§1. Helicopter peashooter hydrogen cracking and the ethical bind [00:01]
All of a sudden the engine manufacturer says oh no, this is ITAR restricted or whatever, can't have some foreign citizen working on this on the other side. So they had to hire me because I was a U.S. citizen and I could look at some of their documents of how they manufacture this. And sure enough, they had procedures for nitriding this and putting it into a furnace within four hours so you get rid of the hydrogen that you might get during some of the processing. It was all there.
Except, as I'm reviewing their procedures, they wanted to do a quality control check after they've been through the procedures that could get the hydrogen in there. They would actually have an exemplar in there and they would cut it off, send it out to the met lab, and they get the results back two days later. And that's when they started counting the four hours — from the day they got the met lab results back. They would start counting the four hours they had to get it into the furnace, within four hours of getting the test lab results back, which was like two days later. I don't think that was really within the spirit of you have to put it in the furnace to get rid of the hydrogen within four hours.
At this point I realized — and in fact this helicopter and a number of others had gone down and killed people, because the splines that held this flexible joint together had cracked. We had looked at them and we saw hydrogen cracks in these things. So at that point I was faced with an ethical problem. As a professional engineer, my number one duty is for the safety of the public. It's not for who hired me. And in fact I was in litigation against the manufacturer, but the manufacturer had a defective spec, and they could kill somebody else.
If I went and blabbed to the FAA — well first of all it's a defense engine in this case, so I could blab to anybody I wanted, but the manufacturer could sue me and claim that I was trying to influence the litigation. So what I did is I called up the attorney I was working for and I said look, this is a defective specification. If they want to do their quality control test and make sure they got their hardened layer the right thickness, they've got to put it in the furnace within four hours of finishing the process, not within four hours of finishing the quality control test. That doesn't make any sense. And of course this sort of would go to whether they produced the defective part.
I had to notify the manufacturer they had a problem. I notified my attorney, I said you've got to tell them. If you don't tell them, I will have to go to the FAA or the Defense Department or whoever and tell them. He contacted the other attorney and said look, our expert says you've got a defective process, and you need to fix it. Turns out the other attorney just laughed, said oh we don't have anything defective. Meantime we go along and finally it goes to trial in Lafayette, Louisiana. I guess it might have been a commercial oil rig helicopter, but in any case, we end up winning the case. And we did learn that in spite of the attorney laughing on the other side, they did go back and get some competent metallurgical expertise and they did fix it. But I couldn't just sit there and read about in the paper some other person dying because these people had a defective process for removing hydrogen in their peashooters.
§2. Alexander Kielland: the fillet weld that sank a rig [04:01]
There's one other example and then we can take a break. I've mentioned Alexander Kielland. This is Easterling's book — in the back he has the Alexander Kielland if you want to see what the jack-up rig looked like. Here's a picture of the Kielland. There were a bunch of pentagon five-legged jack-up rigs, most of them working in the North Sea, built in the 1970s. This was at the French Dunkirk shipyard that they built the Kielland. It was supposed to be a drilling ship.
It's 35 meters tall, so it's 100 feet tall. These are like 30-foot diameter legs, and it's a semi-submersible so it's got big pontoons over here. Right here you actually had a cross brace, and on that cross brace you had a little sonar flange — they called it a flange plate. Here's the flange plate; the crack started here, it ran all the way around the small tube. This is three feet across, 325 millimeters. You can start guessing that that's like 15 feet, and the other big leg is even bigger. The crack started here, ran all the way around, and then started other fatigue cracks at other locations, and a lot of these support structures failed. The thing went down in a big North Sea storm.
They look at the thing afterwards and they see fatigue cracks starting from the fillet weld. You can see they have a little bit of dye penetrant or something. You can see a laminar tear here, you can see a little root crack down here on the weld. These are the actual welds they dug out of the ocean. Here are some other sections — you can see a nice crack right here. Here's the fatigue crack.
Now you get to the hardness values. This is Vickers hardness, or maybe Knoop — doesn't say, just says hardness, but Knoop and Vickers are almost the same numbers. The bracing was 160, which is, you can't crack that stuff with hydrogen. The weld metal was like 250 — that's still less than 300, that's not so bad. But you start looking in the heat-affected zone of the sonar flange plate, and it turns out it's 350, Rockwell C35. Not too surprising. If you look at the main bracing, lots of ductility, area reduction of 30, 35. Not specified for the flange plate. This is the main bracing, this is the flange plate, and you can see one to seven percent area reduction. And lousy Charpy energies.
