§1. The Five Whys and the limits of metallurgy [00:05]
In Japan during the 1980s, when they were learning to do total quality management, they came up with something they called the Five Whys. The idea is, if you ask why five times — there's nothing magic about five — you should get to the root cause of a problem.
Take the Amtrak derailment. Why did the train outside of Philadelphia a couple weeks ago derail? Too fast. It was going too fast into a curve. Why was he going too fast? I was just reading the paper this morning, and on page two they said he wasn't texting on a cell phone. He didn't test positive for drugs. Originally there was a question of whether someone had dropped a rock from an overpass, which happens all the time. Teenagers go to overpasses and drop rocks on top of speeding cars. They kill people. It's not a smart thing to do, and in busy areas they build big fences so the teenagers have to throw overhead.
But if you want to get to the root cause of a problem, you start asking why five times, and each question begets the next. I was going to do that the first day. Why are you taking this course? Because the Navy says so. Why does the Navy say so? Because they've had welding problems and they want you to know about steel. Why do they want a welding metallurgy course? Because most people think — and I have a quote in here — that all welding problems can be solved with metallurgy. No guys, that's not right. I'll give you some examples of that later today.
Why do people think they can solve all welding problems with metallurgy? Because historically, education of welders in this country started big-time at Lehigh University in the 1940s. Bob Stout was dean of the graduate school — Stout and Doty wrote Weldability of Steels in the 1940s. That was a big place doing welding research. Some of those guys went out of Lehigh to US Steel and they developed all the HY steels in the 50s and 60s with Navy money — those high-strength submarine steels — and they had to be weldable because they were going to make submarines out of it. But Lehigh didn't keep their faculty current in that area. They had Al Pense, who's still around, but he rose to become provost of the university. Once you get to the administrative levels, you stop minding the store at the lower levels.
And so what happened is up at Rensselaer Polytechnic, a guy came along named Warren Savage. They called him Doc Savage, after the comic books. Doc was a wonderful person, a bit of an alcoholic, and his students loved him. John Lippold was one of his last students. Doc turned out all kinds of welding metallurgists, and because of this legacy of RPI and Lehigh, everybody thought you solve every problem in the world by metallurgy. My quote is from Maslow: if the only tool you have is a hammer, you see every problem as a nail. So you have to be a little bit broader. If I start asking some of these Five Whys, it's interesting where it leads you. I've gotten up to seven on some of the whys to get to what I thought was the real major thing.
§2. Hydrogen as proton: the soda-can demonstration [05:54]
We're going to start with an experiment today. This is going to get us into hydrogen diffusion. The rate at which hydrogen diffuses out of steel is similar to the rate at which carbon dioxide diffuses out of soda. If I make a weld and there's moisture in the air or cellulose in the electrode, it will be releasing all kinds of hydrogen. Later, second hour, we'll watch a video showing bubbles coming out of steel. [Tom opens a soda container.] You can see the rapid release of bubbles, which would be similar to hydrogen, although this is CO2.
Student: Can the hydrogen that does embrittlement be H2?
The answer is no. If I have moisture in the air, we know it doesn't embrittle steel — every piece of steel in the world has humidity around it. H2O doesn't dissolve into steel. And even as molecular hydrogen H2, it doesn't go into steel. If you go to lab supplies and order a 2,000 PSI tank of hydrogen, it comes in a steel container. You wouldn't put it in there if it were going to ruin it — you'd have a bomb.
What's really bad is atomic hydrogen. Atomic hydrogen is basically just a proton and an electron, and it can dissolve into steel. What's even worse is if you ionize it — it's just a proton. Protons can go through there. What does a welding arc do? It turns it into protons. You've got a plasma with a bunch of protons in there. And there's almost nothing we can do about it because these little protons are small. I've tried to use a broom to pick them up, but it just doesn't work.
§3. Hardness vs. hardenability [08:45]
We had talked about hardness and hardenability. I'm going to give you more precise definitions. Understanding steels, which are 95% of all metals — this is supposed to be a metallurgy course. Some of these principles apply to other metals like aluminum and titanium, and we'll get to those. So even though it seems like I'm spending a lot of time on steel, it's only because it's true. I am.
