WEBVTT

00:00:00.000 --> 00:00:02.430
The following content is
provided under a Creative

00:00:02.430 --> 00:00:03.730
Commons license.

00:00:03.730 --> 00:00:06.030
Your support will help
MIT OpenCourseWare

00:00:06.030 --> 00:00:10.060
continue to offer high quality
educational resources for free.

00:00:10.060 --> 00:00:12.660
To make a donation or to
view additional materials

00:00:12.660 --> 00:00:16.560
from hundreds of MIT courses,
visit MIT OpenCourseWare

00:00:16.560 --> 00:00:17.874
at ocw.mit.edu.

00:00:20.563 --> 00:00:22.730
DAVID HARDT: My name is
David Hardt for those of you

00:00:22.730 --> 00:00:24.140
who I haven't met yet.

00:00:24.140 --> 00:00:27.620
And I'm filling in
for Duane Boning

00:00:27.620 --> 00:00:30.755
today or sitting in or
guest lecturing, whatever

00:00:30.755 --> 00:00:31.630
you want to call it--

00:00:31.630 --> 00:00:32.070
AUDIENCE: You're
not guest lecturing.

00:00:32.070 --> 00:00:32.570
You're co-lecturing.

00:00:32.570 --> 00:00:33.530
DAVID HARDT: Oh, yeah,
I'm a co-lecturer.

00:00:33.530 --> 00:00:33.850
Yeah.

00:00:33.850 --> 00:00:34.460
Yeah, that's right.

00:00:34.460 --> 00:00:36.005
No, I'm guest
lecturing because--

00:00:36.005 --> 00:00:38.190
you know the story.

00:00:38.190 --> 00:00:46.580
So I think my role today is
to in some ways repeat what

00:00:46.580 --> 00:00:49.010
Duane Boning has talked about
in the first two lectures

00:00:49.010 --> 00:00:52.040
from a little bit more of
a mechanical engineering

00:00:52.040 --> 00:00:52.610
perspective.

00:00:52.610 --> 00:00:56.570
But I'm going to also
throw in some things,

00:00:56.570 --> 00:01:00.140
you might just say a personal
perspective on the problem.

00:01:00.140 --> 00:01:05.030
And I've got too much to
try and do in one lecture,

00:01:05.030 --> 00:01:07.790
but that's sort of
typical of me anyway.

00:01:07.790 --> 00:01:09.950
But the notes are
on the web if you

00:01:09.950 --> 00:01:12.643
want for the things
I have to skip over.

00:01:12.643 --> 00:01:14.060
Now, the first
part of the lecture

00:01:14.060 --> 00:01:16.790
is to notice this scene.

00:01:16.790 --> 00:01:18.483
And someone was
pointing out to me

00:01:18.483 --> 00:01:20.900
the other day, said, it must
be tough for you this summer,

00:01:20.900 --> 00:01:25.130
because you go between
these extreme environments.

00:01:25.130 --> 00:01:28.242
So I found this picture of
what I call a tree tunnel.

00:01:28.242 --> 00:01:30.200
Does anybody know where
this picture was taken?

00:01:33.120 --> 00:01:34.772
In January.

00:01:34.772 --> 00:01:37.480
AUDIENCE: [INAUDIBLE]

00:01:37.480 --> 00:01:39.580
DAVID HARDT: It wasn't at MIT?

00:01:39.580 --> 00:01:40.660
It's a little washed out.

00:01:40.660 --> 00:01:44.500
So you can't see what these
things are in the distance.

00:01:44.500 --> 00:01:46.600
If you saw those you'd
probably know in an instant

00:01:46.600 --> 00:01:48.004
if you know--

00:01:48.004 --> 00:01:49.300
AUDIENCE: Barges?

00:01:49.300 --> 00:01:51.040
DAVID HARDT:
Anybody in Singapore

00:01:51.040 --> 00:01:54.600
know what that picture
might be taken?

00:01:54.600 --> 00:01:57.740
Oh, come on.

00:01:57.740 --> 00:02:01.292
All right, where do you
think this picture was taken?

00:02:01.292 --> 00:02:02.240
AUDIENCE: Outside.

00:02:02.240 --> 00:02:03.782
DAVID HARDT: All
right, OK, so here's

00:02:03.782 --> 00:02:08.660
a tree tunnel in Acton,
Massachusetts, in January.

00:02:08.660 --> 00:02:11.390
And here's a tree
tunnel on Sentosa island

00:02:11.390 --> 00:02:15.500
in Singapore in January,
just about the same time,

00:02:15.500 --> 00:02:16.290
same kind of view.

00:02:16.290 --> 00:02:18.740
The difference is that
those are probably

00:02:18.740 --> 00:02:21.410
the second largest collection
of container ships in the world,

00:02:21.410 --> 00:02:26.540
something like that, which is
the main landscape feature off

00:02:26.540 --> 00:02:28.880
of that coast of Singapore.

00:02:28.880 --> 00:02:31.970
OK, just to let everybody
know what the other side

00:02:31.970 --> 00:02:34.790
is enjoying on their weekends.

00:02:34.790 --> 00:02:38.030
So here's what I want
to try to do today.

00:02:38.030 --> 00:02:40.100
I want to go back and
talk a little bit more

00:02:40.100 --> 00:02:42.530
about process definitions
from the point of view

00:02:42.530 --> 00:02:45.110
of geometric change
of an object.

00:02:45.110 --> 00:02:46.610
And I think Duane
has already talked

00:02:46.610 --> 00:02:48.050
about this a fair amount.

00:02:48.050 --> 00:02:52.070
But I want to relate it to that
model or that view of processes

00:02:52.070 --> 00:02:58.610
to a couple key mechanical
processes as examples.

00:02:58.610 --> 00:03:00.200
And then something
that we always do

00:03:00.200 --> 00:03:03.230
is talk about this
taxonomy for control.

00:03:03.230 --> 00:03:04.100
It's a framework.

00:03:04.100 --> 00:03:07.430
It's all these
rather abstract terms

00:03:07.430 --> 00:03:10.280
that say, here's a
way for you to think

00:03:10.280 --> 00:03:13.490
about every process
that's ever been invented

00:03:13.490 --> 00:03:16.910
and try to understand how it
relates to some other process,

00:03:16.910 --> 00:03:19.100
with respect to
things that affect

00:03:19.100 --> 00:03:20.460
how we control the process.

00:03:20.460 --> 00:03:22.670
So that's our
taxonomy for control.

00:03:22.670 --> 00:03:25.100
And then we'll go through a
couple mechanical examples.

00:03:25.100 --> 00:03:28.460
I'll use the process of simple
machining, a turning process,

00:03:28.460 --> 00:03:31.760
a blending process,
and a molding process.

00:03:31.760 --> 00:03:35.060
We used to have a lab in
the course that we did--

00:03:35.060 --> 00:03:38.120
in fact, the guys in
Singapore did it this summer

00:03:38.120 --> 00:03:40.010
as part of their
summer orientation.

00:03:40.010 --> 00:03:41.750
They actually did
this lab where they

00:03:41.750 --> 00:03:46.670
went through and made parts
using these three processes.

00:03:46.670 --> 00:03:49.700
And we studied the process
control aspect to that.

00:03:49.700 --> 00:03:53.060
And they're
ubiquitous processes,

00:03:53.060 --> 00:03:57.560
but they also can be reduced
to some simple enough concepts

00:03:57.560 --> 00:04:00.590
that you can try to understand
this idea of the origins

00:04:00.590 --> 00:04:03.080
of variation, why
did things vary,

00:04:03.080 --> 00:04:08.332
instead of it being a
purely random event,

00:04:08.332 --> 00:04:10.040
instead of sort of
throwing up your hands

00:04:10.040 --> 00:04:12.230
and saying, I have no idea
why anything happened,

00:04:12.230 --> 00:04:14.510
let's just model it
as a random process,

00:04:14.510 --> 00:04:17.300
we say, no, let's try to
see if we can separate

00:04:17.300 --> 00:04:21.170
those things that have a clear
cause, which we may or may not

00:04:21.170 --> 00:04:23.960
be able to control, and those
things that we just don't know

00:04:23.960 --> 00:04:25.730
about.

00:04:25.730 --> 00:04:28.580
And then scrambling
at the end, I'll

00:04:28.580 --> 00:04:33.530
try to get to this concept
of states and properties.

00:04:33.530 --> 00:04:35.690
And again, this is more
of a framework issue.

00:04:35.690 --> 00:04:40.190
But it is an issue of
looking at any process,

00:04:40.190 --> 00:04:43.160
looking at all the
different factors that

00:04:43.160 --> 00:04:45.830
enter into the
outcome of the process

00:04:45.830 --> 00:04:53.090
and just parsing it really into
four areas in this idea of what

00:04:53.090 --> 00:04:55.340
we'll end up calling
thermodynamics states,

00:04:55.340 --> 00:04:58.070
those things that are kind
of transient, come and go,

00:04:58.070 --> 00:04:59.510
and properties,
those things that

00:04:59.510 --> 00:05:01.400
are inherent in the process.

00:05:01.400 --> 00:05:02.870
And then we'll
separate the process

00:05:02.870 --> 00:05:07.040
into the equipment that executes
the process, which kind stays

00:05:07.040 --> 00:05:09.740
there day after day
and year after year,

00:05:09.740 --> 00:05:12.830
and the material
that you're changing

00:05:12.830 --> 00:05:13.820
as part of the process.

00:05:13.820 --> 00:05:16.370
And, of course, that changes
in every cycle of the process.

00:05:16.370 --> 00:05:21.030
And that distinction between
the two is extremely important.

00:05:21.030 --> 00:05:23.240
So here's the model
you've already seen.

00:05:23.240 --> 00:05:27.200
And the first part
of this is to notice

00:05:27.200 --> 00:05:30.080
that we've taken the
manufacturing process and said,

00:05:30.080 --> 00:05:33.270
well, again, there are
these two elements to it.

00:05:33.270 --> 00:05:39.110
There's the material, which is
again sort of coming and going.

00:05:39.110 --> 00:05:42.380
And by its very nature, the
fact that every time you

00:05:42.380 --> 00:05:44.630
run the process,
there's a new piece

00:05:44.630 --> 00:05:46.370
of material in
there, that's going

00:05:46.370 --> 00:05:48.410
to have a different
characteristic

00:05:48.410 --> 00:05:52.640
for its variability than the
equipment, which, again, is

00:05:52.640 --> 00:05:56.660
unless you buy disposable
equipment, you generally

00:05:56.660 --> 00:05:59.810
keep around for years.

00:05:59.810 --> 00:06:02.030
And then there's an
interaction between the two.

00:06:02.030 --> 00:06:06.140
And that interaction, if you
really want to abstract it,

00:06:06.140 --> 00:06:08.870
could be called a
directed energy exchange.

00:06:08.870 --> 00:06:12.380
It always takes some energy to
cause the material to change.

00:06:12.380 --> 00:06:13.860
And what are you
trying to change?

00:06:13.860 --> 00:06:18.410
You're trying to change
its geometry, usually only

00:06:18.410 --> 00:06:19.460
its geometry.

00:06:19.460 --> 00:06:21.710
Often properties change
as a result of that.

00:06:21.710 --> 00:06:23.690
Or sometimes you're
changing properties,

00:06:23.690 --> 00:06:27.000
like a hardening process or
ion implantation, something

00:06:27.000 --> 00:06:27.500
like that.

00:06:27.500 --> 00:06:30.470
You're taking one material
and through this process

00:06:30.470 --> 00:06:34.380
transforming it in some way.

00:06:34.380 --> 00:06:37.300
And, again, just
to remind you, we

00:06:37.300 --> 00:06:39.690
have this simple functional
relationship that says,

00:06:39.690 --> 00:06:42.493
well, guess what, the output is
a function of all the inputs.

00:06:42.493 --> 00:06:43.910
And that's really
all it's saying.

00:06:43.910 --> 00:06:48.920
And defining it right now,
saying that the outputs are

00:06:48.920 --> 00:06:49.700
some vector--

00:06:49.700 --> 00:06:53.510
or the outputs can be
described as a vector, which

00:06:53.510 --> 00:06:55.790
has all of the relevant
geometry and property

00:06:55.790 --> 00:06:58.890
characteristics of the process.

00:06:58.890 --> 00:07:01.430
So it would be dimensions,
material properties,

00:07:01.430 --> 00:07:03.560
things like that.

00:07:03.560 --> 00:07:07.370
And process parameters,
using a really generic word

00:07:07.370 --> 00:07:09.230
for those input
variables, because I'm

00:07:09.230 --> 00:07:12.150
going to distinguish between
inputs and parameters

00:07:12.150 --> 00:07:15.560
a little bit later,
and just saying

00:07:15.560 --> 00:07:18.490
that there's some functional
relationship between the two.

00:07:18.490 --> 00:07:21.110
And, of course, here's the
functional relationship,

00:07:21.110 --> 00:07:23.150
all that stuff that's in there.

00:07:27.230 --> 00:07:33.540
So if we just focus for
a moment on processes

00:07:33.540 --> 00:07:37.210
where geometry change is
the most important one--

00:07:37.210 --> 00:07:40.920
and I don't know, speaking from
my perspective on this class,

00:07:40.920 --> 00:07:44.248
that's going to be 100%
of what I talk about.

00:07:44.248 --> 00:07:45.915
I think when we get
into semiconductors,

00:07:45.915 --> 00:07:49.650
there's a little bit more
on material modification.

00:07:49.650 --> 00:07:51.840
But I think certainly most
of the stuff you've seen

00:07:51.840 --> 00:07:54.540
has been on a
manufacturing process,

00:07:54.540 --> 00:07:58.260
meaning we change the geometry.

00:07:58.260 --> 00:08:00.040
Now, what causes the
output to change?

00:08:00.040 --> 00:08:04.020
Well, there's a directed energy
exchange with the equipment.

00:08:04.020 --> 00:08:07.650
And that energy
exchange can take

00:08:07.650 --> 00:08:10.840
the form of any of the
relevant forms of energy.

00:08:10.840 --> 00:08:12.847
So we could have a
mechanical exchange,

00:08:12.847 --> 00:08:15.180
where you impose a force or
a displacement on something.

00:08:15.180 --> 00:08:16.805
We can have an
electrical change, where

00:08:16.805 --> 00:08:18.540
you run a current
through something,

00:08:18.540 --> 00:08:22.920
or have an electric arc
something like that.

00:08:22.920 --> 00:08:24.565
It could be thermal.

00:08:24.565 --> 00:08:26.190
And certainly, there
are many processes

00:08:26.190 --> 00:08:30.540
where the transfer of
heat is the key way

00:08:30.540 --> 00:08:32.039
that you change a geometry.

00:08:32.039 --> 00:08:33.539
And certainly, it
could be chemical.

00:08:33.539 --> 00:08:35.622
Think of an etching process
or anything like that.

00:08:35.622 --> 00:08:40.200
All of these energies
directed the correct way

00:08:40.200 --> 00:08:46.960
can be thought of as
geometry change mechanisms.

00:08:46.960 --> 00:08:50.260
Now, how do we
direct this energy?

00:08:50.260 --> 00:08:55.410
If I just say, I've got some
thermal energy sitting around.

00:08:55.410 --> 00:08:57.675
How do I make sure it
goes to the right place?

00:09:02.320 --> 00:09:04.570
In effect, that's the
goal of the equipment.

00:09:04.570 --> 00:09:06.820
That's why this one
little thing, which

00:09:06.820 --> 00:09:08.840
my thermodynamics friends
would go crazy with,

00:09:08.840 --> 00:09:10.930
but calling energy
a vector quantity.

00:09:13.940 --> 00:09:17.120
It matters where you
put the energy, right?

00:09:17.120 --> 00:09:21.410
If I want to take this
bar and maybe do something

00:09:21.410 --> 00:09:26.570
to the center of it thermally,
and I've got a thermal energy

00:09:26.570 --> 00:09:29.030
source out here called
a piece of equipment,

00:09:29.030 --> 00:09:31.320
maybe a torch or
something like that,

00:09:31.320 --> 00:09:33.400
how do I make sure it
goes to the right place?

00:09:37.860 --> 00:09:39.165
Equipment, material.

00:09:44.160 --> 00:09:46.950
This is one of my
easy questions.

00:09:46.950 --> 00:09:49.650
I make sure I hold
it in place, right?

00:09:49.650 --> 00:09:50.580
Yeah, go ahead.

00:09:50.580 --> 00:09:52.130
Go ahead.

00:09:52.130 --> 00:09:56.390
AUDIENCE: So I guess by way
of relative positioning,

00:09:56.390 --> 00:09:58.370
these two are in the
correct positions.

00:09:58.370 --> 00:10:00.680
DAVID HARDT: Yeah, I mean if
I had like a point source,

00:10:00.680 --> 00:10:03.830
a laser or an arc, or
something like that--

00:10:03.830 --> 00:10:05.150
laser is a better example--

00:10:05.150 --> 00:10:07.910
and I needed to do
something right here,

00:10:07.910 --> 00:10:11.360
maybe thermally
deform this or join it

00:10:11.360 --> 00:10:13.220
or something like
that, then the key

00:10:13.220 --> 00:10:20.510
would be how both the value of
that energy and where I put it.

00:10:20.510 --> 00:10:24.240
So location would
be really important.

00:10:24.240 --> 00:10:26.420
So that's one way of
thinking of the concept

00:10:26.420 --> 00:10:27.757
of a directed energy source.

00:10:27.757 --> 00:10:29.840
If I want to bend this
piece of sheet metal, which

00:10:29.840 --> 00:10:36.860
was originally flat, and I
have a displacement device,

00:10:36.860 --> 00:10:40.820
it's probably a good idea for me
to put that displacement device

00:10:40.820 --> 00:10:44.210
in the middle, not
over here, and to have

00:10:44.210 --> 00:10:47.660
that displacement device
have a certain shape

00:10:47.660 --> 00:10:49.970
to get what I want.

00:10:49.970 --> 00:10:51.860
Think about this one
for a second this one.

00:10:51.860 --> 00:10:53.235
I don't know if
you can see this.

00:10:53.235 --> 00:10:58.200
Let me try my document camera.