A lot of things came out of that tragedy, but one of the things they learned was that the maintenance people were responsible for making this little three-foot diameter pipe, and the fabrication people made all the braces in the vertical pipes and legs. Those people had very strict welding procedure controls. The maintenance guys, they just go in the shop, pick up a piece of steel, they didn't care, they just weld it up, they didn't have any procedures whatsoever. It was a little fillet weld down hidden. And the other interesting thing is the sonar plate was supposed to be there because this was supposed to be a drilling rig, and they converted it to a hotel where people were just sleeping. It never was used for sonar or anything else, it was just an appendage that had been put on. And they had no process controls for controlling hydrogen content or hardness or anything else.
§3. The West Grafton forging press and lost thick-section welding technology [08:48]
With that I have one last thing which sort of relates to you. It's another forging press problem, which I think I can do fairly quickly. The reason it's interesting is because it shows you what happens and how we lost our technology to weld very thick steel. Back in the 1950s when there was lots of money to protect ourselves against the Soviet threat, the Air Force was trying to get rid of welding in their aircraft, so they wanted bigger and bigger forging presses. They paid to build two 50,000-ton forging presses. These were the two largest forging presses anywhere in what we call the free world. The Soviets might have had a larger one, I don't know. One of them was at the Alcoa works in Cleveland, Ohio, and now it forges aircraft parts out of aluminum and titanium. The other one was in West Grafton, Massachusetts, out here near Worcester, about an hour and a half away. There's a company called Wyman Gordon. This was back in the days you always had redundancy in defense procurement — you always wanted to have two people, so they helped them in whatever way they did, but built these two forging presses.
In the early '80s the one in West Grafton developed a crack in one of the six support beams. Now this is the world's largest forging press. Above ground it's got a big frame up here, but you can't make it out of a solid aluminum casting. It stands maybe 80 feet above ground, and then it goes about 60 feet below ground. You actually had some I-beams down here — six I-beams if I remember. The I-beams were about ten feet tall, the web was 10 foot, the thickness of the web was 10 inches, and the flanges if I remember were like 15 inches thick. They had six of these going across instead of a great big casting. These I-beams created the structure to take the 50,000 tons of force.
The U.S. Air Force was thinking about building a bigger forging press — I guess you're going to forge a whole aircraft all at once or something. The problem was you have to have a foundation to put this whole thing on. This one was cast in concrete. You took an elevator to the bottom — six stories to the bottom if you had to get down there. Somewhat greasy after about 30 years of grease falling down through here. But the Air Force was going to build a bigger one and they weren't going to pour a concrete foundation which might be 100 feet deep. They were going to use a mountain in Colorado — take the bigger press into a mountain. They ended up never building that one. But that was the plans back in the 1950s, to build an even bigger forging press.
At this point in the 1980s they had a 10-foot crack in one of these beams. Six beams, couldn't use it. There were only two of these presses. It had become pretty critical for forging great big aircraft parts and all kinds of other things. The replacement cost of the press was estimated at two billion dollars in 1980 dollars. So they were calling in a bunch of people to figure out how to reweld this ten-foot-high, ten-foot-thick I-beam that had a crack all the way through. I was brought in. I actually got to take the elevator down to the basement of this thing with all the grease. The question was, can we weld it in place, or are we going to have to bring it out of here? To bring it out, there's a lot of equipment in here — it's sort of like bringing out a 40-foot beam, 10 feet high, out of a submarine, if you want to think of how crowded things are. They had to disassemble a lot of this.
One of the questions was, how do we weld it, how do we get almost zero distortion? The top machine surface had to be accurate within an eighth of an inch over like 30 feet. They actually had to machine the surface flat to distribute the loads evenly. What's the welding procedure we should use, and can we do it in place? Well, we knew we were going to have to stress relieve it. I said if you do, you're going to have to stress relieve it down six floors with all this grease around. I said I know it's a pain in the neck, but if I were you I would rig it out of there and bring it up and build a furnace around it so that you can stress relieve it properly and you can do the machining and everything else. Which is what they ended up doing.
But one of the problems is, we didn't know how to weld 10-inch-thick material without stress relieving it. We had done it for hundreds of thousands if not millions of tons of armor plate in World War II and before. But in 1980 we did not have the technology. We still don't have the technology. We lost that technology when a lot of people died of old age.
