§1. Why so many steels? [00:02]
Today I wanted to talk about hardenability — I've mentioned it, but I haven't really discussed why we have this need for it. It goes back to a question a student asked me last year: why do you have so many steels? If you go through a book on steels, you'll find for the carbon steels — carbon, manganese, phosphorus, sulfur, silicon — and then these others: copper and lead can be up there in very small percentages. Copper is there for corrosion resistance, going back to the 1880s. People found that some steels with a little bit of copper had better corrosion resistance in the railroads. So the railroads specified a copper-bearing carbon steel with less than two-tenths of a percent copper, and got double the corrosion resistance in the atmosphere. Lead is sometimes put in steel for machinability — it forms little inclusions. Iron and lead don't mix, so you form these little lead inclusions, and when you're machining, the lead melts and gives you a liquid lubricant. Now we're getting away from lead in what we call free-machining steels.
For alloy steels you've got different specifications: carbon, manganese, sulfur, silicon, chrome, nickel, moly, tungsten, copper, vanadium, aluminum. So it gets more complex. Why do we have these alloy steels? In one simple sense, because we have lots of different product forms — forging, structural sections like I-beams, hot-rolled carbon steel plates, carbon steel bars. If the steel salesman comes to you, he's got an order book and he can look up the prices for each of these classes. There's a whole page of carbon steel types, and the alloy steels expand on it. You can have line pipe — that's like the Alaskan pipeline; oil-country tubular goods — like the pipe I brought in yesterday that had the defect in it, the one that leaks in 12 from Mexico; special-quality tubular goods. Lots of different steels.
It has to do with what part of the steel mill it comes out of. Steel mills are huge. Even now, after forty years, whenever I walk through a steel plant I'm still amazed by the magnitude of it. The last steel mill built on a green-field site in this country was built by Bethlehem Steel — world's second-largest steel company at the time — Burns Harbor, Indiana. Now I think it's owned by ArcelorMittal. But that was in 1965. Fifty years ago. Bethlehem Steel paid about five billion dollars in '65, which would be fifteen or twenty billion today. Since then, plenty of new steel mills have been built in the world, because countries will build a steel mill — it's so important for their economy. What was one of the first things the Japanese did after World War II? They built steel mills, because they needed to build ships. That was part of their tradition and their culture — they're surrounded by water. But since Burns Harbor, no company has built a steel mill. Only countries have invested in them, because the investment is so big. The only people who can afford twenty-billion-dollar investments in manufacturing capacity today are Intel and Boeing. A new state-of-the-art semiconductor fab is twenty billion. A new commercial 777 or 787 — twenty billion. It's a bet-your-company business; you'll go bankrupt if you invest in it.
Anybody know how the 747 came about as a commercial airliner? The Air Force had a competition for the C-5, and Lockheed won with the C-5A. The 747 was Boeing's design for military transport, and they lost. The government had paid for a large part of that development. And Boeing said, we've got to make some money out of this; we've invested too much of our own money plus government money. So they said, let's build a commercial jetliner. And everybody said, the world doesn't need one of those. I remember flying out of Tokyo once — we did a loop around the airport, and I counted forty-seven 747s on the ground. They're distinctive in shape. Forty-seven, just sitting there. And we weren't necessarily the first ones to take off; they take off in waves on the international flights. There could be fifty at Narita Airport. That's a lot of money. The 747 was the savior of Boeing through the 1970s and '80s — big cash cow.
§2. Hardenability and the Jominy test [06:38]
One of the reasons there are lots of different steels is because of two things: the depth of hardening you want if you're going to heat-treat it, and the cost of the alloying elements to do that. If I take a piece of carbon steel and quench it — remember that little thing I sent around with the file — this is 1021 steel, close to the 1018 I had, so the carbon content is .21 rather than .18. They take a bar like this and put it in a fixture — it's called the Jominy test — they have a shield and they spray a jet of water as hard as they can on one end. Then they cut the steel and measure the hardness with a hardness tester across it. This is the distance from the quenched end, and this is the hardness. In a 1021, you can get maybe Rockwell C 40 or 45, but it drops off. This is in sixteenths of an inch. So within a quarter of an inch you can only harden it about an eighth of an inch deep. Even that little half-inch square bar I sent around — you quench that in water and you won't get full-thickness hardness, because the transformation to martensite is just too rapid.