Hardness — right out of a metallurgical dictionary — is a measure of the resistance of a material to surface indentation or abrasion. It may be thought of as a function of the stress. Hardness is actually three times the stress in terms of kilograms per square millimeter, or psi, whatever units you want to use, required to produce some specified type of plastic deformation.
Student: Is that specified type explicitly the top of the curve?
No, it can be the yield point. It can be top of the curve. Depends on the type of correlation, because hardness is just an empirical thing.
Student: That in itself seems like — that explains why you have a difference between 2.8 and 3.2 times the stress.
Exactly. Are you on the top of the curve or somewhere at the end point? And there's scatter in all this. Not every piece of material is exactly the same.
There's no absolute scale for hardness. It's an empirical scale, and each type of test is its own scale. Brinell — that's the one-inch-diameter hardened steel ball, piece of ball bearing. Rockwell is just the name of the company that developed the test that uses either 60 kg, 100 kg, or 150 kg as the load. It can use a diamond indenter, a sixteenth-of-an-inch steel ball, or an eighth-of-an-inch steel ball. Each one is a different scale — A, B, C, D, E, all the way up to M, depending on the load and type of indenter. Vickers started with a different type of diamond cut. Diamond is complex; you can cut it on different planes. Vickers uses a different cut than Knoop. Knoop and Vickers both use diamond. Rockwell uses hardened steel balls; the other two only use diamonds, but different cuts to make different shapes of indentation.
Scleroscope — you take a hardened steel ball, drop it, and measure the rebound. You don't need to do any indentation. The problem is the value you get depends on the stiffness of the thing you're measuring. If you drop it onto a trampoline, it won't bounce very high. So Scleroscope is useful if you need something portable, but it's got problems. So that's hardness — a measure of the resistance to surface indentation.
Student: Why isn't the definition more pinned down? If I were in charge, I would say, look, it's limited to the peak of the elastic region on the stress-strain curve. Done. Then everyone can agree.
Metallurgists just have to have something to argue about. Each measure could be a better measure of what you're trying to get at. This is an indirect measure of what you're really interested in — the whole stress-strain curve.
Dr. Belmar, who gives some of these lectures during the school year, has started a company based on his doctoral thesis at MIT, where he was doing fundamental studies of friction rubbing of metal against metal. He did a PhD on scratching metal. Pretty exciting — if you read his thesis, you'd understand. He's a very practical person, and he's developing a test where you scratch a piece of metal and measure the groove: the depth of the scratch and how the metal rolls up along the side like a plow. You measure all that instead of getting one number, which is the hardness indentation depth. He gets the depth of the groove, the force it takes to plow the groove, and the height of the metal rolled up on the side. His intent — and he's done it theoretically, and now demonstrated experimentally — is to get the whole stress-strain curve by doing scratch testing with an instrumented scratch machine.
He's working with people down in Houston, because the Pipeline Safety Administration of the Department of Transportation has now told the pipeline companies: you've got 45,000 miles of pipe buried in the ground for distribution of oil and gas in this country. Much of it was put in in the 1940s and 50s. We want you to prove the properties of this stuff that's in the ground, because they've had some ruptures, and people don't like 50,000 gallons of crude oil in their backyard or in their lake.
You can go to Instron, which sells mechanical test equipment. For 25 years they've had fairly good instrumented Rockwell testers. In fact, Instron purchased the Rockwell company. They own it now. Instead of selling you a Rockwell tester for $10,000 — and I have one in the lab if you want to see it; apparently it's the last one at MIT. There used to be a dozen in this department. Everybody's doing nano now. We have four nano hardness testers on the infinite corridor worth a couple million dollars, but only one remaining 30-year-old Rockwell tester. They do have instrumented hardness testers that measure the force as the indentation is being made, and you can get more information out of it, but you don't get the full stress-strain.