00:10:58.200 --> 00:10:59.830
Come on, doc cam.

00:10:59.830 --> 00:11:02.110
There we go.

00:11:02.110 --> 00:11:04.650
This is an injection
molded part.

00:11:04.650 --> 00:11:07.530
It's from a precision yo-yo.

00:11:07.530 --> 00:11:14.260
And what's the energy
we're talking about there?

00:11:14.260 --> 00:11:15.413
This was injection molded.

00:11:15.413 --> 00:11:17.830
I think most of you here have
had some hands on experience

00:11:17.830 --> 00:11:19.600
with injection molding.

00:11:19.600 --> 00:11:22.510
If you haven't, I'll
explain it in a few minutes.

00:11:22.510 --> 00:11:27.040
But where's my directed
energy source there?

00:11:27.040 --> 00:11:29.080
Or if you have any
trouble getting your head

00:11:29.080 --> 00:11:31.750
around that, how did I
come up with that shape?

00:11:34.942 --> 00:11:36.375
AUDIENCE: Heat and pressure.

00:11:36.375 --> 00:11:38.000
DAVID HARDT: If I
had heat and pressure

00:11:38.000 --> 00:11:42.900
to put the liquid
plastic in there,

00:11:42.900 --> 00:11:44.340
that kind of filled the mold.

00:11:44.340 --> 00:11:46.050
But what determined the shape?

00:11:46.050 --> 00:11:47.040
AUDIENCE: The mold.

00:11:47.040 --> 00:11:48.450
DAVID HARDT: Yeah, the mold.

00:11:48.450 --> 00:11:51.710
And what's the directed
energy in this case?

00:11:51.710 --> 00:11:55.830
If I'm saying the direct energy
leads to the shape of the part,

00:11:55.830 --> 00:11:57.745
the geometry change--

00:11:57.745 --> 00:11:58.620
AUDIENCE: The forces.

00:11:58.620 --> 00:12:00.030
DAVID HARDT: Pardon me.

00:12:00.030 --> 00:12:01.740
AUDIENCE: The mill
cutting courses

00:12:01.740 --> 00:12:05.402
from whatever bit
used to cut the metal.

00:12:05.402 --> 00:12:07.860
DAVID HARDT: We're talking
about plastic molding right now.

00:12:07.860 --> 00:12:08.110
AUDIENCE: Right.

00:12:08.110 --> 00:12:09.020
But to make the mold.

00:12:09.020 --> 00:12:10.802
DAVID HARDT: Oh,
to make the mold.

00:12:10.802 --> 00:12:12.510
Yeah, I see where
you're going with that.

00:12:12.510 --> 00:12:14.550
But I'm thinking about
just the molding process.

00:12:14.550 --> 00:12:18.540
The tooling somehow
was magically there.

00:12:18.540 --> 00:12:22.680
AUDIENCE: I think the directed
energy in front of the surface.

00:12:22.680 --> 00:12:26.490
So everywhere where we have
surface, we direct energy.

00:12:26.490 --> 00:12:27.750
DAVID HARDT: Exactly.

00:12:27.750 --> 00:12:31.680
So now, the one big
distinction between these two

00:12:31.680 --> 00:12:35.340
is that here it was pretty
clear that I had a pretty

00:12:35.340 --> 00:12:39.330
localized source of energy.

00:12:39.330 --> 00:12:41.970
And, of course, the case of
heating this would be the same.

00:12:41.970 --> 00:12:49.650
Or as you'll see later,
turning with a pointed tool.

00:12:49.650 --> 00:12:52.290
Here, everything's happening
everywhere at once.

00:12:52.290 --> 00:12:56.610
And the direction, if you
will, the spatial aspect of it

00:12:56.610 --> 00:12:59.130
is imbued not by how
I position something,

00:12:59.130 --> 00:13:02.500
but by the shape of the tool.

00:13:02.500 --> 00:13:05.400
So you start to see that they're
kind of two extremes here.

00:13:05.400 --> 00:13:08.340
There's the case where
this geometry is determined

00:13:08.340 --> 00:13:12.900
primarily by a spatial
distribution of energy

00:13:12.900 --> 00:13:15.630
exchange, which is
stretching it a little bit

00:13:15.630 --> 00:13:18.360
here, but a shaped tool.

00:13:18.360 --> 00:13:20.640
And as Simon said,
everything happens everywhere

00:13:20.640 --> 00:13:23.370
at the same time.

00:13:23.370 --> 00:13:27.780
Whereas in this case,
things are changing

00:13:27.780 --> 00:13:32.530
over time and quite
different things

00:13:32.530 --> 00:13:33.780
happening at different places.

00:13:33.780 --> 00:13:36.780
And an more extreme case
again, which I'll get to later,

00:13:36.780 --> 00:13:40.230
is if I'm coming in here with
some sort of a point source--

00:13:40.230 --> 00:13:42.990
let's say heating it or
applying a force here--

00:13:42.990 --> 00:13:44.820
there's nothing
happening out here.

00:13:44.820 --> 00:13:47.220
It's all happening right here.

00:13:47.220 --> 00:13:49.740
And the net shape that
I get has a lot more

00:13:49.740 --> 00:13:53.730
to do with how I move
that interaction port,

00:13:53.730 --> 00:13:57.690
otherwise known as a tool, than
it does with the tool shape.

00:13:57.690 --> 00:14:00.600
Tool shape is important, but
it's of secondary importance.

00:14:00.600 --> 00:14:01.100
OK?

00:14:07.300 --> 00:14:09.900
They're supposed to go away now.

00:14:09.900 --> 00:14:11.980
There we go.

00:14:11.980 --> 00:14:13.120
OK.

00:14:13.120 --> 00:14:15.010
Well, they didn't go away.

00:14:15.010 --> 00:14:18.100
It's good to see you guys.

00:14:18.100 --> 00:14:25.320
So now, we talk
about it and we say,

00:14:25.320 --> 00:14:26.670
it's about geometry change.

00:14:26.670 --> 00:14:29.400
And I've kind of thrown out
some different ones here.

00:14:29.400 --> 00:14:32.813
How do you change the
shape of anything?

00:14:32.813 --> 00:14:34.230
Well, we already
talked about one.

00:14:34.230 --> 00:14:35.063
It's pretty obvious.

00:14:35.063 --> 00:14:37.140
Take out a knife
and a piece of wood

00:14:37.140 --> 00:14:40.650
and you can change its shape
by removing wood chips.

00:14:40.650 --> 00:14:44.760
Whittling, that's your
basic shaping technique.

00:14:44.760 --> 00:14:48.390
Well, that's essentially
one of the most common ways

00:14:48.390 --> 00:14:49.800
of making anything.

00:14:49.800 --> 00:14:53.160
We etch away silicon.

00:14:53.160 --> 00:14:57.660
We machine away material.

00:14:57.660 --> 00:15:01.170
We can oxidize and
burn things off.

00:15:01.170 --> 00:15:04.650
We can do it in a number
of different ways.

00:15:04.650 --> 00:15:06.330
So removal of
material is obviously

00:15:06.330 --> 00:15:08.300
one way of doing things.

00:15:08.300 --> 00:15:09.400
Pretty obvious one.

00:15:09.400 --> 00:15:13.480
How else can I change
the shape of something?

00:15:13.480 --> 00:15:15.635
AUDIENCE: You can add material.

00:15:15.635 --> 00:15:17.240
DAVID HARDT: I can
add material to it.

00:15:17.240 --> 00:15:17.810
Yeah.

00:15:17.810 --> 00:15:19.980
Or join or something like that.

00:15:19.980 --> 00:15:22.880
So I can reform things.

00:15:22.880 --> 00:15:25.647
OK, how else?

00:15:25.647 --> 00:15:27.230
AUDIENCE: You can
deform the material.

00:15:27.230 --> 00:15:29.495
DAVID HARDT: I can deform
the material like this.

00:15:29.495 --> 00:15:32.960
This started out, it's a
fixed piece of material.

00:15:32.960 --> 00:15:34.920
I didn't add or
subtract anything to it.

00:15:34.920 --> 00:15:39.210
But I changed the
shape by bending it.

00:15:39.210 --> 00:15:39.990
How else?

00:15:39.990 --> 00:15:42.090
Let me see what I've got here.

00:15:42.090 --> 00:15:44.010
Plastic deformation
of the material.

00:15:44.010 --> 00:15:47.900
So a constant volume,
constant mass process.

00:15:47.900 --> 00:15:52.220
The first case, you're
throwing material away.

00:15:52.220 --> 00:15:55.190
Adding material, which I think
is what people were already

00:15:55.190 --> 00:15:58.040
talking about.

00:15:58.040 --> 00:16:01.760
So one of my famous
professors here at MIT

00:16:01.760 --> 00:16:05.243
called this machining
and negative machining.

00:16:05.243 --> 00:16:07.160
So it's kind of another
way of thinking of it,

00:16:07.160 --> 00:16:09.810
but adding material,
subtracting material.

00:16:09.810 --> 00:16:11.840
The ultimate
manufacturing tool would

00:16:11.840 --> 00:16:15.230
be one that took from
here and gave to there

00:16:15.230 --> 00:16:17.180
and that sort of thing.

00:16:17.180 --> 00:16:17.760
What else?

00:16:17.760 --> 00:16:20.300
There's one other?

00:16:20.300 --> 00:16:21.950
Where does this fit in?

00:16:21.950 --> 00:16:23.150
It doesn't fit any of those.

00:16:25.892 --> 00:16:27.270
AUDIENCE: Formation.

00:16:27.270 --> 00:16:29.308
DAVID HARDT: Yeah, you
guys took 2810, right?

00:16:29.308 --> 00:16:30.850
So you probably
already covered this.

00:16:30.850 --> 00:16:33.810
But, yes, formation
of material, so where

00:16:33.810 --> 00:16:38.580
you go from a liquid
or vapor state,

00:16:38.580 --> 00:16:43.890
it solidifies against some
energy exchange medium or form

00:16:43.890 --> 00:16:44.400
tool.

00:16:44.400 --> 00:16:45.733
And you come up with that shape.

00:16:45.733 --> 00:16:46.770
OK.

00:16:46.770 --> 00:16:47.670
What else?

00:16:47.670 --> 00:16:48.480
That's only four?

00:16:53.730 --> 00:16:56.010
This is every
manufacturing process

00:16:56.010 --> 00:16:57.600
in the world in the universe.

00:16:57.600 --> 00:16:59.433
AUDIENCE: Well, it's
kind of under addition.

00:16:59.433 --> 00:17:01.170
But you're actually
growing material,

00:17:01.170 --> 00:17:03.650
using chemical,
biological process,

00:17:03.650 --> 00:17:05.040
like a silicon crystal.

00:17:05.040 --> 00:17:06.150
DAVID HARDT: OK, good.

00:17:06.150 --> 00:17:09.480
We'll call that addition.

00:17:09.480 --> 00:17:12.950
It ruins my punch
line if we don't.

00:17:12.950 --> 00:17:15.790
Any others?

00:17:15.790 --> 00:17:21.069
Well, I mean the
point is that again it

00:17:21.069 --> 00:17:23.859
is a matter of definitions, and
you could cut this more finely.

00:17:23.859 --> 00:17:26.589
But this pretty
nicely characterizes

00:17:26.589 --> 00:17:29.650
every manufacturing process
with respect to geometry change.

00:17:29.650 --> 00:17:32.940
We could have a parallel
set for property changes,

00:17:32.940 --> 00:17:36.190
thermodynamic changes
and implantation changes,

00:17:36.190 --> 00:17:38.030
atomic changes, so
forth and so on.

00:17:38.030 --> 00:17:42.950
But for this it seems to
cover pretty much everything.

00:17:42.950 --> 00:17:45.358
So if you look at a
manufacturing process

00:17:45.358 --> 00:17:47.650
or you look at something
going on in a factory and say,

00:17:47.650 --> 00:17:49.775
is this a manufacturing process?

00:17:49.775 --> 00:17:51.400
One of these things
should be going on,

00:17:51.400 --> 00:17:55.280
if it's a geometry
change process.

00:17:55.280 --> 00:17:58.370
So what controls
the geometry change?

00:17:58.370 --> 00:18:01.482
Was, who said, the
location and intensity

00:18:01.482 --> 00:18:02.440
of the energy exchange.

00:18:02.440 --> 00:18:04.510
So we already talked
about this for turning.

00:18:04.510 --> 00:18:07.327
And here's just a schematic
of a simple turning.

00:18:07.327 --> 00:18:09.910
This is actually the experiment
we used to do where you just--

00:18:09.910 --> 00:18:11.952
or you guys actually did
it again, where you just

00:18:11.952 --> 00:18:14.470
come in so-called orthogonal
turning, and you come in

00:18:14.470 --> 00:18:17.095
and reduce the
diameter of this stock.

00:18:20.110 --> 00:18:23.530
And more formally, we would
say this is the location

00:18:23.530 --> 00:18:25.060
of the maximum shear stress.

00:18:25.060 --> 00:18:27.100
You bring in a tool.

00:18:27.100 --> 00:18:29.560
And the material is
spinning in this case.

00:18:29.560 --> 00:18:32.170
And there's essentially
no stresses on it.

00:18:32.170 --> 00:18:34.300
Then you come in with a
tool and start applying

00:18:34.300 --> 00:18:37.390
a significant force over
a very limited area,

00:18:37.390 --> 00:18:40.000
in such a way as to create
a critical shear stress,

00:18:40.000 --> 00:18:45.040
and the material
decides to leave.

00:18:45.040 --> 00:18:48.370
And so in an abstract sense,
I can think of machining

00:18:48.370 --> 00:18:53.620
as locating a very concentrated
maximal shear stress

00:18:53.620 --> 00:18:56.620
point at different
places in such a way

00:18:56.620 --> 00:19:00.270
as to at that location make the
material disassociate with it.

00:19:04.827 --> 00:19:07.160
Yeah, so we already talked
about this one, heat transfer

00:19:07.160 --> 00:19:09.020
at the mold surface
and injection molding.

00:19:14.970 --> 00:19:20.743
So another extreme of
interaction to give you an idea

00:19:20.743 --> 00:19:22.410
that there's more
than one way to remove

00:19:22.410 --> 00:19:26.430
material, and certainly anybody
with a mechanical engineering

00:19:26.430 --> 00:19:29.630
education, any sort of shop
or manufacturing experience

00:19:29.630 --> 00:19:31.962
says, oh, yeah,
machining, that's it.

00:19:31.962 --> 00:19:33.420
And that's how you
remove material.

00:19:33.420 --> 00:19:36.210
But, of course,
we remove material

00:19:36.210 --> 00:19:38.070
in a lot of different ways.

00:19:38.070 --> 00:19:41.340
And, of course, anybody from a
EE or semiconductor background

00:19:41.340 --> 00:19:43.590
says, oh, yeah, removal,
that's chemical etching

00:19:43.590 --> 00:19:46.180
or something like that.

00:19:46.180 --> 00:19:47.160
And that's true also.

00:19:47.160 --> 00:19:50.370
But then you have many things,
these beam type processes,

00:19:50.370 --> 00:19:54.340
which are very important for
a lot of different processes.

00:19:54.340 --> 00:19:58.200
But think the extreme
case of a process where

00:19:58.200 --> 00:20:03.120
things are very localized and
you're just doing something

00:20:03.120 --> 00:20:05.628
in a very small region
and then you move

00:20:05.628 --> 00:20:06.795
that region around the part.

00:20:06.795 --> 00:20:09.720
It might be something
like laser processing,

00:20:09.720 --> 00:20:13.890
laser beams, using laser
beam to cut and moving it.

00:20:13.890 --> 00:20:18.008
This is meant to be a picture
of a robot moving it around.

00:20:18.008 --> 00:20:19.800
So the robot is very
important in this case

00:20:19.800 --> 00:20:22.740
because it's
providing the motion

00:20:22.740 --> 00:20:24.132
and moving the laser tool.

00:20:24.132 --> 00:20:25.590
And, of course,
this has been taken

00:20:25.590 --> 00:20:29.550
to extreme with these
femtosecond lasers, which

00:20:29.550 --> 00:20:34.080
can affect at any given instant,
will only ablate material

00:20:34.080 --> 00:20:37.560
and, I don't know
what, sub-micron areas.

00:20:37.560 --> 00:20:39.600
So you can really localize
it if you want to.

00:20:42.470 --> 00:20:45.580
And then you've got you've
got other things like this,

00:20:45.580 --> 00:20:48.460
like a CVD process where
what you're actually doing

00:20:48.460 --> 00:20:53.560
is locating a chemical reaction
across the entire surface

00:20:53.560 --> 00:20:55.070
of something.

00:20:55.070 --> 00:20:59.650
And again, that location
is determined primarily

00:20:59.650 --> 00:21:03.520
by whatever substrate it
is you're trying to coat.

00:21:03.520 --> 00:21:07.660
In this case, we're
trying to coat.

00:21:07.660 --> 00:21:13.600
But just to throw this out,
if I have a nice clean uniform

00:21:13.600 --> 00:21:15.100
substrate, even with--

00:21:15.100 --> 00:21:17.530
well, let's just take a
flat one, a flat disk,

00:21:17.530 --> 00:21:20.370
and I need to coat it with
a chemical vapor deposition

00:21:20.370 --> 00:21:20.870
process.

00:21:20.870 --> 00:21:26.350
So I've got this vapor, and it's
going to deposit and solidify

00:21:26.350 --> 00:21:28.780
on the surface.

00:21:28.780 --> 00:21:32.363
What's going to determine the
final geometry of that surface,

00:21:32.363 --> 00:21:33.655
of that coating of the surface?

00:21:37.520 --> 00:21:42.200
AUDIENCE: [INAUDIBLE]

00:21:42.200 --> 00:21:43.700
DAVID HARDT: Yeah,
well let's assume

00:21:43.700 --> 00:21:48.590
this is just sort of this
chamber filled with the vapor

00:21:48.590 --> 00:21:51.900
and then you introduce
this in there.

00:21:51.900 --> 00:21:52.400
OK?

00:21:52.400 --> 00:21:53.530
AUDIENCE: Time.

00:21:53.530 --> 00:21:57.178
DAVID HARDT: Time certainly
is a big part of it, yeah.