But we can throw some alloying elements in there and slow it down considerably. This is what we call the TTT diagram — temperature-time-transformation diagram for 1080 steel, eight-tenths of a percent carbon, on a log scale of time versus temperature. Up here we austenitize to form face-centered cubic, hold it there, then quench it down. The cooling curve comes down through here. This is one minute over here, here we're at one second. For this stuff to transform — this is a diffusionless transformation, athermal, it doesn't depend on a specific temperature, it's a thermodynamic thing. The atoms literally shear and transform the crystal structure, and they do it at close to the speed of sound once you get a little nucleus formed. Morris Cohen spent thirty years of his life looking for the martensite nucleus in a microscope — never found it. About one chance in five million. He had a theoretical model of how the dislocation should form, a pseudo crystal structure of body-centered cubic even though it was face-centered cubic, that dislocations could put the atoms in a favorable orientation. He spent half his career looking for a martensite nucleus, and never found it.
Student: If you knew how it forms, you could prevent it from forming, or get much more hardness?
No, he was just a scientist, he was interested. Other people win the Nobel Prize for doing something useful — he just did it for pride.
If I go to a 5140 — 51 is chrome, as I remember — by adding some chrome it slows things down, because chrome diffuses more slowly than just carbon and manganese. If I go to a 1034, a third of a percent carbon, I've got less than a second. If I go to a 9261, I can get ten seconds. I can do a lot more than that depending on the alloy content, but alloys cost money. Nickel: six dollars and fifty cents a pound as of last year.
Student: When you say you have ten seconds — ten seconds to get all martensite without it transforming to the soft phase?
Martensite forms when you quench. How quickly do you have to quench? If you slow-cool it, it'll just go to BCC, it'll be nice soft steel, just like your bridge steel.
Student: And this is forming body —
Body-centered tetragonal. Trapping that carbon, keeping the carbon from diffusing. You form martensite if you trap it in those sites. Remember the little ping-pong balls inside the basketball? That was hydrogen, but you can think of the same thing with carbon trapped in its interstitial sites. When it transforms it forms body-centered tetragonal, and if you can slow down the carbon diffusion, you can get martensite without cooling so quickly.
We did make alloy steels for Henry Ford and Alfred Sloan at General Motors in the 1920s and 1930s, but we didn't have big plate applications until after World War II. One of the reasons is, we didn't know we needed good toughness until after all the Liberty ship failures.
Student: Does martensite form in any kind of metal?
In steel, in some titanium, in some copper alloys, but in general 99.9% of the martensite you'll ever be interested in is steel. Other crystal structures don't have this ability. There are other martensites in other things. A martensite is now defined as just an example of a shear transformation of the metal. The atoms slide past each other due to shear stress, internal shear.
§3. The economics of alloy content [13:53]
If we start talking about nickel at six-fifty a pound — if I put in 1% nickel, that's twenty pounds of nickel for a ton, $130 for 1%. How much nickel is in HY-80? About 3%. You're talking $400. You just doubled the price of steel by putting nickel in there, and you haven't started to account for the chrome or moly or vanadium. Chrome — we actually use ferrochrome, which is cheaper than pure chrome because we already have iron in there. Adding 1% chrome is only $2. Moly, which is in HY-80 or HY-100 at four-tenths of a percent — it's $180 for 1%, so $90 to add the moly. You end up taking a $400 steel and turning it into thousand-dollar steel by adding alloy elements. But now, instead of a sixteenth or an eighth of an inch depth of hardening, I can harden a two-inch or four-inch piece. HY-80 is regularly hardened to four inches thick. Probably not up to six, but four inches is mass.
Student: What are you making that's even —
I'll give you an example. They were using HY-80 — this is Jerry Milgram, who was a professor in 13, which is now Course 2N. He was designing America's Cup yachts. He called me up — this is twenty years ago — and they had a problem. They designed the keel, not keel, but the — what do you call it, the dorsal fin upside down —
Student: The center board?