For non-destructive testing, a lot of people would like that extra information. Early in my career I came up with some new technique to inspect some metal. I was working fairly closely with David Taylor — at Annapolis, now Taylor Carderock, the Navy's research laboratory — and I was working with people like Electric Boat. I came down, talked to Electric Boat, and said, "I've got a new way to test something." They wanted to throw me out of the shipyard. The last thing they wanted was another test they had to meet. There's a lot of truth to that. We need to spend more time coming up with good tests rather than more tests. You can test things forever and never get anything built.
§4. The critical mass of management [17:03]
This breaks the train of thought on steel, but it's a good story. Back in the early days of the LGO [Leaders for Global Operations] program, 25 years ago, I went down to United Technologies to meet with them. There was a guy who was vice president of manufacturing, and he had been vice president of research at United Technologies. Pratt and Whitney, United Technologies — they were a big research-driven company. But he had switched from research to vice president of operations because, as he said, if this company is going to survive, it's not going to be on research. It's going to be on making things better. He decided he was going to go where the real problems of the company were.
And he told the story about the critical mass of management. Hitler had an underground plant that made ball bearings. [Tom draws on the board.] Here's the factory, here's a tunnel, and here are the offices — the management.
The colonel running this factory for ball bearings had his staff in the office, and they would go through the tunnel into the factory. They were a very productive plant. Well, Hitler had a burned-out general from the Russian front, so he decided to send the general to help the colonel manage the factory. There was a problem, because the general staff that he brought with him were always going around the colonel's staff. So the colonel decided to build a new office for the general staff.
Then there was another burned-out general, so he sent him too, and the colonel, having learned his lesson, built another management office over here. Now his staff could manage the flow of troublemakers coming into the plant. Well, then there were so many troublemakers trying to get into the plant that the colonel's staff couldn't do any work. So he had a solution. He formed a meeting room, put tunnels there, and this was the critical mass of management. These two general staffs could keep themselves busy and not interfere with production.
Turns out the Allies found the plant and bombed it out, and the people in charge sent a message back to Berlin: "Send us some money to rebuild our meeting room so we can get this plant back in operation." That's the critical mass of management. Remember that story — you will encounter it time and time again in your organization.
§5. Steel metallurgy: martensite, ferrite, and tempering [20:43]
So hardenability, on the other hand, instead of hardness, is different. It's the relative ability of a ferrous alloy to form martensite. What's martensite? Martensite is a particular crystal structure — the really hard form you get when you quench the steel from a temperature above the critical. Hardenability is commonly measured as the distance below a quenched surface at which the metal exhibits a specific hardness, like 50 Rockwell C — HRC — or a specific percentage of martensite in the microstructure.
What's the upper critical range? It's right here on the iron-carbon phase diagram. This is gamma iron up here. Gamma iron has a different crystal structure: face-centered cubic. Why do we call it face-centered cubic? Because it's a cube with an atom in the center of each face. Face-centered cubic — that's what we call austenitic stainless steel. Gamma iron is also called austenite, after Roberts-Austen [William Chandler Roberts-Austen], a British metallurgist in the late 1800s. At low temperatures, the stable form of iron is body-centered cubic — a cube, but instead of an atom on each face, one in the center. That's BCC. So you have FCC, gamma, or austenite; down here you have alpha iron, or BCC, or what we call ferrite.
When you quench this stuff from high temperature and cool it very rapidly, you won't get exactly this. You'll get another phase that comes out as little needles. It's called martensite after Adolf Martens, a metallurgist in the late 19th century. Why are these things named ferrite, martensite, and so on? Because the beginning of metallurgy was in the 1880s, when a British petrographer geologist — a guy who polishes rock — decided to try polishing steel. He polished it, and it was just a mirror. Looked in a microscope: just a mirror. Then he decided to etch it with some acid, and all of a sudden he started seeing structure, and he started calling these things ferrite, sorbite, martensite. Sorbite's named after Henry Clifton Sorby. He was the first metallurgist.
Metallurgy came to this department in 1888. MIT was formed in 1865 — this was the geology department, geology and mining. Metallurgy didn't exist in 1865. It wasn't until the 1880s that Sorby and others started studying this stuff. That's why I did that term paper I talked about — I went back into the basement of the MIT libraries, caught a cold that fall reading these old original papers from the 1890s, because we were trying to figure out what was going on in the structure of steel.