00:21:57.178 --> 00:21:58.041
AUDIENCE: Topology.

00:22:00.690 --> 00:22:03.780
DAVID HARDT: The topology
of the substrate.

00:22:03.780 --> 00:22:06.123
And I'll just throw
in the other is that--

00:22:06.123 --> 00:22:08.040
because I want to make
the point that with any

00:22:08.040 --> 00:22:12.420
of these processes that
occur with a distribution

00:22:12.420 --> 00:22:16.110
sort of simultaneously, the ones
that Simon was talking about

00:22:16.110 --> 00:22:21.090
earlier, there's
always this concern

00:22:21.090 --> 00:22:23.525
that you don't have uniform
spatial distribution.

00:22:23.525 --> 00:22:24.900
So what I was
getting at is there

00:22:24.900 --> 00:22:29.483
was no intention of having
directed vapor density

00:22:29.483 --> 00:22:30.400
or anything like that.

00:22:30.400 --> 00:22:35.820
The idea is I've got this
uniform cloud of vapor.

00:22:35.820 --> 00:22:37.440
And it's reacting
with the surface.

00:22:37.440 --> 00:22:40.890
So why might it vary spatially?

00:22:40.890 --> 00:22:42.540
It's certainly going
to vary over time,

00:22:42.540 --> 00:22:44.917
because it takes time for
this stuff to solidify.

00:22:44.917 --> 00:22:46.500
But it's also going
to vary spatially,

00:22:46.500 --> 00:22:48.997
because who said it's uniform?

00:22:48.997 --> 00:22:51.330
And one of the things you
want to think about when we're

00:22:51.330 --> 00:22:53.250
talking about manufacturing
is always sort of

00:22:53.250 --> 00:22:58.260
question any statement of
uniformity or constancy

00:22:58.260 --> 00:23:04.050
or things like that,
because inevitably there

00:23:04.050 --> 00:23:09.150
will be variations because the
topography of the substrate

00:23:09.150 --> 00:23:10.750
might change.

00:23:10.750 --> 00:23:11.430
Who knows?

00:23:11.430 --> 00:23:14.490
And in fact, it does
happen, especially

00:23:14.490 --> 00:23:16.200
as you go to larger
and larger wafers.

00:23:16.200 --> 00:23:20.310
You'll get important variations
in the chamber, temperature

00:23:20.310 --> 00:23:21.240
variations.

00:23:21.240 --> 00:23:23.230
You could get
concentration variations.

00:23:23.230 --> 00:23:25.905
All these kinds of things
can cause problems.

00:23:29.710 --> 00:23:31.410
One other that I'll
mention because I

00:23:31.410 --> 00:23:33.660
showed this piece of
sheet metal and talked

00:23:33.660 --> 00:23:37.170
about it as a concentrated
energy exchange.

00:23:37.170 --> 00:23:39.750
Here's another example of
a non-concentrated energy

00:23:39.750 --> 00:23:40.260
exchange.

00:23:40.260 --> 00:23:42.150
This is a large
piece of sheet metal.

00:23:42.150 --> 00:23:46.680
This dimension here is about 6
feet, aerospace manufacturing.

00:23:46.680 --> 00:23:51.405
And it's being pulled across
this tool and deformed in here.

00:23:51.405 --> 00:23:54.540
I actually have a sort of video
of this which I couldn't find.

00:23:54.540 --> 00:23:56.787
But you can see it being
pulled down to the tool.

00:23:56.787 --> 00:23:58.620
And because of the
reflections of the light,

00:23:58.620 --> 00:24:01.500
you can see that there's
not a single portion

00:24:01.500 --> 00:24:05.640
of this large piece of
metal that isn't being

00:24:05.640 --> 00:24:07.420
changed at any instant in time.

00:24:07.420 --> 00:24:10.530
So clearly, this is everything
happening everywhere

00:24:10.530 --> 00:24:12.070
all at once.

00:24:12.070 --> 00:24:16.900
And in effect we're pulling
it across a shaped tool.

00:24:16.900 --> 00:24:19.860
So in the end, we're imposing
this displacement field on it.

00:24:22.710 --> 00:24:26.310
So this gets us to this
concept of the taxonomy.

00:24:26.310 --> 00:24:27.840
You can see that--

00:24:27.840 --> 00:24:30.780
I've been kind of ranging
widely over a number

00:24:30.780 --> 00:24:35.820
of different processes, either
directly or by implication.

00:24:35.820 --> 00:24:39.010
And we want to kind of draw
some conclusions from this.

00:24:39.010 --> 00:24:41.430
So the first is that
there are two extremes

00:24:41.430 --> 00:24:45.750
to these interactions, the
so-called serial interaction.

00:24:45.750 --> 00:24:48.330
And schematically here
it's best described

00:24:48.330 --> 00:24:54.420
as my femtosecond laser,
infinitesimally small area

00:24:54.420 --> 00:24:58.210
of interaction over
the large area.

00:24:58.210 --> 00:25:00.000
So that at any instant
I'm only affecting

00:25:00.000 --> 00:25:03.240
a small part of the material.

00:25:03.240 --> 00:25:05.250
But then I move it.

00:25:05.250 --> 00:25:07.330
I have a time history of this.

00:25:07.330 --> 00:25:09.910
And that's what
determines the geometry.

00:25:09.910 --> 00:25:11.530
Now, of course,
in a real process,

00:25:11.530 --> 00:25:15.210
even with the femtosecond laser,
if you get to small enough,

00:25:15.210 --> 00:25:17.730
there are spatially
distributed effects.

00:25:17.730 --> 00:25:20.550
And as I go to larger
and larger tools,

00:25:20.550 --> 00:25:23.870
this gets worse and worse.

00:25:23.870 --> 00:25:27.100
But if we make a definition
that the area of the interaction

00:25:27.100 --> 00:25:29.890
of the energy
interaction is small

00:25:29.890 --> 00:25:33.710
compared to the total area
that I want to change,

00:25:33.710 --> 00:25:36.970
then it's clearly what
we'll call a serial process.

00:25:39.620 --> 00:25:44.270
If the opposite is true, if
the area of that interaction

00:25:44.270 --> 00:25:46.460
is on the order
of the total area

00:25:46.460 --> 00:25:48.110
that I'm trying to change--

00:25:48.110 --> 00:25:51.350
and again, the icon for that
would be stamping out a car

00:25:51.350 --> 00:25:52.790
hood or something like that--

00:25:52.790 --> 00:25:57.377
the area of interaction
is the same as the area

00:25:57.377 --> 00:25:58.460
that I'm trying to change.

00:25:58.460 --> 00:26:02.850
Same with this part, the area of
interaction is the entire area.

00:26:02.850 --> 00:26:06.110
Then we have a parallel process.

00:26:06.110 --> 00:26:11.870
We run into a couple of
interesting variants on that.

00:26:11.870 --> 00:26:19.670
If we take a very
common process, rolling,

00:26:19.670 --> 00:26:27.800
to create sheet or
just plate reductions--

00:26:27.800 --> 00:26:30.740
I'm coming in with material
that looks like this.

00:26:30.740 --> 00:26:32.960
I'm coming out with a
material that's much thinner.

00:26:46.000 --> 00:26:47.912
Where does this fit
in this taxonomy here,

00:26:47.912 --> 00:26:48.995
this part of the taxonomy?

00:26:48.995 --> 00:26:52.685
Is it a serial process
or a parallel process?

00:26:55.360 --> 00:26:57.240
AUDIENCE: Serial.

00:26:57.240 --> 00:26:58.480
DAVID HARDT: Serial.

00:26:58.480 --> 00:26:58.980
OK.

00:27:02.250 --> 00:27:04.500
OK, so we have 50%
of the electorate

00:27:04.500 --> 00:27:09.240
going for serial in this
mostly democratic primary.

00:27:09.240 --> 00:27:10.200
Parallel?

00:27:10.200 --> 00:27:12.630
50% saying parallel.

00:27:12.630 --> 00:27:14.787
I guess it goes to
the super-delegates.

00:27:18.130 --> 00:27:21.470
OK, you guys in
Singapore and to you,

00:27:21.470 --> 00:27:23.445
you guys, you want to
break the tie here?

00:27:23.445 --> 00:27:24.010
What is it?

00:27:24.010 --> 00:27:26.260
Is it serial or parallel?

00:27:26.260 --> 00:27:28.270
AUDIENCE: I think it's parallel.

00:27:28.270 --> 00:27:31.630
DAVID HARDT: OK, so
51% for parallel.

00:27:31.630 --> 00:27:33.460
Anybody there say serial?

00:27:39.760 --> 00:27:43.248
Yeah, we got some serials here.

00:27:43.248 --> 00:27:44.740
AUDIENCE: It depends
on the light.

00:27:44.740 --> 00:27:46.720
DAVID HARDT: Yeah.

00:27:46.720 --> 00:27:48.020
Yeah, now, we're getting to it.

00:27:48.020 --> 00:27:50.020
OK, it depends.

00:27:50.020 --> 00:27:56.030
If I had drawn this
in 2D like this

00:27:56.030 --> 00:28:00.260
and maybe sort of
drawn it in extreme

00:28:00.260 --> 00:28:01.730
and didn't even
show this necking,

00:28:01.730 --> 00:28:04.970
just kind of showed
it doing this,

00:28:04.970 --> 00:28:07.370
then you'd be pretty
tempted to say,

00:28:07.370 --> 00:28:09.620
oh, that's a serial process,
because all the action is

00:28:09.620 --> 00:28:10.190
right here.

00:28:10.190 --> 00:28:12.050
Look at this huge area
that I'm affecting

00:28:12.050 --> 00:28:14.330
and all the action
is right there.

00:28:14.330 --> 00:28:26.630
If I had drawn it in a front end
view and here's the material,

00:28:26.630 --> 00:28:29.000
you look at that, and
say, oh, that's parallel.

00:28:29.000 --> 00:28:33.710
Everything's happening
everywhere at the same time.

00:28:33.710 --> 00:28:38.870
So just a cautionary
tale, it's not either

00:28:38.870 --> 00:28:42.200
or, it's neither nor.

00:28:42.200 --> 00:28:44.890
It's sort of preponderance
of, if you will.

00:28:47.450 --> 00:28:52.190
So this is a good example of
why they're actually important,

00:28:52.190 --> 00:28:54.320
even though it's mixed on here.

00:28:54.320 --> 00:28:57.110
And I'm going to jump ahead
to the sort of who cares,

00:28:57.110 --> 00:29:00.530
which may not have a
satisfactory answer today.

00:29:00.530 --> 00:29:06.030
But the who cares is if
I'm running, in this case,

00:29:06.030 --> 00:29:08.720
the Alcoa Tennessee--
that's a real town--

00:29:08.720 --> 00:29:11.600
the Alcoa, Tennessee,
cold rolling

00:29:11.600 --> 00:29:15.590
mill that makes all
their beer cans stock,

00:29:15.590 --> 00:29:19.520
and I'm really concerned
about extremely good thickness

00:29:19.520 --> 00:29:22.130
control, because when it goes
into the high speed canning

00:29:22.130 --> 00:29:24.290
machines that thickness
has to be right on.

00:29:24.290 --> 00:29:29.310
Otherwise, the walls will tear
or it won't form properly.

00:29:29.310 --> 00:29:31.040
What do I control?

00:29:31.040 --> 00:29:35.270
Just intuitively, what would I
control to affect the geometry

00:29:35.270 --> 00:29:37.988
parameter called the thickness?

00:29:37.988 --> 00:29:39.300
AUDIENCE: Distance.

00:29:39.300 --> 00:29:41.940
DAVID HARDT: Distance
between the rolls, yeah.

00:29:41.940 --> 00:29:45.120
So I control the
thickness by controlling,

00:29:45.120 --> 00:29:47.670
let's say, this
distance right here.

00:29:51.580 --> 00:29:54.910
Which thickness
does that control?

00:29:54.910 --> 00:29:58.750
Does it control this
thickness right here,

00:29:58.750 --> 00:30:04.930
or this one here, or that one
there, or maybe that one there,

00:30:04.930 --> 00:30:05.800
or that one there?

00:30:09.480 --> 00:30:11.420
Yeah, all you're doing--

00:30:11.420 --> 00:30:14.240
there's one spacing area.

00:30:14.240 --> 00:30:17.570
And if you say, oh, yeah,
it's exactly right here.

00:30:17.570 --> 00:30:20.210
And then I make little
disks and make my cans

00:30:20.210 --> 00:30:22.190
with the material
over here, who's

00:30:22.190 --> 00:30:24.690
to say it's going to be right?

00:30:24.690 --> 00:30:27.090
So even in this
case, I've actually

00:30:27.090 --> 00:30:30.470
got a mixed problem here
and a mixed control problem.

00:30:30.470 --> 00:30:33.690
So, yeah, here's a case
where displacement--

00:30:33.690 --> 00:30:37.353
this trajectory we're talking
about, just like this deal,

00:30:37.353 --> 00:30:38.770
you want to think
of it that way--

00:30:38.770 --> 00:30:44.230
I'm directing the energy
source by displacing it.

00:30:44.230 --> 00:30:45.940
That's part of it.

00:30:45.940 --> 00:30:52.390
But then how do I change
the thickness distribution,

00:30:52.390 --> 00:30:54.730
the parallel part
of the process?

00:30:54.730 --> 00:30:57.040
Well, I'd actually
probably have to change

00:30:57.040 --> 00:30:59.920
the shape of the rolls.

00:30:59.920 --> 00:31:00.970
You do a little bit.

00:31:00.970 --> 00:31:02.290
They actually do both sides.

00:31:02.290 --> 00:31:04.030
And that does help.

00:31:04.030 --> 00:31:05.360
But let's say I do both sides.

00:31:05.360 --> 00:31:06.593
So this is exactly right.

00:31:06.593 --> 00:31:07.510
This is exactly right.

00:31:07.510 --> 00:31:10.660
And the middle looks like
this, which is not uncommon.

00:31:10.660 --> 00:31:13.635
It becomes fatter in the middle.

00:31:13.635 --> 00:31:14.260
What do you do?

00:31:17.760 --> 00:31:19.192
AUDIENCE: Change the rolls.

00:31:19.192 --> 00:31:21.150
DAVID HARDT: Yeah, but
I don't want to do that.

00:31:21.150 --> 00:31:22.410
Yes, you would change the rolls.

00:31:22.410 --> 00:31:24.327
And, in fact, what they
do is they build roles

00:31:24.327 --> 00:31:25.890
with a little bit
of camber in them.

00:31:25.890 --> 00:31:27.390
Because you know
the center is going

00:31:27.390 --> 00:31:28.515
to deflect up a little bit.

00:31:28.515 --> 00:31:30.600
So you put some camber into it.

00:31:30.600 --> 00:31:33.450
Believe it or not at Alcoa,
that's still not good enough.

00:31:33.450 --> 00:31:35.490
This stuff is very precise.

00:31:35.490 --> 00:31:37.440
I shouldn't quote
the tolerances.

00:31:37.440 --> 00:31:42.330
But it's on the order of in
my newly converted SI brain,

00:31:42.330 --> 00:31:47.800
I think it could be 15 microns.

00:31:47.800 --> 00:31:53.710
It could be in the 10,000th,
several 10,000th of inch

00:31:53.710 --> 00:31:55.750
in terms of thickness tolerance.

00:31:55.750 --> 00:31:57.480
It's not very thick
stuff to begin with.

00:31:57.480 --> 00:32:00.830
AUDIENCE: It would be like
locally heated through--

00:32:00.830 --> 00:32:03.080
DAVID HARDT: Yeah, in
fact, that's what is done.

00:32:03.080 --> 00:32:05.920
To get small displacements,
to be able to actually

00:32:05.920 --> 00:32:07.270
arbitrarily--

00:32:07.270 --> 00:32:10.540
not arbitrarily-- but to
put some different profiles

00:32:10.540 --> 00:32:14.910
on that, you'll locally heat
it or cool it in response

00:32:14.910 --> 00:32:17.670
to the feedback to
actually slowly move

00:32:17.670 --> 00:32:19.285
the tool locally a
little bit fatter

00:32:19.285 --> 00:32:20.660
here and a little
bit less there.

00:32:20.660 --> 00:32:23.100
You can imagine if you
only need 10 microns,

00:32:23.100 --> 00:32:26.070
something like that, it
doesn't take a lot of heat.

00:32:26.070 --> 00:32:27.810
And, yes, indeed you're
going to say, oh,

00:32:27.810 --> 00:32:29.130
how localized can that be?

00:32:29.130 --> 00:32:33.270
And that's the trade secret they
won't tell anybody about but.

00:32:33.270 --> 00:32:36.150
Yeah, so in this
particular case,

00:32:36.150 --> 00:32:43.503
notice these two different modes
of modes of interaction lead

00:32:43.503 --> 00:32:45.170
to two different ways
of controlling it.

00:32:45.170 --> 00:32:47.240
In one case, it's very simple.

00:32:47.240 --> 00:32:48.702
I'm going to buy a Servo system.

00:32:48.702 --> 00:32:49.910
It's going to be really good.

00:32:49.910 --> 00:32:51.077
It's going to be very stiff.

00:32:51.077 --> 00:32:55.280
And I'm going to say, make that
2/1,000 inch gap and you get

00:32:55.280 --> 00:32:56.930
2/1,000 inch gap.

00:32:56.930 --> 00:33:00.050
In the other case, it's
like, oh, my gosh, you know,

00:33:00.050 --> 00:33:03.920
I've got to change this
spatial distribution somehow.

00:33:03.920 --> 00:33:05.720
The extreme case is to
shut the thing down,

00:33:05.720 --> 00:33:08.010
put in a new set of roles.

00:33:08.010 --> 00:33:10.260
And the other is to come
up with some clever way

00:33:10.260 --> 00:33:12.660
of deforming it.

00:33:12.660 --> 00:33:15.480
So why do we worry about the
two modes of interaction?

00:33:15.480 --> 00:33:18.780
It leads directly
to how we would ever

00:33:18.780 --> 00:33:22.410
find the knobs on the machine
to change what we're doing.

