§1. Surface ship distortion vs. submarine residual stress [00:02]
When you're about a centimeter thick — and that's the wall thickness of your surface ships, about a centimeter, 3/8 of an inch — you get absolutely the worst distortion. They pick that thickness for good reasons, because you know it'll crawl through before you give it to the Coast Guard. Surface ships have some of the worst distortion going.
On the other hand, submarines have some of the highest residual stresses with big plates. The Boiler and Pressure Vessel Code requires anything you weld for a pressure vessel to be stress relieved if it's more than an inch and a half thick. Up to about an inch you may only get 80 percent of the residual stresses. It depends on the weld sequencing and how much other restraint is around it, whether it's the last side on a six-sided box or the fourth side. You'll have different amounts of restraint. But you can weld-sequence and get around a lot of that.
§2. Thermal stress relief: ovens, on-site furnaces, and why vibratory doesn't work [01:18]
The Boiler and Pressure Vessel Code requires that you stress relieve any welds greater than an inch and a half thick. I won't tell you how thick a submarine hull is, but it is generally thicker than an inch and a half. So how do you stress relieve boilers and pressure vessels, first, before we get to submarines? You anneal them, you heat them. If you went back to Stout and Doty, that other thing I handed out said stress relief, and it said for 8027 steel, desirable 1100 to 1250 degrees Fahrenheit. You put it in a furnace at 1100 to 1250 degrees Fahrenheit for about an hour per inch of thickness, so it heats all the way through, and that softens the steel and the residual stresses will be relieved.
Student: Does it change the microstructure?
You could get reheat cracking that I talked about before. So when you come down from the stress relief, you have cracks to go along with your stress relief. But if you control the composition of the steel — and we know how to do that now — you don't get reheat cracking. It actually usually improves properties if you do it properly. You certainly usually improve toughness, and you improve fatigue resistance. But it's a little pricey to heat it up like that. How do they do it on big vessels?
If you go look up on-site stress relieving on Google, you'll find companies — used to be called Nooter, it's now called Heat Treating and Stress Relief or something, it got bought out by somebody else. So you have a great big vessel, it's going to some oil refinery, you can see they're going to put insulation on it — you have these little lugs, lagging studs on the side. They build themselves a little furnace and they put it on a little railroad track and they just put it into the furnace.
Well, you can't always build a furnace, so sometimes they build the furnace around the vessel. They wrap the whole cylindrical vessel in insulation and heating bars, just like you preheat welds if you've ever seen that in the shipyard. They do the whole vessel and they heat the whole thing up to 1100 degrees for a couple of days. It can take two, three days to do the stress relief. A little pricey, but that's what you have to do to meet the ASME code. They will stress relieve something the size of this room or bigger by building the furnace around it.
It's not easy. You can go to Canada and use the CANMET reactor and do neutron diffraction, except you've got to bring your sample to Canada, and it's not always that easy to bring a pressure vessel to Canada. We know that if you hold that temperature for one hour per inch of thickness and soak it through, and you put thermocouples on there, and you get it up to that temperature, you will soften it enough that it will relieve its own stresses. That's been well proven.
And there are people who are always trying to sell you some other technology to stress relieve. There's something called Bonal Technologies in Columbus, Ohio, and for fifty years they've been trying to tell you they can put all kinds of piezoelectric transducers on there and they can vibrate the vessel at room temperature afterwards and relieve residual stresses. And if you believe that, I've got a bridge in Brooklyn that I'd like to sell you. But they're still in business fifty years later. The U.S. Navy spent over a million dollars trying to prove whether vibratory stress relief works. They don't use it. So what do you think they concluded? The British Welding Institute spent several million dollars for the oil companies trying to verify it — if you could do it, it would cost hundreds of millions of dollars to do stress relief every year on these pressure vessels. You could save a fortune.
Because it's the Holy Grail — let's do something other than heat for stress relief — people are willing to take the chance, and you will sometimes get engineers signing off, we're going to use vibratory stress relief because this salesman told me it works. Just please tell me which pressure vessel you used vibratory on so I can stay away from there. I want to be ten miles from there, because it could fatigue, and it has lousy impact properties, and it's a problem.
A guy just emailed me two weeks ago from an aluminum company, and they're trying to convince this ten-billion-dollar-a-year aluminum company to use vibratory stress relief. He says, what's your opinion, Tom? And I told him sort of like I just told you, in an email. I closed it by saying, I don't think it should be spelled Bonal Technologies, I think it should be Banal Technologies. There's snake oil salesmen all around the world, and nothing says they can't live in Columbus.