The center board, thank you. They were designing the center board. It was six inches thick, HY-80, or it may have been HY-130, I don't remember — it was an HY steel. And when they went on the first sea trials, the center board bent. A lot of force. This was five or six feet long and six inches thick, and had a big lever arm. So he called me. That's when I first realized there were a couple of other faculty who were here at 6:00 a.m. — he called me at 6 a.m. one morning and offered to hire me as a consultant. They were going to go to 4340. 4340 has lots of nickel, chrome, moly, and vanadium. John Lippold in his chapter on hydrogen cracking has the same plots — for a 1046, this is martensite high hardness, over here is bainite plus pearlite when you cool slow. And he's got the same Jominy quench hardenability for 4340. The hardness is nearly 100% martensite full thickness. You extrapolate this out, you can go to eight or ten inches thick. So to your question, if you want to spend enough money and put enough alloy content in, you can harden pretty thick.
There are some hardenable stainless steels that are 30–40% alloy content. You don't even have to quench them — we call them air-hardening. If you want to make them soft, you have to furnace-cool them over several days, because you just take a big forging of stainless steel out of a furnace and depending on its composition, it just cooling naturally in the air, it ends up quenched but not tempered. So you can alloy things a lot. When I said steel was $400, that's not really plate, where we have higher yield. That's sheet metal for automobiles, or for little Red Flyer wagons your kids pull around — the cheapest steel you can get. Rebar, $400 a ton right now. An alloy steel to make bolts for a commercial application might be $1,000 a ton. HY-80 or HY-100, I'm probably talking $2,000, maybe $2,500 a ton today, might be $3,000. Price has been going up in recent years, but it's not cheap.
§4. Fabrication cost and the Nimitz weight problem [19:52]
Then add the fact that all your cutting — you're not going to use everything in nice rectangular sheets, you don't get 100% yield in the shipyard. You've got a lot of scrap left over after you cut paper dolls out and weld them together. The actual cost per pound fabricated is about ten times the cost of the material you started with. The Navy did a million-dollar study down at Newport News. They were looking to reduce the weight of aircraft carriers. The Nimitz-class carriers have been around for fifty years. Why don't we have a new hull design? Because it would cost $50 million — this was twenty years ago — to design a new hull before you even build it. Just to design the shape of a new carrier is a $50 million bill. All the towing-tank stuff, the things you guys come back to learn how to do — they want you to do it for fifty cents. You'll have a summer over at Draper where you get to design a ship. It probably won't be a Nimitz replacement, but anyway.
The Navy made a decision about fifteen years ago. I'm sure they could come up with better hull designs now than they did in 1960 when they came up with the Nimitz class. But they decided they didn't have it in the budget to spend $50 million on just a new design. The problem is, the Nimitz carriers, if you look at the fifty-year history, gain 250 tons a year. I gain a little weight each year, so the carriers do too. And most of it is not down the bottom, it's up top.
Student: Combat systems?
Student: A lot of the combat systems — when they're not in the major RCOH overhauls, like the intermediate ones, if they change out a combat system, they'll install the new one, and they may remove all the components of the old one, but not usually. There's all the extra weight of different little bits and pieces, plus the extra wiring and cabling. They don't want to take it out just in case something doesn't work well. And whenever it gets to its mid-life, they go through and gut it out just to remove all that junk.
If you plot the carriers we've built — about one a year for many years, maybe one every two years now — it's 250 pounds a year on average. That's a lot of weight.
36 ksi is the garden-variety steel for guardrails on highways, Radio Flyer wagons, you name it. Through the normalization heat treatment I talked about the other day — they were using at General Dynamics Quincy — where they put it in a furnace and slow-cool it rather than quench, you can get a finer grain size, and that gets you up to 50 ksi. There are marine grades called EH-32 or DH-32 — the 32 is kilograms per square millimeter, which is 50 ksi. If you go to the structural welding code for bridges now — or in 1985 — you'd find A36 charts and 50 ksi charts. It's not martensite. If you want martensite, you can get 100 ksi with martensitic plate, or 80 ksi in HY-80. But without going to martensite, you could get 50 ksi. The EH steels and DH steels they were using after World War II — I think the Nautilus was one of the first HY-80 subs, but they were using 50 ksi steel before that.
