§1. NDT vs. radiography in Navy shipyards [00:00]
[Tom passes a reference radiograph around the class.] You have to clear out the shipyard while you take your x-rays. I can't get enough intensity here — pass it around, hold it up to the light and you can see bad welds. You really need a high intensity lamp.
By 1995, David Taylor — anybody know what David Taylor is? It's an NDT facility. I used to spend summers there in the 1980s. As you go around the Washington Beltway, at about the 9:30 position when you cross the Potomac River from Virginia into Maryland, right there on the Maryland side just west of the Beltway, is the David Taylor Model Basin. David Taylor was some ship designer from 150 years ago. The Navy ship research and development center is NAVSEA's main laboratory.
The Naval Research Lab, which has turned out a number of Nobel laureates, is the Office of Naval Research's research laboratory. The David Taylor Model Basin of the David Taylor ship research and development center is NAVSEA's research laboratory. And Pax [Patuxent] River, Maryland, is the NAVAIR research center. Each big group in the Navy and the Air Force has its own research center. The big center for research in the Air Force is Wright-Patt — Wright-Patt is sort of the basic research, what they call 6.1 and 6.2 money. 6.1 money is basic research. The 6.1 laboratory of the Navy is the Naval Research Laboratory, where they're studying astronomy and things like that for obvious reasons from the old days of navigation. They were one of the leaders in developing GPS — GPS was because they wanted to do navigation, but it got to be more than that.
David Taylor is what they call a 6.2 lab. These are pots of money that Congress gives out. Congress says you can spend so much on basic research, so much on exploratory research. David Taylor is the 6.2 lab of NAVSEA. They also do a lot of 6.3 and 6.4 — 6.3 is advanced development, 6.4 is prototype. There are all these different pots of money in the DOD accounting system.
By the mid 1990s, David Taylor had done enough work, and ultrasonics had come along well enough, that they said: instead of doing radiography in shipyards, you have your choice — you can do x-rays or you can do ultrasonics. Immediately all of the civilian shipyards switched over to ultrasonics, because it was cost-effective. You didn't have to shut down that area of the shipyard while someone brought out an iridium-192 radioactive source and everyone had to clear out for fifty yards around. You had to have one or two people doing the radiography and ten people on security to make sure nobody except rats went into the radiation zone. They probably are trying to protect the rats too nowadays.
But the Navy shipyards, the ones run by the Navy, didn't want to do ultrasonics, because it would decrease employment. It would save money. And that's the last thing a Navy yard wanted to do, because these are civil service folks — they're their buddies. I had a Navy student do a thesis on this about six or seven years ago, on why the Navy yards didn't switch over. He didn't say it quite as bluntly as I just did, but there was no good reason not to switch over to ultrasonics.
So today ultrasonics is the method of choice for heavy-section welds in a shipyard, because NAVSEA did a lot of work at David Taylor to prove out that ultrasonics works just as well and can find flaws just as well as x-rays. And it also leaves a permanent record. Now that we can store reams of information, you can store the whole ultrasonic test. In the old days, one of the things people wanted to do was store the reference radiographs on a pipeline or a nuclear reactor — you have to keep those radiographs for the life of the nuclear reactor, thirty or forty years. You have to keep your quality control records. Nowadays with ultrasonics you can keep them on a disk drive. It might be terabytes, but you can still do it.
So ultrasonics is the method of choice in your generation, but you only have to go back fifteen years and it wasn't even allowed. It was only about fifteen years ago that NAVSEA made the change and said you could use UT rather than radiography. And then you run into the politics of which shipyards would adopt it — the ones that were trying to save money versus the ones that were trying to spend money. There are important lessons there for you.
§2. Standards as design rules — and as innovation barriers [06:37]
[Tom holds up the NFPA lightning protection standard.] I'll pass this around. This is the 2011 edition of lightning protection systems. I handed out the paper about Ben Franklin and lightning protection systems, and this is what Ben Franklin has evolved to. Seventy pages on how to design a lightning protection system for your house or any other building, or an ammo dump. There's a separate standard for marine — if you've got a sailboat or an aircraft carrier. When you're out in the middle of this big flat ocean and you're the tallest thing around, you actually need a lightning protection system, because you're going to be where it strikes. There's nothing else around for it to strike. It always wants to strike the highest thing around, and you're the highest thing for miles, particularly if you're an aircraft carrier. So you have to have a good lightning protection system, and it's very highly engineered nowadays.
A lot of these standards end up becoming design rules. That's actually where I started this course: if you think you're going to design something, they tell you how to design it in the university, but they don't tell you about all the restrictions on your design. If you're going to design a lightning protection system, that tells you how. And if you don't know anything about how to do it, it tells you how to do a better job than Ben Franklin did. But it also sometimes stifles innovation. The boiler and pressure vessel code: we are still using fifty-year-old steels. The reason is that it would cost several hundred million dollars to qualify a new steel. We have better steels now than we did fifty or sixty years ago, but essentially things are grandfathered in.
If you look at medical implants — polymers that go in the body — back in the early 1970s the Food and Drug Administration came out with a law about medical implants. They were becoming more prevalent, and the law basically grandfathered in all the plastics that had already been used successfully in the human body, things like polylactic acid. Why would polylactic acid be compatible in the body? It breaks down into some of the components of milk — it's lactic acid. That's why people have chosen it. Polyethylene, which is extremely inert. People have done a lot in plastics over the last forty years, but there has not been another plastic certified for use in the human body since the early 1970s, because the law set down the specifications of what you had to do to certify a new material. You'd have to do hamster studies and dog studies and monkey studies — it's going to take years and nobody can afford it. So they always try to engineer around the existing materials that are acceptable.
