12:15.730 --> 00:12:19.769
different sizes of buildings just like
stress and strain are for a solid.

00:12:19.769 --> 00:12:22.720
And so, if you think of interstory drift –
this is a finite element model of a

00:12:22.720 --> 00:12:26.980
building deforming, exaggerated.
It’s simply the relative displacement

00:12:26.980 --> 00:12:29.760
between two floor levels
divided by that story height,

00:12:29.760 --> 00:12:34.400
usually expressed as a percent.
So it is a direct measure of deformation.

00:12:35.500 --> 00:12:38.399
And interstory drift comes
into play in earthquake engineering

00:12:38.399 --> 00:12:41.089
in a number of ways – really,
three principle different ways.

00:12:41.089 --> 00:12:43.360
Number one, it’s used
to define limit states.

00:12:43.360 --> 00:12:46.940
And I’ll refer to the DOE
standard for nuclear facilities.

00:12:46.940 --> 00:12:53.170
And they define limit states of that
facility of being linear-elastic behavior,

00:12:53.170 --> 00:12:56.930
a very small amount of inelastic
behavior, significant inelastic behavior,

00:12:56.930 --> 00:13:01.300
and inelastic behavior short of collapse –
four distinct limit states that they

00:13:01.300 --> 00:13:04.320
think about in performance-based
design for their facilities.

00:13:04.320 --> 00:13:08.430
Those limit states are defined directly
in terms of interstory drift,

00:13:08.430 --> 00:13:10.920
and that’s how you quantify
those states. So that’s one way.

00:13:10.920 --> 00:13:13.250
We think of limit states when we
design our buildings and

00:13:13.250 --> 00:13:15.620
want to characterize the
performance of the building.

00:13:15.620 --> 00:13:19.300
Secondly, we have
not-to-exceed drift limits.

00:13:19.310 --> 00:13:22.790
If a building – you can imagine –
a tall building moves over too much,

00:13:22.790 --> 00:13:26.250
there’s a large P-delta effect
due to gravity loads.

00:13:26.250 --> 00:13:29.579
And that can cause that
building to want to collapse.

00:13:29.579 --> 00:13:33.889
And so – another characterize – way that
these interstory drifts are characterized

00:13:33.889 --> 00:13:38.650
is to limit the interstory drift
for a given earthquake input motion.

00:13:38.650 --> 00:13:42.440
And then finally, where people have
gotten to more recently is they’ve

00:13:42.440 --> 00:13:46.470
used interstory drift as a damage
indicator post-earthquake.

00:13:46.470 --> 00:13:50.069
In other words, an earthquake occurs.
You look at the interstory drift

00:13:50.069 --> 00:13:53.750
that is due to inelastic action
and permanent in any building.

00:13:53.750 --> 00:13:56.779
And you characterize and
limit that to know whether that

00:13:56.779 --> 00:14:00.290
building has excessive damage.
So it’s becoming a – really a broadly

00:14:00.290 --> 00:14:04.260
used quantity and important
in earthquake engineering.

00:14:04.260 --> 00:14:08.120
Measuring interstory drift with
accelerometers is really hard,

00:14:08.120 --> 00:14:11.340
if not impossible.
And I would refer you to two studies –

00:14:11.340 --> 00:14:17.100
one by Wallace and Skolnik at UCLA.
And you can see the bottom line there.

00:14:17.100 --> 00:14:21.060
Double-integrating an accelerometer
time history and trying to get

00:14:21.060 --> 00:14:26.380
reliable displacements is problematic.
And you’re fundamentally conflicted

00:14:26.380 --> 00:14:30.750
by the processing you do to remove
drift in things in that accelerometer.

00:14:30.750 --> 00:14:34.290
You’re removing information that
you’d like to have because you really

00:14:34.290 --> 00:14:37.529
want to have permanent –
measurement of permanent displacement

00:14:37.529 --> 00:14:42.279
as well as dynamic displacement.
And Trifunac has looked at this issue

00:14:42.279 --> 00:14:45.569
and has a couple paper as well,
from USC, where he talked about

00:14:45.569 --> 00:14:48.649
the fact that getting permanent
displacements is very, very hard

00:14:48.649 --> 00:14:50.839
with accelerometers.
This was a really good study,

00:14:50.839 --> 00:14:52.699
so if you have an interest
in this topic, it’s in the

00:14:52.699 --> 00:14:55.889
Journal of Structural Engineering.
I would recommend it.

00:14:55.889 --> 00:14:59.269
So here’s the big idea,
or the idea – the big idea.

00:14:59.269 --> 00:15:03.880
So, to exploit the physics of light,
just like we did on NIF,

00:15:03.880 --> 00:15:08.870
to measure the deformation of a
building during and after an earthquake.

00:15:08.870 --> 00:15:11.920
And so the notion is quite simple.
This thing is moving.

00:15:11.920 --> 00:15:16.430
You have interstory drift.
Propagate a laser across the story height,

00:15:16.430 --> 00:15:21.339
just like it was coming roof-to-floor here.
And on some type of position-sensitive

00:15:21.339 --> 00:15:26.889
detector, measure the shift of that laser.
So conceptually, it’s very, very simple.

00:15:26.889 --> 00:15:29.319
That would be a direct
measurement of interstory drift.

00:15:29.319 --> 00:15:33.709
And, if you did it with a laser, and your
PSD had the attributes of some of those

00:15:33.709 --> 00:15:37.230
earlier sensors I talked about at NIF,
it would be extremely broadband.

00:15:37.230 --> 00:15:40.570
You could measure the dynamics
over a broad frequency range.

00:15:40.570 --> 00:15:43.250
And you could measure
permanent interstory drift.

00:15:43.250 --> 00:15:47.120
And you would have – this is a direct
measure map. There’s no post-processing.

00:15:47.120 --> 00:15:49.160
Right after the earthquake,
you’d have your information.

00:15:49.160 --> 00:15:54.300
Right now, it would be there.
So that’s sort of the underlying concept.

00:15:54.310 --> 00:15:58.899
So we monkeyed around for about a
year or two once we realized that we

00:15:58.900 --> 00:16:03.000
couldn’t use these photoelectric chips.
And so I’m just going to –

00:16:03.000 --> 00:16:05.980
I’m not going to talk about all the
pathways we went down working

00:16:05.990 --> 00:16:09.579
at Livermore, but I’m going to talk about
the end result in what we came up with.

00:16:09.579 --> 00:16:13.350
And we came up with something that
we call a discrete diode position sensor.

00:16:13.350 --> 00:16:18.790
And it is simply –
very simply a staggered array

00:16:18.790 --> 00:16:23.380
of little photoelectric diodes
that are very, very inexpensive.

00:16:23.380 --> 00:16:26.500
It’s not continuous. It’s a grid
of a whole bunch of them.

00:16:26.520 --> 00:16:31.300
And so, fundamentally, what this does,
these serve as on-off switches.

00:16:31.300 --> 00:16:34.870
And I’ll show you in a moment,
when a laser hits one of these diodes,

00:16:34.870 --> 00:16:38.500
it creates a voltage.
And when the laser goes off the diode,

00:16:38.500 --> 00:16:42.850
the voltage disappears very quickly.
So they are simply on-off switches.

00:16:42.850 --> 00:16:46.779
So in your mind’s eye, we take a laser,
we put it through a diffractive element

00:16:46.779 --> 00:16:50.110
to create a line source that impinges
on this array, and then, when the

00:16:50.110 --> 00:16:55.769
interstory drift proceeds, that laser
swoops back and forth over that array,

00:16:55.769 --> 00:16:58.400
and you sample that
array very, very rapidly.

