§1. CSST and the lightning problem [00:02]
[Tom holds up a length of corrugated stainless steel tubing (CSST).] This is used in the United States in homes, and everything's fine until lightning hits your home, in which case it melts a little hole in this and you have a fire inside the wall and it burns down your house. It's only happened about four or five hundred times. Now, this is an improved product. They spent six or seven years before they brought this to market looking at all the codes and standards supposedly that might govern this so that they could get it approved by different agencies. Because this is a plumbing product, and plumbing products are developed by each state. In Massachusetts we have a State Plumbing Board.
The problem is they didn't look at the lightning codes. They'd never thought about the problem of lightning damage. No one had ever had a fire with black iron pipe because lightning had never perforated black iron pipe, so they didn't worry about it. They brought it out to market in 1998, and by 1999 they started hearing reports of homes burning down that had the stuff in it, and they didn't know what it was. They heard stories that it might be lightning. By 2001 they knew it was lightning — that there was perforating, holes in the pipe. And the bottom line story is it's just too thin, and the people who spent six years trying to see if it met all the codes and standards never even thought about lightning as a problem. So it was sort of, oops, we forgot to worry about a natural disaster.
They actually talk about it as a wonderful material. One of the reasons the Japanese developed this twenty-five years ago was because of earthquakes. You have an earthquake and you break your black iron pipe, which is rigid — the ground shakes and you break the connection, now you've got a big gas leak and you burn down your home. So if you have an earthquake you get a fire and burn down your house at the same time. Well, it turns out lightning strikes, thunderstorms, are more frequent than earthquakes, and so a lot of people have been surprised. There's a lot of stuff going on about that, so we'll talk about certain codes and standards.
If you were interested, there's only one code written for that particular product. It was written in 1992. It's called ANSI LC1. They charge $600 for it. But it was written by the industry. It says nothing about lightning. That's not to say that other codes don't say things about fuel gas containers and lightning. The lightning protection code of the National Fire Protection Association says something about lightning. In any case, we're going to go through those types of things.
§2. The course project and what codes contain [03:03]
Your project, your presentation — should you choose to do it, I guess this Mission Impossible — will basically go through a code or standard, whether it's bayonets, anything of your choosing. One person last time did doorknobs. They were interested in doorknobs. There are codes and standards on doorknobs. There are codes and standards on just about anything you can imagine. And I find students do a better job if they get to choose something that they're interested in. So rather than my choosing for you, you're going to have to pick a code or standard, and you're probably going to have to make an appointment, come by and see me, and talk about the code or standard.
Codes and standards have a number of things that go with them, and this little handout on codes and standards that I just gave you — you can talk about the history. I just told you some of the history of CSST codes and standards. The scope and limitations. What's fair use. Conflicts among codes — one society might write a code, another government might write a code, and sometimes they don't agree with each other. What do you do? Is there force of law? Certain things in the plumbing code are written into building code, which is adopted by the legislature in the state, and so some things have force of law.
One of the things I want to talk about in codes and standards is, you were taught as a student that you need to learn how to design things, but no one ever told you that you have limitations on your design which are written into law that say you cannot do certain things. So there's the good of codes and standards: they are a historical repository of things that went bad. The Hyatt Regency walkway collapse, we'll probably talk about — where I think over a hundred people, maybe two hundred people, got killed. The Northridge earthquake in Southern California. When there's a major problem, they often revise the code or standard to make things safer. After 9/11 they revised a lot of the building egress code so people could get out of buildings faster, so the fire wouldn't bring the building down as quickly. Their interpretations of the code — codes are not always the easiest things to read.
Their collaborations and competitions. There is an international competition going on between the European Economic Community and the United States for who controls the world's codes. Historically we've controlled the world's codes, but in the last twenty-five years the Europeans have started to get interested. There's a problem that I've identified that's occurred in the last fifteen or twenty years. You can get government codes by just downloading them for free. Others, if you want a hard copy, you can get a ten-inch-thick code for $150, which is the printing cost. Not-for-profit organizations like the National Fire Protection Association might charge you $100 for a code. For-profit societies like the American Society of Mechanical Engineers, ASME, used to charge a hundred bucks for a code. They've now learned they can get $1,000 for that code because it's required by law and every engineer who's practicing has to have a copy. So all of a sudden, in the last fifteen or twenty years, the for-profits are now charging big time for codes. It's a real money maker.