Those are two examples of specifications stifling innovation. Specifications are good and specifications are bad. They're good because they tell us a lot of the experiential design rules — you don't have to make the same mistakes as the people who went before you. But they're bad because you can't innovate. No one's going to let you take the risk. Somehow, someday, somebody's going to have to resolve that issue.
§3. "You only get what you pay for" — 60 Wall Street and the cost of inspection [10:44]
[Tom sets out the 1979 and 2010 editions of the structural welding code side by side.] Different codes and standards, who writes the codes, and we talked about costs. Just to give you an example — the standard is reference radiographs. There's a written standard you can find in a book, all of two pages long. But what goes with it are all these reference radiographs. I bought this for probably two or three hundred dollars back in the early 1990s. It now costs about $500 or $1,500, because people have decided to make all kinds of money.
This morning I went — to give you an idea how specifications have grown in size — this is the 1979 structural welding code for steel, and this is the 2010. You can bet that almost every line in this new one is the result of some lawsuit. Someone made a mistake, they fought about it, and three or four or five years later someone writes something into the code to explain things. That's where we left off last time, where I said you should only get what you pay for.
This is out of the structural welding code; it's been there a long time. If you decide you want something other than visual inspection, after you've already written your contract that said you just want visual, and you decide you want to do some destructive inspection — it says the owner shall be responsible for all costs of this, but the contractor has to do it. The contractor can't refuse to do the extra inspection, but the owner has to pay the extra money, because it wasn't in the contract to begin with.
This comes up all the time. I'll give you an example: 60 Wall Street. 60 Wall Street is a forty-story building in New York City — it's JP Morgan's building. On the seventh floor they have their computers for stock-market trading. If they lose those computers, they can start losing one or two billion dollars a minute, because of the small changes in price on huge sums of money. If your computer can't keep up with everybody else's, you're not going to be able to do your trades and swaps. Nowadays we're getting down into milliseconds of interest.
They built the building on a very rapid erection schedule. They were building this almost forty-story building, and they were building the riser pipes for the air conditioning system — 16-inch pipe, 16-inch water pipe goes up forty stories. If you calculate the pressure of a forty-story building, about 400 feet, at 0.44 pounds per foot, you're going to have a couple hundred PSI on the ground floor. A couple hundred PSI inside a pipe of 16-inch diameter is fairly high pressure.
They were building it so rapidly that as they're putting the steel up, they were also welding these vertical pipes in place. They got it all built, but the inspectors didn't want to walk out on the I-beams. So they didn't do a lot of inspection of the welds on these riser pipes. About ten years into the system, someone had reason to look at one of these welds. When you're welding a pipe, you're only welding from the outside, and the question is whether you got enough penetration through the thickness. The welds were lousy. Of course they're lousy — if you're on a rush job, the guys are up there hanging off an I-beam with no protection, no wind protection, trying to weld, and they know the inspectors are ten floors beneath them and won't be up there by the time they're done. There wasn't a lot of incentive to do a really great job.
So now they realize they've got really lousy welds in the riser pipe. What do they do? If they lose the air conditioning, they can lose a billion dollars a minute. The building, height-wise, had gone up to about the tenth floor and stopped, and the other thirty floors went above it. So they had this little ledge area, and they spent seven million dollars to move one of the air conditioning compressors. The air conditioning compressor is about the size of this room. They had seven of them in the building. The designers determined they needed two, so JP Morgan put in seven — that's the JP Morgan way of doing business. If you need two, you buy seven. They took one out of the basement and put it on the tenth-floor roof ledge, to cool the computers that were on the seventh floor, as a separate redundant air conditioning system. So that if they did have a leak in their riser pipe and lost all the air conditioning in the rest of the building, they wouldn't lose their trading floor.
Then they got into a fight with the contractor over who was going to pay — not only the seven million but the other twenty million to repair and replace all the rest of the riser pipe. What had happened is they had only bought visual inspection. These welds pass visual inspection — they look fine on the outside, but they were crap on the inside. If it's a half-inch pipe and you go a quarter inch in, there's no more weld. There's only a quarter-inch weld, not a half-inch weld.
This type of fight comes up — I end up getting involved in one about twice a year. The oil-company case I mentioned, with the 600 lengths of pipe: they found one leaking pipe out of 600, and they decided to rip out everything they had already put into their refinery. A three-or-four-million-dollar issue, which could have been solved by a hundred thousand dollars' worth of inspection, turned into a sixty-million-dollar lawsuit, because someone said rip it all out. A lot of times you get a manager who thinks, I'm going to get the other guy to pay for it, so I'm not going to take any risk on myself — I'm just going to rip it out and sue them for the difference. Well, it doesn't work, folks, because the other people will decide you had to be reasonable about what you decided to do. But that's why the codes triple in size. That's why the people who sell the codes, the for-profit people, have decided this is a gold mine.
§4. NDT versus NDE; the Boeing catalytic converter spec [19:17]
So you should only get what you pay for. That's why "should" is in quotes, because you can try to go to court and get something else beyond what you paid for. That pipeline inspection — the hydrostatic test, which was required, sort of equivalent to visual for a weld, might only require 40% of the wall thickness to be good quality. But the non-destructive tests are surrogates for an actual stress test. A hydrostatic test is a real proof stress test on the pipe, and it'll tell you how much load that pipe will take without bursting — a measure of real manufacturing flaws.
Whereas the non-destructive tests — eddy current, x-rays, magnetic particle — will find flaws, but then you have to determine whether the flaw is significant. That's the difference between NDT and NDE. Anybody know what those two acronyms mean? NDT is non-destructive testing, NDE is non-destructive evaluation. A non-destructive test is something the technician does. He goes out and does a test, and he finds — not a flaw — an indication. That's the non-destructive test: to find that indication by whatever technique.