The point is, you can go to higher strengths, but it's going to cost you money. They wanted to go to 65 ksi with high-strength low-alloy steel technology, where you cool the steel as it comes off the rolling mill, red hot. You can't quench it in water, but if you spray it — more than a mist, like a garden hose sprayer, uniformly — you can cool it more rapidly. You can keep the grains from growing as big while it's red hot, and you can get 65 ksi. That's high-strength low-alloy steel. You don't have to put as much alloy in there because you're using fine grain size — could be a hundred times finer grain size. This is what I went over to Japan to see if I could learn from the steel companies. Japan and the Navy wanted to invest in this.
§5. Grain size, single crystals, and HSLA-65 [25:30]
Student: When you say grain, it's like a salt crystal, a collection of —
Similarly facing metal crystals — well, they're not similarly facing, they're random.
Student: What I'm asking is small grain size versus large grain size.
These are clusters of atoms that are all lined up. They're impinging on other grains in the structure. It's the difference between a snow cone and a slurpee. Snow cones have big grains, slurpees — crushed ice — have fine grains. Slurpees stronger, right?
Student: Yes, so basically chopping them while they're small?
If you let it cool on its own, a big heavy plate, you're going to get what we call an ASTM grain size of four — larger numbers are smaller grain sizes, going as the square of the number. If you do accelerated cooling, you can get grain sizes that are ASTM 10. So at ASTM 1 at 100x you can have a grain that's that big; at ASTM 4 you'd have the grain size of a period; at ASTM 10 on a 100x magnification, a grain size of a period at the end of a sentence. Huge differences in grain size.
The only thing you can do to steel or any metal to get both strength and toughness, which is what we want, is fine grain size. You add alloying elements, you're hurting your toughness, but you're getting greater hardenability. Why do we do it? Because if we quench to martensite, the effective grain size in martensite is so small you can't even see it on an optical micrograph at 100x. You have to go to electron microscope.
Student: A lot of small companies are working on crystal technologies, where they grow one crystal that is one single grain — the point being it's stronger and has special properties for electronics.
For electronics, and for high temperature in engines.
Student: Is it a stupid question to say, what about trying to create that within the metal? What prevents you from creating a single grain crystal in any arbitrary large size of metal?
We do it. I'll bring in my single-crystal turbine blade if I can remember. We make single-crystal turbine blades if you want creep resistance at high temperature, so it doesn't flow like Silly Putty does at room temperature. If you don't want it changing shape at high temperature, you want a single crystal. Fine grain size at elevated temperatures is the kiss of death. They make single crystals in silicon because grain boundaries destroy the electronic properties. I just bought some outdoor solar-powered lights for my house, and you can see the solar cells have grains that are an inch across. You can see them just looking at it. It's like a piece of galvanized steel bucket — you see the huge grains. Wherever you have a grain boundary, you're losing electrons, and you're not getting the solar efficiency. So you'd love single-crystal silicon for solar cells, but that's too expensive in general.
Student: It's cost-prohibitive to make large-volume single-grain crystal metal?
To make single-crystal metal, we grow it in a furnace at about a millimeter an hour, maybe two. It's slow. But when you're growing a turbine blade for $7,000, and you're growing twenty-four of them at a time — Rolls-Royce grows two at a time, but they have lots of small furnaces, whereas Pratt & Whitney and General Electric use two or three dozen on one sprue. It's like a Christmas tree with turbine blades on the end, you just break them all off.
So I never did finish the story about HSLA-65. They wanted to go from their 50 ksi — there's a lot of 50 ksi steel on a carrier; you don't need 100 ksi everywhere. We've been using HY-100 on flight decks, which are about four inches thick. That's not classified — the submarine hull is classified, but the flight deck is not. We've been using HY-100 for thirty, forty years, as long as I know — it goes back to the original class of ships, because you need the extra strength for big heavy things on broad spaces on the flight deck. But a lot of the rest of the ship is 50 ksi material. Could they save weight by using HSLA-65? In 1996, NAVSEA gave a million-dollar contract to Newport News to figure out how much money and weight they could save. They found they could save 2,000 tons. That's eight years of carrier weight gain without changing the hull. But it would cost ten dollars a pound. At that time HSLA-65 was selling for $2,000 a ton.