§3. Hydrogen in welds: preheat, post-heat, and electrode moisture [07:21]
Stress relief is a problem, it's an expensive thing. Post-weld heat treatment is a stress relief heat treatment. So someone asked about post-heat after class yesterday. You can preheat the weld to try to have the weld hot so that it drives off the hydrogen with temperature. So you can preheat the weld, or you can post-heat the weld, or in some cases if you've got 4340 steel you probably have to do both, because you can't let the hydrogen stay in there on a very high hardness steel. It could crack within minutes.
There's a publication by the British Welding Institute called Welding Steels Without Hydrogen Cracking. You can pass that around. [Tom borrows a copy from a student.] There's a plot in it — Frank Coe, who's probably passed away now, developed this plot in the 1970s when he wrote the first edition of the book. At the bottom is weld hydrogen, milliliters per 100 grams of deposit, so you can think of this as parts per million. The highest level you might expect in the liquid is like 30 parts per million. If you use rutile electrodes, which is like a 7014 or 7020 — I'd have to go back and look which is rutile — you can have very high hydrogen.
You can have very high hydrogen if you use a basic electrode that's been dried at 100 to 150°C, like this 7018 that I have which is not dried. You can have a wide range of moisture in the flux coating. Typically really good electrodes are 0.4 percent moisture, very extra dry electrodes are 0.2, and so you could easily get 5 or 6 parts per million hydrogen in typical mild steel welding.
You can use flux-cored processes which can give you very high hydrogen or very low hydrogen depending on various things. You can use gas-shielded welding where you use argon and CO2 and you can get down very low. Most of the submarine welding, actually most of the shipyard welding now, is either submerged arc or gas-shielded welding. We don't use as much stick, but we probably still use it in places that are hard to get to.
What does that translate to? A low-strength steel, just a 35 ksi yield, which might be a 70 ksi tensile steel, you probably have to have 30 parts per million to crack that. It's very difficult to get hydrogen cracking on a bridge steel or a building steel. You've got to really try. I've tried — I've soaked those electrodes in water and welded with them and I can't get a hydrogen crack in A36 steel. Maybe if I had higher restraint or something.
At 120 ksi, which might be an HSLA-80 or an HY-80, it might take 10 parts per million. At 10 parts per million I could use stick electrodes — 7018s, or actually 8018s, or 11018s, which is what they typically use in shipyards, and I can weld HY-80. HY-100, well it's going to be in the five or six, and I'd better make sure I really bake those electrodes.
§4. Counting electrode stubs in the shipyard [11:57]
If you've been in a shipyard, they actually count the number of electrodes. You go to the stores at the shipyard, and they will give you 20 electrodes, depending on where you are, because that's what you might be able to weld in the next hour. And when you come back to get another 20 electrodes the next hour, you have to bring the stubs with you. They count the stubs. If you checked out 20, you've got to bring back 20 stubs. If you left a stub out there — they have had pressure vessels and pressure vessel shops blow up because some guy left a three-inch stub on the ground and some other guy came along two days later and welded with it and created a hydrogen crack, and the vessel blew up later.
If you're being audited by the ASME code inspectors and they see a stub on the ground, you lose your license. It's serious. In the Navy shipyard they count the stubs when you come back to get extra electrodes. And they time you — you can't be out there for four hours. If you get sick, those electrodes go in the trash that you didn't use. Or they've got to go back to the baking oven. They have little hot rod pouches, insulated things that you keep the electrodes in to keep them warm. So we do a lot to make sure we have low-moisture electrodes if we have flux-shielded electrodes. If we do shielding gas, we make sure the gas is dry. Unless you're a wet welder, you're just full of hydrogen anyway. But you have to do something.
§5. The Seawolf hydrogen cracks and the dirty wire conclusion [13:44]
What happened with the Seawolf: they had weld metal that was on the high side of the chemistry, had enough hardenability, they were at about 150 ksi yield, which would be more like 180 ksi tensile, two to three parts per million hydrogen — and guess what, you have a hard time achieving that even with the best control in the shipyard. That's why they got hydrogen cracking. They thought they were welding HY-100 and they should have been around 130 tensile, but in fact the weld metal was 180 tensile because it had lots of extra alloying, was on the high side of the chemistry.
