§1. Course wrap-up and write-up philosophy [00:00]
Today is my twelfth and last day for this module. Dr. Belmar will be doing all the days next week, and one day into the week after. He might have to miss one day, so he'll give you a day off — he'll let you know the schedule. It's on Stellar. And we'll start the presentations on the 19th.
Everybody is supposed to watch twenty student presentations. They're going to be videotaped. Doing your own counts as one, so you have to watch another nineteen. There are going to be about thirty-three of them. You don't have to watch them all. A number of students have said they like the presentations as well as the lectures, and frankly the students do a good job on these presentations because it's something of interest to them. When people know more about things that are of interest to them, in general, they do a better job.
You also have to come up with a half- to one-page outline of each of four different modules — although unfortunately in the past the students have been doing two- and three-page outlines. I mentioned in the introduction that there's usually one or two themes a professor can get across in an hour, and that's all we wanted. What was the theme that you selected? If I gave twelve lectures, there are only eighteen to twenty-four themes in my whole twelve lectures. Can you actually distill it down to that kind of basic stuff? I had one student once complain that he thought it was busy work, and I disagree. I think it's good for you to say, okay, what's the basic idea here — particularly when I ramble with stories that are semi-relevant.
You can do it any way you want, because it's for you, not me. I want to see that you did it, but I'm not going to study it. It's really for your benefit. But we do look at them to see what the students really get out of these modules. One of my former doctoral students, now a postdoc, basically says, well, Joel Clark teaches materials economics one way, and someone else teaches it another way. What I've been teaching here in terms of structural materials economics is at a senior-manager level. A senior manager should know that it costs $10,000 a pound to get weight in orbit. So when they hear in the Wall Street Journal that Elon Musk is going to start sending people to Mars — well, you can't even get them into orbit except for $10,000 a pound. Someone weighs 200 pounds, we're talking two million bucks just to get them into orbit, and we haven't even sent them to Mars yet. We could just leave them in orbit if they can't come up with the rest. Or drop them home, see how they survive.
§2. What is a composite? [04:30]
Composites. This is a typical type of overblown picture of composites. The only way you can tell it's composites is that they've got some sort of graphite fiber shell for the car, and you've got this young girl essentially lifting up the car. What's wrong with this picture? What about the weight of that axle? Is she really that strong? I don't care how light the frame is — there's a lot of weight in that axle and wheel and brake, and there might be an engine in there somewhere. So this is an overblown view, but you see these things all the time, and they're not always as obvious as this one.
A problem with composites is, what do you mean by composites? We could be talking about concrete at 20 cents a pound, or we could be talking about some fancy aerospace composite. We're actually going to do a case on the X-33 this morning. This is $12,000 a pound as fabricated 15 years ago, and we can talk about how you get to that price. Composites can have five orders of magnitude in different material properties. We can have five orders of magnitude in the price per pound for materials. And a dentist would call this a composite tooth — I actually have two in my mouth now as I get older. Titanium stud and a zirconia, zirconium oxide crown. They'll outlast the rest of my teeth, and they only cost about $5,000 per tooth. Some people get rid of their dentures, and if you want to have a $200,000 mouth, you can replace all your teeth with these. It's embedded in the bone.
This is a composite. It's a piece of artwork done by the CLT wood group, composite laminated timber, somewhere in London. It's called the Smile. If you look on the side, it's just made out of laminated wood product. The problem with composites is, it's one of the broadest terms in the field of engineering. It can be 27-cents-a-pound concrete, it can be super aerospace composites, it can be all kinds of things.
§3. Boeing aircraft and the composite trajectory [07:06]
We can take an aerospace example and look at the last few aircraft Boeing has built. They started with the 707 and the 727, which no longer exists, and the 737, which still exists but was completely redesigned in the late 80s, early 90s. The 737 of today is not the same as the 737 of 1975. The Boeing 777, which was started around 1990, was said to be a twenty-billion-dollar investment by one company. This is a bet-your-company project.