I could have told them that off the top of my head. I have a rule of thumb that the fabricated cost of a piece of metal — steel, aluminum, titanium — is ten times the cost of buying the material. I gave you that little Defense Department 21st century, that number is in there. I've been using that even before they gave the million-dollar grant. If NAVSEA had given me the million-dollar grant, I could have written them the report the next day and cashed the check. I always liked that report, because I read it and thought, I've been saying this for years, and that's exactly what they found with their big study.
§6. Stout and Doty and the welding procedure look-up [33:00]
With high hardenability, lots of alloying elements, lots of money, you can get all kinds of hardenability — more than we really need. You can make something twenty inches thick with high hardenability. So we have lots of steels, but with all these steels we have all kinds of problems welding them when they get thicker. When Professor Milgram said, we want to use 4340 — that very high hardenability one I showed you — and asked, can we weld it? I said, if you've got enough money, you can weld almost anything. And they did. I helped them. But you have to preheat it — not to 400 degrees, this oven you can cook pizzas in, this is a 600-degree preheat for 4340. That's not a pleasant thing to work around. You don't necessarily wear blue jelly suits, but you do wear radiation protection when you're standing next to that, and you have fans blowing air through your suit. Not pleasant. Can be done. Needs lots of quality control to make sure they did it right.
This is out of a book you'll get when we do stainless steels, on weldability of stainless steels, but it's useful here. It looks at the diffusion coefficient in centimeters squared per second in alpha-iron and gamma-iron — body-centered cubic, face-centered cubic. I pulled this out for hydrogen at room temperature, 20 degrees C, or 1100 degrees where we stress-relieve. Hydrogen diffusivity: 10⁻⁵ or 10⁻³, doesn't change much. But look at the diffusivity of carbon — it's actually a fast diffuser, but at room temperature it's 10⁻¹⁷. If you look down here at tungsten, which we don't use, you have 10⁻⁶⁰. Hydrogen diffuses through steel just like water. I ought to bring in a demonstration.
In gamma-iron, hydrogen is 10⁻¹⁰ rather than 10⁻⁵. Hydrogen diffuses slower in gamma-iron than in alpha, but its solubility is ten times as great. At high temperatures it's got the same diffusivity, but it's slow at lower temperatures. Carbon: 10⁻¹³ versus 10⁻²² or 10⁻¹⁵. Carbon diffuses fast through iron. I can slow it down by putting in things like manganese, which has an affinity for carbon, or chromium, which has an affinity for carbon — so the carbon atoms aren't as free to move. They like to stay near their neighbors; they like chromium atoms.
That might sound fatuous, but a lot of the computational metallurgy people do today is getting down to the quantum level, and it's not much different than how many nearest neighbors do you have that are chromium or nickel, and how does that influence the vibrational jump frequency of the atom going to the next site. You can start to predict the rate of phase transformations. And that's all this is, a phase transformation.
If I go to good old Stout and Doty, which came out — the first edition I think was in the 1940s. Bob Stout was the professor at Lehigh, which was the big welding school at the time. In the back of Stout and Doty there are about twenty pages. Starts out with steel number one, which is ASTM A27, steel castings. Steel number two is ASTM A36, structural steel — the garden-variety we make bridges out of. It goes all the way up to steel number 200. A lot of steels. They all have different weldability. When people would call me up over the last twenty-five years and say, Tom, we've got some steel, we need a welding procedure — all I would do is pull Stout and Doty off the shelf, go to the index, find the steel, and depending on the carbon content and the thickness, it tells me minimum preheat and interpass temperature. This is how much you have to heat up the steel so the hydrogen will diffuse out while you're welding it.
Remember the little video we did, a finite element analysis on the hydrogen in two different thicknesses of steel, quarter-inch and inch, preheated and non-preheated, and how fast the hydrogen diffuses through when you weld on it? You go back and watch that — you'll find this is basically a 1940s table version of what to do. So I'd look up the steel, ask what's the thickness and the chemical composition. There's a firm out here that does a lot of architectural work. They were working on the Charles Street MBTA station, built in 1910. Didn't even have steel specifications back then for some of the steels they were using, but they had to weld to them when rebuilding. For 500 bucks, I would write them a welding procedure if they told me the chemical composition and the thickness. I would go to this book — I even told them this much, but they never wanted to learn how to do it. They wanted to pay me the 500 bucks. It took me five minutes. $500 for five minutes' work. It tells me what preheat should be, if you need a postweld heat treatment, and whether you need to peen it — hammer the surface to relieve residual stresses. Postweld heat treatment would be a stress-relief heat treatment. Lots of different welding procedures, right there in Stout and Doty.