00:33:22.410 --> 00:33:25.070
Now, that will
become most important

00:33:25.070 --> 00:33:26.820
in the last part of
the class when we talk

00:33:26.820 --> 00:33:28.200
about process optimization.

00:33:28.200 --> 00:33:31.057
And you're saying,
what can I do?

00:33:31.057 --> 00:33:32.640
What can actually
change on a process?

00:33:32.640 --> 00:33:37.860
What can I change real
time, millisecond?

00:33:37.860 --> 00:33:40.050
What can I change, maybe--

00:33:40.050 --> 00:33:43.200
in this case, this is an
unusual example-- what can I

00:33:43.200 --> 00:33:46.660
change on a frequent basis?

00:33:46.660 --> 00:33:48.163
But in most cases,
the major change

00:33:48.163 --> 00:33:49.830
you do with tooling
and things like this

00:33:49.830 --> 00:33:55.510
is on hourly shift,
yearly basis.

00:33:55.510 --> 00:33:57.420
So your ability to
affect the output

00:33:57.420 --> 00:34:00.900
once you've kind of
literally cast it in stone

00:34:00.900 --> 00:34:03.690
is greatly changed
by whether it's

00:34:03.690 --> 00:34:05.790
a parallel or serial process.

00:34:14.570 --> 00:34:17.330
OK, so just to
summarize this again,

00:34:17.330 --> 00:34:21.760
so for this lumped
or serial case,

00:34:21.760 --> 00:34:24.010
the time trajectory of
that interaction port

00:34:24.010 --> 00:34:25.550
is one of the most
important things.

00:34:25.550 --> 00:34:29.500
So robotic manufacturing,
which was all the rage for many

00:34:29.500 --> 00:34:32.050
a year, until we found
out it wasn't really

00:34:32.050 --> 00:34:36.159
such a thing to
be rageful about,

00:34:36.159 --> 00:34:40.929
it works for a limited class
of processes pretty well.

00:34:40.929 --> 00:34:45.310
And your classic CNC
machine, the CNC machine

00:34:45.310 --> 00:34:48.580
of the '60s, which sort of
got generalized in the '70s

00:34:48.580 --> 00:34:50.895
and put everywhere,
essentially means

00:34:50.895 --> 00:34:53.020
if you want to put in the
terminology of what we're

00:34:53.020 --> 00:34:56.050
talking about today
usually means the ability

00:34:56.050 --> 00:35:00.010
to control trajectories
in a very repeatable way.

00:35:00.010 --> 00:35:03.820
So it's not just a machine
that can follow a trajectory,

00:35:03.820 --> 00:35:05.890
it's also the ability to
generate the trajectory

00:35:05.890 --> 00:35:07.490
and things like that.

00:35:07.490 --> 00:35:09.250
And then in this distributed--

00:35:09.250 --> 00:35:11.440
I've used two different
terms here-- but distributed

00:35:11.440 --> 00:35:15.040
or parallel case,
more than anything,

00:35:15.040 --> 00:35:17.470
it's the shape of the energy
distributor, you know,

00:35:17.470 --> 00:35:21.082
patterns, molds, masks,
heat distributions,

00:35:21.082 --> 00:35:21.790
things like that.

00:35:24.310 --> 00:35:25.900
I think we've seen
enough examples.

00:35:25.900 --> 00:35:28.630
We're sick of these now.

00:35:28.630 --> 00:35:31.090
Oh, yeah, well, you'll
see this in a moment,

00:35:31.090 --> 00:35:35.440
but a lot of the so-called
rapid prototyping

00:35:35.440 --> 00:35:39.310
methods that popped up in
the '90s, or solid free form

00:35:39.310 --> 00:35:43.930
fabrication as the
government likes to call it,

00:35:43.930 --> 00:35:47.050
stereolithography, sort of
the first big one on that 3D

00:35:47.050 --> 00:35:52.100
printing and developed
here, are serial processes.

00:35:52.100 --> 00:35:55.370
And they figured out
how to put down--

00:35:55.370 --> 00:35:57.950
they turn out to be usually
serial addition type

00:35:57.950 --> 00:35:59.090
processes--

00:35:59.090 --> 00:36:00.770
figured out how to
put a dot of material

00:36:00.770 --> 00:36:05.870
down at a time to
build up a structure.

00:36:05.870 --> 00:36:09.800
And that dot itself was cohesive
enough that if you put it down,

00:36:09.800 --> 00:36:12.350
it stayed there and you didn't
need anything to hold it.

00:36:15.890 --> 00:36:18.090
And I'll talk more about
that in just a moment.

00:36:18.090 --> 00:36:22.130
But I think we've kind of
talked about-- these are all

00:36:22.130 --> 00:36:23.220
interesting examples.

00:36:23.220 --> 00:36:28.760
So if we talk about a taxonomy
to kind of bring together

00:36:28.760 --> 00:36:36.200
the energy source, if you
will, for the geometry

00:36:36.200 --> 00:36:41.700
change, the way we apply it,
and a couple other things that

00:36:41.700 --> 00:36:44.660
will come up in just a
second, then we can--

00:36:44.660 --> 00:36:49.655
and I think I've said why for
the mode, the change mode--

00:36:49.655 --> 00:36:50.870
oh, actually I haven't.

00:36:50.870 --> 00:36:52.640
I'm sorry about that.

00:36:52.640 --> 00:36:54.710
Why does it matter--

00:36:54.710 --> 00:36:56.360
let me go back to this.

00:36:56.360 --> 00:36:58.910
I think I won't bother
with the slides.

00:36:58.910 --> 00:37:03.560
What I called mode before
was really my energy source.

00:37:03.560 --> 00:37:11.750
Why does it matter whether I'm
using mechanical, electrical,

00:37:11.750 --> 00:37:16.400
thermal, or chemical
as my change agent?

00:37:19.940 --> 00:37:23.690
What difference does that make?

00:37:23.690 --> 00:37:24.760
It's just energy, right?

00:37:30.530 --> 00:37:33.650
And again, back to lecture
1, what do we care about

00:37:33.650 --> 00:37:34.400
in the process?

00:37:34.400 --> 00:37:37.560
Quality, number
one in this class.

00:37:37.560 --> 00:37:39.710
But also we worry about rate.

00:37:39.710 --> 00:37:42.110
And we worry about cost.

00:37:42.110 --> 00:37:45.990
And we worry about flexibility.

00:37:45.990 --> 00:37:53.890
So with respect to which
energy mode is involved,

00:37:53.890 --> 00:37:58.860
what difference does it
make in this consideration?

00:37:58.860 --> 00:37:59.530
Simon.

00:37:59.530 --> 00:38:01.780
AUDIENCE: Well, in the way
we can control the process,

00:38:01.780 --> 00:38:04.100
we need to know what
[? eventually ?] is involved,

00:38:04.100 --> 00:38:06.017
so we can change the
[? size ?] of the energy.

00:38:06.017 --> 00:38:07.900
DAVID HARDT: Yeah,
sure, absolutely,

00:38:07.900 --> 00:38:09.715
so I would control
these in different ways.

00:38:12.670 --> 00:38:14.260
But let's assume
I could do that.

00:38:14.260 --> 00:38:16.540
I know how to control
mechanical energy.

00:38:16.540 --> 00:38:19.060
I can change forces and
displacements and velocities,

00:38:19.060 --> 00:38:22.540
electrical currents
and voltages.

00:38:22.540 --> 00:38:25.360
Thermal, that's a
little bit different.

00:38:25.360 --> 00:38:27.760
I can certainly
control temperatures.

00:38:27.760 --> 00:38:30.190
I can also try to control
their distributions.

00:38:30.190 --> 00:38:33.490
I can sometimes control-- heat
flux is a little bit harder

00:38:33.490 --> 00:38:34.450
to do.

00:38:34.450 --> 00:38:37.450
And chemical, I can control
concentrations and temperatures

00:38:37.450 --> 00:38:40.090
and maybe reaction rates
as a result of that.

00:38:40.090 --> 00:38:40.825
Yeah.

00:38:40.825 --> 00:38:43.250
AUDIENCE: [INAUDIBLE]
if you're affecting

00:38:43.250 --> 00:38:47.592
a large area or smaller
wafer versus a big stack of

00:38:47.592 --> 00:38:48.960
[INAUDIBLE]

00:38:48.960 --> 00:38:52.588
DAVID HARDT: Yeah,
I thought you were

00:38:52.588 --> 00:38:53.880
going someplace else with that.

00:38:53.880 --> 00:38:56.610
But, yes, there
is a size scaling.

00:38:56.610 --> 00:38:59.250
But the example you
gave actually, though,

00:38:59.250 --> 00:39:04.590
a wafer, in terms of the
relative size of the features

00:39:04.590 --> 00:39:07.855
that I care about
is actually huge

00:39:07.855 --> 00:39:09.510
when you think about
a wafer compared

00:39:09.510 --> 00:39:12.510
to a tiny little
sub-micron line,

00:39:12.510 --> 00:39:13.920
and then I think
about a car hood

00:39:13.920 --> 00:39:17.400
with a little feature
in the middle,

00:39:17.400 --> 00:39:20.130
the wafer is actually bigger.

00:39:20.130 --> 00:39:24.040
AUDIENCE: I was thinking of the
[INAUDIBLE] with the changes

00:39:24.040 --> 00:39:24.540
in--

00:39:24.540 --> 00:39:25.040
I mean--

00:39:25.040 --> 00:39:26.920
DAVID HARDT: Yeah, yeah, yeah.

00:39:26.920 --> 00:39:30.305
OK, I think you're getting
to what I wanted to do.

00:39:30.305 --> 00:39:32.680
Let me see if I can get any
more here and then we'll see.

00:39:32.680 --> 00:39:33.410
Yeah, [? Kimmy. ?]

00:39:33.410 --> 00:39:34.827
AUDIENCE: The rates
are definitely

00:39:34.827 --> 00:39:37.436
change of [INAUDIBLE] mechanical
is usually faster than

00:39:37.436 --> 00:39:38.930
[INAUDIBLE]

00:39:38.930 --> 00:39:40.890
DAVID HARDT: Yeah,
yeah, believe it or not,

00:39:40.890 --> 00:39:43.370
when you think about this, you
always think of electrical,

00:39:43.370 --> 00:39:46.040
oh, that's the gigahertz range,
and all that kind of stuff.

00:39:46.040 --> 00:39:48.860
And that's true if all you
need to do is move electrons.

00:39:48.860 --> 00:39:50.910
But we have to move
material around here.

00:39:50.910 --> 00:39:54.200
So this tends to be very fast.

00:39:54.200 --> 00:39:58.470
If I apply a mechanical
load to something,

00:39:58.470 --> 00:40:02.820
I can modulate that load pretty
fast these days if I want to.

00:40:02.820 --> 00:40:08.550
And when I bring a tool
into a piece of sheet metal

00:40:08.550 --> 00:40:12.263
like this, how long does
it take for this to react?

00:40:12.263 --> 00:40:14.180
Well, essentially, at
the kinds of time scales

00:40:14.180 --> 00:40:16.790
we're talking about it, it
happens instantaneously.

00:40:16.790 --> 00:40:19.940
Compare that to how long
it takes to etch something

00:40:19.940 --> 00:40:21.950
or do certain
deposition processes.

00:40:21.950 --> 00:40:25.790
We're doing these nickel
plated micron scale tools.

00:40:25.790 --> 00:40:27.650
I just saw a
presentation last night,

00:40:27.650 --> 00:40:33.290
three days to make a tool
with 50 micron features on it.

00:40:33.290 --> 00:40:36.530
That's not what
you'd call smoking.

00:40:36.530 --> 00:40:38.400
So pretty low stuff.

00:40:38.400 --> 00:40:40.807
So rate-- we can argue
different things.

00:40:40.807 --> 00:40:43.140
You can have fast chemical
stuff and so forth and so on.

00:40:43.140 --> 00:40:46.530
But in general, there's
a speed issue here.

00:40:46.530 --> 00:40:47.030
What else?

00:40:53.360 --> 00:40:54.000
That's rate.

00:40:54.000 --> 00:40:56.666
What about quality?

00:40:56.666 --> 00:40:59.175
AUDIENCE: There are also
side effects of the process.

00:40:59.175 --> 00:41:00.508
DAVID HARDT: Yeah, side effects.

00:41:00.508 --> 00:41:01.090
I like that.

00:41:01.090 --> 00:41:02.950
Say a little bit more
about side effects.

00:41:02.950 --> 00:41:04.300
AUDIENCE: Well, you
got the theoretical.

00:41:04.300 --> 00:41:05.850
Maybe I want to cut
a perfect square.

00:41:05.850 --> 00:41:08.860
But the tool is not going
cut perfectly square edges.

00:41:08.860 --> 00:41:10.990
So depending on
what process I use,

00:41:10.990 --> 00:41:13.260
it may have undercutting
or rounded edges.

00:41:13.260 --> 00:41:14.260
DAVID HARDT: Absolutely.

00:41:14.260 --> 00:41:14.960
Absolutely.

00:41:14.960 --> 00:41:18.850
So there's something
here inherently related

00:41:18.850 --> 00:41:22.180
to quality, at least two
things that I can think of.

00:41:22.180 --> 00:41:27.700
One of them is related to
what I'll call diffusivity

00:41:27.700 --> 00:41:31.120
in the general sense, not just
purely in the thermal sense.

00:41:31.120 --> 00:41:35.455
If I apply a load to
something, as you're

00:41:35.455 --> 00:41:37.830
saying, if I apply a load to
something, even with a point

00:41:37.830 --> 00:41:40.412
tool, there's a stress
field that develops in it

00:41:40.412 --> 00:41:41.370
and that sort of thing.

00:41:41.370 --> 00:41:43.490
And this will be a
little bit facile,

00:41:43.490 --> 00:41:46.130
but I can make an argument
that says, OK, if I apply

00:41:46.130 --> 00:41:48.290
that load to here, it diffuses.

00:41:48.290 --> 00:41:52.375
And so I wanted to put
the load in a tiny point,

00:41:52.375 --> 00:41:53.750
but it actually
spread out and so

00:41:53.750 --> 00:41:55.130
I didn't get
exactly what I want.

00:41:55.130 --> 00:41:57.560
How about if I put a
thermal source in there?

00:42:00.920 --> 00:42:03.510
And the worst part is not
only does it dissipate,

00:42:03.510 --> 00:42:05.100
it just keeps dissipating.

00:42:05.100 --> 00:42:06.630
Probably never
reaches equilibrium.

00:42:06.630 --> 00:42:07.580
So it's really going.

00:42:07.580 --> 00:42:11.120
How about a chemical reaction?

00:42:13.998 --> 00:42:16.040
One of the problems we
have with that, of course,

00:42:16.040 --> 00:42:18.700
is if you do thick etching, you
have undercutting, because it--

00:42:18.700 --> 00:42:19.520
don't go there.

00:42:19.520 --> 00:42:20.440
But it's going there.

00:42:20.440 --> 00:42:23.240
It wants to go everywhere
and other things like that.

00:42:23.240 --> 00:42:27.010
So, again, you have different
diffusion characteristics here,

00:42:27.010 --> 00:42:29.060
depending on what's going on.

00:42:29.060 --> 00:42:31.810
So your ability to
control the shape,

00:42:31.810 --> 00:42:34.750
to get a minimum feature size,
to get a level of precision,

00:42:34.750 --> 00:42:39.370
will vary depending
on what you have here.

00:42:39.370 --> 00:42:44.080
And what else do you think
is important to quality?

00:42:44.080 --> 00:42:48.730
Obviously, my ability
to sort of direct things

00:42:48.730 --> 00:42:51.050
exactly where I want
them to be is important.

00:42:51.050 --> 00:42:53.500
What else do you think
is important to quality?

00:42:53.500 --> 00:42:56.890
AUDIENCE: How about
repeatability?

00:42:56.890 --> 00:43:03.000
DAVID HARDT: Yeah, sort of
the how well I can repeatedly

00:43:03.000 --> 00:43:06.120
apply this energy and have
the same thing happen.

00:43:06.120 --> 00:43:09.990
And that will vary again,
depending on the nature

00:43:09.990 --> 00:43:11.130
of the interfaces here.

00:43:11.130 --> 00:43:15.570
For example, if
thermal conductivity

00:43:15.570 --> 00:43:18.570
is a big part of it
and you're continually

00:43:18.570 --> 00:43:20.998
bringing new material into
contact with an old tool,

00:43:20.998 --> 00:43:23.040
it could be a different
heat transfer coefficient

00:43:23.040 --> 00:43:25.260
every time.

00:43:25.260 --> 00:43:26.940
In mechanical
cases, if I'm always

00:43:26.940 --> 00:43:30.390
coming back with the same tool,
except for long term where,

00:43:30.390 --> 00:43:34.742
maybe it's a little
bit more repeatable.

00:43:38.120 --> 00:43:39.760
There's one other
aspect to it now.

00:43:39.760 --> 00:43:42.470
Now, think a little
bit more about where

00:43:42.470 --> 00:43:44.090
we're going with the
class with things

00:43:44.090 --> 00:43:46.610
like statistical process
control, which says,

00:43:46.610 --> 00:43:48.170
let's find out
things that are not

00:43:48.170 --> 00:43:50.480
quite right with the
process and fix them

00:43:50.480 --> 00:43:52.400
and that will
improve our quality.

00:43:52.400 --> 00:43:56.330
And then in the more
active process optimization

00:43:56.330 --> 00:43:58.520
stuff, where we say
let's adjust things

00:43:58.520 --> 00:43:59.780
until they're really right.

00:43:59.780 --> 00:44:01.340
And then actually
at the end, we'll

00:44:01.340 --> 00:44:03.215
talk a little bit about
active control, where

00:44:03.215 --> 00:44:05.430
you're continually adjusting.

00:44:05.430 --> 00:44:10.800
But think of it this way,
because you will be thinking

00:44:10.800 --> 00:44:12.930
about this, here's the process.

00:44:12.930 --> 00:44:16.990
I just measured its result.
It's not good enough.

00:44:16.990 --> 00:44:19.380
I need to improve its quality.

00:44:19.380 --> 00:44:21.480
And I can't go
buy a new machine.

00:44:25.380 --> 00:44:28.910
So I have to have the
ability to adjust something.