00:16:58.400 --> 00:17:01.579
And you know where the laser is.
And you know what the time history

00:17:01.579 --> 00:17:05.050
of the drift is, as well as
any permanent drift, in theory.

00:17:05.050 --> 00:17:07.110
So this is what the
response of a diode looks like.

00:17:07.110 --> 00:17:09.640
This is actual data from
one of these little diodes.

00:17:09.640 --> 00:17:13.490
These things cost a buck or two.
You sweep the laser over the active

00:17:13.490 --> 00:17:17.120
area of the diode. You can see
what the actual voltage looks like.

00:17:17.120 --> 00:17:21.100
It picks up as the diode sweeps
across, and then it falls off.

00:17:21.100 --> 00:17:25.459
We put in a comparative circuit.
So we turn that into a signal

00:17:25.459 --> 00:17:29.010
that is a top hat signal.
The damn thing’s either on or off.

00:17:29.010 --> 00:17:32.470
We can’t have this kind-of-on,
kind-of-off stuff in that ramp.

00:17:32.470 --> 00:17:36.289
We set bounds and thresholds, and we
go through with this comparator.

00:17:36.289 --> 00:17:41.260
We see that voltage, and bingo. That
thing is binary. It’s either on or off.

00:17:41.260 --> 00:17:45.200
And so that’s how we sense how
that laser is moving across there.

00:17:45.210 --> 00:17:49.770
There’s a madness and a method
to how we stagger these things.

00:17:49.770 --> 00:17:52.950
We stagger them because we want to
increase the accuracy measurement.

00:17:52.950 --> 00:17:55.250
And we’ll talk about accuracy in
a few moments when we get into

00:17:55.250 --> 00:17:58.440
the experimental campaign.
As this laser moves across there,

00:17:58.440 --> 00:18:01.750
we want to know where that laser is,
so we stagger this into array.

00:18:01.750 --> 00:18:05.980
And I’ve just indicated, if that’s our laser
beam that’s come down and diffracted,

00:18:05.980 --> 00:18:11.210
as it moves from t-1 to t-2 to t-3,
these things light up, and successive

00:18:11.210 --> 00:18:15.250
diodes light up, and we know
where that laser is, and we translate that

00:18:15.250 --> 00:18:19.730
directly into drift. So I’m going to
talk a lot about the validation of this.

00:18:19.730 --> 00:18:25.070
We’ve gone through a very, very
extensive campaign of testing this

00:18:25.070 --> 00:18:28.539
concept and building these things.
First thing we did was we built a

00:18:28.539 --> 00:18:34.900
very simple test bed at Chico with
a very well-controlled precision table.

00:18:34.900 --> 00:18:38.909
And you can just think of this as being
a miniature earthquake shake table,

00:18:38.909 --> 00:18:42.780
but with very, very high accuracy
and a great deal of controllability.

00:18:42.780 --> 00:18:46.840
We strap the sensor down to that.
We put a laser on the roof.

00:18:46.840 --> 00:18:51.290
We put the table on the floor.
And we began to test the overall

00:18:51.290 --> 00:18:55.780
diode and the sensor by imposing
various motions to this.

00:18:55.780 --> 00:18:59.210
And so we wanted to impose –
ultimately, we did a bunch of

00:18:59.210 --> 00:19:02.450
sweep tests – very simple sine waves.
But we ultimately wanted to begin to

00:19:02.450 --> 00:19:07.720
understand how this thing would
perform for realistic earthquake motions.

00:19:07.720 --> 00:19:10.970
And so Professor Astaneh and I had
developed some very detailed

00:19:10.970 --> 00:19:14.630
nonlinear models for steel frame
structures a few years ago.

00:19:14.630 --> 00:19:19.210
In this case, a three-story and a 40-story.
We took some very strong

00:19:19.210 --> 00:19:23.170
near-field earthquake motions.
We applied them to the base of those

00:19:23.170 --> 00:19:27.770
structures, and we looked at some of the
extreme interstory drifts that occurred.

00:19:27.770 --> 00:19:32.630
This was the Landers/Lucerne record
with a correction that Iwan and

00:19:32.630 --> 00:19:35.260
his student did to try and
retrieve the long-period motion.

00:19:35.260 --> 00:19:39.480
So it’s got a very big displacement pulse
in it. You can’t see it, but if you band

00:19:39.480 --> 00:19:43.700
pass this and remove the high frequency,
you’d see a really big pulse here.

00:19:43.700 --> 00:19:47.240
So we shake this building to it, and up
here at about third part of this building,

00:19:47.240 --> 00:19:51.299
you can see the interstory drift.
There’s a great deal of inelastic action.

00:19:51.299 --> 00:19:54.640
The building shifts, and then
there’s a significant permanent offset

00:19:54.640 --> 00:19:57.780
as the building rings down.
So this has both dynamic

00:19:57.780 --> 00:20:00.799
low frequency – it’s about
a 4- or 5-second-period building –

00:20:00.799 --> 00:20:04.320
with a large displacement
offset – DC offset.

00:20:04.320 --> 00:20:06.350
We took a three-story building –
much higher frequency.

00:20:06.350 --> 00:20:09.250
We subjected that to the
Turkey earthquake record from

00:20:09.250 --> 00:20:13.669
the near-field. It also as a big pulse.
It also has a permanent displacement

00:20:13.669 --> 00:20:17.159
and offset, but much higher frequency.
So we took motions like this –

00:20:17.159 --> 00:20:22.179
synthetic motions, and we said, if that
was our interstory drift, could the sensor

00:20:22.179 --> 00:20:26.880
measure that in that test bed? So we
imparted that motion via this test bed.

00:20:26.880 --> 00:20:29.510
This was the input motion.
That’s the output motion.

00:20:29.510 --> 00:20:33.690
The sensor could do it. And moreover,
the sensor could to it really, really well.

00:20:33.690 --> 00:20:38.309
So I’m overlaying the imposed
interstory drift, and I’m – with the

00:20:38.309 --> 00:20:43.010
interstory drift measured by the DDPS.
You can see they track very, very

00:20:43.010 --> 00:20:46.890
accurately, both in terms of waveform,
amplitude, and frequency content.

00:20:46.890 --> 00:20:51.080
If you difference those, you can see that,
because of the grid layout we have,

00:20:51.080 --> 00:20:55.380
you can measure that drift within
0.1 centimeters, or a millimeter.

00:20:55.380 --> 00:20:57.740
Really, really good accuracy
of measuring the drift.

00:20:57.740 --> 00:21:02.000
So for both records, you see that.
So that was encouraging.

00:21:02.460 --> 00:21:06.320
So then we asked ourselves
in the laboratory, you know,

00:21:06.320 --> 00:21:09.220
we’ve got to think about applying
this to a representative structure.

00:21:09.220 --> 00:21:14.220
Just looking at, you know,
simple motion is not sufficient.

00:21:14.220 --> 00:21:17.600
That’s really good.
It shows us the inherent sensor works.

00:21:17.600 --> 00:21:20.450
But when you think about applying
it to a structure, all the sudden,

00:21:20.450 --> 00:21:23.570
you have to worry about
a number of other things.

00:21:23.570 --> 00:21:29.820
You know, moving simply uniaxially
this way is only part of the problem.

00:21:29.820 --> 00:21:34.100
When a structure vibrates, there are local
deformations of the structural members.

00:21:34.100 --> 00:21:38.190
And there are local rotations, and so
forth, that you have to account for.

00:21:38.190 --> 00:21:42.240
So we really wanted to construct a
building in the laboratory, so we made

00:21:42.240 --> 00:21:46.250
a simple two-story frame – aluminum
frame with simple cross-sections.