The upper critical temperature — this is temperature versus composition. Here's the austenite, gamma iron austenite, then alpha iron ferrite. Here is the upper critical temperature. At 0.77 carbon is the lowest temperature at which austenite is stable. This is the solvus line for carbon in austenite; this is the solvus line over here for carbon in ferrite. In this region you have both. You have to heat up above this line. So if it's very low carbon steel, you've got to be above 900°C. If you're high carbon steel, you only have to be above 723°C. You quench it down, in water, and then it's glass brittle.
What do you have to do? You have to bring it back up to around 400 to 600°F, or above 1,000 — like 1,000 to 1,200°F — to temper the steel, to take some of that hardness out. You don't want to be Rockwell C60. That's too brittle. It has a strength of 300,000 pounds per square inch, but so brittle that if you hit it hard with a hammer it'd shatter. So you want to get some ductility back, and you temper it. If you're tempering at a relatively high temperature like 1,000 or 1,200, that's what we do with HY80. We make a steel that can be tempered at a high temperature, so that when it sees the heat of the welding arc, it can tolerate it.
Student: Doesn't tempering release internal stress in the structure of the metal as well?
Yep.
Student: I always thought that was the primary reason you temper. Is that not the case?
Relieving that stress also gets rid of the brittleness. It's the same explanation, just from a different direction. If you did this with absolutely pure iron, you would not form martensite. Steel is an alloy of iron and carbon. The carbon is an interstitial small atom found in the interstices. There are octahedral sites — eight-sided sites — where the carbon can sit between the larger iron atoms.
When you quench it, it wants to transform to BCC, but it does it in a diffusionless, displacive way. We know more about steel than any other material in the world, including silicon. Why? Because we use so much of it. It's such a huge industry, and we've been studying it since 1888. There have been billions of dollars spent on understanding the microstructure of steel, and it's a very rich field. When you start looking at a phase diagram this complex, and you throw all these different alloying elements in there, you end up with something very sophisticated.
Student: Is body-centered or face-centered cubic the stronger metal?
Neither. They're both very soft if they have no carbon. If you put carbon in, it'll be stronger proportional to the carbon. If you quench it, the carbon — because of the displacive allotropic way this thing transforms — goes into preferential sites, and it will not form body-centered cubic. It forms body-centered tetragonal, because the carbon atoms make one axis a little longer than the other two. That's the strain — the residual stress, if you want to call it that — that makes it so strong, and so difficult to deform. If you temper it, that carbon moves around and relieves some of that stress, and it starts to revert back toward body-centered cubic.
Student: So quenching it — in a way it's almost like an old aqueduct arch, where you have a compressive stress that helps hold it together. So when you quench it, it's creating internal stress that makes it harder, because of the structure of each, and more brittle.
Right. Unacceptably brittle.
Student: Then why quench it in the first place?
To get the strength. BCC has a strength of 35,000 PSI until you quench it. Heat it up and quench it, now it's got ten times the strength, 300,000 PSI. Martensite isn't exactly body-centered — it's body-centered tetragonal. To a PhD metallurgist, that's very important in understanding the details. You don't need to know that detail. It won't be on the quiz. There are people who spend their lives studying this. Morris Cohen, a professor here, spent his life studying martensite. He was kind of Mr. Martensite.
§6. Weld defect taxonomy and the legacy of dirty steel [30:26]
That's the difference between hardness and hardenability. I know you guys get tired of steel, but we won't be doing it forever. If the only tool you have is a hammer, you see every problem as a nail. I talked to you about some things like hot cracking. Solidification cracking or hot cracking is because we have to have manganese in there to tie up the sulfur — otherwise we get a structure where, near the melting point, it just pulls itself apart as it contracts when the temperature comes down.