00:44:28.910 --> 00:44:31.175
I have to have the ability
to change something.

00:44:31.175 --> 00:44:33.050
So think about what it
takes to change these,

00:44:33.050 --> 00:44:38.330
just as I was saying before with
the rolling case in the-- yeah,

00:44:38.330 --> 00:44:39.680
go ahead.

00:44:39.680 --> 00:44:42.050
AUDIENCE: I think for
some mechanical processes,

00:44:42.050 --> 00:44:44.230
it's easier to stop.

00:44:44.230 --> 00:44:49.520
But for the chemical, thermal,
even if you know the problem,

00:44:49.520 --> 00:44:50.990
it can be hard to
stop the process.

00:44:50.990 --> 00:44:51.770
It's hard to stop.

00:44:51.770 --> 00:44:52.850
DAVID HARDT: Yeah.

00:44:52.850 --> 00:44:55.225
I mean, I'm not trying to come
up with winners and losers

00:44:55.225 --> 00:44:57.633
here, because they all have
advantages and disadvantages.

00:44:57.633 --> 00:44:58.800
But you're absolutely right.

00:44:58.800 --> 00:45:00.570
There are big differences.

00:45:00.570 --> 00:45:03.830
If, for example, there's
a time aspect to it,

00:45:03.830 --> 00:45:06.980
mechanically I can pull that
force off, and it goes away.

00:45:06.980 --> 00:45:09.420
There's a wave that goes
through the material.

00:45:09.420 --> 00:45:12.320
But it goes at sonic rates.

00:45:12.320 --> 00:45:16.400
The thermal could take a long
time to change anything there

00:45:16.400 --> 00:45:20.750
and certainly the chemical
depending on scale

00:45:20.750 --> 00:45:22.760
could be like that too.

00:45:22.760 --> 00:45:25.280
So it is important to know
what these different energy

00:45:25.280 --> 00:45:29.108
sources are, if only to say,
OK, these types of things

00:45:29.108 --> 00:45:30.650
are more important
when you're really

00:45:30.650 --> 00:45:32.942
faced with the decision of
what's the best I could ever

00:45:32.942 --> 00:45:34.370
expect this process to do.

00:45:34.370 --> 00:45:35.330
I've worked.

00:45:35.330 --> 00:45:36.200
I've done SBC.

00:45:36.200 --> 00:45:37.978
I've done process optimization.

00:45:37.978 --> 00:45:39.770
There's a fundamental
limitation because it

00:45:39.770 --> 00:45:41.540
uses mechanical energy.

00:45:41.540 --> 00:45:46.070
I'll never be able to do
the following or so on.

00:45:46.070 --> 00:45:48.740
So you have these types
of things interacting

00:45:48.740 --> 00:45:51.380
with how we apply them.

00:45:51.380 --> 00:45:53.697
And, of course,
you could say it's

00:45:53.697 --> 00:45:56.030
impossible to use all these
in all their different ways.

00:45:56.030 --> 00:45:57.710
But actually that's
not entirely true.

00:46:01.490 --> 00:46:04.510
So kind of said all that.

00:46:04.510 --> 00:46:05.690
So let's go on.

00:46:05.690 --> 00:46:09.920
Now, this eye
chart is an attempt

00:46:09.920 --> 00:46:12.020
to take this entire
taxonomy and then

00:46:12.020 --> 00:46:18.520
just populate it, this
chart, with some examples.

00:46:18.520 --> 00:46:21.340
And some of these
are no-brainers.

00:46:21.340 --> 00:46:23.890
And some of them could be
a little bit controversial.

00:46:23.890 --> 00:46:26.350
So what this says
is cross here, this

00:46:26.350 --> 00:46:29.720
is the removal mode
of transformation.

00:46:29.720 --> 00:46:32.500
And here is addition, and
here's formation, and here's

00:46:32.500 --> 00:46:33.850
deformation.

00:46:33.850 --> 00:46:36.340
And in each case
then, it's divided

00:46:36.340 --> 00:46:40.150
in half with serial and
parallel and then four different

00:46:40.150 --> 00:46:41.120
energy sources.

00:46:41.120 --> 00:46:43.480
So you look at each one
and say, OK, where where's

00:46:43.480 --> 00:46:45.700
my favorite process?

00:46:45.700 --> 00:46:48.360
So serial, removal.

00:46:48.360 --> 00:46:52.260
Mechanical, cutting, grinding,
broaching, polishing,

00:46:52.260 --> 00:46:54.370
water jet.

00:46:54.370 --> 00:46:55.570
Water jet is mechanical.

00:46:55.570 --> 00:46:57.490
It happens to be
fluid mechanical.

00:46:57.490 --> 00:46:58.810
But it's mechanical.

00:46:58.810 --> 00:47:02.620
Electrical, serial,
wire EDM, take

00:47:02.620 --> 00:47:04.280
a single skinny
little wire, drill

00:47:04.280 --> 00:47:06.760
a hole, something like that.

00:47:06.760 --> 00:47:08.320
Put a whole bunch
of them together.

00:47:08.320 --> 00:47:11.140
Do it all at once, say
with a formed electrode.

00:47:11.140 --> 00:47:15.130
Now, it's parallel EDM.

00:47:15.130 --> 00:47:19.510
Photolithography with
a wet etch if you

00:47:19.510 --> 00:47:23.510
want to think of it that way is
basically a parallel process.

00:47:23.510 --> 00:47:25.960
It's etching everywhere at once.

00:47:25.960 --> 00:47:28.210
But you just happen to put
a mask in a few places.

00:47:28.210 --> 00:47:29.410
That's your tool.

00:47:29.410 --> 00:47:31.465
So that it can't resist.

00:47:35.260 --> 00:47:37.930
I think Duane has already
talked about chemical mechanical

00:47:37.930 --> 00:47:41.530
polishing, which I'm told by the
experts in the field is really

00:47:41.530 --> 00:47:44.140
mechanical polishing.

00:47:44.140 --> 00:47:46.460
So if we take the
C off of there,

00:47:46.460 --> 00:47:50.950
then CMP is a mechanical
parallel process.

00:47:50.950 --> 00:47:52.682
If it's chemical
mechanical process,

00:47:52.682 --> 00:47:54.640
then I guess it has to
show up in both of them.

00:47:54.640 --> 00:47:55.150
Which is it?

00:47:55.150 --> 00:47:56.290
Is it both?

00:47:56.290 --> 00:47:58.884
AUDIENCE: The chemistry
softens the surface, and not

00:47:58.884 --> 00:48:01.344
the mechanical abrasion.

00:48:01.344 --> 00:48:02.393
So you have both.

00:48:02.393 --> 00:48:03.060
DAVID HARDT: OK.

00:48:05.730 --> 00:48:08.430
But the chemical by itself
would not remove any material.

00:48:08.430 --> 00:48:09.600
AUDIENCE: That's correct.

00:48:09.600 --> 00:48:13.350
DAVID HARDT: Chemically
assisted mechanical processing.

00:48:13.350 --> 00:48:15.690
AUDIENCE: Strangely enough,
the abrasion by itself

00:48:15.690 --> 00:48:17.730
would no remove any material.

00:48:17.730 --> 00:48:19.204
DAVID HARDT: Any at all?

00:48:19.204 --> 00:48:21.150
That's interesting.

00:48:21.150 --> 00:48:22.770
OK.

00:48:22.770 --> 00:48:25.020
Well, that's actually
an example here

00:48:25.020 --> 00:48:29.550
of where the chemical part is
actually affecting properties,

00:48:29.550 --> 00:48:31.330
not unusual.

00:48:31.330 --> 00:48:32.880
So this is still
true if I'm only

00:48:32.880 --> 00:48:40.752
talking about the geometry
change part of it.

00:48:40.752 --> 00:48:42.210
There's a bunch of
other ones here.

00:48:42.210 --> 00:48:45.630
One of the interesting things
is if you come over here

00:48:45.630 --> 00:48:51.300
to something like hot isostatic
pressing, a powder metallurgy

00:48:51.300 --> 00:48:57.660
or powder ceramic process,
where you take uniform powders

00:48:57.660 --> 00:48:59.640
and compress them in a mold.

00:48:59.640 --> 00:49:01.500
So that's obviously
an addition process.

00:49:01.500 --> 00:49:04.330
And then I sinter them and
then I do other stuff to it.

00:49:04.330 --> 00:49:09.960
And I do it all at once
in a form tool, a can,

00:49:09.960 --> 00:49:10.770
and I'm all set.

00:49:13.540 --> 00:49:16.620
That's clearly a
parallel process.

00:49:16.620 --> 00:49:23.760
And the result depends primarily
on the shape of the can.

00:49:23.760 --> 00:49:28.860
Then Ely Sachs at MIT kind
of invented this process

00:49:28.860 --> 00:49:30.720
called 3D printing.

00:49:30.720 --> 00:49:33.870
And if you look at 3D printing,
it's not exactly this.

00:49:33.870 --> 00:49:36.270
But it's essentially the
same as saying, OK, I'm

00:49:36.270 --> 00:49:40.110
going to put down each
piece of that powder one

00:49:40.110 --> 00:49:42.600
piece at a time.

00:49:42.600 --> 00:49:46.170
He does it by joining the
powder one drop at a time.

00:49:46.170 --> 00:49:47.153
He puts a binder on it.

00:49:47.153 --> 00:49:48.570
Instead of binding
it all at once,

00:49:48.570 --> 00:49:49.890
he puts a drop down at a time.

00:49:49.890 --> 00:49:53.430
It's an ink jet type process.

00:49:53.430 --> 00:49:55.200
Now, it becomes
a serial process.

00:49:55.200 --> 00:49:59.355
And think about the difference
between these two processes.

00:49:59.355 --> 00:50:01.200
If I've got this
big press that has

00:50:01.200 --> 00:50:06.300
to push this powder into
a green compact, big piece

00:50:06.300 --> 00:50:10.140
of equipment, going to
have an expensive tool,

00:50:10.140 --> 00:50:12.750
whereas this 3D printing,
it's a desktop--

00:50:12.750 --> 00:50:14.430
thing are desktop
versions of it--

00:50:14.430 --> 00:50:18.180
desktop thing, all I have
to do is type in a CAD file.

00:50:18.180 --> 00:50:20.430
It does a section file of it.

00:50:20.430 --> 00:50:22.050
It creates this thing.

00:50:22.050 --> 00:50:25.272
And I've got essentially
equivalent parts, not exactly

00:50:25.272 --> 00:50:26.980
equivalent, essentially
equivalent parts.

00:50:26.980 --> 00:50:32.230
So clearly this serial
process is much more flexible,

00:50:32.230 --> 00:50:33.070
really great.

00:50:33.070 --> 00:50:36.485
And I probably can
control the quality

00:50:36.485 --> 00:50:38.110
better because of
it's not quite right,

00:50:38.110 --> 00:50:40.085
I just change the
program a little bit.

00:50:40.085 --> 00:50:42.460
If it was a little bit too
wide the last one I made,

00:50:42.460 --> 00:50:44.312
I said, well, let's
reprogram the geometry

00:50:44.312 --> 00:50:45.520
and bring it in a little bit.

00:50:45.520 --> 00:50:46.680
And it just does it.

00:50:46.680 --> 00:50:50.800
Whereas if I'm doing the
hot isostatic pressing,

00:50:50.800 --> 00:50:54.310
I've got this big expensive
tool, and it's not quite right

00:50:54.310 --> 00:50:58.330
and I can't heat it and cool
it like I did with the rolls

00:50:58.330 --> 00:51:02.250
on the aluminum, I'm
kind of stuck with it.

00:51:02.250 --> 00:51:06.900
So why would anybody do HIP,
when you've got 3D printing?

00:51:06.900 --> 00:51:08.660
AUDIENCE: Rate and cost.

00:51:08.660 --> 00:51:11.000
DAVID HARDT: Rate
and cost, absolutely.

00:51:11.000 --> 00:51:15.290
You do the hot isostatic
pressing in minutes,

00:51:15.290 --> 00:51:18.460
seconds, that sort of thing.

00:51:18.460 --> 00:51:20.440
The 3D printing, the
fast ones, are going

00:51:20.440 --> 00:51:23.590
to take you minutes to hours.

00:51:23.590 --> 00:51:25.390
And it's volume dependent.

00:51:25.390 --> 00:51:29.440
Whereas with the HIP,
it's much, much less so.

00:51:29.440 --> 00:51:33.370
In general, why does anybody
do parallel processing?

00:51:33.370 --> 00:51:35.230
Because it's really fast.

00:51:35.230 --> 00:51:37.000
And the example I
always like to give

00:51:37.000 --> 00:51:41.380
is imagine if someone took
a block of steel or better

00:51:41.380 --> 00:51:44.170
yet aluminum and
said, OK, I'm going

00:51:44.170 --> 00:51:46.990
to make all the body panels for
your car, the thin little body

00:51:46.990 --> 00:51:52.240
panels, with my
machine tool over here.

00:51:52.240 --> 00:51:56.110
Just imagine what it would
take to build that, as opposed

00:51:56.110 --> 00:52:01.190
to going to a modern
high speed stamping shop.

00:52:01.190 --> 00:52:03.800
Has anybody seen automobile
body stamping going on?

00:52:03.800 --> 00:52:04.300
Yeah.

00:52:04.300 --> 00:52:07.570
How long does it take to
make a side panel for a car

00:52:07.570 --> 00:52:11.770
with like integral door frames
and all that other stuff?

00:52:11.770 --> 00:52:14.990
Each given operation probably
takes about 2 seconds.

00:52:14.990 --> 00:52:16.397
There's multiple stages.

00:52:16.397 --> 00:52:18.730
So maybe we'll give it 10
seconds by the time it's done.

00:52:18.730 --> 00:52:21.730
But blank metal in one end, and
these huge complex structures

00:52:21.730 --> 00:52:24.670
coming out the other
end at incredible rates.

00:52:24.670 --> 00:52:28.387
You're not going to do
that with a serial process.

00:52:28.387 --> 00:52:30.220
Although I will say
they've done some things

00:52:30.220 --> 00:52:32.678
with laser cutting and laser
welding where the thing really

00:52:32.678 --> 00:52:33.910
whips along.

00:52:33.910 --> 00:52:36.010
But it's still got a ways to go.

00:52:36.010 --> 00:52:37.730
And it's much more costly.

00:52:37.730 --> 00:52:40.510
So all these things do matter.

00:52:40.510 --> 00:52:42.850
And they come in different ways.

00:52:42.850 --> 00:52:48.258
Oh, just because I always like
to do this-- stereolithography,

00:52:48.258 --> 00:52:49.050
how does that work?

00:52:52.860 --> 00:52:54.240
It was the original--

00:52:54.240 --> 00:52:56.800
I think-- the original
rapid prototyping process.

00:52:56.800 --> 00:52:58.500
Anybody know what stereo--

00:52:58.500 --> 00:53:00.250
what's the trade
name for that now?

00:53:00.250 --> 00:53:01.980
That's the most
commercial of them.

00:53:01.980 --> 00:53:03.873
There's a machine
right down the far end

00:53:03.873 --> 00:53:05.790
of this hallway in the
architecture department

00:53:05.790 --> 00:53:08.520
that's an STL machine.

00:53:08.520 --> 00:53:09.900
Stereolithography?

00:53:09.900 --> 00:53:11.590
I guess it's become passe now.

00:53:11.590 --> 00:53:12.090
Yeah, Adam.

00:53:12.090 --> 00:53:14.344
AUDIENCE: It's just-- I
don't know what liquid is,

00:53:14.344 --> 00:53:17.620
but you select part of that
liquid material with the laser.

00:53:17.620 --> 00:53:19.560
And then it runs that.

00:53:19.560 --> 00:53:21.660
DAVID HARDT: Yeah, that's right.

00:53:21.660 --> 00:53:24.450
There are these photo
curable polymers

00:53:24.450 --> 00:53:28.230
that will be in a liquid
state at room temperature

00:53:28.230 --> 00:53:30.310
and with just regular
light on them.

00:53:30.310 --> 00:53:32.790
But if you hit them with
the right wavelength light,

00:53:32.790 --> 00:53:34.740
they will polymerize.

00:53:34.740 --> 00:53:37.560
They're UV curable
polymers that are

00:53:37.560 --> 00:53:41.400
used for a lot of things,
like contact lenses and things

00:53:41.400 --> 00:53:43.720
like that.

00:53:43.720 --> 00:53:47.010
But this particular
case, what they said

00:53:47.010 --> 00:53:49.710
is, well, instead of
putting it in a mold

00:53:49.710 --> 00:53:52.020
and flashing it with
UV light everywhere,

00:53:52.020 --> 00:53:54.960
what if we put a little
tiny spot of UV light on it?

00:53:54.960 --> 00:53:58.080
Will it form a little
ball of polymer?

00:53:58.080 --> 00:54:00.390
And the answer was sort of yeah.

00:54:00.390 --> 00:54:03.030
So you have this bath.

00:54:03.030 --> 00:54:04.950
I haven't walked by
there this month.

00:54:04.950 --> 00:54:08.967
But there used to be one of
these glass walled classrooms

00:54:08.967 --> 00:54:10.800
down here at the far
end of the fourth floor

00:54:10.800 --> 00:54:13.740
in the architecture
department this white box

00:54:13.740 --> 00:54:15.780
with a clear window in it.

00:54:15.780 --> 00:54:17.050
And it's kind of lit up.

00:54:17.050 --> 00:54:19.050
And there's usually
something going on in there.

00:54:19.050 --> 00:54:21.870
And what you'll see is a bath
of nearly colorless liquid

00:54:21.870 --> 00:54:23.575
with a platen on it.

00:54:23.575 --> 00:54:25.950
And the platen starts at the
top with the very thin layer

00:54:25.950 --> 00:54:28.620
of fluid on it.

00:54:28.620 --> 00:54:30.810
And this laser,
usually not visible,

00:54:30.810 --> 00:54:33.000
but this laser will be
dancing across the surface,

00:54:33.000 --> 00:54:35.340
tracing an outline.

00:54:35.340 --> 00:54:38.190
And that outline
will become solid.