Does anybody know how Boeing got the 747? They lost the C5-A contract to Lockheed Martin — actually it was a different aerospace company then, in Marietta, Georgia. I can't remember; one of my best friend's fathers worked there when I lived in Atlanta as a kid. The 747 was a twenty-billion-dollar bet-your-company project. The earlier bet-your-company project, four or five billion dollars, was the competition to build the super-size air transport for the Air Force, and the C5-A Galaxy won. Before that, they used the C-130, which is much smaller. The C5-A, you can drive a couple of M60 or M1 tanks onto. The army wanted an airlift to take tanks around the world, and the 747 was Boeing's entry into that competition back in the 60s. Boeing had lost the competition, had invested billions of dollars of their own money, and they said, well, why don't we make a commercial air transport. And everybody thought, whoever wants to have 400 people on a plane? Well, I remember taking off out of Tokyo Narita once, and we circled over the airport and I counted 47 747s on the ground. Forty-seven 250-million-dollar airplanes — you're talking ten billion dollars worth of planes right there at Narita. It's one of the largest hubs for 747s.
So the 737 was essentially an all-aluminum airplane. The 777 was going to be all composite when they put it on the drawing board and made the pitch to the Board of Directors. When they finally finished, it was 12 percent. By the way, the guy in charge of the 777 program was Alan Mulally, an MIT grad — actually a Sloan fellow, but close enough — and then he became chairman of Ford. So they never hit all-composite aircraft. When they got to the 787, they decided it was going to be a hundred percent.
And lo and behold, the 787 ended up being 50 percent composite. The darker blue, the royal blue you see there, is basically carbon fiber composite — 50 percent by weight. Just like my comment about polymers not being stiff enough, these composites would cost a small fortune to make the ribs in the keel beam and so on, so that's still got aluminum and titanium. You see 20 percent aluminum, 15 percent titanium. The internal structure is still primarily metals, but you can make the skin that goes on it lighter weight. In order to compete and have low enough fuel costs, you've got to have a lightweight vehicle. If Boeing comes up and says they're going to build another major aircraft for twenty billion dollars, another bet-your-company project, then they may get to 75 or 80 percent composite. There's still going to be other things.
§4. Questions to ask about any composite [12:25]
Whenever you hear about composites, because they are so broad, you ought to start asking a few questions. Here are the questions I came up with. Are we talking about structural or functional? Structural means mechanical properties. I'm taking you back to the very second slide on the first day of my lectures: we have structural materials and we have functional materials. Functional materials are optical, electronic, photonic, semiconductors. Or they could be both — smart materials, where you use some sort of functional sensor that is a property of the material, but the material is also structural. People want to put fibers into helicopter wings that are strong enough to strengthen the wing and make a composite wing, but also have properties, whether piezoelectric or whatever, that can tell exactly what's happening to the helicopter wing in service and adjust things. That's a smart material that's both functional and structural. The advantage of composites is that they're not limited to the same set of properties as conventional materials.
You also need to know the target application and cost. Ships, aircraft, aerospace — we talked about $20,000-a-pound material. Are we talking $2 a pound over the life of the vehicle? This is the question a manager needs to ask. If I'm in the automotive business, don't tell me about $20,000-a-pound aerospace composites. But people do all the time. Amory Lovins in the 1990s — this guy out of Colorado, an environmentalist — used to say the automotive companies are idiots, we could build hundred-mile-per-gallon vehicles today. Yeah, we could. Indianapolis 500 racers for a million dollars apiece. You could have made a 100-mile-per-gallon vehicle if you weren't going for performance and you were going for economy of fuel. Twenty-five years ago I could have built a hundred-mile-per-gallon vehicle, but I couldn't have sold it for $25,000. You're giving up something.