00:21:46.250 --> 00:21:49.600
Wanted it to be flexible.
We had a stepper motor and

00:21:49.600 --> 00:21:51.630
an automatic control
system so we could impart

00:21:51.630 --> 00:21:54.450
earthquake motions to this,
just like we did at the table.

00:21:54.450 --> 00:21:57.530
And then this thing vibrates like a
building, as you’ll see in a minute.

00:21:57.530 --> 00:22:00.900
To get ground truth on drift,
we had two string potentiometers –

00:22:00.900 --> 00:22:06.200
a very high quality – here and here, so
we could – we knew the input motion.

00:22:06.200 --> 00:22:09.620
We knew the displacement here.
We know the displacement here.

00:22:09.620 --> 00:22:11.880
So we can calculate ground
truth for interstory drift.

00:22:11.880 --> 00:22:14.240
We can compare that to
the sensor measurement.

00:22:16.140 --> 00:22:17.779
Here’s part of the problem.

00:22:17.779 --> 00:22:21.880
And this has been something we’ve been
thinking and working at for a while now.

00:22:21.880 --> 00:22:27.970
This is a picture from a finite element
model of a segment – a bay of a frame

00:22:27.970 --> 00:22:31.920
undergoing earthquake excitation that
we’ve popped out and exaggerated.

00:22:31.920 --> 00:22:35.820
If your sensor is mounted here,
you got a little bit of a problem

00:22:35.820 --> 00:22:39.470
because there is a rotation
of the frame at that point.

00:22:39.470 --> 00:22:44.730
So if this was the drift that
you want to measure – the real drift,

00:22:44.730 --> 00:22:49.850
this would be the erroneous
observed drift you have here

00:22:49.850 --> 00:22:54.120
because of this local rotation.
And this gave us fits early on.

00:22:54.120 --> 00:22:58.510
Because this is really a correction
you have to make to that measurement

00:22:58.510 --> 00:23:00.710
if you really want to
measure the overall drift.

00:23:00.710 --> 00:23:03.580
So the way we did that in
our first study is, we said, okay,

00:23:03.580 --> 00:23:07.220
we can use these things.
We’ll have the DDPS measurement here,

00:23:07.220 --> 00:23:10.540
and we’ll put a second one over here.
And we’ll split that laser beam,

00:23:10.540 --> 00:23:12.250
and we’ll measure the
vertical displacement here,

00:23:12.250 --> 00:23:14.309
and we’ll know exactly
what the rotation is.

00:23:14.309 --> 00:23:17.930
So we can account for that.
That worked out great.

00:23:17.930 --> 00:23:20.380
This is an example of
that test of that frame.

00:23:20.380 --> 00:23:24.929
If you look here, you can see that
laser swooping across that sensor.

00:23:24.929 --> 00:23:27.409
This is, I think, El Centro
input ground motion.

00:23:27.409 --> 00:23:32.000
And then we have a second DDPS
sensor here to get that rotation.

00:23:32.600 --> 00:23:35.980
Everything seemed to work really well.
If you look on the bottom,

00:23:35.980 --> 00:23:39.840
you can see that fracted laser beam
sweeping across that array.

00:23:39.840 --> 00:23:44.290
We sample this 384 times a second.
So we have a field-programmable

00:23:44.290 --> 00:23:47.020
gate array that swoops through
and checks every single one

00:23:47.020 --> 00:23:49.090
of those diodes and the
voltage in those diodes,

00:23:49.090 --> 00:23:52.450
and runs it through a comparative
circuit 384 times a second.

00:23:52.450 --> 00:23:55.010
So we’re resolving way higher
frequencies than we need.

00:23:55.010 --> 00:23:58.380
But that sort of shows you how
this thing actually operates.

00:23:58.380 --> 00:24:03.100
So this is data from that test.
Up here, in the upper right-hand corner,

00:24:03.100 --> 00:24:09.991
I’m showing you ground truth versus
sensor measurement if you don’t do

00:24:09.991 --> 00:24:16.360
that correction for the local rotation
of the laser, and that affects that.

00:24:16.360 --> 00:24:19.710
Down here, I’m showing you,
once we measure the rotation,

00:24:19.710 --> 00:24:24.840
this is a comparison of the interstory
drift from the string encoders,

00:24:24.840 --> 00:24:27.870
which is ground truth,
compared to the DDPS sensor.

00:24:27.870 --> 00:24:30.381
And again, really good comparison.
You look at the error.

00:24:30.381 --> 00:24:34.050
It’s on the order of 0.15 centimeters.
Why is that error a little higher?

00:24:34.050 --> 00:24:35.809
Because you have an error in
your rotation measurement

00:24:35.809 --> 00:24:38.320
that you didn’t have when
you just looked at this.

00:24:38.320 --> 00:24:41.570
But I’m telling, if you can measure
to 0.15 millimeters, you’ve sort of

00:24:41.570 --> 00:24:46.450
got the drift problem knocked.
This is – that was El Centro.

00:24:46.450 --> 00:24:50.240
We ran other motions like Landers.
Similar observations.

00:24:50.240 --> 00:24:54.539
So one other thing in our technology
development that we wanted to do is

00:24:54.539 --> 00:24:59.600
we wanted to develop a predictive
capability for these sensor systems.

00:24:59.600 --> 00:25:04.570
We didn’t just want to be reliant
on doing a full-up new experiment

00:25:04.570 --> 00:25:07.670
every time we did something.
So we began to look at the ability

00:25:07.670 --> 00:25:11.159
of finite element models to
predict these sensor performances.

00:25:11.159 --> 00:25:13.710
In other words, if we had a finite
element model of a building or a

00:25:13.710 --> 00:25:18.630
test structure, and we could get
all the local rotations and displacements

00:25:18.630 --> 00:25:22.250
from that, could we use the
finite element model to predict –

00:25:22.250 --> 00:25:24.710
be a predictive tool
for sensor performance?

00:25:24.710 --> 00:25:28.330
So we took that test frame
they built at Chico, we pulled it,

00:25:28.330 --> 00:25:30.760
and we released it,
and let it do a ring-down.

00:25:30.760 --> 00:25:33.860
We got a logarithmic decrement so
we could calibrate our damping.

00:25:33.860 --> 00:25:35.680
And then we created
a finite element model.

00:25:35.680 --> 00:25:41.240
These are pictures of the finite
element model in red predicting

00:25:41.240 --> 00:25:44.770
what the sensors should be
measuring, which is the blue.

00:25:44.770 --> 00:25:48.520
So with no rotation correction,
we can take our computational model,

00:25:48.520 --> 00:25:51.460
and we predict that. And you can
see what the sensor actually did.

00:25:51.460 --> 00:25:54.250
And then you can see
with the rotation correction.

00:25:54.250 --> 00:25:59.179
So that was good. So to be clear,
there are two elements balled up in here.

00:25:59.179 --> 00:26:01.960
Can we use the displacements
and rotations from the

00:26:01.960 --> 00:26:07.200
finite element model as a good surrogate
to predict what the sensor will do?

00:26:07.200 --> 00:26:10.110
And can we actually represent
the dynamics of that frame?

00:26:10.110 --> 00:26:12.830
And the answer to that is, yeah,
we can do a pretty good job on both.

00:26:12.830 --> 00:26:15.669
And I’ll get back to why
that’s important in a minute.

00:26:15.669 --> 00:26:19.500
This is just another record. Same-same.
So we were feeling pretty good.

00:26:19.500 --> 00:26:25.720
As a bonus, you know, even if you
measure deformations in the structure –

00:26:25.720 --> 00:26:27.500
if you think about nuclear
facilities and so forth,

00:26:27.500 --> 00:26:30.520
you’d still like to have the
accelerations in that structure.