Okay, so let's take a six or seven minute break and then we'll start another topic. I think that's most of my war stories of hydrogen cracking, but it still occurs on a fairly regular basis. When the world started deciding in the 1980s that metallurgy was passé and we didn't need metallurgists anymore — everything was going to be electronic materials — I used to tell Professor Sata[wa]y [Saint-Way?] that we would have plenty of work in our old age if they weren't going to produce metallurgists anymore.
§4. After the break: serialization marks and the Kodiak helicopter return [16:08]
Yeah, we brought them home on the ship instead of flying home. Seeing the helicopter, it was kind of fun. They took the rotors off and craned it off the ship. We were downtown Kodiak, Alaska, about four or five miles from the air station. They brought one of the airport mules, towed the helicopter on the city streets all the way back, with the rotors. It's a big helicopter.
It turned out at the warehouse inventory supply center in Carolina, North Carolina, they had stamped or etched a unique identifier code onto this high-speed drive shaft, and that's what started the decrease. They did an actual experiment during the investigation — they did like 50 of them, put 50 of these things through fatigue simulation. They crack-etched in 25 more, and it was 100% on every etched one. First of all it's a high strength steel. If they etched it they were introducing hydrogen from the etching process.
The other problem people have is they went to laser scribing, where you melt the surface. All they're doing is a super rapid cool in a high strength steel, and they would end up with high hardness, not necessarily a hydrogen crack, but just very high hardness. The thing would start a brittle crack and they would start fatigue cracks. So now they actually have some laser engraving techniques that don't melt the surface. Without melting the surface you don't get to high enough temperatures to transform things. But they have a lot of different techniques now, because this was back in the '70s and '80s when lasers — it was also part of programs that were becoming popular then, PMI, positive materials identification, and serializing parts. So now you serialize the part, and you know exactly where it came from, and so you can tell exactly why it failed. Well in this case it failed because you're serializing it.
§5. Crack stoppers, riveted vs. welded construction, and brittle fracture in pipelines [19:05]
Okay, we ready to start again? The one thing I would say about this 10-foot-high beam on this last forging press: the cracks started at a fillet weld where they had little blocks. If you look at the beam sideways it was like 40 feet long, and it tapered for whatever reason to the bottom flange. So you had your flanges and this is your web. In order to rig it in there, they had welded on a little block. I say a little block — it was a piece of steel that probably weighed about 50 pounds, and it was about 10 inches square, actually a 45 degree cut. They had welded it on, and they decided we don't need to take that off. Well that became the stress concentration, that was the beginning of the fatigue crack that caused the thing to fracture.
There's an interesting brittle fracture story here. The crack was about 10 inches long before it finally, on one of the press loadings, just ran the other nine feet. So that's critical flaw size. I've told you before that good steel you could have critical flaw sizes that are pretty big, and this one we knew the critical flaw size was about 10 inches, because we could see the fatigue crack grew about 10 inches from here. And then it ran in a way that cut the beam in two.
Student: Did the flange stop the crack or did it affect the flange?
No, it just started at the toe of the fillet weld. When it ran to the top, it went all the way through the flange because the flange was full penetration welded to the web. So the crack ran all the way through. One of the advantages of old riveted ships is you never had a crack run further than the plate. When you get to the rivets, crack stops. It was the all-welded construction in World War II — and before that we always riveted ships together. If you had a brittle fracture, we had brittle fractures in World War I but they only ran the length of the plate. You get to the end of the plate where the plate stops and you got rivets and the crack stops. It's when you get to all-welded construction that all of a sudden you can get a crack run for great distances.
The Titanic had brittle steel, but the problem was the iceberg cut through six compartments and all six of them flooded, and it became heavy on one end. It wasn't until the early 1930s that we started building critical structures of all-welded construction. The first one was the Big Inch pipeline from Louisiana up to New Jersey. It's a 30-inch diameter gas pipeline, and it was all-welded construction — first really critical thing that had been built. As far as I know it was successful; it probably was taken out of service for corrosion, or it just wasn't big enough — we have much bigger pipelines and higher strength pipelines now that can take more gas pressure. But it was World War II when we started to go into all-welded steel construction that allowed the Liberty ships, if you get a crack start it could run all the way around the ship and split the ship in two, because there was nothing to stop the crack.