Then the non-destructive evaluation — someone has to come along and interpret it. The guy who actually runs the x-rays can say, wow, it's bigger than this standard in that little green book back there. In terms of doing an engineering interpretation — actually those guys can do some interpretation. A lot of times they can tell whether it's an artifact of the x-ray or the ultrasonics, depending on surface roughness or a surface gouge, rather than something internal — whether it's a false indication or non-relevant or relevant. If it's relevant, now you get down to NDE, the evaluation. You really need someone with engineering training to determine whether the flaw is acceptable or rejectable. That depends on the stress level, the environment, the wear allowance, the corrosion allowance. Obviously the guy who runs the x-ray machine can't tell you any of those things, so it really has to be an engineering type who can do the NDE based on the finding of the NDT.
Many of these codes — the structural welding code for steel or aluminum — tell you what's an acceptable indication or an unacceptable indication. We talked about an eighth-of-an-inch flaw being typically what is allowable in many codes.
About twenty years ago, I came back from a trip to Japan, I'd been gone one or two weeks. I get back and I have a note on my desk to call this company down in New Jersey — I think it was Engelhard, makes precious metals. They were building a catalytic converter for the 747-400. The 747 comes in different dash numbers — they're up to the dash 900 now, but back twenty years ago they were building the 400, an extended-range 747. You can be even more miserable on an eleven-hour flight than a nine-hour flight. The first time I ever went to Japan we had to stop in Anchorage, Alaska, because you couldn't go all the way across the Pacific. They did a great circle, you'd stop in Anchorage, you'd see this big stuffed polar bear, hang around for an hour and a half, then they'd refuel the plane and you'd get on. With something like the 747-400, you could go from Los Angeles to Sydney, which is one of the longest city routes on a great circle in the world. There are two or three longer, but now the 787 Dreamliner can do any of them, unless you're the Air Force and you want to go all the way around the world in one trip without refueling.
They had a titanium welding problem. Did you know there were catalytic converters on an aircraft? They're there for the air you breathe up there. There's a lot of ozone at forty thousand feet, and if you breathed that ozone you would get a headache on one of these long flights. So they take the air coming in when you're flying and put it through a catalytic converter, just like in your car, to get rid of the ozone. So you have two catalytic converters on a 747 — forget the engines, this has nothing to do with the engines. It's the air you breathe.
They have these pipes, a dual set, dual system, in case one breaks down. They had been making them out of stainless steel, and they decided for the extended range they had to save more weight, so they were going to make them out of titanium. My first welding contract for the US Navy, back in the 1970s, was welding of titanium, so I knew something about it. I get this phone call: they were not able to meet the Boeing specs. There were four welds in this catalytic converter, and they were getting a 50% reject on the x-rays. If you've got to make four welds and 50% of them fail, and you're only allowed one chance to repair — no second or third option — you can start figuring out the statistics of getting four good welds. All of a sudden it was going to hold up the rollout of this aircraft, which at the time was a hundred-million-dollar aircraft, a quarter-billion-dollar aircraft today. All these orders and Boeing's reputation were on the line, and here little old Engelhard couldn't meet the Boeing spec.
[Tom sketches the catalytic converter cross-section on the board.] What was the Boeing spec that was so critical? The story is that the spec made no sense whatsoever. Somebody at Boeing who was obviously not really competent — the catalytic converter looked like this in cross-section. Engelhard made the catalyst that goes in here, and here are the tubes for the airflow coming through. You had to make a weld here and a weld here, and they had to pass an x-ray inspection — like 1/16-inch-thick titanium. The spec said you were not allowed a flaw any larger than ten thousandths of an inch. That's about four human hairs.
Where did they get that spec? If you take an x-ray and use a five-times magnifying glass — which is what most specs allow, no more than a five-times magnifier on an x-ray, because if you go to twenty times you can find the whole Rocky Mountain range in the x-ray — at high enough magnification, someone found that was the smallest flaw they could find, and that's how they decided in their specification at Boeing to say: I want a perfect weld, I don't want any flaws. There was no acceptance criterion that made engineering sense. It was just what could be found. With welding of titanium, cleanliness is next to godliness. Fingerprints can give you enough hydrocarbon residue to create a pore on that weld.
I'd just gotten back from two weeks away, I didn't have time to go to New Jersey, so I told him on the phone: how are you cleaning it? We're just cleaning it in some acetone. I said, what grade of acetone? Get some reagent-grade acetone and clean it. Because most acetone that comes in a 55-gallon drum, if you pour it on a glass slide and let it evaporate, you'll see Newton's rings. There's a film of oil on there, just like fingerprints. Get some reagent-grade acetone, clean it, and try your welds and see if you can get better than 50% pass on your 360-degree x-ray.
They tried for a week, and they called back and said, you've got to come down, we're still not passing. I had to be in New Jersey for something else, so I said, if you can stay until six o'clock at night, I'll come by after my other business. So they stayed, and I went by and I looked at their operation, and I said, well, how are you cleaning it? They had not gone to get any reagent-grade acetone. They had not taken my advice that I gave them for free on the phone. They just said okay and tried to keep doing the same thing. The current definition of stupidity is to continue doing the same thing you've been doing and expect a different result.
So I walked through the plant and I told them: get yourself some reagent-grade acetone. That's how I left. They did it, because I had met with them face to face, and I told them the thing I'd told them on the phone, but telling them face to face made them believe it. They were delighted. A week later they had only failed one out of ten rather than one out of two. They sent me a check.
Three weeks later they called me up and said, we're falling off the cliff again. I said, well, are you following the procedure? Can you come down. This time they had to pay the whole freight of going down. I went down, looked at what they were doing — they'd quit doing what I'd told them to do and gone back to their old procedures. So I told them a third time how to clean it, and I guess 747s are flying.