00:54:38.190 --> 00:54:41.820
And once it's polymerized, it'll
move it down and do it again.

00:54:41.820 --> 00:54:46.290
So it's basically causing
a chemical reaction,

00:54:46.290 --> 00:54:50.850
or in this case a polymerization
reaction, at a very small area

00:54:50.850 --> 00:54:55.650
by hitting it with the right
amount of activation energy.

00:54:55.650 --> 00:54:57.480
So that's stereolithography.

00:54:57.480 --> 00:55:01.710
That would be a serial
chemical formation process.

00:55:01.710 --> 00:55:03.210
You're going from
liquid to solid.

00:55:06.480 --> 00:55:09.180
We also do-- in fact,
I was here last night

00:55:09.180 --> 00:55:11.400
listening to a PhD
thesis from Singapore,

00:55:11.400 --> 00:55:15.300
where what they did is they
took a mold, poured a UV curable

00:55:15.300 --> 00:55:19.200
polymer on it, flashed
it with a UV lamp

00:55:19.200 --> 00:55:23.460
over the whole thing for
about 30 seconds or something,

00:55:23.460 --> 00:55:26.610
I think, and then
pulled it apart.

00:55:26.610 --> 00:55:31.060
And it was all about the mold,
the mold quality, and filling

00:55:31.060 --> 00:55:33.030
and that sort of thing.

00:55:33.030 --> 00:55:38.580
And 90% of her thesis was
about-- well, not 90%--

00:55:38.580 --> 00:55:42.270
a good percentage of her thesis
was about making the molds

00:55:42.270 --> 00:55:44.520
and how well could I extract
this thing from the mold?

00:55:44.520 --> 00:55:49.525
So physically, the same
material, the same process.

00:55:49.525 --> 00:55:51.150
But in this case, a
really good example

00:55:51.150 --> 00:55:52.950
of going from a
directed energy source

00:55:52.950 --> 00:55:56.220
that was a point to a
directed energy source that

00:55:56.220 --> 00:55:58.530
was really this entire surface.

00:55:58.530 --> 00:56:00.480
So I guess my
point is, and now I

00:56:00.480 --> 00:56:02.520
guess it's getting
a little bit dated,

00:56:02.520 --> 00:56:07.920
but when these rapid prototyping
processes were all the rage,

00:56:07.920 --> 00:56:12.180
it was interesting to notice
that on this taxonomy,

00:56:12.180 --> 00:56:16.478
they simply, with the fact
that some people realized

00:56:16.478 --> 00:56:18.520
and it's been done for a
lot of different things,

00:56:18.520 --> 00:56:20.940
not just these, but
people realized that I

00:56:20.940 --> 00:56:22.870
can take a parallel process.

00:56:22.870 --> 00:56:27.030
And if I serialize
it, it all of a sudden

00:56:27.030 --> 00:56:30.270
becomes a rapid
free form method.

00:56:30.270 --> 00:56:32.430
It doesn't need any tooling.

00:56:32.430 --> 00:56:33.810
And so you've got that example.

00:56:33.810 --> 00:56:36.810
You've got what
turns out to be--

00:56:36.810 --> 00:56:38.280
I guess I'd call it--

00:56:38.280 --> 00:56:40.620
it's sort of like
a reaction molding

00:56:40.620 --> 00:56:45.250
parallel reaction process to
a serial reaction process.

00:56:45.250 --> 00:56:48.870
There's also with
powdered metals,

00:56:48.870 --> 00:56:51.090
you can sinter them
thermally to get the shape.

00:56:51.090 --> 00:56:54.420
And when people decided
they could sinter it locally

00:56:54.420 --> 00:56:58.110
with a laser, you could
take a powder of metal

00:56:58.110 --> 00:56:59.850
and zap it with a
laser and just get

00:56:59.850 --> 00:57:03.390
things to sinter to
each other locally.

00:57:03.390 --> 00:57:06.390
So they got they changed
the whole equation on how

00:57:06.390 --> 00:57:08.910
the process is controlled,
which has effects

00:57:08.910 --> 00:57:11.820
on all these different things.

00:57:11.820 --> 00:57:14.310
Last thing, because
we're not going

00:57:14.310 --> 00:57:16.470
to talk about it much
for the rest of the term,

00:57:16.470 --> 00:57:20.010
but because it has been an
extremely important part

00:57:20.010 --> 00:57:22.500
of manufacturing,
interestingly fading

00:57:22.500 --> 00:57:24.570
in recent years for good
and sufficient reasons,

00:57:24.570 --> 00:57:25.470
is flexibility.

00:57:28.250 --> 00:57:31.850
What do you think
about if I were

00:57:31.850 --> 00:57:33.950
to go to different
corners of this taxonomy

00:57:33.950 --> 00:57:40.490
and say, all right, so let's put
a flexibility overlay on this?

00:57:40.490 --> 00:57:44.660
And we'll talk about
flexibility here--

00:57:44.660 --> 00:57:49.490
I guess, well, two common
ways of looking at flexibility

00:57:49.490 --> 00:57:51.430
are sort of product flexibility.

00:57:51.430 --> 00:57:54.150
So in our case, it's going to be
geometry flexibility, geometry

00:57:54.150 --> 00:57:56.270
and material flexibility.

00:57:56.270 --> 00:57:59.630
And the other would be
quantity flexibility,

00:57:59.630 --> 00:58:05.240
ability to vary the capacity
or output of something

00:58:05.240 --> 00:58:08.000
without having big
cost variances.

00:58:08.000 --> 00:58:09.920
But let's do the first one.

00:58:09.920 --> 00:58:12.200
We won't get into
that second one.

00:58:12.200 --> 00:58:14.660
For flexibility of
this process to work

00:58:14.660 --> 00:58:18.110
with different materials
in different shapes,

00:58:18.110 --> 00:58:20.690
obviously, the extreme
example of flexibility

00:58:20.690 --> 00:58:24.380
is every part's different.

00:58:24.380 --> 00:58:28.800
Every part is different,
and the setup time is zero.

00:58:28.800 --> 00:58:33.780
So mass production rates,
mass production cost savings,

00:58:33.780 --> 00:58:41.650
but custom part capability.

00:58:41.650 --> 00:58:43.400
These are lies somewhere
in between there.

00:58:43.400 --> 00:58:46.940
But if I look at this
broadly, what do you think

00:58:46.940 --> 00:58:49.420
is the most flexible?

00:58:49.420 --> 00:58:52.240
Or most flexible
or more flexible?

00:58:52.240 --> 00:58:55.000
If I look at the
taxonomy, what area

00:58:55.000 --> 00:58:58.705
would you look at
for high flexibility?

00:58:58.705 --> 00:59:00.420
AUDIENCE: Serial processes.

00:59:00.420 --> 00:59:01.670
DAVID HARDT: Serial processes.

00:59:01.670 --> 00:59:03.270
And why?

00:59:03.270 --> 00:59:08.440
AUDIENCE: It's like if you're
working on a local [INAUDIBLE]

00:59:08.440 --> 00:59:13.770
So I think it gives you more
flexibility [INAUDIBLE] I

00:59:13.770 --> 00:59:17.640
guess feature by feature
and also part by part.

00:59:17.640 --> 00:59:18.850
DAVID HARDT: Yeah, exactly.

00:59:18.850 --> 00:59:23.886
I mean think of, again, the
ubiquitous example is a--

00:59:23.886 --> 00:59:30.530
I'm trying to think of some
consumer-oriented parts

00:59:30.530 --> 00:59:31.030
of this.

00:59:31.030 --> 00:59:34.380
But I'm not coming up with them.

00:59:34.380 --> 00:59:37.260
But, yeah, a machine tool.

00:59:37.260 --> 00:59:38.760
Take a machine tool
with an end mill

00:59:38.760 --> 00:59:40.802
and it can write your name
in a piece of plastic.

00:59:43.950 --> 00:59:46.680
Once I programmed your name
in there and then your name

00:59:46.680 --> 00:59:48.840
and then your name,
there's absolutely

00:59:48.840 --> 00:59:52.530
no overhead to me switching,
because all I have to change

00:59:52.530 --> 00:59:54.000
is the trajectory of the tool.

00:59:54.000 --> 00:59:58.650
And nowadays with CNC control,
that's a zero cost changeover.

00:59:58.650 --> 01:00:02.040
There's a minimal cost
in creating the programs

01:00:02.040 --> 01:00:03.730
for simple things like that.

01:00:03.730 --> 01:00:08.280
So, yeah, obviously, something
that is purely serial

01:00:08.280 --> 01:00:09.420
makes a big difference.

01:00:09.420 --> 01:00:12.360
And I'm going to add one other
thing, because if I hurry up

01:00:12.360 --> 01:00:15.150
I'll get at least
to mention this.

01:00:15.150 --> 01:00:23.190
I'll make the case that cutting
metal mechanically is maybe

01:00:23.190 --> 01:00:27.060
the purest form of geometry
transformation, pure geometry

01:00:27.060 --> 01:00:30.460
transformation with
no side effects.

01:00:30.460 --> 01:00:34.750
So that also makes it very
flexible in that sense.

01:00:34.750 --> 01:00:37.060
It's almost as if
all I have to do

01:00:37.060 --> 01:00:40.060
is reprogram the trajectory
and nothing else changes

01:00:40.060 --> 01:00:42.560
and everything's fine.

01:00:42.560 --> 01:00:43.880
So we'll start from that.

01:00:43.880 --> 01:00:49.450
And then we move to
other changes here.

01:00:49.450 --> 01:00:54.280
Let's go to deformation,
serial defamation, bending.

01:00:58.113 --> 01:00:59.946
I'm going to bend this,
three-point bending.

01:01:02.980 --> 01:01:05.470
I'm calling it serial,
because most of the action

01:01:05.470 --> 01:01:07.990
is happening right here.

01:01:07.990 --> 01:01:10.210
And I'm not using form tools.

01:01:10.210 --> 01:01:12.460
You can even think of this
as using knife-edge tools,

01:01:12.460 --> 01:01:14.980
and I would get sort
of the same shape.

01:01:14.980 --> 01:01:17.680
And the main thing
that determines

01:01:17.680 --> 01:01:22.930
the shape, the main thing is how
far I push down in the middle.

01:01:22.930 --> 01:01:25.540
So that's pretty flexible.

01:01:25.540 --> 01:01:29.320
I mean if I want 45 degrees,
I just take it down less.

01:01:29.320 --> 01:01:33.200
If I go down farther,
I take it down more.

01:01:33.200 --> 01:01:36.580
But as can be shown--

01:01:36.580 --> 01:01:38.890
and it is in the notes, but
we probably won't have time

01:01:38.890 --> 01:01:39.640
to go through it--

01:01:39.640 --> 01:01:43.140
as can be shown, it
makes a huge difference

01:01:43.140 --> 01:01:47.100
what this material is and
what the underlying machine

01:01:47.100 --> 01:01:48.990
structure looks like.

01:01:48.990 --> 01:01:52.050
And it also-- and
this really kind of

01:01:52.050 --> 01:01:54.870
gets us to the next
theme here-- it also

01:01:54.870 --> 01:01:59.370
depends very strongly on the
uniformity of this material.

01:01:59.370 --> 01:02:02.640
So I'll make the statement
that the natural variations out

01:02:02.640 --> 01:02:06.900
of an aluminum mill, maybe not
the Alcoa, Tennessee, mill,

01:02:06.900 --> 01:02:10.290
but the other ones,
of aluminum is enough

01:02:10.290 --> 01:02:13.830
that it can completely
confound precise control

01:02:13.830 --> 01:02:15.570
of a process like this.

01:02:15.570 --> 01:02:17.820
The natural variation
of a piece of aluminum

01:02:17.820 --> 01:02:20.250
like this that I'm
going to turn just

01:02:20.250 --> 01:02:21.720
doesn't make any difference.

01:02:21.720 --> 01:02:23.310
Sure, if I put a
different alloy in,

01:02:23.310 --> 01:02:26.820
or I sneak in steel or
titanium instead of aluminum,

01:02:26.820 --> 01:02:28.770
that's a big mess up.

01:02:28.770 --> 01:02:33.360
But if I just say,
well, you know, here's

01:02:33.360 --> 01:02:34.200
a piece of aluminum.

01:02:34.200 --> 01:02:35.790
We did it in this
mill over here,

01:02:35.790 --> 01:02:39.630
and it came out with the
following properties.

01:02:39.630 --> 01:02:41.100
It's elasticity.

01:02:41.100 --> 01:02:43.050
It's yield strength.

01:02:43.050 --> 01:02:45.390
It's strain hardening
characteristics.

01:02:45.390 --> 01:02:47.893
And then I get supposedly
the same material

01:02:47.893 --> 01:02:48.810
from a different mill.

01:02:48.810 --> 01:02:50.970
And they're 10% different.

01:02:50.970 --> 01:02:54.390
Not the elasticity, by the
way, but all the others.

01:02:54.390 --> 01:02:56.670
10% different.

01:02:56.670 --> 01:03:00.210
Huge difference here,
virtually no difference

01:03:00.210 --> 01:03:01.360
for the machining.

01:03:01.360 --> 01:03:03.060
So there's an
inherent sensitivity

01:03:03.060 --> 01:03:05.050
of the processes to that.

01:03:05.050 --> 01:03:09.240
So if I look just at this
purely serial parallel thing,

01:03:09.240 --> 01:03:11.910
yeah, serial is
better than parallel.

01:03:11.910 --> 01:03:14.310
I think we can show in
general that removal

01:03:14.310 --> 01:03:17.358
tends to be the least sensitive
to material properties.

01:03:17.358 --> 01:03:19.650
It tends to be the least
sensitive of all these things.

01:03:19.650 --> 01:03:23.040
So serial removal kind
of wins in most cases

01:03:23.040 --> 01:03:24.210
in terms of flexibility.

01:03:26.760 --> 01:03:29.280
Deformation is
generally a killer

01:03:29.280 --> 01:03:31.050
in terms of flexibility
for these reasons

01:03:31.050 --> 01:03:32.320
that we've talked about.

01:03:32.320 --> 01:03:35.370
And in many cases--

01:03:35.370 --> 01:03:40.500
well, certainly anything that
involves a parallel energy

01:03:40.500 --> 01:03:44.310
exchange sort of starts
off behind the eight ball

01:03:44.310 --> 01:03:45.580
with respect to flexibility.

01:03:45.580 --> 01:03:46.110
Simon.

01:03:46.110 --> 01:03:48.610
AUDIENCE: What about the
chemical [INAUDIBLE]??

01:03:48.610 --> 01:03:51.360
I mean, they throw in
all sorts of parts.

01:03:51.360 --> 01:03:56.070
And they come out
with the coating.

01:03:56.070 --> 01:03:58.470
And I mean there
is no dependency--

01:03:58.470 --> 01:04:00.370
DAVID HARDT: That's
a very good point.

01:04:00.370 --> 01:04:02.040
That's a really good point.

01:04:02.040 --> 01:04:03.243
I'm glad you raised that.

01:04:03.243 --> 01:04:04.410
I hadn't thought about that.

01:04:04.410 --> 01:04:08.730
But a process like this, some
of these addition processes,

01:04:08.730 --> 01:04:11.910
any sort of a plating
or coating process

01:04:11.910 --> 01:04:20.010
where you put stuff into
a liquid or vapor bath,

01:04:20.010 --> 01:04:22.080
it doesn't care about
the shape, does it?

01:04:22.080 --> 01:04:24.053
I mean at a microscopic
level it might.

01:04:24.053 --> 01:04:24.970
But, yeah, it doesn't.

01:04:24.970 --> 01:04:25.960
That's a good point.

01:04:25.960 --> 01:04:29.815
So in terms of
geometry, which there's

01:04:29.815 --> 01:04:31.440
the underlying
geometry, the substrate,

01:04:31.440 --> 01:04:35.280
plus whatever thickness
you add, to my knowledge,

01:04:35.280 --> 01:04:38.280
if I have a big piece over
here and right next to it

01:04:38.280 --> 01:04:40.693
a little piece over here,
another one over here,

01:04:40.693 --> 01:04:42.110
they're independent
of each other.

01:04:42.110 --> 01:04:43.990
So they all deposit
at the same rate,

01:04:43.990 --> 01:04:45.490
unless you do
something really dumb.

01:04:45.490 --> 01:04:46.800
It's a good point.

01:04:46.800 --> 01:04:47.340
Good point.

01:04:47.340 --> 01:04:50.720
I don't know how
to pull that out.

01:04:50.720 --> 01:04:54.580
But in that sense--

01:04:54.580 --> 01:04:56.330
well, to relate it to
what we said before,

01:04:56.330 --> 01:05:00.880
the energy exchange
area is determined

01:05:00.880 --> 01:05:04.980
by the shape of the substrate,
not the shape of the tool.

01:05:04.980 --> 01:05:07.820
And so in a situation like that,
the shape of the substrate,

01:05:07.820 --> 01:05:12.080
if it's right and I can change
it each time I put it in,

01:05:12.080 --> 01:05:14.060
I get high
flexibility with that.

01:05:14.060 --> 01:05:14.980
Yeah, good point.

01:05:17.800 --> 01:05:20.450
I have to remember that.

01:05:20.450 --> 01:05:23.060
OK, enough on the taxonomy.

01:05:29.310 --> 01:05:35.030
Now, the last thing I want to
talk about is what goes wrong.

01:05:35.030 --> 01:05:36.885
What we've been
talking about so far--

01:05:36.885 --> 01:05:38.260
we've made some
reference to it--

01:05:38.260 --> 01:05:39.980
what we've been
talking about so far

01:05:39.980 --> 01:05:42.500
is how to make major changes.

01:05:42.500 --> 01:05:47.742
I want to go from this diameter
down to half that diameter.

01:05:47.742 --> 01:05:48.950
Well, I have to push this in.

01:05:48.950 --> 01:05:50.330
I have to make chips.

01:05:50.330 --> 01:05:52.520
And I have to push
this down so far.

01:05:52.520 --> 01:05:54.920
But what I want to talk
about is, why doesn't this

01:05:54.920 --> 01:05:58.340
always come out to the
same angle every time?