There have been cases — I've heard of cases — that there were small diameter pipelines, I don't know if they're 12 inch or whatever, where a crack would start and run for 30 miles in a buried pipeline, a brittle fracture. When I worked for Bethlehem Steel, they were starting to build the Alaskan pipeline, and Bethlehem Steel made big diameter pipe for pipelines, and there were other pipeline projects people were building in the world. The guy in the office right next to me was our pipeline expert at Bethlehem Steel. He was the one I think told me about the 30-mile-long running crack. They were looking at using inserts in the pipelines to prevent the fractures. Every couple of hundred yards they were going to weld in a heavy wall, high toughness steel. So this would be your regular pipe, but they might put a 12-inch-long piece that they would just weld in there that was thicker and very high toughness steel.
The type of steel they were looking at was A710, which is the same thing that was the predecessor to HSLA-80 that the Navy uses now. That was a steel developed by International Nickel. It had like one or two percent nickel. As a young mid-20s engineer I said well why don't they just build the whole thing out of A710. He says, well first of all you couldn't afford it, with all the alloy content to get that high toughness. But the other thing, there wasn't enough nickel in the world to build a whole pipeline out of something with that high nickel content. Well there probably was enough nickel, but it was really the cost issue. Nonetheless, you do run into problems of whether you have enough material. So they actually sometimes design crack stoppers, crack arresters — just thicker, higher toughness, and if a crack starts running, you can stop it.
Now you only get this in gas pipelines in general. Because it turns out, if you get a brittle fracture in a pipeline and it's carrying oil or a liquid, the speed of the brittle crack running is a substantial fraction of the speed of sound. It's like one-third or one-half the speed of sound that the crack is running. And if the compressed fluid that's stressing the tip of that crack is a liquid — well the speed of sound in the solid steel is like 5000 feet per second, in the liquid it's like 3000 feet per second. In the liquid you actually decompress: if you start a fracture right here in the pipeline, the escaping liquid will create a pressure wave that actually can outrun the crack. So you end up relieving the stress at the tip of the running crack, in a liquid pipeline.
In a gas pipeline, what's the speed of sound in a gas? As I remember from my high school, the speed of sound at STP is 343 meters per second — a thousand feet per second in a gas. So that's a lot less than the running speed of the crack. In a gas pipeline, the pressurizing gas is always stressing the crack tip at the full pressure, and so in a gas pipeline you can run for 30 miles. But in an oil pipeline you're actually going to stop. This is important for you Coast Guard people. If you're going to build gas pipelines from Alaska, you need to understand the fracture mechanics. There's a difference between gas and oil in terms of whether you're stressing the crack tip.
§6. The IMT hot isostatic press explosion [27:08]
I won't tell you the other — well maybe I will tell you the other story. This is about almost 20 years ago now, 18. I remember I was department head at the time because I remember sitting in my office getting a phone call. Up here in North Andover they had the world's largest hot isostatic press, and the world's largest hot isostatic press was 60 inches in diameter. It was a 300-ton forged vessel, and it had a shape on the outside. It was threaded top and bottom. They had a plug — the plug weighed 50 tons. This plug weighed 50 tons, and this was a 200-ton cylinder, with a plug top and bottom, with 60-inch diameter. This was 11 inches thick, this was like 17 inches thick. You had to have the thicker thing up here.
It had been made by a Japanese steel company. There was no one in the United States that could make these things in the 1980s. The Japanese had some of the best technology. A hot isostatic press is something where you fill it full of argon and you compress the argon to 20,000 psi. So this thing's going to hold 20,000 psi of argon gas. If you start calculating that, you now take gaseous argon and you bring it into a state where it's almost as dense as liquid argon. You're squeezing it at that pressure, because that's over a thousand atmospheres, like 1300 atmospheres. You put powder parts in here, or castings, like the disc of a jet engine — the rotor disc — and you want to get rid of the imperfections. You heat it up to 1500 degrees Fahrenheit under 20,000 psi, and you basically just take those pores and imperfections and you just forge them out of it, you just isostatically squeeze the thing until everything welds back together at the temperature and in this argon gas.
They had this thing — a company called IMT up here in North Andover, Massachusetts — and they had a number of these presses. They only had one press like this. They were building another one, fortunately, because when this thing let go, it let go with a crack right here, at that stress concentration from the outside. They designed this on a supercomputer. In the mid-1980s, the supercomputer was not quite as good as a PC of five years ago. They had assumed everything was steady state heat flow — but you have this thing heated on the inside and you have to water cool the outside to keep it from getting too hot. They had this brittle fracture. This top piece let go with 16 tons, and it was found a quarter mile away.