And not only that, when you can charge for a code you'd like to write more codes. And it's called regulations. If you listen to the Republican National Convention they want to decrease the regulations. A lot of people think we're getting over-regulated, and we'll talk about that in terms of codes and standards. We won't talk necessarily about Republicans and Democrats, but it is part of the whole regulation thing. What you should get inspectors. So your paper, your presentation, can take a code and include not necessarily all of those, but any of those that you want. There's sort of an outline.
§3. Course logistics, grading, and exams [07:31]
Student: [Asks about course requirements.]
Anybody have any questions on what's required of the course? You can collaborate on anything you want. I'm supposed to tell you that — I'm supposed to tell you whether you can or cannot collaborate. But in this type of course you can collaborate on anything. If you're doing C and D and doing the problem set, I let the students collaborate on the problem set. The record is nineteen students turning in one problem set. I don't think you'll do it in this class this year, we don't have that many students.
Also, don't worry about your grade. In twenty-five years I've only given out two grades. Some people do get F's. Actually, three grades — I used to give out incomplete, but the registrar said you can't do that, you have to give them F's. So I give F's and then I give the other grade at the other end. Because this course is not required, right? When I was a young professor, the way to get students to take your graduate course was to get it to be a course that you had to take in order to answer a question on the written exam for the doctorate. If you wanted a lot of students to take your course, the only way to motivate them to pick your course as opposed to another course was to always write a question for the general exam that anybody who took your course could answer but no one else could.
I'll tell you what happened to me when I took the general exam around here. There was a solidification question, and I flunked it big time — not the whole exam, but that question I flunked big time. Because I figured it was Professor Flemings who taught a solidification course, and I gave his answer to the solidification question. But unfortunately it was Professor Russell who asked the solidification question. So I got it wrong, because Professor Russell's answer to that question would be different than Professor Flemings'. At that point I started calling the general exam the specific exam. It had nothing to do with general knowledge of the field, it had to do with the specific knowledge of which professor taught what.
It's one of the reasons I have negative views of exams. The other reason I have negative views of exams is because they give you problems that are — I call them closed forms — they give you just enough information to solve the problem, no more and no less. And the smart student — and guess what, everybody at MIT is a smart student — they learn the patterns and they realize that if they didn't use all the information there's something wrong with their approach. Or if there's not enough information and you give it to them as a homework problem, they don't know how to go look up the modulus of steel. They've never had to do it, it's always given to them in the problem. So if you do the problem set, I will warn you that the problem set for joining one and two actually has open-ended problems. You may have to go look up a piece of data, you may have more information than you need in some cases. Students hate it. I tried that with undergraduates in 1982 and they killed me on the evaluations. How could a professor change from the established pattern of giving you just the information you know that you need? Another reason why I don't like exams.
But it's not required for anything. Now, when I teach Thermo, I actually give exams and I grade them and I give A's, B's, C's, D's, and F's, because I had to — it's a required course and you have to separate the students who are going to go on to graduate school and who are going to go on for doctorates. But when you have a completely entirely volunteer subject like this, we don't have to worry about the next hurdle in your life. So don't worry about your grade. You will have to do something. The Institute requires it.
§4. Manufacturing science and the structure of this third of the course [12:01]
Dr. Belmar will be lecturing on safety and safety factors — there are really three parts. You're going to have to decide on these seven things. I think everyone's going to take the live one, six, and you're going to have to pick two more, and you're going to need to let me know sometime in the next week or so which other two you want to take. They should be online on my website, but my secretary has DVDs if you'd like to get them that way.
To talk a little bit about what the live lecture is — the one we're going to do right now. I called this manufacturing science. When I was sitting down a couple of weeks ago with Dr. Belmar and he looked at what I was going to talk about, he said, this doesn't have anything to do with what's in the catalog. I sort of agreed with him. He said, this is just Tom Eagar talking. And I said, well, that's true. At my age I'm going to teach you what I want to talk about. But in fact there is an order to this. I called it manufacturing science because for the last twenty years I actually have been working in manufacturing, and there are certain things I've seen that I don't know that anyone else teaches.