The other problem with the Seawolf — they brought in six of us, two individuals and four companies, brought in by Admiral Fireball [Bruce DeMars?], who was Chief Engineer in the Navy. He had designed the Seawolf when he was a captain. And so now they have this huge problem. Eighteen percent of the hull built — this was 1991 or '92 — and they come back and report, we've got to rip it all out, we're finding these little micro cracks in the weld metal. It wasn't in the heat-affected zone. Most of these cracks, hydrogen cracks, occur in the heat-affected zone. In fact, they're often called under-bead cracks. If you go to the same book on welding steel without hydrogen cracking, they'll show you pictures, and the cracks are in the heat-affected zone, and they call them under-bead cracks because they don't penetrate the surface — pretty bad, you can't even find it with magnetic, you could find it with ultrasonics. But in 1992 the Navy hadn't qualified ultrasonics for the shipyard. This one is a toe crack, did penetrate the surface, but in any case, these are under-bead cracks, and it's a problem.
The way they discovered it — in the weld metal, instead of having long cracks that might be a quarter of an inch or half an inch long — everything in the shipyard is designed to look for eighth-inch cracks or larger because the probability of detection is reasonably good if you have an eighth-inch crack. The Air Force has done all kinds of studies on non-destructive testing, and they can find with greater than 90 percent probability things that are greater than an eighth of an inch in size. These things were a sixteenth of an inch in size in the weld metal, and they were all through it because the weld metal had a finer structure.
They'd been doing their own tests and their radiography, but none of the tests were designed to find cracks that small. A grinder was grinding the outside hull — you think about submarines, you don't even like to have weld reinforcement on the outside because it causes ripples in the water, which with sonar you can find, makes more noise than a whale. The guy was grinding the weld smooth, and he happened to notice that the grinding swarf — the little powder, we call it swarf, very good word if you play Scrabble, S-W-A-R-F, not a lot of your compatriots will know what it means — the swarf was actually aligning in these little sixteenth-inch lines. He was doing a magnetic particle test with the grinding swarf. He was making magnetite out of steel, and it was lining up, and he said I've never seen that before, so he asked some engineer. They went and they did a more complete, more sensitive test. Eighteen percent of the hull was riddled with these little cracks. And the decision was to rip it out.
They hired two individuals — myself and a guy who had helped invent gas metal arc welding, he used to be at Airco, which is now part of Lincoln Electric — and they hired Lincoln Electric as a company, and a couple of other companies, four companies. We were supposed to advise Admiral Fireball and his staff on what Electric Boat was doing — was it the right thing to fix this? The other individual talked to the press. We were told don't talk to the press. So he was booted out. There were only five of us left.
We had to write a report. I wrote a report, and there was almost no data, just a little bit of data. They had measured hydrogen. They were using three different diameter welding wires — 3/32, eighth inch, and I think sixteenth, depending on the thickness of what they were doing. They had done all kinds of hydrogen tests on the wires — there'll be some residual high hydrogen. I plotted this, and I found the highest hydrogen in the smallest wire, next highest in the middle, and least in the largest.
What was the conclusion? These ranges actually overlapped if you looked at all the individual measurements they had made. Can anyone guess what my conclusion was?
Student: [inaudible]
Well, we knew it was hydrogen cracking. My conclusion was it was dirty wire. This is milliliters hydrogen per 100 grams of weld — rutile electrodes high hydrogen, basic which is 7018, that's what they're trying to weld with. Dirty wire can give you regions higher in hydrogen than you would like to see in 180 ksi weld metal.
Now why would they have dirty wire as a function of diameter? Which one has the highest surface-to-volume ratio? Surface-to-volume goes as one over r. The volume goes as pi r squared times l, and the surface goes as 2 pi r times l. So everything's proportional to one over r. Smaller diameter wire, more surface to volume. So dirty wire has more surface, and what's on the surface? Drawing lubricant. They had to draw that wire, and they had to use a lubricant which is made of grease.
When I wrote my report, this was my conclusion. I had to meet with two captains — they introduced them to me, I can't remember their names, but one owns the boats and one operates them, for all the submarines. They told me, yours was the only report that said anything, gave us any conclusion. Everyone else just spat back the data they saw. The Navy could tell the data, that was their data, Electric Boat's data. But I actually drew a conclusion. I analyzed the data and drew a conclusion that this could have been dirty welding wire. Lincoln Electric was not happy that I was impugning their wire and their cleaning procedures. But since we weren't all supposed to know what other people's conclusions were, they didn't know it was me. I don't think they did. They may have found out.
§6. Don't derate the existing HY-100 sections [22:21]
The question they asked me, which was an interesting question, was: okay, we have the Seawolf and we're going to rebuild that, but we have two ships out there that we've already built with an HY-100 section. They actually took baby steps in the submarine navy. In some of the prior 688-class attack subs, rather than building all HY-80, they used same thickness and everything but they put in HY-100, which was what they were building the Seawolf out of, just to get the experience and see if they would have any fabrication problems with the HY-100 as opposed to HY-80. So they built a 20 or 30 or 40 foot long section, I don't know.