Porosity — we get gases in. Porosity is not usually a huge problem unless it gets really gross, like 10% by volume, and then it weakens the steel. Hydrogen cracking — we're going to spend a lot of time on. Lamellar tearing was a problem back when we had steels that had 400 parts per million sulfur and big oxide inclusions. This was the steel-making technology of the 1960s and before. Big problem. When I worked at Bethlehem Steel in the mid-70s, the oil companies were building a lot of big offshore oil platforms. Big heavy steel plates 3 and 4 inches thick. They were getting lamellar tearing problems because the steel wasn't very clean. The Japanese taught us you can make steel inexpensively if you have modern facilities — not if you have 1911 blast furnaces, like Bethlehem Steel had. So the United States fell behind.
Lamellar tearing is not a problem today. Instead of 400 parts per million sulfur, it's easy to buy steel at 50 ppm sulfur, and you don't get lamellar tearing. My office mate at Bethlehem Steel — that was one of his projects. I don't know that I've ever seen more than one problem in the real world of lamellar tearing. Forty years ago it was a big problem, not today. The metallurgists will tell you about it because it's part of their history, but it's not necessarily part of their problem set today.
Reheat cracking. My supervisor at Bethlehem Steel had done his doctoral thesis at Lehigh on reheat cracking of steels. If you stress-relieve a steel — we put steels in a furnace at 1,200°F for an hour to get rid of the residual stresses from welding — sometimes they'd come out, and you'd actually hear a pow inside the furnace, and there'd be a great big crack going through the steel, a very intergranular crack. It turns out it was small amounts of sulfur, arsenic, phosphorus, antimony — things that shouldn't have any business being in the steel, but they'd do it. That's what Karl Meyer did his doctoral thesis on. He was one of the last in the line of the great metallurgists at Lehigh. He was my supervisor at Bethlehem Steel. We learned how to solve the reheat cracking problem. It was a problem all the way up through the 1970s, but not really today.
Solidification cracking — we talked about the low-melting sulfur compounds. You put manganese in there. When we get to stainless steels you'll find some other problems, and when you get to aluminum alloys you'll find lots of solidification cracking.
§7. Charpy toughness and the Liberty Ship legacy [34:10]
Low heat-affected zone toughness due to grain growth. Here's my HY80 weld from 30 years ago. You can see the heat-affected zone — this little black region next to the fusion line. Let's go back to that Navy report from 1946 about the Liberty ships. They found that the ships that sustained the big cracks in the hull structure had less than 10 foot-pounds of Charpy toughness.
What's Charpy toughness? The Charpy test has been around for 105 years. Five years ago they celebrated the 100th anniversary. How many people have seen a Charpy bar? It's a V-notch. It's a French test — Charpy, C-H-A-R-P-Y, a French metallurgist of 105 years ago. It's the new metric: 1 cm by 1 cm by 10 cm, with a 2 mm notch right in the center.
This is like tearing a piece of paper. You put a notch in it. The best way to make the notch is to broach it — run a cutter through there — because it has very specific requirements on how sharp the notch is, so you get the same stress at the tip. It's a stress concentration. You put it in a big pendulum machine, raise this 100-pound weight on its arm, put the Charpy bar in, and let the pendulum go by gravity. It swings down, breaks the Charpy. It hits the Charpy at the tup, which is the hammer head. The big piece of steel hits this thing, breaks it, and goes flying off. If you've ever been in a room where they're testing Charpys, they've got little screens to catch the bar.
When I was at Bethlehem Steel, we had a 264 foot-pound Charpy machine. It's about 75 foot-pounds per joule. Foot-pounds is a measure of energy. Toughness is a measure of the energy of fracture. Tensile stress is the best measure of the force of fracture. This isn't really a toughness measurement, but it's an energy measurement that people have tried for years to relate to fracture toughness. There are correlations. We have fancier, more expensive tests today. Each Charpy bar cost about a hundred bucks when I was at Bethlehem Steel. Because there's scatter, you usually have to test three at any given temperature. For the full Charpy curve with temperature, you might have to do five different temperatures, three or five different sets of three.
We had a machine shop up at the research labs, but we had to get our Charpy bars — company rules — machined down at the regular big machine shop at the steel plant, because they were doing these bars all the time for test plates. We learned that you'd call up the foreman down there and say, "Are you busy?" If he said yes, you'd send it to him. If he said, "No, didn't have a lot of work," you'd wait, because you didn't want to pay $150 for a Charpy bar. If he was really busy, you might get them for 75 bucks apiece. The guys just wanted to fill up his work book. There's a lesson in that story.