00:26:30.520 --> 00:26:32.809
Be able to get the accelerations.
So we asked ourselves,

00:26:32.809 --> 00:26:36.520
sort of as the last step in this
first traunch, is there a way that

00:26:36.520 --> 00:26:38.990
we could use these
deformation measurements

00:26:38.990 --> 00:26:44.000
to back-calculate acceleration?
And so it turns out that you can.

00:26:44.000 --> 00:26:48.600
And if you think about it,
this is – I’ve blown up a portion

00:26:48.600 --> 00:26:54.260
of the time history from this sensor
here in a small deformation piece.

00:26:54.260 --> 00:26:59.120
And if you look at this, these are really –
because of the character of this

00:26:59.120 --> 00:27:04.299
laser sweeping across these diodes,
you really get a quantized time history.

00:27:04.299 --> 00:27:07.809
You get little tiny step functions.
Right? Because it’s jumping.

00:27:07.809 --> 00:27:11.590
You only have finite measurements.
It’s jumping from diode to diode.

00:27:11.590 --> 00:27:14.410
So you have these
little tiny step functions.

00:27:14.410 --> 00:27:16.730
So we want to get rid of those
if we want to – if we want to

00:27:16.730 --> 00:27:20.120
go after accelerations because
those can be associated –

00:27:20.120 --> 00:27:22.580
if you double-differentiate that,
that’s a problem, right?

00:27:22.580 --> 00:27:25.890
You’re going to have these
huge fictitious accelerations.

00:27:25.890 --> 00:27:31.450
So we low-passed those signals.
And then we treated those as

00:27:31.450 --> 00:27:34.730
full analytical signals,
and we double-differentiated them.

00:27:34.730 --> 00:27:38.559
This shows that process.
And it turns out, if you do that carefully,

00:27:38.559 --> 00:27:42.529
this is, from the finite element model,
the acceleration response spectra

00:27:42.529 --> 00:27:44.529
in the structure.
And this is the acceleration

00:27:44.529 --> 00:27:49.260
response spectra calculated from the –
from the – directly from the diode data.

00:27:49.260 --> 00:27:52.970
So it turns out, if you’re careful
about processing that diode data

00:27:52.970 --> 00:27:55.899
and the interstory drift data,
you can double-differentiate that

00:27:55.900 --> 00:27:59.000
and get back to accelerations and
getting the structure accelerations.

00:27:59.000 --> 00:28:01.680
So that’s a –
that’s a bonus point.

00:28:01.690 --> 00:28:05.080
So this was – this was all,
you know, sort of summarized

00:28:05.080 --> 00:28:09.400
in a Earthquake Spectra article
that came out in November 2017,

00:28:09.400 --> 00:28:14.059
and that was sort of the
logical end of that first traunch.

00:28:14.059 --> 00:28:17.799
After that, DOE Office of Nuclear
Safety – we got them interested in this.

00:28:17.799 --> 00:28:22.510
And they supported a much larger,
more realistic test of these sensors

00:28:22.510 --> 00:28:26.690
that I want to speak to
to wrap up this thought process.

00:28:26.690 --> 00:28:30.380
And so we took a
one-third-scale steel frame.

00:28:30.380 --> 00:28:33.160
It was already at the
University of Nevada-Reno.

00:28:33.160 --> 00:28:36.840
And we placed it – we modified it a bit,
and we placed it on their shake table.

00:28:36.840 --> 00:28:39.820
And here’s a – here’s a human to
give you some idea of scale.

00:28:39.820 --> 00:28:43.320
And we kind of
did a similar test.

00:28:43.320 --> 00:28:48.720
Off there, we have a diagnostics tower.
We stretched tension cables from

00:28:48.720 --> 00:28:53.340
that tower with string potentiometers.
Now shown, we also had cables through

00:28:53.340 --> 00:28:57.000
the diagonal so we could have a
second measurement of interstory drift.

00:28:57.000 --> 00:29:00.630
The point is, we instrumented this
large structure now in a way that

00:29:00.630 --> 00:29:06.169
we could get to interstory drift.
Second thing we did is, up to this point,

00:29:06.169 --> 00:29:09.700
we sort of cobbled together
the hardware for this system

00:29:09.700 --> 00:29:12.370
in our work at Chico.
So we had a field-programmable

00:29:12.370 --> 00:29:16.929
gate array, the comparator bank,
separate from the main sensor board,

00:29:16.929 --> 00:29:18.700
separate from
the photodiode array.

00:29:18.700 --> 00:29:21.679
It made a lot of sense because we
were mixing and matching and

00:29:21.679 --> 00:29:26.039
changing things out and trying to vet
this technology and get it to work.

00:29:26.039 --> 00:29:29.140
But for the tests at Reno, we said,
look, you know, that doesn’t

00:29:29.140 --> 00:29:31.760
look like a system [chuckles]
you’d ever think about deploying.

00:29:31.760 --> 00:29:35.320
So we need to begin to think about,
what would the form factor

00:29:35.320 --> 00:29:38.130
and the construction of a real,
deployable sensor look like?

00:29:38.130 --> 00:29:42.559
So we moved all that stuff
to a single board at Chico.

00:29:42.560 --> 00:29:47.180
And so now, the microcontroller, the
field-programmable gate array, all of the

00:29:47.180 --> 00:29:52.040
diodes, the comparator bank, everything
is on a single board that’s about 9 inches.

00:29:52.050 --> 00:29:55.730
And I brought one with me. And this
is a really light, flimsy little thing.

00:29:55.730 --> 00:29:59.650
If you have interest, you can look at it.
So we migrated this to something

00:29:59.650 --> 00:30:03.840
that might begin to look like
something you would deploy.

00:30:03.840 --> 00:30:07.710
I would argue there’s a lot
we could do here even beyond this.

00:30:07.710 --> 00:30:09.860
You don’t need a
comparator bank anymore.

00:30:09.860 --> 00:30:13.370
You can do all these comparators on
a chip that can handle multi-circuits.

00:30:13.370 --> 00:30:16.720
So all this real estate can go away,
and this thing could get much, much

00:30:16.720 --> 00:30:20.539
narrower. And so you could really get
down to a compact form function.

00:30:20.539 --> 00:30:22.600
Nevada test setup –
this just shows you

00:30:22.600 --> 00:30:25.590
we were still measuring
rotations at that time.

00:30:25.590 --> 00:30:28.980
So the laser is here, splits,
there’s a sensor here, sensor here.

00:30:28.980 --> 00:30:32.570
We had the tension cables
here and here to get ground truth.

00:30:32.570 --> 00:30:36.100
And we really shook this thing.
When we did this experiment at Nevada,

00:30:36.100 --> 00:30:40.890
we were on a fairly tight schedule,
so we set aside one day, and we took

00:30:40.890 --> 00:30:44.760
a large number of earthquakes,
and we started them at low scale

00:30:44.760 --> 00:30:46.620
and scaled them up to
a very high amplitude.

00:30:46.620 --> 00:30:50.140
And we ran all those tests back to
back to back in about eight hours.

00:30:50.140 --> 00:30:52.409
They had their data
acquisition system all set up,

00:30:52.409 --> 00:30:54.950
which, if you think about it,
was kind of realistic because,

00:30:54.950 --> 00:30:58.549
in a real environment, you have an
earthquake, and you have aftershocks.

00:30:58.549 --> 00:31:03.090
And one of the things we didn’t know
going in is whether this was going to

00:31:03.090 --> 00:31:06.600
be adversely affected by
really, really strong shaking.

00:31:06.600 --> 00:31:10.440
Was this thing going to have some
weird dynamics or some failure mode

00:31:10.440 --> 00:31:14.440
that we didn’t anticipate or weren’t
aware of? Turns out, it didn’t.