01:05:58.340 --> 01:06:01.640
Why is the third part
out of the run of 50

01:06:01.640 --> 01:06:05.300
different from the 24th
part and that sort of thing?

01:06:05.300 --> 01:06:11.060
And why, when I machine this,
is it not the same diameter

01:06:11.060 --> 01:06:13.740
everywhere?

01:06:13.740 --> 01:06:19.280
And so it gets into looking
at the sources of variation.

01:06:19.280 --> 01:06:23.840
And by our definition
of a process parameter,

01:06:23.840 --> 01:06:26.140
we're going to say,
well, it-- oops.

01:06:26.140 --> 01:06:29.040
Don't do that.

01:06:29.040 --> 01:06:30.980
I'll say yes, but what did I do?

01:06:38.498 --> 01:06:49.410
By our definition of this,
it's all wrapped up somehow--

01:06:49.410 --> 01:06:51.840
well, in the end,
it's all wrapped up

01:06:51.840 --> 01:06:54.060
in this functional
relationship, but primarily

01:06:54.060 --> 01:06:57.180
through these
process parameters.

01:06:57.180 --> 01:06:59.910
So we're going to use those
to try and understand it.

01:06:59.910 --> 01:07:02.400
So let's break down
these process parameters

01:07:02.400 --> 01:07:10.058
in a little bit more
of a logical fashion.

01:07:10.058 --> 01:07:11.850
And I'm going to do it
in the following way

01:07:11.850 --> 01:07:13.933
with the following definitions.

01:07:13.933 --> 01:07:15.600
First of all, we're
going to have alphas

01:07:15.600 --> 01:07:17.230
associated with the equipment.

01:07:17.230 --> 01:07:19.230
And then we're going to
have alphas associated--

01:07:19.230 --> 01:07:20.438
let me go back here actually.

01:07:20.438 --> 01:07:21.970
I need my picture.

01:07:21.970 --> 01:07:24.930
So we're going to have these
parameters associated just

01:07:24.930 --> 01:07:26.673
with the equipment.

01:07:26.673 --> 01:07:28.590
And I think I've already
explained why we want

01:07:28.590 --> 01:07:30.460
to distinguish between the two.

01:07:30.460 --> 01:07:33.780
So now, we're talking about
properties, variables,

01:07:33.780 --> 01:07:36.330
things I can adjust,
things that happen

01:07:36.330 --> 01:07:38.820
that are staying with that
piece of iron that's bolted down

01:07:38.820 --> 01:07:39.880
to the factory floor.

01:07:39.880 --> 01:07:44.232
And then a similar
set, but for the stuff

01:07:44.232 --> 01:07:45.690
that goes in and
out of the machine

01:07:45.690 --> 01:07:47.107
or that flows
through the machine.

01:07:47.107 --> 01:07:48.160
It doesn't go in and out.

01:07:48.160 --> 01:07:50.740
So that's why we
distinguish between them.

01:07:50.740 --> 01:07:52.890
So let's just focus--

01:07:52.890 --> 01:07:56.520
I don't know, let's focus on
the machine, on the equipment.

01:07:56.520 --> 01:08:00.180
If I think of two broad
classifications of a machine--

01:08:00.180 --> 01:08:03.060
well, let's see,
try this exercise.

01:08:03.060 --> 01:08:06.000
I think everybody's familiar
with a machine tool.

01:08:06.000 --> 01:08:10.190
Simple lathe, it's sitting
here in front of us.

01:08:10.190 --> 01:08:12.090
And it's not doing a thing.

01:08:12.090 --> 01:08:12.930
It's not moving.

01:08:12.930 --> 01:08:15.780
It's not even plugged in.

01:08:15.780 --> 01:08:18.330
What alphas, what
characteristics

01:08:18.330 --> 01:08:20.939
can you tell me about it
just by looking at it?

01:08:20.939 --> 01:08:21.779
It's not moving.

01:08:24.487 --> 01:08:27.029
If you were to describe it from
an engineering point of view,

01:08:27.029 --> 01:08:29.237
what are the kinds of things
you'd worry about on it?

01:08:29.237 --> 01:08:30.100
AUDIENCE: Stiffness.

01:08:30.100 --> 01:08:31.710
DAVID HARDT: Stiffness.

01:08:31.710 --> 01:08:32.460
How stiff is it?

01:08:32.460 --> 01:08:33.235
Does it deform?

01:08:33.235 --> 01:08:33.735
Yeah.

01:08:36.710 --> 01:08:37.979
That's almost it.

01:08:37.979 --> 01:08:39.229
That's a good one right there.

01:08:39.229 --> 01:08:45.560
Stiffness, straightness,
some details

01:08:45.560 --> 01:08:47.973
of mechanical
characteristics of it.

01:08:47.973 --> 01:08:49.640
AUDIENCE: Thermal
expansion coefficient.

01:08:49.640 --> 01:08:51.682
DAVID HARDT: Yeah, thermal
expansion coefficient,

01:08:51.682 --> 01:08:52.609
very important stuff.

01:08:52.609 --> 01:08:55.010
All those things that
basically go into

01:08:55.010 --> 01:08:56.990
is the structure
and where I'm going

01:08:56.990 --> 01:08:59.600
to direct that energy source
where I think it's going to be?

01:09:02.390 --> 01:09:04.340
Now, I turn it on.

01:09:04.340 --> 01:09:05.779
It starts moving.

01:09:05.779 --> 01:09:08.140
And I start machining.

01:09:08.140 --> 01:09:10.140
Forget about the chips
and all that other stuff,

01:09:10.140 --> 01:09:11.550
but just the machine itself.

01:09:11.550 --> 01:09:15.064
What other characteristics
of the machine

01:09:15.064 --> 01:09:16.439
are you now going
to worry about?

01:09:19.229 --> 01:09:20.550
AUDIENCE: The turning speed.

01:09:20.550 --> 01:09:23.670
DAVID HARDT: Yeah, the
turning speed, exactly.

01:09:23.670 --> 01:09:24.600
The speed.

01:09:24.600 --> 01:09:25.109
What else?

01:09:29.640 --> 01:09:31.290
AUDIENCE: Accuracy
of the rotation.

01:09:31.290 --> 01:09:35.430
DAVID HARDT: The accuracy of
the rotation, yeah, OK, yeah.

01:09:35.430 --> 01:09:38.580
But now we've got the
turning speed, now

01:09:38.580 --> 01:09:41.609
what's the difference from
the machine's point of view

01:09:41.609 --> 01:09:45.189
if I've got the tool
and I'm coming along--

01:09:45.189 --> 01:09:47.910
I think you guys can see this--
if I'm coming along like this?

01:09:47.910 --> 01:09:50.550
Right now I'm
characterizing the machine.

01:09:50.550 --> 01:09:52.200
And now I'm characterizing
the machine.

01:09:52.200 --> 01:09:55.140
I've hit the material, and
I'm starting the machine.

01:09:55.140 --> 01:09:59.240
What's the difference
between when I'm here

01:09:59.240 --> 01:10:01.562
and when I'm here?

01:10:01.562 --> 01:10:03.354
AUDIENCE: Negative vibrations.

01:10:03.354 --> 01:10:05.310
DAVID HARDT: Yeah,
you've skipped something.

01:10:05.310 --> 01:10:06.150
Go ahead, I think--

01:10:06.150 --> 01:10:07.863
AUDIENCE: [INAUDIBLE]

01:10:07.863 --> 01:10:09.780
DAVID HARDT: Yeah, you
talked about the speed.

01:10:09.780 --> 01:10:13.860
So I have a speed here, and
I have a rotation speed.

01:10:13.860 --> 01:10:17.178
But if there's no force,
like there's no power,

01:10:17.178 --> 01:10:17.970
there's no nothing.

01:10:17.970 --> 01:10:19.980
No nothing much happening.

01:10:19.980 --> 01:10:21.510
That's good grammar.

01:10:21.510 --> 01:10:26.050
But if I come along here, all
of a sudden the force changes.

01:10:26.050 --> 01:10:29.820
So as a minimum, you've got
these two sets of variables.

01:10:29.820 --> 01:10:32.320
You've got how stiff
is the machine?

01:10:32.320 --> 01:10:35.160
How does it react to
things like forces?

01:10:35.160 --> 01:10:38.220
How straight is the machine
if I move this way relative

01:10:38.220 --> 01:10:41.160
to the axis of rotation--
that's your accuracy thing--

01:10:41.160 --> 01:10:41.850
is it that way?

01:10:41.850 --> 01:10:43.590
And those are properties
that they're not

01:10:43.590 --> 01:10:45.473
independent of whether
it's moving or not.

01:10:45.473 --> 01:10:46.890
But you kind of
look at it and you

01:10:46.890 --> 01:10:49.320
say it's the basic
characteristic of the machine.

01:10:49.320 --> 01:10:51.780
And then you've got these
transient ones that are only

01:10:51.780 --> 01:10:54.180
related to the energy exchange.

01:10:54.180 --> 01:10:55.860
So you'd agree that
if there's no force,

01:10:55.860 --> 01:10:57.240
there's no energy exchange.

01:10:57.240 --> 01:10:59.880
But as soon as I have
a force and a velocity,

01:10:59.880 --> 01:11:02.940
I'm putting mechanical
power into that thing.

01:11:02.940 --> 01:11:05.340
And something's going to happen.

01:11:05.340 --> 01:11:07.260
So from that we can
actually say, OK, there

01:11:07.260 --> 01:11:08.648
are two types of variables.

01:11:08.648 --> 01:11:11.190
Think of these sort of ones that
are intrinsic variables that

01:11:11.190 --> 01:11:16.198
characterize the basic
properties of the machine.

01:11:16.198 --> 01:11:18.240
And if we were doing
applied mechanics, we'd say,

01:11:18.240 --> 01:11:21.780
those are the constitutive
properties of the material,

01:11:21.780 --> 01:11:24.900
or they're constitutive
properties of our machine.

01:11:24.900 --> 01:11:28.020
And then there are
the energy states,

01:11:28.020 --> 01:11:32.550
how things change over time
as the energy waxes and wanes

01:11:32.550 --> 01:11:34.990
and interchanges and
things like that.

01:11:34.990 --> 01:11:38.790
And those are useful divisions,
because, for example, how

01:11:38.790 --> 01:11:40.860
easily can I change the
stiffness of a machine?

01:11:43.860 --> 01:11:44.890
It can be done.

01:11:44.890 --> 01:11:46.450
I could buy a new one.

01:11:46.450 --> 01:11:49.490
I could go in and put
in some more supports

01:11:49.490 --> 01:11:50.920
and that kind of stuff.

01:11:50.920 --> 01:11:54.990
Or I can take some away
and make it less stiff.

01:11:54.990 --> 01:11:56.160
Kind of hard.

01:11:56.160 --> 01:11:56.810
Yeah.

01:11:56.810 --> 01:11:59.570
AUDIENCE: It depends
a lot on where

01:11:59.570 --> 01:12:02.640
I have my [INAUDIBLE]
if I have a mill,

01:12:02.640 --> 01:12:06.990
I can take the table out very
far and use it [INAUDIBLE]

01:12:06.990 --> 01:12:09.758
DAVID HARDT: Oh,
yeah, yeah, stiffness

01:12:09.758 --> 01:12:10.800
is not a single quantity.

01:12:10.800 --> 01:12:11.100
You're right.

01:12:11.100 --> 01:12:13.290
I can find stiffer regions
and less stiff regions.

01:12:13.290 --> 01:12:15.000
But I guess what
I'm saying is if you

01:12:15.000 --> 01:12:18.810
need to elevate the stiffness
of-- there's no acceptable spot

01:12:18.810 --> 01:12:20.610
in it, what do you do?

01:12:20.610 --> 01:12:22.380
You have to change the iron.

01:12:22.380 --> 01:12:23.910
You have to go in
and make a change.

01:12:23.910 --> 01:12:26.130
It's not something that
we would call really,

01:12:26.130 --> 01:12:29.400
at this case, a
control variable.

01:12:29.400 --> 01:12:34.170
On the other hand, if I need to
change the speed at which I'm

01:12:34.170 --> 01:12:37.250
cutting, change the speed
at which I'm cutting

01:12:37.250 --> 01:12:39.140
and it changes.

01:12:39.140 --> 01:12:44.510
So again, there's an important
distinction between those two.

01:12:44.510 --> 01:12:47.810
And our ability to know them
as well as to control them

01:12:47.810 --> 01:12:49.460
is vastly different.

01:12:49.460 --> 01:12:54.680
Now, the same thing holds true
for this, this piece of sheet

01:12:54.680 --> 01:12:56.920
metal in its undeformed state.

01:12:56.920 --> 01:12:59.570
This is really hard stuff.

01:12:59.570 --> 01:13:00.320
It's like butter.

01:13:04.400 --> 01:13:07.650
If I just put this piece
of sheet metal down here,

01:13:07.650 --> 01:13:08.900
how would you characterize it?

01:13:14.220 --> 01:13:14.880
Yeah, please.

01:13:14.880 --> 01:13:15.990
AUDIENCE: Yield strength.

01:13:15.990 --> 01:13:19.220
DAVID HARDT: Yield strength,
extremely important.

01:13:19.220 --> 01:13:21.293
Anything else?

01:13:21.293 --> 01:13:24.180
AUDIENCE: Elastic.

01:13:24.180 --> 01:13:25.356
DAVID HARDT: Go ahead, Adam.

01:13:25.356 --> 01:13:26.580
AUDIENCE: Elasticity.

01:13:26.580 --> 01:13:27.907
DAVID HARDT: Yeah, elasticity.

01:13:27.907 --> 01:13:28.740
AUDIENCE: Thickness.

01:13:28.740 --> 01:13:31.001
DAVID HARDT: Thickness.

01:13:31.001 --> 01:13:32.170
AUDIENCE: Temperature.

01:13:32.170 --> 01:13:34.750
DAVID HARDT: Temperature.

01:13:34.750 --> 01:13:37.420
Actually, I'm going to
disagree with you on that.

01:13:37.420 --> 01:13:39.520
No.

01:13:39.520 --> 01:13:41.530
Why does temperature matter?

01:13:41.530 --> 01:13:44.086
AUDIENCE: Because it
changes the dimension.

01:13:44.086 --> 01:13:45.490
DAVID HARDT: OK.

01:13:45.490 --> 01:13:48.880
So there's a thermal expansion
characteristic on this.

01:13:48.880 --> 01:13:52.930
That's very important point.

01:13:52.930 --> 01:13:55.540
When I say characterizing
this, you don't characterize it

01:13:55.540 --> 01:13:56.800
by its temperature.

01:13:56.800 --> 01:14:00.160
You characterize it by its
reaction to temperature.

01:14:00.160 --> 01:14:02.770
I don't characterize it
by its stress or strain.

01:14:02.770 --> 01:14:05.710
I characterize it by its
reaction to stress or strain.

01:14:05.710 --> 01:14:06.460
So those are all--

01:14:06.460 --> 01:14:08.710
again, very clearly
for piece of metal,

01:14:08.710 --> 01:14:10.638
it is constitutive properties.

01:14:10.638 --> 01:14:13.180
And we could also worry about
its chemical reactivity and all

01:14:13.180 --> 01:14:13.888
this other stuff.

01:14:13.888 --> 01:14:16.600
But its inherent
characteristic of it,

01:14:16.600 --> 01:14:23.050
which describes how it
reacts to impinging energy.

01:14:23.050 --> 01:14:26.950
So the mechanical
energy I put in here

01:14:26.950 --> 01:14:29.860
combined with its
elasticity, its yield stress,

01:14:29.860 --> 01:14:35.170
and other things, determines how
far it bends and how it bends.

01:14:35.170 --> 01:14:36.680
So we distinguish
between the two.

01:14:36.680 --> 01:14:40.000
Again, so when this
baby is sitting here

01:14:40.000 --> 01:14:43.750
waiting to be bent, it's got
its constitutive properties.

01:14:43.750 --> 01:14:49.750
As it's being bent, it's
now bearing external loads

01:14:49.750 --> 01:14:51.130
and displacements.

01:14:51.130 --> 01:14:54.970
It's stressing and straining.

01:14:54.970 --> 01:14:58.730
And that's a transient state.

01:14:58.730 --> 01:15:00.130
It's an energy transfer.

01:15:00.130 --> 01:15:02.330
And those are two
different things.

01:15:02.330 --> 01:15:04.840
They're both important to
determining what comes out

01:15:04.840 --> 01:15:07.660
here, because if I
said, OK, go back

01:15:07.660 --> 01:15:09.940
to your computers
or your textbooks

01:15:09.940 --> 01:15:14.410
and give me an analysis of what
it takes to get this shape,

01:15:14.410 --> 01:15:16.750
you're going to need to
know the characteristics

01:15:16.750 --> 01:15:18.685
of the material,
stiffness, yield

01:15:18.685 --> 01:15:19.810
stress, that sort of thing.

01:15:19.810 --> 01:15:21.430
You're also going
to need to know

01:15:21.430 --> 01:15:23.110
what forces and
displacements I put

01:15:23.110 --> 01:15:26.217
into it, which will tell
you what happened internally

01:15:26.217 --> 01:15:26.800
to this thing.

01:15:26.800 --> 01:15:30.380
What stresses and
strains did it have?

01:15:30.380 --> 01:15:34.120
Same thing with this,
I needed to know

01:15:34.120 --> 01:15:37.480
what the characteristic
of the polymer was.

01:15:37.480 --> 01:15:38.860
How did it react to temperature?

01:15:38.860 --> 01:15:40.810
What was its viscosity
with temperature?

01:15:40.810 --> 01:15:42.820
What's its heat transfer
characteristics?

01:15:42.820 --> 01:15:45.230
And then I needed to know
what temperature it was,

01:15:45.230 --> 01:15:47.230
what pressure it saw, and
what the heat transfer

01:15:47.230 --> 01:15:50.500
characteristics
of the mold were.

01:15:50.500 --> 01:15:53.200
So what does this all lead to?

01:15:53.200 --> 01:15:58.880
These four babies-- to two major
things, states and properties--

01:15:58.880 --> 01:16:00.650
states and properties.