To give you an example, the first third of the course — it's only four lectures — but the first part is something called measurement science. Last spring I was serving on a committee to evaluate the manufacturing program at the National Institute of Standards and Technology in Gaithersburg, Maryland. This is a $600 million federal laboratory, and they reorganized things and created a manufacturing program out of a bunch of other things. So the old fire protection stuff, the people who do the atomic clocks, are now part of manufacturing. Really? Yeah. Because measurement is fundamental to making something, and measurement is fundamental to codes and standards. Most codes and standards, you have to measure what you're talking about so that the buyer and the seller who are using the standards can decide whether they actually got something that was in this case fifteen inches long or fourteen inches long or sixteen inches long. If you bought a fifteen-inch ruler you need to know, is it fifteen inches plus or minus how much? So measurement is fundamental, and we're going to talk about the seven fundamental quantities.
I was down there for three days in March of this year, and I mentioned to this woman who's in her early eighties, but very vibrant and dynamic — she gave the best talk of the three days. She's head of the group. She's a physicist. She started at Bryn Mawr [?], came to MIT for undergraduate, finished out her undergraduate degree. MIT physics department wouldn't admit their own undergraduates to be graduate students, so she went up the river to Harvard and got her PhD up there, and she's head of the group that develops the atomic clocks and all this other stuff. I mentioned to her that I taught a course on codes and standards. She said, you do? I said, on measurement. You do? Because she didn't know anywhere else that anyone would teach a course on how to measure something.
We're going to go from the sublime to the ridiculous. We're going to talk about the most precise measurement ever attempted, one part in 10 to the 23rd. Limits to materials properties — there's only so much you can get out of a material. The highest melting point element is — anybody know what the highest melting metal is? It's tungsten. But we'll talk about that later. It has to do with the bonding. There are certain properties and you just can't get a modulus greater than 60 million or so PSI. Then we'll talk about codes and standards, sort of what I just talked about. And then we're going to talk about design for manufacturing and the different types of design. Part of that will be safety factors in design. So that's the outline for this third of the course. That's the syllabus. I'm going to turn that into Anita as the syllabus. Any questions?
§5. Lord Kelvin, observation, and the uncertainty principle [17:06]
If not, then after killing the first half hour, I'm going to start talking about measurement science. If you walk into my office, right as you walk in there's a little quote from Lord Kelvin, and this was on Morris Cohen's wall. He had this framed, and I mounted it and framed it. A number of his graduate students said, well, when he passes away I want his framed quote from Lord Kelvin. I just made my own rather than asking for it. He says: "When you can measure what you are speaking about and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind. It may be the beginning of knowledge, but you have scarcely in your thoughts advanced to the stage of science, whatever the matter may be."
So if you come to MIT to learn science, you really should have learned how to measure something. But how many times has anyone talked to you about measuring something? Sometimes in the labs. But how many times in a classroom has anyone ever talked to you about measurement? Almost never, right? If you're going to measure something you have to observe it, and if you go to fundamental physics, what did you learn about the limits of observing and measuring something? You learned about it somewhere in freshman physics. The Heisenberg —
Student: Uncertainty principle.
Uncertainty principle, right. The Heisenberg uncertainty principle says that there are limits to what you can measure, and you can write it lots of different ways. The way it's typically written is the change in position and the change in momentum have to be greater than Planck's constant over two. Fortunately Planck's constant is 10 to the -34, so it's a pretty small number, which is the limit of your measurement. We don't usually, except in quantum mechanics, have to worry about the uncertainty principle because we're not measuring things so small. But we do have to measure lots of things, and we have to recognize there's a difference between precision and accuracy.
Does anyone know the difference between precision and accuracy?
Student: Precision is how close you are to the target, or accuracy — um...
Well, yes, sort of. I think I know what you meant, but a picture is worth a thousand words. Actually you may have said it backwards, but who knows. I couldn't explain it in words either, but I can explain it this way. [Tom draws on the board.] If the real value — or they call it the reference value, because who knows what is real, right, this is something you'll get in the humanities — if the real value is some reference value here, how accurate you are depends on how far your average measurement comes from it.