We'll look next week at one of the welding codes that sort of superseded this, and there are other ways to calculate all that. So now you know why we have lots of different steels and lots of different alloys. When you get into the railroad business or the automotive business, they'll sell their souls to give you a steel that's $20 less per ton, because they're making a million tons of that steel. $20 a ton is worth $20 million. So they'll have a special steel optimized just for that thickness and that hardness, because you're talking millions of dollars for tenths of a percent of alloy content. That's why you've got hundreds of steels. When we get to nickel alloys, you'll find you can put all the nickel alloys on two pages, whereas the steel takes ten pages, and aluminum alloys take three pages. It's just how much of these materials we use and how specialized it is.
§7. The Seawolf and the high-side wire [42:01]
So let's go back to what happened with the Seawolf. They were welding HY-100 in the Seawolf. It was the first full-scale submarine with HY-100. They'd been welding HY-80 for forty years, successfully for thirty of those forty years. The first ten years were a little hairy. Whenever they switch to a new steel, they have all kinds of problems. They'd built two modules. Submarines are built in modules. Say it's thirty feet in diameter and a module is thirty or forty feet long. They can lift these whole modules — they might weigh five hundred or a thousand tons. They have big trains to move them. Building in modules speeds up production and saves money. They had built modules and put HY-100 modules into HY-80 hulls because they wanted the experience. My first student ever, John Galligan, was working at Electric Boat, and he was assigned the responsibility to build the first HY-100 module hulls — not for the Seawolf, but for the couple of ships that preceded it.
There's always a range of alloy content. The weld metal is modified to optimize for the weld zone, and you buy the wire from someone like Lincoln Electric, the world's largest manufacturer of welding electrodes and wire. When it's gas-metal arc, it's just a bare wire, and you use argon shielding. NAVSEA had a spec, and it turns out the wire they got was what they called high-side wire. It had higher hardenability because of higher alloy content. It met spec, but everything was on the high side. Chromium on the high side of the range, nickel on the high side, everything on the high side. Which meant that even though it was supposed to be HY-100 weld metal for HY-100 — shooting for 120 or 130 ksi yield, because you want slightly overmatching weld metal rather than undermatching, for fracture reasons — they actually got something more like 130 or 140 ksi. This would have been good for HY-130 steel. And that meant you had to hold your hydrogen to much tighter tolerances. Typically in a shipyard you can get about four parts per million deposited weld metal hydrogen.
I'll put up this little table, which you will not find in a book — this is a Tom Eagar table, but it's approximately correct. At full restraint, which when you build these eggshell constructions is pretty restrained — that metal is not moving anywhere — if you've got a 70 ksi tensile strength steel, like the 50 ksi submarine steel of the 1950s, you can tolerate 30 parts per million hydrogen. That means you can weld with a cellulosic electrode and not get cracking. At 120 ksi, you could probably tolerate ten parts per million. So they had some leeway with HY-80, which is actually about 100, 105 ksi in the weld metal — they could tolerate a fair amount, and they had learned to get down to four parts per million consistently. That was one of the reasons they were encouraged to go to HY-100.
Well, it turns out they were really at about 140 — this circle was getting bigger. Remember those three circles? Microstructure, stress, and hydrogen. You want to keep all of these as small as possible. Unfortunately, this martensitic structure is the most susceptible to hydrogen. I'm not sure anyone really knows exactly why martensite's the worst, but we've done plenty of studies — we have the worst type of structure, but it gives us some of the best toughness and some of the highest strengths. When we have higher strength, we have higher stress, because it goes up to the yield stress when you start welding and all the residual stress. What happened to the hydrogen? They actually measured the hydrogen content when I was brought in. I don't remember the exact numbers, and they told us, don't tell anyone — not because it was so classified, but because it was embarrassing they had this problem. As I remember, they had numbers in the range of five to fifteen parts per million in welds they'd taken out as known samples from the Seawolf. You look at this little table — even at HY-180 I can only tolerate two or three. I can't weld that stuff even with my best welding processes in the shipyard. I'm going to be at too high a hydrogen level; I'm going to have to use higher preheats to drive the hydrogen out afterwards.