The real problem here was that someone at Boeing had written a stupid spec. Just because you can find a flaw doesn't mean the flaw is harmful. Here's an example of a specification that should never have seen the light of day. It was going to tie up getting the entire aircraft out the door because someone was writing a CYA spec without knowing any better.
There are a lot of specs like that, which gets down to the question of whether the design code or specification is good enough. If you've got a well-thought-out and peer-reviewed code like the boiler and pressure vessel code — it's been around for a hundred years, has about three or four hundred people on the committee, and those people keep each other honest — then it's a pretty good code. Maybe not always good enough, but usually good enough. But when you've got one guy fresh out of school who went to work for Boeing, doesn't understand anything about fracture mechanics or non-destructive testing, and he's told to write a specification — he writes one that requires people to do things that are a total waste of money.
§5. Inspectors with a little i and a capital I [32:21]
After those stories, let's talk about inspectors with a little i and a capital I. Anybody have any idea what I'm talking about?
Student: Inspector Gadget?
There is an Inspector Gadget, and if you've got Inspector Gadget you don't need codes and specifications. That's actually a good definition. If you look in the structural welding code, it says all welds shall be visually inspected. Does that mean the capital-I Inspector has to do it? No. That means someone has to inspect it.
Who usually inspects with the little-i? The welder himself. He looks at it, and he can tell — if it looks like a turd on a plate, it's probably not a good weld. If it's nice and smooth, it may not have full penetration depth, but it's got a chance of being a good weld if it looks good. The code requires certain types of capital-I inspectors.
[Tom turns to chapter 6 of the structural welding code.] This is under inspection, chapter six. Contains all the scope — remember these things tend to include things like scope. Contains all the requirements for the inspectors' qualifications and responsibilities. There's the contractor's inspection and the verification inspection. Those of you who've worked in shipyards know this. If you're at Bath Iron Works, they have their own inspectors. They have their own welders who are supposed to look at it — that's little-i inspection. They're not doing the structural welding code, they're doing some Navy spec, but nonetheless the contractor — the people building the vessel or the ship or the bridge — have their inspectors running around.
But there's the verification inspection. The verification inspector is someone like Lloyd's Register of Shipping, or Bureau Veritas, or American Bureau of Shipping, or the U.S. Navy SUPSHIP. The contractor's inspection may have to be 100%, or 10%, for different welds — critical areas have different requirements. The verification inspection is quality assurance; the contractor's is quality control. The capital-I inspector has to be a certified welding inspector if you're doing it to this code — which is another way for the American Welding Society to make money, by certifying welding inspectors. But it's a worthwhile thing, because now you have a national standard — another standard that's come along in the last fifteen or twenty years — that everybody knows this person has met a certain minimum level of qualifications. They've taken a test.
Little-i and capital-I inspectors — that's my terminology. If you look at the boiler and pressure vessel code, it gets very specific about who this inspector can be. There's actually a whole organization, the National Board of Boiler and Pressure Vessel Inspectors, which I mentioned before. Boilers and pressure vessels have to be inspected, like every three years — depends on the state. Each state has different laws and requirements, but a typical average is about every three years a guy has to go in and pull things away and look at things.
Some of those inspections get fairly involved. You have to remove the insulation if it's an insulated pipe. They don't have to inspect everything, but they have to inspect enough. In their standards there's a book this thick that tells them what they have to do. They can tell the owner, you've got to remove the insulation on this pipe so I can look at how pitted it is. There's something called corrosion under insulation, which is a big problem — moisture builds up. I've seen a number of failures because of corrosion under insulation. They go in with an ultrasonic thickness monitor and see how much wall thickness is left.
One of the problems is that it's often a lot of work and expense to remove insulation. I have a situation right now up in Salem Harbor: a forty-year-old boiler for a utility plant generating electricity. For about fifteen years, the Massachusetts state inspector decided it was too much trouble to make them clean all the ash off the manifold header at the bottom of the boiler. There was ash built up several feet deep, and they were getting corrosion under ash — not under insulation, but the same type of attack, where you collect moisture. The boiler code said this was one of the critical areas that had to be inspected on a regular basis. And every three years for about twelve, fifteen years he decided, I will wait till the next time to do it — right until it blew up and killed four guys. A lot of times there's a little money under the table to save the expense of removing all the ash or the insulation.
Remember the boiler and pressure vessel code under scope excludes human-occupied vessels and your home hot water tanks. One of the exclusions was anything that operated below 210 degrees Fahrenheit. Where did you get a number like that? It just happens to be two degrees less than the boiling point of water.
There's a company in Tennessee making Manwich. Anyone ever had a Manwich sandwich? A little tomato barbecue sauce you mix with some ground beef. My grandson loves it. They're making Manwich, and they have this stainless steel hot water tank to hold — well, it's supposed to be water, not steam — that, if the Manwich line stops, has to keep the Manwich tank above 180 degrees, because if it drops below 180 those little bugs can start growing and my grandson's going to get sick. To stop the line for an hour or two if something's gone wrong, they have this hot water tank.
There might have been some problems with the controls. The tank was stainless steel. It's a food processing plant, so what do they use to clean everything? Bleach — it's a great disinfectant for those little bugs. What does bleach do to stainless steel? It corrodes it. And what happens underneath the insulation — if some of that bleach gets under the insulation, you've got corrosion going on that no one can see, and no one takes the insulation off. This vessel looked like a dried riverbed in terms of all the cracks all over it, until it blew up and killed a couple of people.
Inspections have to be done with some common sense, and they can end up costing money, but a lot of the inspection requirements are there because we have a history — someone got hurt once somewhere. We have these codes growing in size, doubling every — maybe we'll come up with a Moore's law for specifications and codes, how many years it takes to double in thickness.