Are you constrained by a specific industry? If you're going to an aluminum company, they want an aluminum composite. They're not interested in other materials. Some industries don't have a material constraint, but some companies by the nature of their business do. If you're a Boise Cascade, you're in the wood business.
Trade-offs. This is the one that people don't get most of the time. When you make a composite, you might be improving one or two or three properties, but you are probably destroying four or five or six other properties. I can put silicon carbide into aluminum and make a stiffer, more wear-resistant, stronger material. But I can only cast it. I can't machine it because it's full of these silicon carbide particles. If I want to machine it, I've got to use diamond tools, and it's very expensive. If I want to form it, I'm going to wear out my dies because it's full of all these abrasive particles. If I want to repair it — well, there's no good way to do fusion welding of aluminum silicon carbide composites.
Back in the late 80s, when aluminum silicon carbide composites became much less expensive to produce, a company came to me and said, we want you to take our aluminum silicon carbide composites and electron beam weld them. I said nope, not going to do it. I said, if you want to do that, you can go to Ohio State University, they've got a welding engineering department, they'll be happy to take your money. They said, why not? I said, because when you melt silicon carbide in the presence of aluminum in a fusion weld, you're going to form aluminum carbide, and aluminum carbide reacts with the humidity in the air and causes cracking. They said, but we want MIT to do this. I said, fine, you can find someone else at MIT, I wouldn't take your money for this job. It's a stupid project — I didn't say that to them, but I've said it since. Well, they went to Ohio State, and one guy who didn't get admitted to MIT to do his work on welding did that work at Ohio State, and I know him now. We talk about it every now and then. He thought it was a lousy project when he finally learned about it, and he will admit it was a stupid project, because you give up certain things.
What's your time frame? Bringing a new material to market can take twenty years. There are lots of processes that have to be worked out. People look at composites as continuous fibers, or short fibers, or particles. The fibers can be in different plies — they call them zero, 45, and 90 degree plies. When you see something like that carbon fiber vehicle that the young girl was lifting up, that's basically multi-ply.
§5. Composite forms — laminates, honeycombs, and adhesives [18:25]
[Tom moves to a different visual aid.] Composites are not really very repairable. That's one of the things you give up. They're not very joinable. This X-33 space plane, this is zero, forty-five, ninety ply of carbon fiber laminates. You can also make — and we passed around copies — this honeycomb. You're doing a ninety-degree to the vertical, and you can do all kinds of other structures.
We can have laminar composites. If we think of the Smile, the wood sculpture I showed you in the beginning, they've taken plywood to a new level. All these engineered wood products are basically composites of polymer and wood — adhesive and wood, where the wood is the cheaper material. If you look at those things I brought in, the adhesive in there is very thin because it's very expensive. The cheapest polymer adhesives are probably about two dollars a pound. You can get them for ten or twenty cents a pound — things like cornstarch — really cheap adhesives, but they also get attacked by moisture. If you want adhesives that won't be attacked by moisture, you'll find — in my welding lectures, when I talked about adhesives — typically you've got to get up to four to ten dollars a pound to get moisture-resistant adhesives. I'm talking about things that are going to be moisture-resistant for 10 or 20 years. You can get plenty of things that are good for three or four years, but not everybody likes their whole thing to delaminate, just like the Qantas airline reinforced composites. So they learned about the problem of adhesion.
You can have a honeycomb sandwich, and this honeycomb sandwich allows me to get into the X-33 space plane in a little more detail, unless people have questions. This is a case study.
Concrete is a composite. It's stone and Portland cement. Reinforced concrete is steel, stone, and Portland cement. That's at the low end. This is Nomex, a high-strength polymer fiber that's been made into a paper-type material with a resin in it. Mostly you think about carbon fiber composites — why do you think of that? Are you in the automotive business? You'll see that Boeing has a lot of carbon fiber composites. We had a similar fiber composite for 60, 80 years before that. It was called fiberglass. Before we had carbon fibers we had glass fibers, which — I told you yesterday — if you take those glass fibers as soon as they're formed, they'll be stronger than steel. They'll be just as strong as the carbon fiber, but carbon fiber is lighter, and the costs have come down dramatically.