They didn't have a lot of data. They brought us in early on. There were six of us — two individuals and four companies. One of the individuals, who had been head of welding at one of the big electrode manufacturers, got bounced early on because he spoke to the press. So there were five of us left. I looked at the data — they had three different wires. The wire diameters are .045, which is standard, sixteenth (.063), and 3/32 (.093). Standard welding electrode sizes. I plotted it, and on a very crude average, I would see higher hydrogen the smaller the wire. So I concluded it was drawing lubricant. Why? If it were inherent hydrogen in the welding process, the same argon in the shipyard no matter what the wire size — what's different about these? The smaller wires have a higher surface-to-volume ratio; it goes as 1/R. So I get more surface per pound of weld metal with a smaller-diameter wire, and I saw higher hydrogen. My report, which was not classified but we weren't supposed to tell the press — and now it's almost twenty-five years later, so who cares — concluded it was drawing lubricant on the wire.
Drawing lubricant — when you draw the wire, you have to use a grease.
Student: From the school?
In the mill. And then you're supposed to clean it off. Really good wire today — not necessarily back then — they actually skive the wire. They draw the wire with the lubricant, and then for extra cost they put it through another wire-drawing die that machines off the surface. They call it skiving. They clean off the top thousandth or two, to make sure no drawing lubricant got down in little crevices and got rolled into the surface. Lincoln Electric didn't like this theory, for obvious reasons.
§8. Fracture mechanics and what to tell the captains [51:30]
Another question came up at the second meeting in Crystal City. They had two captains. They said, these two captains want to talk to you, because they'd read the reports. It was a big room, fifty people, and I went over in a corner. One captain was responsible for operating all subs at sea; the other was responsible for maintenance of all the subs. They wanted to know what they had to do about the two ships that had been built with these modules of HY-100 in the hull. Did they have to bring them into port and replace the modules, or derate them — you can't go as deep, can't go as fast — or what? I said, the good news is you don't have to do either, but when they do come in, you have to do more inspection. I told them that because of fracture mechanics. They didn't prep me for this; I did it off the wily.
If I look at the fundamental equation of fracture mechanics — the toughness must be greater than the stress times the square root of pi times the crack length. We knew the crack length of these things, they were all less than an eighth of an inch. I told you about how they were looking for eighth-of-an-inch flaws by magnetic particle inspection — these were like a sixteenth of an inch or less, half that size. You don't have to do either one, because the stress is whatever the stress is. Let's say the stress is half the yield stress. You've got to have a safety factor — let's say they designed for 50 ksi, it's got to be something like that, or you'll run into fatigue problems.
Student: Are you talking about the induced stress of the metal due to joining, or the stress induced by the pressure?
The designer doesn't even know how to weld or that welds exist. What average stress do I have in that structure? This is stress.
Student: What stress?
It's the design stress. This is when the guy first pulls out a sheet of paper. They're getting some consultant under NAVSEA's guidance — okay, we've got 100 ksi material, we're going to design this ship for 50 ksi operating stress as the maximum it should ever see in service. The actual stress it might see may only be 30 ksi, but nowhere in the structure will have more than 50 ksi, because if it does, we'll run into fatigue problems. You need a factor of two safety to avoid fatigue. Commercial, military, doesn't matter, you need a factor of two.
So if I take 50 ksi as my maximum stress, and the toughness of HY-100 should be on the order of — well, you have no reason to know this — about 50 ksi square root of inch. It has unusual units because it has to have the same units on both sides of the equation. I can do this in my head. I did it in my head for them, I didn't have to pull out a calculator. Stress squared is equal to toughness squared over pi C — the critical flaw size comes out to over an inch. There's no way these cracks are going to grow in a couple years before the ship comes back for some maintenance. Not a major overhaul, but routine maintenance. A sixteenth-of-an-inch crack is not going to grow to one-inch size and cause the ship to fail in that period.