[Tom opens the NDT handbook to the management organization chart.] This is out of the overview of the non-destructive testing handbook. If you're interested in non-destructive testing, I've got a nice little series of books, about nine volumes, takes up about twenty inches of shelf space. You've got the general manager of the plant; the chief engineer, responsible for design and manufacturing processes and product specs; the purchasing agent, who keeps everybody from spending money — he's the Preventer, like Dilbert, the Preventer of Information Services. The purchasing agent is the Preventer of Productivity. You have a plant manager in operations, and you have a quality manager. Many of the codes require that the quality manager report to the head of the business and cannot report to the chief of operations. Why? Because there's an obvious conflict. Why do we have SUPSHIP in addition to the shipyard inspector? Because you can't have the fox watching the henhouse.
You have a chief inspector — the capital-I inspector — and he may have other capital-I inspectors with him. They report to a quality manager who can go to the general manager to beat up on the plant manager if the plant manager refuses to do what they say. The general manager arbitrates between the operations guy and the quality guy. That's how, if you go to SUPSHIP, you'll have the pleasure of spending half of your life in meetings where people are arguing over whether something was acceptable or not — and on things like the good old Boeing spec, where half the time the spec doesn't even make sense. Someone just created it out of whole cloth, but it's there.
§6. Levels of design — conceptual, architectural, detailed [44:41]
Let's switch gears and go back to design. We've gone through lots of things on codes and standards, but many of the codes actually have design rules. The boiler and pressure vessel code has specific design rules. Chapter two of the structural welding code is always on design.
Three levels of design. One is conceptual. What I'm going to do here comes out of buildings or bridges — steel construction type design, making something big and heavy. There are others. What's the difference between conceptual design and architectural design?
Student: Conceptual design represents the requirements together in broad stroke.
Exactly. You've got some concept — you're going to build a cable-stay bridge that's never been built before. We have cable-stay bridges now, but if you went back twenty-five years we didn't. So you had a conceptual design that — rather than having catenary bridges with big cables, like the Brooklyn Bridge, or the Verrazano-Narrows, or the San Francisco Bay Bridge, where you hang a cable between two towers and drop cables straight down to hang the roadway — the cable-stay bridge is like the Bunker Hill bridge, with tension cables, a bunch of them, and it makes a more pleasing thing architecturally, a more beautiful bridge.
Each one of the Bunker Hill bridge towers is topped like the Bunker Hill Monument, if you haven't noticed. The cable stay was a technology developed in France, where they actually use sensors to tension each cable. Those cables have to be equally or properly tensioned. If one was highly tensioned and the others slack, the highly tensioned one is holding up the whole bridge and it's going to snap. The thing that made the technology viable was the French technique of putting sensors on the cables when they install them and tightening as they keep building the bridge, so the cables bear the load in a shared way as intended.
Student: Do they re-tension the cables over time?
I don't think so. They may go back and check from time to time, but there's not a continuous adjustment every year. I don't know enough to say they don't check after five or ten years. It's a fatigue-loaded situation, but the tension should stay fairly constant. Plus there's a big safety factor — safety factors are tomorrow's lecture with Dr. Belmar, and safety factors take care of a lot of mistakes.
[Tom crumples up a piece of paper while the student is talking.] This is a conceptual design of an MIT building. Can anyone imagine which? The Stata Center. Frank Gehry, one of the world's top architects. He does concepts; he doesn't worry about details. If you've been through the inside of the building you'll realize he doesn't worry about details. He does concepts.
I heard this story from Vicki Sirianni, who was head of MIT physical plant. She's an architect; her husband's an architect in downtown Boston. When I was department head, at Christmas time I'd give the night custodians a Christmas breakfast between six and seven in the morning. I get in at 5:30 or six, and I park where the nighttime custodians punch their time clock at seven, so I knew a lot of them. We had to do it on the clock, and I had to get permission from Vicki, because a lot of these people work two or three jobs. They work nighttime at MIT and they finish at seven and go to another job at 7:30. That's the life of some of these people.
People used to say, why are you doing this, you're a department head? Well, we got better service at physical plant in the department when I was department head than any other department. If you show appreciation for people they will respond. There was a reason we did it.
Vicki had come back the day before, while they were thinking about the concept of the Stata Center. They'd hired Frank Gehry, because Chuck Vest was president of MIT and wanted to leave some legacy. The Stata Center was going to be that legacy. It was supposed to be the new main entrance to MIT on that side of campus — we had 77 Mass Ave, but this was over by the subway station. They'd built the Whitehead Institute and the biology buildings, and now we'd have the Stata Center. Frank Gehry, one of the world's top architects — Bilbao Spain art museum, the Getty Museum in Los Angeles — titanium-surface structures.
Vicki was sitting in his conference room in Los Angeles the day before, and he comes in, crumples up a piece of paper, throws it in the middle of the table. He says, that's what your new building will look like. So the Stata Center's concept was crumpled paper. Now as you look at the Stata Center, you'll understand why it's designed the way it is. You had an idiot for an architect, and you paid him a lot of money. The building was supposed to cost $150 million; it came in at $430 million. All of us took a pay freeze in 2003 or 2004 to help pay for it. Every time I walk through that building I'm thankful.
Some of you have been in the building. What design features struck your eye?
Student: I've only been on a couple of floors, and I don't like to go through the building very often.
That's MIT — it fits right in with the rest of MIT. I could say that about a lot of MIT buildings. There's no real uniformity in design — you'd want all your bathrooms in the same spots.
Walk through the lobby and look at the floor. This is a $400 million building — it's bare concrete. I think they put some lacquer on it. If you notice the gaps between the concrete slabs — I had a ruler in my pocket, or I went back with one a couple of days later — there's a gap a sixteenth of an inch wide and about three-quarters of an inch deep. Whoever designed that has never swept a floor. You know what'll collect in that sixteenth of an inch. So it's really helpful.