00:31:14.440 --> 00:31:17.800
We just slaughtered the test right
straight through in one day on this.

00:31:17.800 --> 00:31:22.799
We used El Centro and Rinaldi motions.
Rinaldi has some near-field pulses.

00:31:22.799 --> 00:31:25.960
And so these are actually what –
we’re working on these right now

00:31:25.960 --> 00:31:30.200
in a second paper.
These are the drift measurements.

00:31:30.200 --> 00:31:34.490
The ground truth versus the DDPS.
You can see the interstory drift at

00:31:34.490 --> 00:31:38.110
each floor, much like in
our previous test, looks good.

00:31:38.110 --> 00:31:42.269
Rinaldi looks good.
We even took it up to 250% of

00:31:42.269 --> 00:31:47.320
El Centro shaking, which [chuckles]
the frame was just rattling like crazy.

00:31:47.320 --> 00:31:50.700
Because we wanted to see if
we encountered anything untoward

00:31:50.700 --> 00:31:53.580
in the sensor for that.
Nothing. Just plowed right through it.

00:31:53.580 --> 00:31:55.480
This was actually easier.

00:31:55.480 --> 00:32:00.860
This one-third steel frame was built for a
different purpose, so it was awfully stiff.

00:32:00.860 --> 00:32:06.990
And so the 250% El Centro actually
was giving us more realistic motions.

00:32:06.990 --> 00:32:12.340
If you look at the error in this thing,
again, this is the 250 motions –

00:32:12.340 --> 00:32:15.780
250% interstory drifts.
This is the difference.

00:32:15.780 --> 00:32:18.539
And this is the, quote,
error in the system.

00:32:18.539 --> 00:32:23.039
You can see the error here is on the
order of 2, 3, or 4 millimeters.

00:32:23.039 --> 00:32:27.029
And it’s a little higher than what
we observed in the small-scale test.

00:32:27.029 --> 00:32:30.510
And I believe that’s due to the
inaccuracy in the measurement

00:32:30.510 --> 00:32:34.170
system for the ground truth here.
These are very, very long cables,

00:32:34.170 --> 00:32:38.309
and they’re not as – quite as
high-quality potentiometers.

00:32:38.309 --> 00:32:41.830
And so this is both sensor error
as well as measurement system error

00:32:41.830 --> 00:32:45.929
of the ground truth.
So still, if we can measure interstory

00:32:45.929 --> 00:32:50.370
drift within 4 millimeters, you know,
we’ve got it. We’re doing good.

00:32:50.370 --> 00:32:53.760
So what we’re working on now,
just to sort of move to a couple things

00:32:53.760 --> 00:32:57.220
to wrap up here, and then we can
take some questions, is we want to

00:32:57.220 --> 00:32:59.880
get rid of that
second line of sight.

00:32:59.880 --> 00:33:04.940
That’s a thorn in our side
to correct for that rotation.

00:33:04.940 --> 00:33:07.920
It’s pretty easy to get a vertical
line of sight in a building.

00:33:07.920 --> 00:33:10.710
It’s not so easy unless you’re in an
industrial facility, to always get

00:33:10.710 --> 00:33:13.770
that second horizontal line of sight.
So we wanted to get rid of that.

00:33:13.770 --> 00:33:16.710
So we’ve been going through –
because we’ve built up confidence

00:33:16.710 --> 00:33:20.399
in our computational model to
be predictive to these sensors,

00:33:20.399 --> 00:33:22.470
we’re looking at different
mounting configurations.

00:33:22.470 --> 00:33:28.159
We’re looking at Alternative 1, 2, and 3.
So we ran this numerical with no rotation

00:33:28.159 --> 00:33:33.240
correction to try and understand
the sensitivity to the local rotations.

00:33:33.240 --> 00:33:37.400
And much as we observed in our
experiments, alternative number 1,

00:33:37.400 --> 00:33:43.529
you can see that there is a big error
between the actual and the, quote,

00:33:43.529 --> 00:33:48.049
simulated, or what we would say
that the sensor would actually see.

00:33:48.049 --> 00:33:49.950
Because that’s the
error we had to correct.

00:33:49.950 --> 00:33:54.460
But if you move across to alternative
locations, you begin to see that

00:33:54.460 --> 00:33:58.519
error asymptotically goes
away in certain locations.

00:33:58.519 --> 00:34:03.750
And the reason being is, if you go to
the 1/4-point on that structural member,

00:34:03.750 --> 00:34:05.860
that thing really
doesn’t have any rotation.

00:34:05.860 --> 00:34:10.660
It has simply a vertical up and down
as that building wags back and forth.

00:34:10.660 --> 00:34:16.040
You know, points here
and here have big rotations.

00:34:16.040 --> 00:34:20.280
Those rotations go to zero as you get
to this idealized L-over-4 point.

00:34:20.280 --> 00:34:22.910
And so you have no rotation there,
and you just sort of nail it.

00:34:22.910 --> 00:34:27.800
So that’s one thing we’re interested in.
We’ve also looked at providing a

00:34:27.800 --> 00:34:33.060
pin-pin strut to mount these things on.
And by that, I just mean a little

00:34:33.060 --> 00:34:38.230
mechanism that’s about yeah-big
that’s just a metal bar that has

00:34:38.230 --> 00:34:41.240
pin-pin connections at the –
at the joints of that.

00:34:41.240 --> 00:34:46.530
And that would isolate that mount
of the laser from the local rotations.

00:34:46.530 --> 00:34:50.060
And so we’ve done numerical studies of
that too, and that works wonderfully.

00:34:50.060 --> 00:34:53.710
And so the point is, we think we’ve
got this issue of having to correct for

00:34:53.710 --> 00:34:57.390
these rotations – we have practical ways
of getting around that, I believe.

00:34:57.390 --> 00:35:01.200
So – and that’s sort of what we’re
working on and finishing up right now.

00:35:01.200 --> 00:35:04.710
So let me sort of get sort of to
the punch – a couple punchlines here

00:35:04.710 --> 00:35:08.790
to leave you a thought with.
So here’s how we view the end state.

00:35:08.790 --> 00:35:11.730
You have critical facilities –
and we’re focused on DOE facilities –

00:35:11.730 --> 00:35:14.900
a lot of nuclear facilities.
You would like to develop

00:35:14.900 --> 00:35:18.870
understanding through your analysis
and your engineering design of

00:35:18.870 --> 00:35:22.220
how that particular structure behaves
and what the limit states are

00:35:22.220 --> 00:35:24.610
for that structure
in terms of drift.

00:35:24.610 --> 00:35:28.560
So you have those in hand
and in your pocket a priori.

00:35:28.560 --> 00:35:31.230
You have an earthquake.
You measure – in real time,

00:35:31.230 --> 00:35:34.540
you measure those drifts with
a sensor like we’ve showed you.

00:35:34.540 --> 00:35:38.170
So you have both a transient time
history, the peak dynamic drift,

00:35:38.170 --> 00:35:41.130
as well as the residual drift.
And you compare that –

00:35:41.130 --> 00:35:44.940
immediately can compare that,
in a stoplight chart sense,

00:35:44.940 --> 00:35:48.510
to all of your codes and standards
that have established these drift limits.

00:35:48.510 --> 00:35:50.890
And that would really be
the way we would envision

00:35:50.890 --> 00:35:53.720
utilizing this technology ultimately.
So if you have an earthquake,

00:35:53.720 --> 00:35:57.110
either go out to the building
and look at a stoplight chart,

00:35:57.110 --> 00:36:00.840
or have your smartphone that this
data is piped to so you would

00:36:00.840 --> 00:36:03.650
immediately know if that
building exceeded the limit states.