They don't have to derate it, but when it comes in, you've got to select 10, 20% of those welds on the hull, and do a very good inspection — better than usual — to make sure you know how big those cracks have gotten. If they're at 3/8 of an inch at that maintenance, you may have to do some repairs. But the odds are, with sixteenth-of-an-inch flaws, you do the extra inspection, and those ships can stay out there until you're ready to scrap them, and you'll never have to do anything. But you don't know for sure. You certainly weren't going to send out their number one ship when you knew it was full of cracks — if anything ever did happen, they would probably fire the entire U.S. Navy at that point. Some heads have got to roll. Not senators in Congress's head — they always except themselves.
That's what happened in Seawolf, and that's where fracture mechanics can tell you what to do. Doing extra inspection is what you need to do.
§9. Aloha Airlines and the fatigue-leader problem [57:09]
To give you a similar example — anybody remember Aloha Airlines? Got to do something for Adrian, the aerospace industry, because other people during the school year could be watching this and they might be in the automotive business. Aloha Airlines, the one with the window —
Student: Crack?
What you're thinking of is the Comet, which was the 1950s. They had square windows on this British-designed aircraft, flying across the Atlantic, and they would just disappear in the Atlantic, because they were getting fatigue cracks from stress concentration in the corner. You will not see square windows in airplanes since then. But this — the top of the aircraft, more than 180 degrees, top half — section just blew off. The only person who died was one stewardess who wasn't belted. Everyone belted in survived. 40,000 feet, the top blows off. I was impressed the whole thing didn't break in two. If you Google Aloha Airlines, there'll be a picture of it sitting on the ground, and over half the top is gone.
This was in the early 1990s. There are more fatigue pressure cycles in airlines in Hawaii than anywhere else on the 737, because no flight is more than forty minutes. The islands are all close to each other. Aloha flew between the islands, and they got 40,000 cycles in like the first 30,000 hours of life. Typically a Boeing aircraft is designed for 100,000 hours of service. You figure out how many fatigue cycles you'll get. They hadn't accounted for someone just island-hopping in Hawaii, so they had accumulated more cycles. Boeing had requirements for inspection after so many hours, increasing inspection requirements after so many hours, but it was for the fleet on average. They were taking averages for the whole world of 737s, not for Hawaii.
[Tom puts the image up on the projector.] See, there it is — everybody standing there on top of the aircraft. They're in the middle of the aircraft. Pretty scary. To think — flying along, and the roof flies off like in a tornado.
Aloha Airlines was the fatigue leader. Around the rivet holes, there was no fracture that Boeing didn't know about. Boeing has some of the top fracture people in the world. The guy who taught me fatigue and fracture was Professor Reggie Pelloux, who got his ScD here, then went to Boeing, and came back as a faculty member and specialized in fatigue and fracture. Boeing has top-notch people, and they pay attention to safety. This was one where they got broadsided. Only one person died, but it was a wake-up call for how to do your averaging of cycles. You have to look at the outlier — Aloha Airlines. The critical flaw sizes for the sheet metal skin are like several feet — they had big cracks. But nobody was looking for them because nobody expected them. They were the outlier leader.
I do similar work for Cape Air. They fly Cessnas, and they put more hours on Cessnas than anybody else. Their head of maintenance, whenever he saw a crack or a problem he had never seen before, would call me up, send me the part, and say, tell us what happened. I'd look at it metallurgically, look at the fatigue, and give my assessment. I'd write a one-page letter, sometimes a page and a half. Over the years I learned that what he would do is send it to Cessna, because Cessna needed to know what was going to happen to all their other aircraft. Cape Air was getting more fatigue cycles than anybody else. They were the market leader in cracks. They would also send a copy to the Federal Aviation Administration, which for that particular type is in Burlington, Massachusetts, right across from Burlington Mall. I never had to get into some sort of fight, and I never got direct feedback from Cessna or the FAA, but I know from other things that happened that they were reading my reports. I was just doing a failure analysis of why it occurred — was it a stress concentration, was it higher stresses than expected. In most cases it was just really old, and things wear out after time. And that's basically what happened at Aloha Airlines.
One day we will finish steels — but a lot of this stuff does apply to things other than steel, and we're almost done with hydrogen. If I hadn't told you a few stories, we could have probably finished hydrogen in steels today. But we'll start to go faster through the other metals, because you are learning some metallurgical principles that apply to more than just steels. It's just easier to tell the story with steels when I'm talking to a bunch of people who make ships out of steel.