01:16:00.650 --> 01:16:03.820
Energy states,
transient things, things

01:16:03.820 --> 01:16:06.055
that don't hang
around, things that

01:16:06.055 --> 01:16:08.430
are just there to accomplish
what we wanted to accomplish

01:16:08.430 --> 01:16:09.600
and then they go away.

01:16:09.600 --> 01:16:13.232
And properties, which, if
you will, we can change them.

01:16:13.232 --> 01:16:14.940
But they kind of are
there when we begin,

01:16:14.940 --> 01:16:16.357
and they're still
there when we're

01:16:16.357 --> 01:16:24.570
done, and are more inherent to
either the equipment itself,

01:16:24.570 --> 01:16:26.910
essentially the material
or the equipment

01:16:26.910 --> 01:16:31.450
or inherent to the
material itself.

01:16:31.450 --> 01:16:33.090
So when we talk
about energy states,

01:16:33.090 --> 01:16:35.610
this one can be made
much more precise

01:16:35.610 --> 01:16:37.650
than we have time to do here.

01:16:37.650 --> 01:16:42.540
And those of us who are
students of system dynamics,

01:16:42.540 --> 01:16:47.100
in the mechanical, not
in the Sturman sense,

01:16:47.100 --> 01:16:50.730
but in the Painter
and Richardson sense

01:16:50.730 --> 01:16:54.360
of system dynamics,
understand that you can really

01:16:54.360 --> 01:16:56.400
do a good job of describing
the world by simply

01:16:56.400 --> 01:17:01.830
talking about energy or power
and constitutive relationships.

01:17:01.830 --> 01:17:07.117
And you can model anything as
how it reacts to these things.

01:17:07.117 --> 01:17:08.700
So that's really
what we've done here.

01:17:08.700 --> 01:17:13.320
We said the equipment
either provides or absorbs

01:17:13.320 --> 01:17:15.060
some form of energy--

01:17:15.060 --> 01:17:18.990
force, velocity, pressure,
flow, force displacement,

01:17:18.990 --> 01:17:23.760
that sort of thing, voltage
current, heat transfer,

01:17:23.760 --> 01:17:25.770
heat flow.

01:17:25.770 --> 01:17:30.800
And those are the things
that determine specifically

01:17:30.800 --> 01:17:35.180
what happens in the
interaction between these two.

01:17:35.180 --> 01:17:36.590
And then the
properties, which is

01:17:36.590 --> 01:17:38.780
sort of what's going to
happen when that energy is

01:17:38.780 --> 01:17:40.880
transferred, are
all these things

01:17:40.880 --> 01:17:43.490
we've talked about-- margins
of elasticity, plastic flow

01:17:43.490 --> 01:17:45.980
properties, viscosity, sort
of almost the same thing,

01:17:45.980 --> 01:17:47.870
resistance, inductance,
capacitance,

01:17:47.870 --> 01:17:49.490
chemical reactivity,
heat transfer

01:17:49.490 --> 01:17:53.690
coefficient, thermal
diffusivities,

01:17:53.690 --> 01:17:55.490
intensive constituent
properties.

01:17:55.490 --> 01:17:58.190
Plus, because someone
mentioned this before,

01:17:58.190 --> 01:18:00.690
someone mentioned thickness,
so this is extremely important.

01:18:00.690 --> 01:18:02.750
So are the other two dimensions.

01:18:02.750 --> 01:18:04.550
These dimensions are important.

01:18:04.550 --> 01:18:06.590
And these dimensions
are important.

01:18:06.590 --> 01:18:09.710
And so it turns out the
object of all our affection

01:18:09.710 --> 01:18:15.620
is actually the
geometry in the end.

01:18:15.620 --> 01:18:19.380
So it's sort of an extensive
property of the material.

01:18:19.380 --> 01:18:22.140
But we also have to worry about
the geometry before and after.

01:18:22.140 --> 01:18:24.560
And we worry about
the geometry of--

01:18:24.560 --> 01:18:26.930
obviously, the geometry of
the machine is important.

01:18:26.930 --> 01:18:27.560
Is it straight?

01:18:27.560 --> 01:18:31.190
Is it square that's
sort of thing.

01:18:31.190 --> 01:18:33.440
So let me finish
up by just saying

01:18:33.440 --> 01:18:37.700
that we can have this
model and then talk about--

01:18:37.700 --> 01:18:39.550
and why did I bother doing this?

01:18:39.550 --> 01:18:41.390
I mean it's kind of fun.

01:18:41.390 --> 01:18:44.120
It's intellectually
stimulating a little bit.

01:18:44.120 --> 01:18:46.550
But I think there's one
real benefit to it, which is

01:18:46.550 --> 01:18:51.020
to understand why things vary.

01:18:51.020 --> 01:18:55.250
So if I just say, all right,
I'm going to subdivide all--

01:18:55.250 --> 01:18:59.370
by the way, all
four of these things

01:18:59.370 --> 01:19:03.810
our subdivisions of this
vector alpha that we said

01:19:03.810 --> 01:19:07.470
is what makes everything happen.

01:19:07.470 --> 01:19:08.405
So I've got alpha.

01:19:08.405 --> 01:19:09.780
And then I'm going
to divide that

01:19:09.780 --> 01:19:12.420
into what I'll call
equipment states, equipment

01:19:12.420 --> 01:19:16.300
properties, material states,
and material properties.

01:19:16.300 --> 01:19:18.450
And why do I bother to do that?

01:19:18.450 --> 01:19:19.200
Well, in the end--

01:19:19.200 --> 01:19:20.730
I'll give you the punch line
because we're just about

01:19:20.730 --> 01:19:21.690
out of time--

01:19:21.690 --> 01:19:28.040
is some of these are really
easy to control, to keep fixed,

01:19:28.040 --> 01:19:32.100
and to basically know
deterministically.

01:19:32.100 --> 01:19:35.490
Most of them,
certain categories,

01:19:35.490 --> 01:19:37.800
are really hard to do that with.

01:19:37.800 --> 01:19:42.840
As an example, material
properties, yeah,

01:19:42.840 --> 01:19:45.420
material property
of this, how well

01:19:45.420 --> 01:19:48.090
do I know the properties
of this bar of aluminum

01:19:48.090 --> 01:19:50.730
or this hunk of plastic?

01:19:50.730 --> 01:19:53.740
Well, I can do some tests
and do this kind of stuff.

01:19:53.740 --> 01:19:55.440
But, again, it can
be shown that these

01:19:55.440 --> 01:19:57.450
can be highly
variable, particularly

01:19:57.450 --> 01:20:00.400
as I go from this bar
to that bar to that bar.

01:20:00.400 --> 01:20:04.200
So we're going to be able to
argue that unless you spend

01:20:04.200 --> 01:20:06.690
a lot of time upstream
eliminating variations,

01:20:06.690 --> 01:20:14.370
like unless you do really good
process control on the casting

01:20:14.370 --> 01:20:16.500
and rolling and other
processes that go into,

01:20:16.500 --> 01:20:19.630
this will have
variable properties.

01:20:19.630 --> 01:20:23.940
So this tends to be--
material properties

01:20:23.940 --> 01:20:27.030
tends to be variable.

01:20:27.030 --> 01:20:32.030
And as a result, depending
on the process mechanics,

01:20:32.030 --> 01:20:37.010
has a huge effect on
what your result can be.

01:20:37.010 --> 01:20:39.830
Unless-- there are some
wonderful processes out there,

01:20:39.830 --> 01:20:42.570
again, like I said, like
machining, that say,

01:20:42.570 --> 01:20:43.560
I don't really care.

01:20:43.560 --> 01:20:45.410
I'm very insensitive to that.

01:20:47.990 --> 01:20:52.460
Equipment properties--
equipment property,

01:20:52.460 --> 01:20:57.020
so the stiffness of the machine,
the temperature in a reaction

01:20:57.020 --> 01:20:58.820
chamber, other
things like that--

01:20:58.820 --> 01:21:00.020
I'm sorry, not temperature.

01:21:00.020 --> 01:21:06.860
Sorry, the thermal insulation
in an oven, something like that.

01:21:09.520 --> 01:21:12.162
Within limits, of course,
there's some uncertainty there.

01:21:12.162 --> 01:21:13.120
It's not just variable.

01:21:13.120 --> 01:21:18.850
I should say variable
or uncertain.

01:21:18.850 --> 01:21:22.570
Either one is a source of
uncertainty or variation

01:21:22.570 --> 01:21:23.070
for us.

01:21:26.550 --> 01:21:28.950
Equipment properties, well,
again, if I pay a lot,

01:21:28.950 --> 01:21:31.290
I'm careful with
the machine, if it

01:21:31.290 --> 01:21:34.750
doesn't degrade a lot over
time, then I'm in good shape.

01:21:34.750 --> 01:21:36.160
Machines degrade over time.

01:21:36.160 --> 01:21:40.320
So tool wear is a degradation
in an equipment property.

01:21:40.320 --> 01:21:41.850
It's an expected
dimension that's

01:21:41.850 --> 01:21:44.140
actually changing over time.

01:21:44.140 --> 01:21:53.145
So this can be variable
over, many cases, long times.

01:21:53.145 --> 01:21:55.520
And that's a maintenance issue,
as much as anything else.

01:21:55.520 --> 01:21:58.255
But again, that's within
a question of how good

01:21:58.255 --> 01:21:58.880
the machine is.

01:21:58.880 --> 01:22:02.120
If you build a machine
with very sloppy bearings,

01:22:02.120 --> 01:22:05.718
a sloppy bearing is
basically a noise generator.

01:22:05.718 --> 01:22:07.260
You just don't know
where things are.

01:22:07.260 --> 01:22:10.350
If you build a machine that is
supposed to distribute a vapor

01:22:10.350 --> 01:22:14.750
and it's got some sort of
a vapor handling system

01:22:14.750 --> 01:22:18.050
that's supposed to distribute
it evenly, but it clogs easily

01:22:18.050 --> 01:22:21.260
or something like that, or it
just wasn't designed properly,

01:22:21.260 --> 01:22:24.080
then you've got bad properties.

01:22:24.080 --> 01:22:26.095
It could cause variability.

01:22:29.980 --> 01:22:37.950
Equipment states-- forces,
pressures, temperatures, flows,

01:22:37.950 --> 01:22:40.980
velocities, what do
you think about those

01:22:40.980 --> 01:22:43.770
in terms of our ability to
control those and eliminate

01:22:43.770 --> 01:22:44.850
uncertainty?

01:22:44.850 --> 01:22:46.493
Good, bad, or indifferent.

01:22:50.437 --> 01:22:52.410
AUDIENCE: Good.

01:22:52.410 --> 01:22:54.070
DAVID HARDT: Good.

01:22:54.070 --> 01:22:57.040
We have a vote for good.

01:22:57.040 --> 01:23:00.220
I would opine that any machine
built in the past 20 years,

01:23:00.220 --> 01:23:02.320
at least, it's really good.

01:23:02.320 --> 01:23:06.490
And that, if you will, is the
victory of CNC type controls,

01:23:06.490 --> 01:23:07.420
of computer controls.

01:23:07.420 --> 01:23:10.060
One of the biggest things
is that this is now

01:23:10.060 --> 01:23:17.840
very minimal, minimal
variation or uncertainty.

01:23:17.840 --> 01:23:22.190
I know exactly-- if I have
a good Servo and I say,

01:23:22.190 --> 01:23:25.320
go to this velocity, it
goes to that velocity.

01:23:25.320 --> 01:23:27.620
If I have a good
temperature regulator

01:23:27.620 --> 01:23:29.060
and I say go to
this temperature,

01:23:29.060 --> 01:23:31.940
it'll hold that easily
within a degree,

01:23:31.940 --> 01:23:34.470
and in some extreme cases,
a lot less than that.

01:23:34.470 --> 01:23:37.520
So particularly with the
use of active control,

01:23:37.520 --> 01:23:40.280
we've made that really good.

01:23:40.280 --> 01:23:42.470
So the last one--

01:23:42.470 --> 01:23:47.700
and I would say end
controllable, in other words,

01:23:47.700 --> 01:23:49.820
not only is it not
variable, I can actually

01:23:49.820 --> 01:23:53.400
tell you what I want it to be.

01:23:53.400 --> 01:23:56.120
So I'm going to skip discussion
of the material states,

01:23:56.120 --> 01:24:04.190
except to say that
they're often--

01:24:04.190 --> 01:24:12.110
well, let's say they are
variable and hard to know.

01:24:12.110 --> 01:24:14.900
And I'll give you my one
favorite example of that.

01:24:14.900 --> 01:24:16.400
Let's say that I
have a polymer that

01:24:16.400 --> 01:24:17.900
needs to be at a
certain temperature

01:24:17.900 --> 01:24:20.030
to do what it's supposed to do.

01:24:20.030 --> 01:24:22.940
Typical way of doing that would
be put it in an oven and let it

01:24:22.940 --> 01:24:24.350
sit in that oven.

01:24:24.350 --> 01:24:26.270
So the oven has
certain properties

01:24:26.270 --> 01:24:28.880
that makes it keep
uniform temperature.

01:24:28.880 --> 01:24:31.420
And it has a really good
temperature control on it.

01:24:31.420 --> 01:24:35.000
So if I say heat this to 100
C, it's right there at 100 C.

01:24:35.000 --> 01:24:37.250
So I've got this
heated air, and it's

01:24:37.250 --> 01:24:42.290
at 100 C. I take a piece of
polymer, and I put it in there.

01:24:42.290 --> 01:24:45.260
And I wait sufficiently long
that it's at equilibrium

01:24:45.260 --> 01:24:47.090
and the polymer is now
at 100 C. And then I

01:24:47.090 --> 01:24:48.882
pull it out and put it
in a warming machine

01:24:48.882 --> 01:24:51.830
or something like that.

01:24:51.830 --> 01:24:55.120
What's the temperature of
the polymer when I formed it?

01:24:55.120 --> 01:24:57.210
That's a state of
the polymer, right?

01:24:57.210 --> 01:24:59.140
It's a thermodynamic
state of the polymer.

01:24:59.140 --> 01:25:01.140
I don't know.

01:25:01.140 --> 01:25:03.820
Unless I measure
it, I don't know.

01:25:03.820 --> 01:25:05.920
I'm not controlling it anymore.

01:25:05.920 --> 01:25:07.450
Now, there are also
cases, and this

01:25:07.450 --> 01:25:09.200
is true I think in the
semiconductor world

01:25:09.200 --> 01:25:12.460
now more than anything else
is that if I want to control

01:25:12.460 --> 01:25:16.820
a temperature, for example, I
don't measure the temperature

01:25:16.820 --> 01:25:19.070
of the oven or the chamber.

01:25:19.070 --> 01:25:23.630
I measure the temperature of the
thing that I'm going to make.

01:25:23.630 --> 01:25:26.632
And an example of that
is-- in fact, the machine

01:25:26.632 --> 01:25:29.090
that this was made on if you
guys have used this, the ENGEL

01:25:29.090 --> 01:25:32.450
down in Building 35, you
control the temperature

01:25:32.450 --> 01:25:34.880
usually on this
process by controlling

01:25:34.880 --> 01:25:37.190
the nozzle temperature and
the barrel temperature.

01:25:37.190 --> 01:25:39.770
And you say, oh, that's the
temperature of the polymer.

01:25:39.770 --> 01:25:41.990
But what really matters
is the temperature

01:25:41.990 --> 01:25:46.370
of the polymer in the mold,
which you never measure.

01:25:46.370 --> 01:25:46.880
Right?

01:25:46.880 --> 01:25:50.690
But there's actually a little
temperature sensor in the mold

01:25:50.690 --> 01:25:52.372
that you can measure that.

01:25:52.372 --> 01:25:54.330
And there's also a pressure
sensor in the mold.

01:25:54.330 --> 01:25:56.690
So you could, if you
wanted to, actually

01:25:56.690 --> 01:25:58.850
measure the pressure on this.

01:25:58.850 --> 01:26:00.500
But it's hard to do.

01:26:00.500 --> 01:26:03.170
So this tends to
be-- it's knowable,

01:26:03.170 --> 01:26:05.600
but sort of expensive to do.

01:26:05.600 --> 01:26:08.210
So here's the
moral of the story.

01:26:08.210 --> 01:26:09.890
Biggest source of
variation tends

01:26:09.890 --> 01:26:14.670
to be material uncertainties,
just inherent in the material

01:26:14.670 --> 01:26:18.770
and, if you will, the coupling
between this energy source

01:26:18.770 --> 01:26:21.710
and how well it actually
gets into the material

01:26:21.710 --> 01:26:24.000
through sort of an
uncertain interface.

01:26:24.000 --> 01:26:26.900
So all of these things
here, these three,

01:26:26.900 --> 01:26:29.000
are your sources of
variation typically

01:26:29.000 --> 01:26:30.450
in differing measures.

01:26:30.450 --> 01:26:33.710
This is the one saving grace.

01:26:33.710 --> 01:26:35.390
You're going to
actually control--

01:26:35.390 --> 01:26:37.310
way back at the
beginning of the process,

01:26:37.310 --> 01:26:40.640
you're going to control
equipment stakes.

01:26:40.640 --> 01:26:45.440
So when it's all said and
done, what we manipulate here

01:26:45.440 --> 01:26:48.845
are typically, not always, but
typically equipment states.

01:26:51.500 --> 01:26:54.380
And our ability to do
that and how well I

01:26:54.380 --> 01:26:56.720
can change an
equipment state to get

01:26:56.720 --> 01:26:59.090
a change, the sensitivity
of the process,

01:26:59.090 --> 01:27:01.820
has a lot to do with what
I can do with its control.

01:27:01.820 --> 01:27:04.040
And the rest of the lecture,
which is in the notes,

01:27:04.040 --> 01:27:06.450
is going through
these three processes

01:27:06.450 --> 01:27:08.700
I've talked about and kind
of addressing those issues.

01:27:08.700 --> 01:27:11.200
So if you have a chance, take
a look at the rest of those.

01:27:11.200 --> 01:27:12.390
OK?

01:27:12.390 --> 01:27:14.090
OK, thank you.

01:27:14.090 --> 01:27:15.590
Thank you all.

01:27:15.590 --> 01:27:16.880
And I won't see you next time.

01:27:16.880 --> 01:27:18.580
He will.