In any case, the Navy in 1946 found out that the ships that sustained brittle fracture had less than 10 foot-pounds. The lowest I've ever seen in my career is 2 foot-pounds. The most brittle steel I've ever seen. Typically 5 foot-pounds or less is extremely brittle steel.
I meant to bring — Don Sadoway, a faculty member of this department, sent me a link to a problem they're having on the Oakland Bay Bridge. They have some big steel tie rods, and they're corroding and breaking, and they're holding this bridge up. It's a multi-billion-dollar bridge. It's hydrogen cracking, because it doesn't show up right away — it takes a few days to show up. It's due to corrosion. Another way to get hydrogen into the steel is not just a welding arc plasma — it can be by a corrosion process. You can take a bolt that has no welds, stick it in seawater, put a stress on it, and the bolt will crack in a couple weeks because of hydrogen cracking. So it's not just welding, it's lots of other things. One of steel's Achilles' heels is hydrogen embrittlement. Other materials have hydrogen embrittlement too.
§8. The Coast Guard, the LNG tanker, and the test-plate trick [40:04]
What happened with these Charpy tests? In the 1950s, after the 1946 report came out and took a while to digest, the Navy said, "We're going to require 15 foot-pounds. We're not just going to require a tensile test on our steel. We're going to require 15 foot-pounds of Charpy energy, and we're going to require it at the weld, because a lot of these things are found at weld defects." So: base metal, weld fusion line, weld metal — they'd look everywhere.
Then the Coast Guard comes along in 1960. The Coast Guard wants to be safer than the Navy. So they decide — if 10 foot-pounds was the problem level and the Navy wanted 15, the Coast Guard would ask for 20. And whereas the Navy looked at the weld metal, fusion line, and base metal — each one of these is $300, $5,000 worth of testing — the Coast Guard decided they were going to be carrying liquefied natural gases on some of these ships, and they wanted to be extra safe. So they'd require it at the 1 mm position, the 3 mm position, the 5 mm position, and the base metal. And you'd have to meet not 15 foot-pounds, but 20.
So when I was working on the skirts to hold up these big aluminum tanks for the LNG tankers down at Quincy — they were building them at Quincy — that's what I had to do. I had to measure each one of those locations along the weld. I had to weld a plate a foot long just to get all the specimens out of it, because you had big tensile specimens and everything else.
Now guess what the shipyard did. This was a big problem — even more tests that had to be done. The last thing they wanted was to pass more qualification tests. There was variability in the steel plate. Bethlehem Steel at that time — the base metal would meet 20 foot-pounds 70% of the time when it came out of the mill on first pass. You didn't want to scrap 30% of your steel. So they'd send it back through a heat-treatment furnace — in this case a normalization furnace, where you heat it up to upper critical and slowly cool it without quenching, and you get a finer grain size, and it should give you better toughness.
After they did that, maybe two-thirds of the ones that didn't pass the last time — the 18 or 19 — would now pass at 20, 21, 22, whatever. Some that still didn't pass — the 10% — they'd put through a second renormalization, and some would pass and some wouldn't. They'd scrap the ones that didn't pass after three tries. If they were Japanese, they would have just lowered the sulfur. But it was going to cost us money to do that, so we didn't want to.
Every now and then, the shipyard would get a report from Sparrows Point that this particular plate had 30 foot-pounds of toughness. A really good steel. A really good steel could stop the hammer at 264 foot-pounds. When I was developing my steel, I used to stop the hammer on a regular basis. If you stop the hammer, you have to recalibrate with about $2,000 worth of Charpy bars. You used to buy them from Watertown Arsenal, which is a mall now out here in Watertown, Massachusetts. Now you buy them from NIST, the National Institute of Standards and Technology — they sell standard Charpy bars, supposed to be calibrated. So the steel could be 264, or 5, or 15, or 20, or 30.