Some of the people in the town of Andover were not too happy. It didn't land on anybody, but people weren't too happy to think that they should have projectiles. One five-ton piece — this thing broke into like 70-some pieces. I had to reconstruct the whole fracture. One five-ton piece landed and crushed the chair of the operator. It occurred at two o'clock in the morning. This plant ran three shifts a day, and he had just gotten up to change the chart paper, and he was about two or three feet away from it when the five-ton piece crushed his chair. He was sort of rattled.
This thing fractured for a number of reasons, one of which had to do with corrosion and hydrogen. The critical flaw size of this steel should have been about 12 to 15 inches. It was a very good nickel-chrome-moly-vanadium, same type of steel you make great big generator rotors out of, that are four feet in diameter and weigh 300 tons. This thing weighed 200 tons. Very high quality steel, supposedly. And water cooled jacket.
But what they didn't do — it's actually not a bad example for material, stress, and environment. Each one of these things conspired to change the critical flaw size from about 15 inches of what it should be. You'd like it to be 15 inches so it'll actually crack all the way through and leak before it breaks. That's what we call in fracture mechanics, leak before break. You'd like to have in a submarine, will leak before it breaks, instead of like a Liberty ship. It's good to know — you can try to plug the leak, and hopefully the crack won't grow very quickly or very far. And that's why we like high toughness hulls for submarines.
But the material turned out to be somewhat deficient. They did a heat treatment on this, and typically on a big forging like this you will have what they call a prolongation. You actually forge an extra little piece on the end of it, that's part of it, you cut it off and you do all your mechanical test specimens on this prolongation. That's standard technology. It turns out the first heat treatment, they measure the impact toughness and tensile strength, it didn't meet the impact toughness. The heat treatment wasn't very good — this first one they'd ever built, the only one they really ever wanted to build initially anyway. So they decided they had to renormalize it, and so they did another heat treatment, but they didn't have a prolongation, so they just stuck a little piece in the furnace next to it. They tested that. Well the little piece is not going to cool at the same rate as the big piece, duh. They measured the properties and it just barely made spec in terms of impact toughness and strength. They said well, we'll assume that this is the same as that — which is not. But they at least had a quality control spec that they could turn into everyone else and say see, it meets spec.
So it didn't really meet spec. Afterwards we actually did — since the whole thing was a bunch of little pieces, we cut a piece out of here, because this piece had never been heated. The furnace was up in here, and the thing was 15 years old, but this part had never been heated, so this still had the original properties. We found it still didn't meet spec. This little piece may have met spec, but that one didn't. So the material was deficient — you can shrink this circle. The stress: they had calculated the stresses at steady state when this thing was operating. In steady state when this thing is operating, you should actually have your maximum tensile stress on the inside, depending on where the heat zone is, but it was on the inside. In fact, the crack had started on the outside.
I remember coming back from that first day of going up there, or actually the second day when I actually got to see the piece, I went back and told our administrative officers, beware the transient thermal stresses. This thing had been heating up for about an hour and a half, and something this thick — when you go through my fusion welding course I'll tell you how to calculate through the Einstein formula the rate of heating — but this was still in the transient, it was still being heated through the thickness. It turns out that was going to create tensile residual stresses on the outside right at that location, because that's a stress concentration. So you can increase the size of the material because it had worse properties, the stress was larger so you can increase the size of that.
And the last thing is, they had the world's largest water treatment company developing the water chemistry control for the cooling water on the outside of this, and they were using phosphates, and it turns out it was corroding the whole thing. They went and looked at a sister thing that had the same stuff, and it looked like a Dalmatian, with all these corrosion pits all over the surface from the water chemistry. So you can take your environment and blow that up, and so all of a sudden you have hydrogen brittleness, stress corrosion cracking. The critical flaw size dropped from 15 inches, to — I measured half an inch. It had a fatigue crack in here half an inch long, that was the initiating crack that caused the whole thing to blow up.
Fortunately, to get a new one would take about three years lead time. And every large aircraft engine in the western world had to go through this furnace. We would have had a problem. Pratt Whitney, General Electric, Rolls-Royce would have had to quit making aircraft engines for Boeing and people like that, except they had another one.