The other thing along with the concrete slabs and the concrete block wall — which they did paint — what's the furniture? Plywood. Custom-built one-inch-thick plywood with twenty laminations per inch. You know what this cost? No wonder it cost $430 million. We have custom-built plywood benches at fifty times the price of one you could have bought from a catalog. The other thing that's good about plywood — aside from the fact that it doesn't work — you can't refinish it, you just go down through the laminations.
Over in the lobby they have — what for a coffee shop? Starbucks. This is for the students. A lot of students spend three dollars a day for a cup of coffee. Turn around from the coffee shop and gaze upon the restrooms — the doors to the restrooms. And they have water fountains. How many water fountains are there? Five water fountains between the men's room door and the women's room door. This allows for high volume — if you've been to an airport where they have two water fountains, you know the long lines. We have five water fountains, staggered at different levels. I knew my pay raise had been well spent when I walked through there the first time.
Which brings us to the next level of design, called detailed. Conceptual, architectural — where someone says, we're going to use a certain size beam. Typical floor load in a building today for human occupancy, where you're not doing manufacturing, might be 100 pounds per square foot. A manufacturing building maybe needs 150 or 200 pounds per square foot, because you'll have heavy machinery. This building probably has 300 pounds per square foot capability, because it was designed by a bunch of concrete junkies back in the early 1900s — they didn't have the fancy computers.
[Tom opens the AISC steel construction manual.] In the American Institute of Steel Construction manual, there are column shapes — W shapes for I-beams. This is a W14 by — let's say weight per foot 22, that's pounds per foot. 14 would be a 14-inch web height. You can get different weights of these beams, and they give you the section properties. If you're a civil engineer, you look in this manual and someone else has already done the work.
The conceptual design says we're going to have so many floors, a shape that fits this footprint, and if it's a big skyscraper, we'll top it off and it'll look like this on top. Go look downtown in Boston and see how people have done different things. Architectural design is where some architect or draftsman or civil engineer says, if we're going to have this floor load, we need this size beam, this is how many beams, this spacing. But they just put in a beam — they don't tell you how to connect that beam to other beams.
There'll be a drawing — same type of thing on a ship. Some concept: a littoral craft, a submarine, an aircraft carrier. Some architecture lays out the geometry. This is, on a ship, the CAD system design. In the CAD system you don't have the welds, the bolted connections, the clip angles. You just have this size beam, this type of bulkhead.
Detail design, which today might follow on from the CAD system, is where someone says, this is the size weld I need here for this shear loading or tension loading or compression. I'm going to put a clip angle here. Because I also have to worry about another type of drawing — the erection drawings.
Anytime you're building a big thing you have to know the sequence to put it together. Particularly building in downtown Boston, you don't have a football field next door to lay down all your I-beams in order. You have to have them delivered every half hour during the day, and you lift them straight up off the truck by crane to the top of the building. They've got to come in the right order, because if not, where do you store them? The erection drawings tell you when to plumb the building, when to put in what bolts, what torques. You're getting to finer and finer detail.
Then there's a last set of drawings. Anybody know what the last set is? When you finish the building, someone goes back and says: how did we actually build it? Did we do it this way? Invariably the answer is no. There were changes made along the way. Many times the contract requires the contractor provide a final set of drawings — the as-builts. Sometimes someone goes back later, needs to do a repair, and you can't assume it was built as designed. In the real world, people run into conflicts and problems in erection and detailed design, and they find a way around it, or they decide not to put that system in and put another one in instead. When you go back to figure out how it was built, you can't use the original drawings — you need a set of as-builts.
Student: [story about a building dimension being different than designed]
Seven inches longer than it's supposed to be — that's how it was built. Did they let you through? Did they charge you extra? Bribes work.
§7. Hyatt Regency and the Pennsylvania roof collapses [61:25]
Things are not always the way they're supposed to be, and that usually is not a problem. But there's a fairly famous problem — the Hyatt Regency Hotel in Kansas City.
Student: I'm from Kansas.
So you already studied it. [Tom sketches the walkway support detail on the board.] It was a steel rod that was supporting the walkway. Those of you who have been through Hyatt Regencies — they have these huge lobbies that go all the way up thirty or forty floors, and people can go out and look. Great place to commit suicide. I grew up in Atlanta when they built the first Hyatt Regency, the first in the world. People would just walk through and see this huge lobby. The Kansas City Hyatt had walkways that went from the ends, where the rooms are. They had walkways at different levels, about twenty stories tall, five or six different walkways at different levels from different floors, and you could walk across a shortcut path through the middle of the air. You're walking in the clouds.
Student: [explains the design — the rods would be difficult to install as designed, so they extended them.]
Those rods would be difficult to install. When they extended the way on the left, it created a stress concentration, a shear right here. This little distance between here creates a shear load. The original was straight tension. The tension through the rod was sized properly, but not for a shear load on this little box beam. The box beam was already welded together, so you couldn't just put the clamshell together. These were supposed to be threaded rods, and someone was supposed to thread the nut on for twenty stories. How'd you like to be the guy who had to turn the nut? Talk about carpal tunnel syndrome.
So someone in erection decided that's too complex, let's do this. They just didn't bother to tell the design engineers, who were probably incompetent to begin with. Now, there is a procedure: if there's a change, it's supposed to be signed off by all the engineers, everyone up through the ranks. That didn't happen here. There were checks and balances in the procedure, but they didn't follow them, and so no one learned about this problem.
It was the connection that failed. It wasn't the rod, it wasn't the bolt — it was the C-channel; two pieces of C-channel just sheared, because you had a shear loading on them that was never designed for.