00:36:03.650 --> 00:36:07.020
And that’s sort of the thing we’re
really – we’re really working towards.

00:36:07.020 --> 00:36:10.950
Couple of corollary things here.
And this has to do with

00:36:10.950 --> 00:36:15.710
both accelerometers
as well as our sensors.

00:36:15.710 --> 00:36:21.600
We’ve done a lot of numerical
look at what drifts – the drift profiles

00:36:21.600 --> 00:36:26.500
look like in critical structures for
realistic earthquake ground motions.

00:36:26.500 --> 00:36:29.550
And it turns out, in our view,
they’re extremely complex.

00:36:29.550 --> 00:36:33.700
It’s a little hard to see, but we
had a very, very detailed model

00:36:33.700 --> 00:36:38.120
of a tall steel frame structure
that Professor Astaneh designed –

00:36:38.120 --> 00:36:43.340
he came up with. And we put a whole
large suite of near-field records in that.

00:36:43.340 --> 00:36:48.900
And what is shown in the black bars are
the peak interstory drifts at every story

00:36:48.900 --> 00:36:52.100
for the entire duration of that
earthquake versus what you would

00:36:52.100 --> 00:36:56.100
see in a linear model –
a simple linear-elastic model.

00:36:56.100 --> 00:37:00.550
And this is a comparison of the two.
And what you see is that the

00:37:00.550 --> 00:37:06.820
large inelastic drifts can tend to be
very localized, and they can tend to

00:37:06.820 --> 00:37:12.320
be much larger than – and much more
distributed and complex than the profile

00:37:12.320 --> 00:37:16.010
you get from a simple linear model.
So this has important implications

00:37:16.010 --> 00:37:21.180
for how you instrument these buildings.
Because it’s giving you a hint that

00:37:21.180 --> 00:37:26.100
sparse instrumentation
is probably not very good.

00:37:26.100 --> 00:37:29.320
You probably are going to have
fairly dense instrumentation arrays

00:37:29.320 --> 00:37:33.140
on these types of building structures.
And so this is something else that

00:37:33.140 --> 00:37:36.270
we’re thinking about in terms of
moving forward with deployment.

00:37:36.270 --> 00:37:41.000
So finally, we are – there are
two things we want to do next.

00:37:41.000 --> 00:37:47.400
We want to reconfigure this sensor and
its platform onto a smaller form factor.

00:37:47.400 --> 00:37:49.340
And we’re looking for
deployment opportunities

00:37:49.340 --> 00:37:52.610
to actually put this on real structures.
And so we are working with

00:37:52.610 --> 00:37:57.600
UCSF and the Parnassus campus.
They have a couple of tall buildings

00:37:57.600 --> 00:38:01.340
that they’re retrofitting, and we would
hope to start to begin to deploy these

00:38:01.340 --> 00:38:05.720
types of sensors for the first time
as part of their retrofit of their

00:38:05.720 --> 00:38:08.450
critical structures and be able to
demonstrate the practicality

00:38:08.450 --> 00:38:13.350
over long-term with monitoring
with this type of sensor.

00:38:13.350 --> 00:38:16.670
So I will stop there,
and maybe we have

00:38:16.670 --> 00:38:18.860
a few minutes for
questions. I don’t know.

00:38:19.280 --> 00:38:25.160
[Applause]

00:38:28.660 --> 00:38:31.560
- [inaudible] with the first question?

00:38:34.100 --> 00:38:41.060
[Silence]

00:38:41.580 --> 00:38:45.320
- Thanks for a super-interesting talk.
Have you thought about

00:38:45.330 --> 00:38:49.510
MEMS gyroscopes for
measuring the local rotations?

00:38:49.510 --> 00:38:52.880
- You know,
we did think about that.

00:38:53.640 --> 00:38:58.280
We’ve looked at a number of ways of
trying to think about the local rotation.

00:38:58.280 --> 00:39:04.020
And what we kept running into is the
long-term reliability in drift of some of

00:39:04.020 --> 00:39:08.760
these things and their ability to continue
to measure those rotations accurately.

00:39:08.760 --> 00:39:14.320
So we have gone down a path
currently – and not saying we’ll

00:39:14.320 --> 00:39:17.850
always stay there – look,
we’d love to have a point rotation

00:39:17.850 --> 00:39:20.060
measurement system.
No question.

00:39:20.060 --> 00:39:24.740
I mean, that would – I mean, that
would make this just lickety-split, right?

00:39:24.740 --> 00:39:29.200
From a practical standpoint, we have
not run across – although you can do it

00:39:29.200 --> 00:39:33.540
conceptually – a good robust solution
for measuring those local rotations.

00:39:33.540 --> 00:39:35.821
If you have better ideas,
we’d love to hear about them,

00:39:35.821 --> 00:39:38.210
but we pulled on
a number of threads.

00:39:38.210 --> 00:39:41.920
And it turns out that, with a lot of
those gyros, they drift, and people

00:39:41.920 --> 00:39:45.619
have to apply a Kalman filter,
and nah, nah, nah, nah, nah.

00:39:45.619 --> 00:39:48.050
And so we have gone the
path of trying to figure out

00:39:48.050 --> 00:39:51.700
a way to not have to
make those rotations.

00:39:51.700 --> 00:39:55.480
We looked at a number
of MEMS inclinometers.

00:39:55.480 --> 00:39:58.780
They’re all essentially
gravity-based.

00:39:58.790 --> 00:40:01.970
And so you impart the additional
accelerations from an earthquake,

00:40:01.970 --> 00:40:04.580
and they just – you know,
you try to think about

00:40:04.580 --> 00:40:06.830
how you could filter,
blah, blah, blah.

00:40:06.830 --> 00:40:10.650
We just couldn’t get to a good solution.
It was a dead end for us.

00:40:10.650 --> 00:40:14.590
But it’s a great idea, and if the
technology comes along the pike that

00:40:14.590 --> 00:40:19.650
solves that, we’d love to have a point
rotation sensor measurement. [laughs]

00:40:19.650 --> 00:40:21.260
Good question.

00:40:24.900 --> 00:40:28.340
- [inaudible]
- No, no, you can go.

00:40:28.340 --> 00:40:32.670
- David, what are the prospects
of using this for tall buildings?

00:40:32.670 --> 00:40:35.640
- I think – I think it’s
great for tall buildings.

00:40:35.640 --> 00:40:40.760
- But I don’t mean interstory drift.
I mean, an average for the whole height.

00:40:41.480 --> 00:40:44.360
- An average –
an average drift?

00:40:45.620 --> 00:40:49.720
I think – you may have stepped out.
I showed some drift profiles.

00:40:49.720 --> 00:40:52.640
I think – I think you
want this on every story.

00:40:52.650 --> 00:40:55.700
You want a cost function that
allows you to put this everywhere.

00:40:55.700 --> 00:40:57.810
Because when you begin to
think of inelastic deformation

00:40:57.810 --> 00:41:03.070
under a large event, you get
very, very localized large drifts.

00:41:03.070 --> 00:41:06.530
And so I worry about drift averaging.
And you can see – Mehmet,

00:41:06.530 --> 00:41:11.100
we can talk about this this afternoon.
This is a 40-story steel building

00:41:11.100 --> 00:41:16.340
subjected to near-field motions.
If you look in the linear-elastic analysis,

00:41:16.340 --> 00:41:20.230
you tend to get nice, smooth
drift profiles to some degree.

00:41:20.230 --> 00:41:22.690
When you begin to look,
particularly at near-field records with

00:41:22.690 --> 00:41:29.610
very, very large motions, you really
tend to drive very localized high drifts.