What did Quincy shipyard do with that 30 foot-pound plate? They turned it into test plates. What are you going to test in the laboratory? You're going to weld this huge plate, but you've got to have a piece to test. So you take two little pieces of steel, put them on the end of the weld, and weld a one-foot tab of test plate on the end. When they found one outlier at 30 — and the other sheets coming from Sparrows Point averaged 20.5 foot-pounds, with 20 as the minimum — they'd cut all their test plates out of the 30 foot-pound plate, and that's what they sent to the lab to meet all the properties.
So, Mr. Coast Guard, what do you think of that?
Student: Sounds too good.
This story tells you there's always a way to get around. This is your job in the future.
Student: I'm just impressed that they weren't just writing 20 and saying it's fine regardless.
I'll give you the name of the test lab down south of Boston. If you've got pre-stressed concrete cores, no need to send the core in from the shop — they'll just send you the test report.
Student: That's criminal.
Try to prove it. You could, if you had subpoena power. There are labs that do that. You probably said that because you know of situations where that's been done — qualifying test reports. I can neither confirm nor deny. And now we find out why some people in the shipyards in the ship don't always get along.
§9. Fingerprinting steel: the Gulf of Mexico pipeline and the bridge steel [47:08]
Variability in steels. The steel companies know a lot about the variability of their steels. This is Contribution to the Metallurgy of Steel, published by the American Iron and Steel Institute. This one is September 1974: "The Variation of Product Analysis and Tensile Properties, Carbon Steel Plates and Wide Flange Shapes." Wide flange shapes are I-beams. This is the June 1984 issue — wonderful reading: graphs, tables, the variability of hundreds, if not thousands or tens of thousands. The central machine shop at Bethlehem Steel was turning out Charpy tests on plates. They had all the data. Every now and then, in an anonymous way, the steel companies would give the data to the American Iron and Steel Institute, and they'd publish the sanitized version of variability.
Here's some of the variability. Lateral expansion is related to foot-pounds, and you can see the variability from 16 to 71. Here's absorbed energy in foot-pounds, 16 to 61. Pretty wide range. Oh, this is a good plate — look at this. All the way up to 26 from actually 2.94. Don't buy that steel, baby. So there's data on this variability. And there's data on the composition, and I'll tell you a story of the composition.
[Tom holds up a section of seamless steel pipe.] Here we have a piece of steel pipe. We cut through it. There's a big missing piece here. This is seamless steel pipe. It's been painted on the outside. It was buried in the Gulf of Mexico to carry oil from a production platform to shore. This never carried any oil, because what happened is the steel mill that made it — which will remain nameless, but it's in northeast Ohio and it's been there for 100 years — produced a piece of steel and sold it, and they shipped it to Houston to be used in building this pipeline under the ocean, or on the floor of the ocean.
Before it left the steel mill it had to go through a full ultrasonic inspection. It had to go through a pressure test — a hydrostatic test at 75% of its yield strength. We actually call that a proof load test. Every length of pipe. This is seamless pipe. No welds. They basically take a piece of hot steel at 2,200°F and a steel mandrel that's cold, and they shove it through there 40 feet long. They take a solid piece of steel and turn it into a big red piece of hot steel pipe, and it can be heat-treated afterwards. That's steel tubing.
They had pressure-tested it, done a proof load test to show it would hold the pressure. They took it down, sand-blasted it, took off all the identification markings, painted it for corrosion resistance — that's very good paint, by the way — laid it on the seabed, and did one final pressure test after it had all been welded together on the seabed. It failed. It cost them $5 million to retrieve it from the bottom of the seabed. These supply ships and maintenance ships and diving ships cost $100,000 a day. They're not cheap.
They recovered it, sent it to me to look at, and what most likely happened is one of those mandrels that goes pushing through the piece of hot steel finally, after multiple uses, developed a fatigue crack and spalled off a piece, which got rolled into the surface. It passed the first two tests because we had a piece of embedded steel plugging the hole. Over time that piece of embedded steel — basically the poor bond — broke loose, and on the third pressure test it leaked. Cost them $5 million to replace.
Student: Did it break through right here where you also notched it?