The way BMW was able to build the i3: they took carbon fiber, which is so expensive that it could only be used in the aerospace industry — at $200 a pound for a pound saved — and they started with the premium vehicle. The BMW i3 is like a $50,000 mini car. They put the construction of the fiber in the Pacific Northwest, where they have lots of hydroelectric power. To form carbon fibers, you've got to put the polymer in the furnace at 2,000 degrees centigrade. That's a tremendous electrical energy cost. So they had to move their manufacturing to where the energy was cheapest in order to get the cost down. There were all kinds of other things they did. It's a very interesting systems thing. It's one of the students' presentations in this class — one of the LGO students who happened to have been interested in it for other reasons did a wonderful presentation on how the BMW i3 took about five or six different ways to drive the cost down. But it's still a $50,000 small car, when you could buy another small car made out of steel for $15,000. You're paying three times as much for carbon fiber.
Still generally too expensive. I talked to one of my doctoral students, who's a partner at McKinsey now. He bought an i8, which is also carbon fiber BMW. Chris did his thesis on polymers, and he knows all about composites and resins. I told him I was putting in ten kilowatts of photovoltaics on my house when they finish it this month, and so I might buy an electric vehicle. I could become net energy zero with ten kilowatts of power on my roof and a Tesla battery in the basement — until Belmont tells me to go pound sand. The state of Massachusetts has told them they have to buy back my power. Chris told me that the quality of the resin transfer molding of the BMW i8, which will go zero to sixty in under three seconds — fantastic performance, that's what Chris wanted — he said it stinks. He said you couldn't build an airplane out of something that crappy. So it's junk. Looks good enough for the shell of an automobile, because all it has to do is deflect the wind. It's not really structural. But it's structural on Boeing aircraft, and the fibers have to be in the right place and the right orientation. Lots of people can make a cheap thing, and they learn to make it cheap, and they get better and better until they finally end up with something that's really structural.
In the meantime, there are some people who decide they want to build a composite structure — the X-33 space plane. The whole thing is composite, but on different levels.
§6. The nacelle question [26:09]
Student: [Question about the nacelle on the aircraft.]
Yep, here's the nacelle right here, one of the green ones. You'll see if you go to the 787, the nacelle is also — in this case it's titanium, or it may be carbon sandwich, probably carbon sandwich in titanium. I'm color blind. It's not just an encasement. The cost of the engine is about five or ten million dollars; the cost of the nacelle is also about five or ten million dollars. So on a 787 it's probably ten million dollars an engine and ten million dollars per nacelle. The nacelle, you're right, is just a shell on the outside, but it's got the major structural components that have to take all those hundred thousand pounds of thrust out of that engine and transfer it into the wing. There are interesting structural parts. I've seen some failures of nacelles, and you've got pieces of high-strength steel that are three inches square, and in some accidents this stuff is twisted and torn. Big forces on the nacelle. A lot more technology than I realized went into a nacelle, until I started looking at some of the failures. You'll learn a lot from the failures.
Some of the earliest composites was the outside of the nacelle. You'll see, with the 12 percent on the 777, they were putting it in the nacelle, because at that point it's just a wind break. That's what it was on the BMW i3 — just a wind break.
§7. The X-33 space plane [28:13]
There are lots of different levels of composite. Here is a composite structure made of sub-composite structures. That was a 1.3-billion-dollar program that failed because they were a little too aggressive in trying to push new technology, including composite technology. The X-33 space plane, I mentioned it before, had liquid oxygen in an aluminum tank. Originally they were going to try to get it to a composite tank, but there are real problems with holding liquid oxygen in a composite. There are problems with holding it in aluminum too. The two hydrogen tanks ended up being — this is a piece from one of those hydrogen tanks, from the X-33. They had to build two of them. They were in four lobes apiece. They would lay up these things in four individual lobes and put them in the autoclave, and then they would essentially join these together with a bunch of adhesives. These struts and little round tubes were Inconel and titanium. But these multi-lobe hydrogen tanks — this was the final decision to kill the program.