00:41:29.610 --> 00:41:34.490
And if you don’t have a dense
enough array to capture some of those,

00:41:34.490 --> 00:41:39.100
you know, it’s really, really tough.
So, in my mind, we’d like to drive

00:41:39.100 --> 00:41:42.440
the cost function to a point where you
could afford to put these in a dense

00:41:42.440 --> 00:41:46.060
array throughout the structure.
That’s my thinking.

00:41:46.060 --> 00:41:53.200
- [inaudible]
I think the main – I think the

00:41:53.200 --> 00:41:59.400
main problem in installing these
is the construction companies.

00:41:59.400 --> 00:42:05.800
- Mm-hmm.
- Because they almost never allow

00:42:05.800 --> 00:42:09.520
people like you or me [feedback noise]
to go in there and try to install

00:42:09.520 --> 00:42:12.520
such things in a bay of a …
- Yes.

00:42:12.520 --> 00:42:15.740
- Forty-third floor [inaudible], 57th floor.
- Yes.

00:42:15.740 --> 00:42:19.120
- They will think that
you are in [inaudible].

00:42:19.120 --> 00:42:21.320
- Yeah.
- So that is realistic?

00:42:21.320 --> 00:42:23.400
- I think it is.
And we’re having that

00:42:23.400 --> 00:42:26.850
kind of discussion around
this building right now.

00:42:26.850 --> 00:42:31.830
And so we’re trying to think about how
to make the installation as least-obtrusive

00:42:31.830 --> 00:42:35.950
as we can. And so I think we’re
having those kind of discussions.

00:42:35.950 --> 00:42:39.580
We’ll see how that plays out.
But I do think we have a shot

00:42:39.580 --> 00:42:43.620
at being able to mount these
things in certain locations.

00:42:43.620 --> 00:42:46.900
So I think – my answer is,
I think we can do it.

00:42:50.740 --> 00:42:56.380
- I have a couple of questions.
The first is, based on all the test results,

00:42:56.390 --> 00:43:01.400
and so it looks like you only look
unidirectional, right? All the testing?

00:43:01.400 --> 00:43:05.690
So your slip is longer in the one way,
but it’s shorter in the other way.

00:43:05.690 --> 00:43:09.600
But the structure is – you know, during
earthquakes, they’re going both ways.

00:43:09.600 --> 00:43:13.920
You think you can – your laser is
going to be out of the receptors?

00:43:13.920 --> 00:43:21.020
- No. So the tests at Reno were biaxial.
The only problem with those tests is

00:43:21.030 --> 00:43:26.350
they really didn’t exercise the biaxial
nature because the frame was stronger

00:43:26.350 --> 00:43:31.140
in one direction than the other direction.
But we have made this beam large

00:43:31.140 --> 00:43:36.480
enough to accommodate the out-of-plane
deformations you get in the

00:43:36.480 --> 00:43:39.050
orthogonal direction.
That’s the idea.

00:43:39.050 --> 00:43:41.940
And I’m confident, based
on our analysis and our –

00:43:41.940 --> 00:43:48.120
what we observed at Reno that these –
you can split and defrac that beam –

00:43:48.120 --> 00:43:52.500
let me go back – like we see here.
This is – it’s actually bigger than that

00:43:52.510 --> 00:43:56.620
so that when you get that out-of-plane
deformation, this will translate

00:43:56.620 --> 00:43:59.210
this way some, but it’s not going to
go off the [inaudible].

00:43:59.210 --> 00:44:02.160
That’s the intent.
That’s the design intent.

00:44:02.160 --> 00:44:05.210
And so the design intent
would be to have one of these

00:44:05.210 --> 00:44:08.000
in orthogonal directions
on the building structure.

00:44:08.000 --> 00:44:11.800
And you’re measuring the
in-plane motion in one direction.

00:44:11.800 --> 00:44:16.040
- Well, my next question is,
how are you seeing the production?

00:44:16.640 --> 00:44:21.560
I’m assuming that you cannot use that
in – when it’s visible, so it should be,

00:44:21.560 --> 00:44:23.380
you know, within the partitions?
- Yeah.

00:44:23.380 --> 00:44:24.830
- Of the building?
- Yeah.

00:44:24.830 --> 00:44:28.140
- So I’m assuming you shouldn’t
have any braces like that on the –

00:44:28.140 --> 00:44:30.980
on the corner.
So you should have straight sides.

00:44:30.980 --> 00:44:34.590
- Yes. You …
- How about the long-term dust?

00:44:34.590 --> 00:44:36.430
The buildings …
- That’s a good question.

00:44:36.430 --> 00:44:39.770
We don’t know about that yet.
And so that’s why we want to

00:44:39.770 --> 00:44:42.580
do some long-term deployment.
We don’t know whether you’re going to

00:44:42.580 --> 00:44:47.760
have to clean these things off once a
month or once a year or whatever.

00:44:47.760 --> 00:44:50.170
I mean, those are the kind of reliability
things we’re going to have to look at.

00:44:50.170 --> 00:44:54.520
I will say – you know, this thing will
not go into the field looking like this.

00:44:54.520 --> 00:44:58.120
There will be some
sort of a window over this.

00:44:58.120 --> 00:45:02.580
And, for my laser jock friends at
Livermore, you can put a coating

00:45:02.580 --> 00:45:07.250
on that that is very selective
in terms of the wavelength of light

00:45:07.250 --> 00:45:09.970
that it will let through.
So there are things you can do

00:45:09.970 --> 00:45:14.410
to make that cover work
very, very well for the laser

00:45:14.410 --> 00:45:16.850
and not other kinds of light,
for example, right?

00:45:16.850 --> 00:45:20.050
Those are the kind of things we’ve still
got to monkey with a bit. Absolutely.

00:45:20.050 --> 00:45:23.810
It’s a good question.
Which we’ve thought a little bit about,

00:45:23.810 --> 00:45:27.220
and we just don’t know until we
have some long-term deployments.

00:45:32.580 --> 00:45:37.700
- Yeah. I have a question about –
you showed some examples

00:45:37.700 --> 00:45:45.240
comparing the finite element model
acceleration with the data.

00:45:45.240 --> 00:45:48.180
- Yep.
- But have you done any –

00:45:48.180 --> 00:45:52.450
did any of the tests that you do
also have a conventional, you know,

00:45:52.450 --> 00:45:55.460
force-balance accelerometer?
- This one did.

00:45:55.460 --> 00:45:58.190
And so we are going to compare
the accelerations on this.

00:45:58.190 --> 00:46:00.960
We’re doing that – about to do that now.
- Okay.

00:46:00.960 --> 00:46:04.480
- So we are going – at each floor level,
we had accelerometers on this.

00:46:04.480 --> 00:46:06.000
- Gotcha.
- So we are going to

00:46:06.010 --> 00:46:09.880
go through the same exercise.
That’s sort of our last task

00:46:09.880 --> 00:46:13.670
in this new paper is to
go back and repeat

00:46:13.670 --> 00:46:16.860
that acceleration where
we have acceleration data.

00:46:16.860 --> 00:46:21.050
- And then, one other question.
- You know, just – I’m sorry to –

00:46:21.050 --> 00:46:24.680
when we did the first experiment,
thinking of using these to

00:46:24.680 --> 00:46:27.460
get acceleration was not in
our consciousness, right?

00:46:27.460 --> 00:46:30.860
So we didn’t put the accelerometers on.
And, you know, afterwards, we’re, like,

00:46:30.860 --> 00:46:34.380
you know, god, we could probably use
this more clever to get accelerations.