Precision is how reproducible your measurement is. If you make the measurement a hundred times you're going to get a bell curve and you will get some average value, but the average value differs from the reference value, or let's call it the real value. The people who wrote this up would not like that — they're statisticians, they don't believe in reality. The reference value is what it should be, and accuracy is how far you deviate from the real value. Precision is how reproducible. If you make the measurement multiple times, what's the standard deviation from the mean? With a picture it makes it easier, right? I couldn't have said it in words either, so don't worry about it. So that's accuracy and precision.
§6. The standard kilogram and the seven fundamental quantities [21:43]
Now let's talk about — I didn't know until Katherine Gabby, this woman at NIST, mentioned in her talk that the challenge right now is redefinition of the kilogram. The standard kilogram. Has anyone read about the standard kilogram? Does anyone know what the standard kilogram is? It's a platinum-iridium weight locked in a vault in Paris, France. It's brought out about once every ten years and reweighed. It is the world standard for mass. Does anyone know what the problem with this is? It's getting smaller. I asked her at one of the breaks, I said, how do they know it's getting smaller if it's the standard? She says, well, they have other masses, platinum-iridium spheres, which are the secondary standards, and when they compare the weight of them to the standard, they find that the standard is losing something like 50 micrograms per century. Well, 50 micrograms out of a one kilogram mass is one part in 10 to the 12th, or one part in 10 to the 11th or something. For people who study these things that's a huge number.
So they would like to have all the standards based on some constant. And as they say in this paper — I didn't give you the whole paper, the paper gets into lots of other details, but I gave you the first part that's important — if you just read the abstract it will tell you that they'd like to have international standards that anyone in the world with an infinite checkbook could reproduce and get the same number. They don't want it based on something that's sitting in Paris because who knows how long the French are going to let us come to Paris. Or maybe the Germans will attack them with a nuclear weapon or something and blow up the standard mass, and then what would you do?
She said in her talk that they were going to replace it with Planck's constant. I was sitting there thinking, and I said, so how are they going to do that? Are they going to take E = hν, and E = mc², and combine them? You'll get the speed of light, right? And so if you want to get mass, then mc² = hν, and you can measure ν very well because it's time-based, it's one over time, and we can measure time more accurately than anything else. She says to me, oh, it's not that simple, I'll give you a paper. This is the paper she gave me. It turns out, at my level, it is that simple. At her level it's a lot more complex. But in fact this paper shows you that there are seven fundamental physical constants that we need to try to measure.
You learned a number of them when you were a freshman, three of them. What are the fundamental things that they taught you when you were looking at dimensions, the three fundamental quantities? Length, mass, and time, right? So we've got mass, length, time. Mass is a problem because it's this platinum-iridium sphere in Sèvres, France, in a vault, and people can only look at it once every ten years. Length — how do we measure length today? Right, it's related to the speed of light and time. And time is measured today by what? Cesium atoms. Cesium-133. It's atomic clocks. There's a signal coming from Golden, Colorado at the NIST laboratory — you get a radio signal and it'll synchronize everything. You can wear it on your wrist.
§7. NIST F1, GPS, and the atomic clock you can buy [26:05]
How important is that signal coming from NIST? Well, your cell phones use it. Actually they don't use it directly, they use atomic clocks located where? There's a set of thirty-some atomic clocks rotating around a couple hundred miles above our heads, called the Global Positioning System. Each one of them has an atomic clock. But those are calibrated to NIST F1, which is located in Golden, Colorado. They actually will send out a signal several times a day. It's very difficult to get it in my office. I usually have to take my atomic clock home, and even that's a problem because I live in the valley underneath Belmont Hill, and so the signal coming from the west doesn't get down to the valley every night very well, depends on things. And here it doesn't get past the Mechanical Engineering building across the Great Court. If I was up on the fourth floor maybe I could get the signal, but I can't get the signal.
[Tom shows a slide of NIST F1.] Here's NIST F1. This is the time standard for the world. You need kind of a room about this size to hold it. And here is what they've been able to do since the early '50s — NBS-1 (it used to be National Bureau of Standards) through NIST F1. There was a NIST 7. This is the improvements: one part in 10 to the 15th. This is the frequency uncertainty. So it's 10 to the -15. One part in 10 to the 15th. Anybody know what that's equivalent to in terms of seconds? There's about 10 to the 8th seconds in a year. So this is one second in thirty million years — my notes say thirty billion. Anyway. Someday they'll find another one even better. But we can measure time better than anything else because we're looking at the frequency of a cesium atom vibrating.