It had two aerospike engines that had been developed by NASA Stennis Center in Mississippi. There were metal parts, but the whole structure — the main strength of the structure included these three tanks, and the stuff on the outside was just the skin to give it aerodynamics. Not only did that skin give aerodynamics, but these little wings — this thing was supposed to be single-stage-to-orbit, which meant you carry all your fuel with you to get up to space, and then you come back down and you land it. Do you think these little wings were going to be sufficient to land 140 tons? Of course not. It turns out the whole structure was the wing. The shape of that triangle gave lift. It didn't have to give fantastic lift, because it would take off vertically like a rocket, and it had to have just enough lift to be able to land in a way other than a rock. It had to glide. It's a glider.
So if you look this up, Lockheed Martin — I guess it was Lockheed before Lockheed Martin that did the C5-A and beat Boeing out for the aircraft transport back in the 60s — the X-33 Lockheed Martin was supposed to replace the space shuttle. It was supposed to be single-stage-to-orbit. The problem with $10,000 or $20,000 a pound in orbit is that you have to throw away your vehicle on every trip, and that's expensive. That's one of the things SpaceX has done. SpaceX has a returnable vehicle, in theory, as long as it doesn't blow up on the pad, which they do. We can talk about those statistics. If you can bring back your structure and reuse it — that's what the Space Shuttle was supposed to do. If you go back and look at the claims and the justification, the Space Shuttle was going to take $10,000 a pound in orbit and drop it to $1,000 a pound. It turns out they had enough problems that they actually raised the price to about $40,000 a pound. The Space Shuttle was never cost-effective. But it was our main way of getting things in space for thirty years. We had a fleet of five, and when we lost one we built another. Finally it was just too expensive to maintain and they decided to go back to the old disposable rockets.
§8. Reliability statistics, Feynman, and SpaceX [32:43]
One of the ways they justified the Space Shuttle: NASA did big studies. The guy Ed Roberts over at the Sloan School was one of the people who helped NASA learn how to manage huge projects like the Apollo project, putting a man on the moon. People at MIT in the management school basically said, how do you manage something as huge as putting someone on the moon? This is a very complex project. They developed techniques to look at reliability. They told Congress that the Space Shuttle had one chance of a disaster on takeoff out of 10,000 operations. Does anybody know what the actual failure rate was? When the Space Shuttle Challenger went up and failed, it happened to be the 25th shuttle flight.
One of the guys on the commission to evaluate what happened in the Space Shuttle Challenger was Richard Feynman, who won the Nobel Prize for quantum chromodynamics [quantum electrodynamics] at Caltech, but who was an MIT undergraduate in physics. Feynman pointed out to the commission that since Wernher von Braun had come from Germany in the second half of World War Two, all the rocket flights out of Huntsville and Cape Canaveral and Edwards Air Force Base, 4 percent had failed on liftoff. But NASA, supposedly because of these reliability calculations, claimed the Space Shuttle was going to be 0.01 percent, not 4 percent. Feynman pointed out that it just happened to be the 25th shuttle flight, so that's your 4 percent. In those statistics it could have been any time between one in zero and one in fifty, but it just happened to be at 4 percent. It was a good way of making the point to lay people who don't understand statistics: these things fail 3 or 4 percent of the time.
SpaceX had been touting that their aircraft were going to be single-stage-to-orbit, and much more reliable — I was reading this stuff a week ago, it was something like one out of three hundred. 99.7 percent reliable or something. Well, it turns out the SpaceX shot that failed recently. So we haven't really improved much from the 4 percent. People can wish that things would be more reliable, but that doesn't make them more reliable. That's another management principle. This X-33 had so many new pieces of technology in it that it failed, and the reason this particular one failed is interesting.