00:46:34.380 --> 00:46:37.470
And so the only avenue we had
at that point was to – was to really

00:46:37.470 --> 00:46:39.570
look at the accelerations from
the finite element model,

00:46:39.570 --> 00:46:42.550
which is why we did that.
- Yeah.

00:46:42.550 --> 00:46:45.330
What about six degree of freedom,
so adding the vertical in,

00:46:45.330 --> 00:46:50.020
and how do you think that will
affect things like your Alternative 2

00:46:50.020 --> 00:46:54.180
and it being a quarter of the
way down, or just in general,

00:46:54.180 --> 00:46:57.400
how it will affect the …
- Adding the vertical acceleration?

00:46:57.400 --> 00:47:00.690
- Yeah. And rocking and …
- So the only thing I would say

00:47:00.690 --> 00:47:03.890
that worries me about the
vertical acceleration is

00:47:03.890 --> 00:47:08.840
we’re thinking of using a
pin-pin strut to mount the laser on.

00:47:08.840 --> 00:47:11.250
And the reason we’re doing that is
you have the local deformation

00:47:11.250 --> 00:47:16.780
of the structure. We’d like to have
a fairly long, you know, simple pin-pin

00:47:16.780 --> 00:47:20.420
strut that holds the laser.
You don’t want that strut

00:47:20.420 --> 00:47:23.740
to vibrate under vertical
or other accelerations, right?

00:47:23.740 --> 00:47:27.440
So you have to put very careful
design considerations into that.

00:47:27.440 --> 00:47:30.530
So you’d want to push that
frequency up to – out of the

00:47:30.530 --> 00:47:32.620
frequency range of
response in the system.

00:47:32.620 --> 00:47:37.360
And that consideration, I think,
is important for that very issue.

00:47:37.360 --> 00:47:40.320
But other than that,
as long as you’re careful

00:47:40.320 --> 00:47:43.440
how you design that,
I think you’re going to be okay.

00:47:46.000 --> 00:47:48.260
- What do you mean
by a pin-pin strut?

00:47:48.260 --> 00:47:54.300
- So I mean a strut that has a –
pin-pin is an engineering term.

00:47:54.300 --> 00:47:57.130
You would – you would have a
strut – just think of it as a bar –

00:47:57.130 --> 00:48:01.070
where you have a connection at
the end, and it’s like two ball joints.

00:48:01.070 --> 00:48:06.600
So that when the structure deforms
and rotates, that bar doesn’t rotate.

00:48:06.600 --> 00:48:09.360
It translates.
It just moves like this with

00:48:09.360 --> 00:48:12.500
the laser on it, but it does
not have a moment in it

00:48:12.500 --> 00:48:16.040
or a – a moment,
so it doesn’t deform.

00:48:16.040 --> 00:48:20.310
So I didn’t bring those results, but we’ve
done numerical studies of that as well.

00:48:20.310 --> 00:48:23.080
And that looks like
a perfectly good concept.

00:48:23.080 --> 00:48:26.570
The thing that’s nice about that
concept is it can be migrated to

00:48:26.570 --> 00:48:29.790
other types of structures,
like a shear wall structure

00:48:29.790 --> 00:48:32.400
and combination moment frame
and shear wall structures.

00:48:32.400 --> 00:48:35.520
So we’re big on that concept,
and we’re just sort of finishing up

00:48:35.520 --> 00:48:38.170
the sensitivity to that
right now as part of this.

00:48:38.170 --> 00:48:43.020
Does that answer your question?
It’s simply – just think of it as a bar

00:48:43.020 --> 00:48:47.210
with two ball joints at the end
so that you can’t rotate from

00:48:47.210 --> 00:48:52.040
the structure any moment or rotation
or bending into that strut.

00:48:52.040 --> 00:48:56.200
- Thanks.
- It’s really simple. It’s a simple concept.

00:48:58.020 --> 00:49:03.860
- I was concerned with how quickly
you glossed over the possibility that,

00:49:03.860 --> 00:49:09.220
at the half – halfway along a beam,
something is zero, and at a quarter way

00:49:09.220 --> 00:49:12.600
along the curvature or the
bending, moment is zero.

00:49:12.600 --> 00:49:17.180
This is going to depend – I mean,
these precise locations of these are

00:49:17.190 --> 00:49:22.200
going to depend on the loading
on the beams and also whether or not

00:49:22.200 --> 00:49:29.560
the column itself in the – in the story
height you’re considering are doing the –

00:49:29.560 --> 00:49:32.020
these columns doing
the same double curvature.

00:49:32.020 --> 00:49:33.800
- Yes.

00:49:33.800 --> 00:49:37.450
- I don’t think it’s very important
for your whole story, but …

00:49:37.450 --> 00:49:39.940
- Yeah. So …
- What do you think?

00:49:39.940 --> 00:49:43.820
- So I think this is
certainly an idealization.

00:49:43.820 --> 00:49:47.940
The question would be, and this ought to
be validated with experiment, you know,

00:49:47.940 --> 00:49:52.260
if you think about – the issue
I would be worried about is

00:49:52.260 --> 00:49:59.610
changes in load on that structure.
But during dynamic motion, you know,

00:49:59.610 --> 00:50:02.460
these lasers would be mounted,
after all the dead load and everything

00:50:02.460 --> 00:50:07.240
is deformed, that does want to have
the deformation pattern that

00:50:07.240 --> 00:50:10.930
we’re observing – that beam.
The question is, if there’s any

00:50:10.930 --> 00:50:13.900
variability along the length,
how precise is that going to be?

00:50:13.900 --> 00:50:15.920
And is it going to be
exactly at the quarter point,

00:50:15.920 --> 00:50:20.090
or is it going to be 10% over
from the quarter point?

00:50:20.090 --> 00:50:23.880
So that’s why the strut is a –
is a very robust, we think,

00:50:23.880 --> 00:50:27.070
appealing way to mount the laser.
Because you would get around

00:50:27.070 --> 00:50:30.240
those types of considerations
that you’re describing.

00:50:30.240 --> 00:50:31.630
But we don’t know yet.
We’re going to –

00:50:31.630 --> 00:50:35.100
we’ve seen this numerically,
and we need to do some experimentation

00:50:35.100 --> 00:50:39.730
that really tests this concept out.
But you raise a good question.

00:50:39.730 --> 00:50:45.070
I mean, the point is, if you’re exactly
at Alt 2, you know, is the –

00:50:45.070 --> 00:50:47.561
is the displacement going
to be precisely vertical there?

00:50:47.561 --> 00:50:50.300
Or is there going to be
a little, tiny bit of rotation?

00:50:50.300 --> 00:50:52.940
And there could be.
There very well could be.

00:50:52.940 --> 00:50:56.580
It’s going to be a hell of a lot better than
putting it at Alt 1 and Alt 3. [laughs]

00:50:56.590 --> 00:50:58.930
But whether it’s going to be
as good as our numerical simulation

00:50:58.930 --> 00:51:01.640
tells us or not,
not sure.

00:51:06.700 --> 00:51:12.540
- Anyone else before we take off?
No? Well, thank you again, David.

00:51:12.540 --> 00:51:14.850
- Thank you for your time.
- If you’d like to meet with him

00:51:14.850 --> 00:51:17.380
before he takes off for the day,
please get a hold of me.

00:51:17.380 --> 00:51:19.440
And I think there’s
some time in the afternoon.

00:51:19.440 --> 00:51:22.220
But otherwise, thanks for coming,
and thanks for the great talk.

00:51:22.220 --> 00:51:24.040
- Oh, thank you for having me.
I appreciate it.

00:51:24.040 --> 00:51:28.420
[Applause]

00:51:31.660 --> 00:51:34.760
- Take that mic whenever you’re ready.
- Oh, yes, I don’t …