We thought that was the best location. We cut it out to do metallography to see what we could see. We saw that it was decarburized there. We knew we'd lost some carbon. We knew that surface was present at the steel mill. The steel company had two defenses. First: there's nothing wrong with the pipe. That's not a very good defense. Second: it's not our pipe. You have no identification. You took all the identification off. You bought steel from other people, and it must have been one of their pipes.
Even though it's got carbon, manganese, sulfur, silicon, phosphorus — it also has residual small amounts, unintentional amounts of nickel. Maybe 0.15% nickel. It could have 0.08% copper. It has maybe 10 different elements. I can go to this thing which gives me the composition variation of various steels. If I know the statistical variation of the composition produced on thousands of heats of steel, and I do a chemical analysis on that piece of steel, I can do a statistic. If I have a 90% confidence level statistically that it meets the carbon requirement — and these are narrower ranges than the requirements in the specifications, looking at a specific number — if I measure that, it's only a 10% probability that it's not the same piece of steel. Times 10% for manganese, times 10% for silicon. I proved that was the same piece of steel produced by that steel company with a probability of one part in 100 million.
They weren't expecting that when they said it's not our steel. You're right — there's a one in 100 million chance it's not your piece of steel. You can fingerprint the steel. I get to do that about once every two years. That's a standard defense: who, my steel?
That was one of my first jobs at Bethlehem Steel. They had a steel pipe — actually it was steel plate — that had terrible toughness, and all I got was a small little piece. It was part of a highway bridge on the interstate highway system that had just been built. I got one little piece, we measured the composition, and it was what's called weathering steel. It had small amounts of intentional nickel, copper, and chromium. Bethlehem Steel used nickel and copper for its corrosion resistance. US Steel used chromium and nickel with no copper. Just differences between the steel companies. The company that had used it and had this big expensive repair on the bridge sent it back to Bethlehem and said, "Your steel is junk." I analyzed it, and I wrote back — I wrote the letter report to my boss's boss — and I wrote, "Yes, you're right. It is junk. It's got chromium in it. Go talk to US Steel."
§10. Untempered plate and the unheat-treated fasteners [55:51]
There are lots of ways to fingerprint things. I didn't have to go to statistics there. Bethlehem didn't sell that grade of steel with chromium in it — that was US Steel. So you can often tell who the manufacturer is. You can sometimes get down to a specific lot of steel by going back to the fingerprint. In the aerospace business they keep all those records. In the medical business they keep all those records. In the nuclear energy business, they keep all those certifications all the way back. If they have a problem, they have to have a product recall.
We had a problem with a piece of quenched and tempered plate, 100 ksi steel like HY100, but it was a lower chemistry because it wasn't quite as thick. It was a bridge steel. I was reading about it in the Bethlehem Steel report — I don't remember if it was Bethlehem Steel plate or just a problem the Federal Highway Administration had. The Charpy toughness was like 5 foot-pounds. It was terrible. I didn't understand this. I was asking Bob Summers, the grand old man of welding — a Lehigh grad at Bethlehem Steel, one of our grand old men of welding. I said, "Bob, how can this steel be so bad?" He said, "Well, scuttlebutt is it went through the mill without tempering. It had 160 ksi strength and it was glass brittle because they hadn't bothered to temper it. It slipped through the process without getting tempered."
Student: I've had that happen. My job was to go check all the paper trail, and sure enough, these fasteners were tied to paperwork from the manufacturer that said this is entirely compliant. Then we chased their paperwork back to the company that made the steel and used for evidence the actual properties from the miller. We noticed that their paperwork said it has the following properties if heat-treated. It was never heat-treated. They assumed it was. They machined it, manufactured it, produced it, and produced paperwork with their company logo on it saying it's qualified to these standards. There are many reasons why that should never have made it that far. They're supposed to destructively test a certain percentage. It should never have made it as far as it did. Sure enough, all through our submarines, through the rocket systems, were a bunch of these that were never heat-treated.
Not good news, but these things happen. That's why you guys are out there — to keep them from happening. If that happens on your watch, you start looking for a new job. Why don't we take a break for a little bit.