§9. The Gulf War, helicopter engines, and the engine air particle separator [36:10]
After the first Gulf War — we took six months to mobilize in Saudi Arabia to invade Iraq and beat back Kuwait. Now not every enemy is going to watch you mobilize for six months and wait for you to attack. The first Gulf War lasted either three days or seven days. It's like the Israelis or the Egyptians — it's all over in just a few days once they decided to attack. They needed to mobilize, and they had all kinds of problems. All the helicopters had not been designed to operate in a sandy desert environment. When they got over there and started doing some of their initial stuff, they found the helicopters were losing their engines in twenty-four hours, because you start putting abrasive sand through an engine and it sandblasts itself apart. So they had to develop better engine air particle separators.
You might know what an engine air particle separator is called. EAPS. Well, you have one yourself. It's called your nose. If you're in a dusty environment — out working in the garden, all dusty, or in a coal mine, all dusty — you breathe air, and it has to go up over and around. It turns out the particles that can get into the worst parts of the lung are between 1 and 10 microns. Particles that size and larger will not be able to turn the corner. They have enough momentum that when the air has to go 180 degrees, in several 90- and 180-degree turns, their momentum just keeps them going straight, and they hit the mucus on the walls of your nose. When you blow your nose, you see all the dust that the mucus is designed to catch. You have an engine air particle separator — two of them — right here.
The reason asbestos gets down in the lungs is that it has an aspect ratio of chalk. A five-micron particle of dust is heavy enough that it will hit the wall when it tries to turn the corner. Asbestos fiber can be much longer — 20, 25 microns long and only one micron in diameter — and guess what, it will go with the flow, because it has a different drag coefficient. Your nose was not designed to breathe asbestos, and so the asbestos fibers get down in the lungs. It's just aerodynamic stuff.
§10. Rapid prototyping and the X-33 contract [39:29]
After the first Gulf War, they had learned about the importance of rapid prototyping. That was the big buzzword in Washington and on Capitol Hill. NASA sold this to Congress as a 1.3-billion-dollar program that was going to be able to rapidly prototype from letting the contract to first flight in thirty-three months. Design it, build it, fly it in thirty-three months. It was going to be a demonstration of how quickly you could prototype a very complex thing. It was going to replace the Space Shuttle. It was half the size of the Venture Star, which was the design that was going to be large enough to actually replace the Shuttle. The cargo bay here was 5 feet by 10 feet, whereas on the Venture Star it was going to be as large as a Space Shuttle. You had to get those lasers in space for the Air Force.
The people who got it were Lockheed Martin Skunk Works at Edwards Air Force Base in Palmdale, California. They proposed using three-dimensional woven carbon fiber composites. This is just a ring-stiffened cylinder, like a submarine — you've got a cylinder and rings on the inside and beams to stiffen it. They were going to make those beams out of carbon fiber composites that were a three-dimensional weave, so you'd have your two-dimensional weave, you'd lay them up, and then you'd have people with needles and thread literally hand-sewing carbon fibers across these plies to get the three-dimensional structure. That's the only way people knew how to do it.
They went out to price it, and they found this was going to consume the whole budget. They said, well, that won't work. They only had about a month or two left to design the hydrogen tank. They said, we can buy off-the-shelf Nomex honeycomb, we can buy adhesives, we can buy carbon fibers that we can laminate. We already know how to make carbon fiber composites from the aerospace industry, and we could build these hydrogen tanks. Up at Morton Thiokol in Logan, Utah, we have autoclaves that are large enough to autoclave one quarter of the tank — one lobe.
§11. The autoclave, the 1.05 safety factor, and the 3M epoxy data sheet [42:06]
They decided to build 10 of these lobes, put them in the autoclave, heated them up, and two of them were junk. A couple of the others had some delaminations. It wasn't a good day. So they sharpened their pencils, measured the strength of these delaminated areas with ultrasonics, and decided that instead of a safety factor of two, they had a safety factor of 1.05.