If I go and look at the other physical quantities — you learned about mass, length, and time, but there are seven fundamental physical quantities, and they're right there in the abstract of the paper from NIST. Actually it's not just a paper from NIST, it's a paper from scientists at NIST in the United States, the United Kingdom's measurement laboratory, and France's measurement laboratory. So if you look at where those authors are, those are kind of the people who are leading the development of standards for the whole world. It really is something that we, the French, and the British are doing for the whole world.
As we develop standards, they can give us the seven constants. There's the cesium atom for time, there's the speed of light in a vacuum, there's the luminous efficiency — lumens per watt. There is the Planck constant, the elementary charge e, the Boltzmann constant, Avogadro's number. These are defined, but they're not all defined exactly. One, two, and seven are defined exactly. We actually have defined the speed of light in a vacuum in meters per second. The speed of light is not constant — the speed of light is less than that in this room because light interacts with the atoms. And when I talk about the most precise measurement in the world ever attempted, it's done in the best vacuum they could obtain, because if you're trying to get one part in 10 to the 23rd, you've got to get rid of the air — light interacts with air. There is some uncertainty in these numbers, but in general the uncertainty in these numbers is about one part in 10 to the 7th. For time it's one part in 10 to the 15th.
But that's for NIST F1. For $1,500 today, you can buy your own atomic clock, which is less than the size of a grain of rice. It was developed at NIST. [Tom shows a slide.] Here's the little printed circuit board with the lights going on it, and for $1,500 you can buy one of these things, your own little time standard. This is what it looks like split up. There's a little heater where you heat up the cesium, and this is a boron nitride container, and there's a little laser down here that shines up through here and measures the frequency of the vibrating atom. This is only good to one part in 10 to the 10th, which is one second in three hundred years. So if you want to be on time you can have your own little personal atomic clock. It only takes 75 milliwatts. Pretty neat, huh?
§8. Temperature standards and the Vienna mean ocean water [32:04]
There are other standards you can get. There's a temperature standard. Does anyone know what the temperature standard is based on? It's actually based on Boltzmann's constant, because energy is equal to Boltzmann's constant times temperature, right? So if you know energy, which you know from Planck's constant, you could get temperature. If you really want to know temperature to a thousandth of a degree centigrade at close to 0°, you can buy yourself, for $1,500 or $2,000, depending on what size you want, a triple point of water standard, sold by Fluke Corporation. Hart Scientific division of Fluke will sell you a temperature standard. I was actually talking to Mike Tarkanian about whether we should buy one for the undergraduate labs. This will be accurate to one ten-thousandth of a degree, triple-point.
In order to do this, they basically are taking something that thermodynamically is an invariant point where water vapor, ice, and liquid water coexist. You can basically take these little glass tubes and you can put some dry ice down this little opening, and you will get to the point where all three of them will coexist. The masses of the three phases may change over time. You can put this in a little incubator-type thing and it will hold the temperature for several weeks. You put your little thermometer that you want to calibrate down inside this thing, and you will know — it turns out the triple point of water is at about a tenth of a degree centigrade. But they don't just use any water. What do you think they use? It's not just deionized — it's the Vienna Standard Mean Ocean Water sample.
You never even heard about that before, did you. They take seawater because — there are several different types of water they could have used to make their standard. When you're trying to get one ten-thousandth of a degree accuracy, you've got to be careful what water you use. So repetitively distilled ocean water: the Vienna Standard Mean Ocean Water. They could have used ocean water, polar water, or continental water (which is rivers and streams and stuff). Ocean water is the heaviest. It has more oxygen-18 and more deuterium than other waters. And the amount that is there is a standard. So they just take ocean water and they distill it and distill it, because impurities can have the biggest effect on all this. The errors in isotopic concentration are essentially what limits you to this thousandth of a degree accuracy. However, the calibration cell is good to 7 microKelvin. So if you want to be accurate, you have to go to great pains.
This is kind of looking at some of the most accurate things we can measure. Tomorrow we will talk — we'll be at 2-131 or 2-132. We will email you.