I've seen the same thing on roof collapses in buildings. In Pennsylvania, we had a snowstorm in the 1980s — the 300-year storm. Go to the weather channel, they have stories about this storm. I had work for the next two years on roof collapses. In that particular case, this was before computers got quite as sophisticated, and you're going to learn about safety factors tomorrow, but the building safety factor is 1.67. The 1.67 came about historically — people found that was good enough that we didn't have lots of buildings falling down.
In the early 1980s, some steel companies, some mini-mills, decided they would go into making bar joists for roofs — for malls and shopping centers. They could beat the system and sell lighter-weight joists if they started doing things like — [Tom sketches a bar joist on the board.] if this is your bar joist, it's a truss, with open spaces, angle steel or rods. Instead of making the whole thing three-quarters of an inch thick steel, they could make it thinner over here and thicker up there. They'd weld steel on steel to take up the bending stresses more efficiently, taking weight out from all the people who'd been building bar joists before. This was their competitive advantage. They could calculate these things on computers, because computers were getting more powerful in the mid-1980s. You had PCs with disk drives with 20 megabytes of storage and 128k of memory. These numbers may sound silly to you, but that's what we had.
The problem was, you could model a perfectly symmetric system in 1985 — you could not model an asymmetric beam. In order to actually make the thing, there was a little gap. The computer program had all three of these things coming together at a point, but in fact one came in here and another came in here, and there was a shear load similar to the Hyatt Regency, and that's why the building came down. You go look at the beams after the snow load, and the thing just sheared, like someone put it in a vise and pushed. Same type of thing as the Hyatt Regency: shear loading in the as-built.
Little changes, seemingly minor, that may not seem important to you if you're not the stress designer, can be very important to the piece of steel. The steel wins — they get the ultimate vote of what they can carry.
114 people died in the Hyatt Regency collapse, 200 were injured. Today the building is still there, but the lobby is two stories tall. They built another building beside it, and you check in there. The area where the twenty-story lobby was, where everybody died, is now filled in. There's nothing any taller than a ballroom there. Who wants to be in part of a lobby where 300 people got injured?
§8. NDT levels and the Ford Taurus air conditioner [69:52]
The other thing I want to talk about right now is levels of inspectors. On the American side of non-destructive testing, there are three levels, very cleverly named Roman numeral one, Roman numeral two, and Roman numeral three. An inspector with a level one certificate might have been working for six months as an apprentice, and he's learned how to do certain things. A level two inspector is probably several years further along and has taken some tests. Everybody has to take tests. It's like apprentice and journeyman in plumbing, if you know the codes and trades. You would think level three is someone even more sophisticated in non-destructive testing.
No, it doesn't work that way. Level three is the management guy — he probably doesn't even know how to turn on the machine. Level one is the operator, who knows how to operate the machine on the basics. Level two is the highest proficiency level you can get as an American Society of Non-Destructive Testing inspector, in terms of running the magnetic particle or ultrasonic testing. Level three keeps the paperwork. He's a clerk. He's often the owner of the inspection company, and he likes to say, I'm a level three inspector. Unless you happen to know the business, you don't realize that just means he's a clerk. He shuffles paperwork — not that the paperwork isn't important; someone has to send out invoices and cash the checks. But he probably doesn't know squat about how to do an inspection.
I see this all the time — some guy says, I'm a level three inspector, and people who aren't in the business think that's really something. It's not anything — it's actually less than the others. In the American Welding Society interpretations book on the structural welding code, there are questions about level one, level two, and level three inspectors, and whether a level three inspector can actually perform an inspection. The code says no — they haven't been tested. Let them stay in the office. You should be aware of those types of things.
Other examples of screw-ups that have happened because engineers don't talk to mechanics. About twenty years ago, Ford came up with a new air conditioner compressor for the Ford Taurus. A very clever design — a single piston that shuttled back and forth and acted as two pistons. You had two cylinders but a single piston: compressing one side while sucking in on the other, then compressing the other while sucking in on this side. To make it work, there were very precise machining tolerances — a fraction of a thousandth of an inch on the clearance. The engineers at Ford had conceived a less-expensive compressor design, fewer moving parts but greater precision in assembly.
They knew when they machined the cylinders they had to do it properly — if they weren't flat, if there was a bow, the shuttle piston couldn't go back and forth, because the tolerances were so tight. They wanted to send it out for prototype. They sent it to a machine shop — might have been a Ford machine shop — with a fixturing on how to hold the cylinder when doing the reaming and lapping and grinding. They got the prototype back, worked great. They spent a hundred million dollars to build a line to make lots of these compressors.
I happened to buy a Ford Taurus that year, in February. Come May, I go to turn the air conditioner on, and it doesn't work. I go to the Ford dealership and say my air conditioner doesn't work. Typical: how can we inconvenience the customer? They'll make you bring it in to diagnose, then another time to fix it, then a third time. If you're taking your car back to a dealership you know what I'm talking about.
I bring it in, they say, oh, it needs a new compressor, but Ford doesn't have any. I said, what do you mean Ford doesn't have any? I haven't heard they quit selling Tauruses. They had found a problem in the production. Lots of these compressors weren't working, and they were making sure all the new ones they had fixed were going to production, so they could sell cars. Those of us who bought a car in February — tough luck. They said, oh well, we may have a compressor for you in a couple of months. Gee, I can run my air conditioner in September? I said, the car is under warranty, I suggest you get me a whole new air conditioner or a whole new car if you like — this could be a Massachusetts lemon law. I did take a car in once back in the 1980s and won against General Motors. They finally got me a compressor — it was early June.
I learned the story from a friend. They built this hundred-million-dollar line, designed the fixturing just like the engineer had specified. Then they started finding nothing fit and they weren't getting their clearances. They went back and tracked down the root cause. Someone went to the machinist and said, did you machine it with this fixture? He said, oh no, I knew that wouldn't work. I changed the fixture. The prototype worked because the machinist knew the engineer's design was crap, and he fixed it. He just didn't tell anybody. Ford spent a hundred million dollars building a plant based on the crappy design.