Anybody know about strength safety factors, typical values? This building you're in, by law, has a safety factor of 5/3 on mechanical strength — 1.67. The bridges you drive over, by law, have a safety factor of two, because of vibration and things like that. You climb a ladder — everybody's been up on a ladder — that's got a safety factor of four, because OSHA requires a factor of four for scaffolding. The more hazardous something is, the bigger the safety factor.
They designed this thing — they can't go too big because it's heavy, but to make things safer. They designed it for 2.0, but because of the delamination during autoclaving they found they only had 1.05. But this is a billion-dollar project; 1.05 is greater than one, so it should work. They took one of these tanks and put it in test in Huntsville, Alabama.
This was the laminate. Instead of the three-dimensional woven composite, which was way too expensive, they were going to take the carbon fiber 0, 45, and 90 degree ply, then have this sheet of epoxy, then have the honeycomb that you can buy off the shelf, and make a sandwich layer. They went and looked at a 3M data sheet for epoxy adhesive. The epoxy is already made up in a big sheet by 3M, kept in a refrigerator, so the epoxy won't cure until you get it to the autoclave. They looked at the PowerPoint slide that 3M had on their website, and it said good for ten days out of the refrigerator. They figured they could make those lobes, get them all built and into the autoclave within ten days. And they did — they got them all in within four to nine days.
What they didn't stop to think about was the actual data. This is not the actual data, but it's a typical curve for epoxy shelf life — it's actually for marine epoxy, in minutes rather than days, but think of this as time out of the refrigerator versus days, one through ten. The strength of the epoxy that hasn't cured is starting to cure at room temperature as soon as you take it out of the refrigerator. They've got to lay all this stuff up, form it, and then get it in the autoclave to join it and heat it up. And lo and behold, the curve looks like that. When 3M had their bullet point, they were talking about 10 percent of the remaining strength after 10 days. NASA had assumed that 3M meant 80 or 90 percent of the strength, because it said right there on the bullet point that it had a shelf life of 10 days. But 3M defined the shelf life, as good marketing people, as 10 percent of the expected strength remaining. NASA assumed, well, they must mean it has most of its strength left.
§12. Hydrogen permeability and the failure [47:10]
They found the strength had delaminated, and they said, okay, it still should be good. After measurements on the other two cores that had failed, they calculated from statistics that they had a safety factor of 1.05. So they take it down to Huntsville, Alabama, because not everybody has a test facility with enough liquid hydrogen to fill up something this size. This is scaffolding here — this is about 20 feet tall. Here's a lobe. They put it in test with the liquid hydrogen. Everything's video; you don't have people around liquid hydrogen, because if it goes boom, it's not a good day.
Everything passes. As they cooled it down and started warming it back up, they were giving themselves high-fives. Then, on the video, about two to five minutes after it started warming up, they saw this pop, and a bunch of the frost came off. What had happened: they had done some tests of permeability of hydrogen through this graphite. Hydrogen diffuses through anything. They had done some tests at Stennis in Mississippi, and they had some numbers, but the numbers were fishy. When they went back to ask how they got the number, no one could tell them. Apparently someone pulled the number out of the air and gave them a number, and they did their calculations on that number.
When they had the failure afterwards and were trying to analyze it, no one could tell them where they got the number. When he redid the test for the first time, the numbers were off by about three orders of magnitude for permeability. Hydrogen — liquid hydrogen — diffused through, was inside these pores, and when that liquid hydrogen started to expand, it built up too much pressure, and things blew apart. They realized they could not fly this thing. That was the final Achilles heel that killed the program.
So that's the X-33 space plane — a number of mistakes. They couldn't reproduce the numbers when they were challenged. And when you start looking at bullet points on data sheets, sometimes you'd like to see the real data. It only cost a billion dollars.