The point is, it helps to communicate, it helps to have some respect for the hourly workers out there who are actually doing the work — sometimes they know a heck of a lot more than the engineers.
§9. Failures driving code change — Liberty ships and naval disasters [77:43]
[Tom holds up Petroski's To Engineer Is Human.] I wanted to talk about how failures lead to code changes — this is basically Henry Petroski. In the early-to-mid-1980s, he's a civil engineer at Duke University, and he wrote this book To Engineer is Human. He has something in here about the Hyatt Regency, and mostly civil engineering things — bridges and buildings, famous failures. To Engineer is Human comes from to err is human. His thesis is that the only way we really progress in engineering is by failure. We build a bridge, we build a new way, it collapses, and we decide that's not a good way to build it. There's the Tacoma Narrows Bridge, Galloping Gertie — you've all seen the vibration study where only the dog died. The casualty was the dog.
[Tom holds up the 1946 Maritime Commission report.] There are lots of other studies on how codes change. Next Monday we'll have class, which may be the last. One of my students got this out of the MIT Library when they were selling old books that no one had checked out in years: the 1946 report on design and methods of construction of welded steel merchant vessels. Report of Investigation, 15 July 1946. This is the Liberty ships. It has great photos in here, some of which you have not seen before.
There's the famous photo of the Schenectady sitting at dry dock. This is in many textbooks — a Liberty ship sitting docked, minding its own business, along comes a crack, and it splits in two. This is the Schenectady. The best one isn't this one, it's the SS Manhattan, which happened in the middle of the North Atlantic. I'd rather have it happen at the dock if you're going to split the vessel in two.
The statistics: 4,694 welded steel merchant vessels were built by the Maritime Commission in the United States and considered in this investigation. 970 of those vessels — out of 4,700 — suffered casualties involving fractures. That's a pretty high failure rate. 24 sustained a complete fracture of the strength deck. One vessel sustained a complete fracture of the bottom. Eight vessels were lost — four broke in two, four were abandoned after fracture occurred. 26 lives were lost.
After World War II, three places in the world decided to figure out why these ships failed. One was the Naval Research Laboratory, with William S. Pellini, chief of metallurgy. When he retired from NRL, he came to MIT ocean engineering department, where he wrote a pamphlet on guidelines for fracture-safe and fracture-reliable design of steel structures. Another was the British Welding Institute — the UK decided they needed something to look at why these things failed. The third place was here at MIT in the metallurgy department, with Professor Cohen and Averbach and others. I TA'd for Professor Cohen in the 1970s. He was a big-time metallurgist; he'd worked on the Manhattan Project during World War II.
Pellini and NRL came up with a new spec that all submarine steels had to meet — the explosion bulge. Take some big heavy plates, cut a hole about a foot or 16 inches in diameter, put a plate on top, and set off an explosive charge on the test plate to try to drive it through the hole. If it cracks in an acceptable way, the steel is good for building naval ships. [Tom shows test images.] Here's one that fractured like glass — not acceptable. These are increasing temperatures. At 160 degrees Fahrenheit you actually form a dome — explosively forming this dome before it cracks. Today, surface ships don't use the explosion bulge test, but the Nuclear Navy still does. Very expensive — probably $100,000 a pop to qualify your materials.
[Tom holds up Great Naval Disasters of the 20th Century.] Just like Petroski and the civil engineering failures — if any of you haven't seen this, Great Naval Disasters of the 20th Century, it's fun to read. Gives you two, three, or four pages on disasters going back to the Spanish-American War. What great naval disasters do you know of?
Student: Edmund Fitzgerald.
That's not really — it's maritime, it's civil. When it says naval disasters, it means U.S. Navy. The Edmund Fitzgerald was a coal carrier in the Great Lakes, and it failed. I can't remember exactly the reason — bad weather and other things. I don't know if it's a structural failure. You don't know any of the —
Student: Thresher?
Thresher. If you go back to the Spanish-American War, the Maine — explosives blew up. After every one of these there are new codes and standards. Just like after the Liberty ships they came up with the explosion bulge test, and other people picked these things up. So Petroski's whole theory is that we progress by failures. Something fails, we study it, and we do something else to keep it from happening again.
You don't know the Belknap disaster in the Navy? I probably have talked about it — maybe in the video. You've gotten that far in the video.
The Belknap, a destroyer, ran into the John F. Kennedy in one of his early operations. It hit right below the hangar deck or the elevators that take aircraft up from the hangar deck to the flight deck. Some of the jet fuel landed on the Belknap and started a fire in the aluminum superstructure, and the Belknap was toast. The joke in the Navy at the time was that a couple of gallons of aviation fuel would wipe out any capital ship in the fleet. This was not long after the British Sheffield, which got hit by an Exocet missile — it's a cruiser. Aluminum fire in the superstructure wiped out the whole ship.
Back in the mid-1980s at David Taylor Annapolis, which is closed now, they were doing a lot to replace aluminum superstructures with waffle steel, so it wouldn't catch fire. I've had several students do papers on the Thresher and the Belknap — not the Belknap, but the aluminum superstructure fires. On the web, people are still debating whether the aluminum caught fire on the Sheffield and the Belknap. But I guarantee NAVSEA thought it did, or they wouldn't have been spending $100 million figuring out how to make high-strength steel waffle members to get rid of aluminum superstructures.
Today, what are your littoral ships made out of? Aluminum. So I guess a couple of gallons of jet fuel will still wipe out any ship — you just don't want anyone to get that close to you with the jet fuel. If it's just a helicopter pad, those are small.
Dr. Belmar will be here tomorrow; I'll be here next Monday, and then the live lectures will be done. Kathleen, you've got to figure out whether you're doing the problem set or the presentation.