They said, we've got a couple of subs out there that have HY-100 in them in the hull, what should we do? We have to derate them. My answer was actually what they wanted to hear — they wanted the truth — but the truth was, no, you don't have to derate them. You may have to do a little more inspection when they come back in port for inspections, but those little sixteenth-of-an-inch-long cracks, how fast are they going to grow? They're starting right down here and they've got plenty of time between inspections, between overhauls.
When they come back, increase your inspection budget on those welds and see if the cracks are growing. You probably can't even find the cracks for 10 or 20 years. They may even go the whole life of the ship. Until you put it in mothballs, or send it — you don't send those to Bangladesh, do you. You used to blow them up off of Newport News, or Nags Head — that's actually sort of the story in Hunt for Red October, right. Patrick Henry. They used to scuttle them and bury them in the ocean with the reactor too. That was before some people decided that wasn't a good thing to do.
I had to explain fracture mechanics to these two captains — you don't have to worry about it. You don't have to derate them, but you might want to increase your inspection budget in the shipyard, because those cracks could not grow fast enough. Did you know that virtually every airplane that's flying is full of cracks?
There's a story about the Air Force wanting to do an acoustic emission test. They were going to put an acoustic emission monitor on an aircraft, and they were going to introduce a flaw, a crack, into one of the structures, and they wanted to see if the acoustic emission could pick this up as part of a research project. Since they were going to introduce a crack that would make the plane unairworthy, they had to get a general to sign off that they could fly an airplane with a crack in it, and he refused. I'm not going to fly any airplane with a crack in it. Excuse me, General, but every one of them out there has a crack, and you just don't know where they are. This one at least we know where it is and we're going to monitor it. They had to fly it on an Australian jet, because this Air Force general — not gonna fly airplanes with cracks. We have cracks all through planes and submarines and other places. Steel is very forgiving material. So they didn't have to derate it.
It did cost a lot of money and Congress wasn't very happy, and there's other stories that follow on for the next ten years from that.
§7. Mechanical stress relief and the first deep dive [26:31]
Getting back to where I was — we were talking about how do you stress relieve a submarine. It's got a wall thickness, it's going to have yield-level residual stresses, and you don't like residual stresses because they can run into fatigue cracks and everything else, poor toughness. We do stress relieve submarines. Does anyone know how we stress relieve the submarine? We don't stick the whole thing in an oven.
Student: [inaudible]
Nope, we don't do that either, because if you do that, you're going to have a region between the cold and the hot that's in this region where we have all kinds of temper embrittlement and other things, which would just destroy the properties of that part of the steel, in the heat-affected zone of the heat treatment. So local stress relief has to be done. We do local stress relief sometimes. So that's a good thought, but that's not what we do on submarines. Not every type of stress relief is thermal. I can do mechanical stress relief.
Every aircraft wing on every Boeing or anyone else's jet, when they make the aluminum plate in Davenport, Iowa — it could be four-inch, six-inch thick plate, and they quench it and heat treat it to get high strength — they have terrible residual stresses in that. And if you made that wing out of that material with those residual stresses, it would fatigue within hundreds of hours of operation. So they obviously don't do that. How do they stress relieve it? I've been to the room — it wasn't in operation, but they take that whole plate, that could be 50 feet long and four inches thick, and they have this three-million or ten-million pound machine, great hydraulic jaws, and they grab it and they stretch it three percent. They plastically deform it, and that relieves the residual stresses. So many of your aluminum alloys are stress relieved by adding mechanical stretching plasticity.
When does the submarine see plasticity in its welds? The very first deep dive. And who gets to go on that dive? All the management of the shipyard. One of the best quality control techniques I've ever heard is, you send the managers on the first deep dive. They have to go — not all of them, but most of them, and I don't know if they draw straws to see who takes the chance, but they have real incentive to make sure, in the submarine shipyard, that they do good quality control, at least on the hull welds. Because they get to go on the first deep dive, and on that first deep dive you mechanically stress relieve the welds. So they do stress relieve them.
It might be a long story, but it sort of tells you a lot of principles about stress relief and management. There actually are some morals to these stories. You may have to hunt for them, but they're there somewhere. Why don't we take a break. Come back at 8:45.
§8. Apprenticeship at JG Sylvester and Associates [31:31]
So a couple of you guys are welders. How much of this stuff did you already know about steel?
Student: Not much, but I know a little bit about hydrogen-induced cracking, just because that was something we were very interested in.
Yeah, but you didn't all know the scientific reasons or the principles behind it. I had you try to read portions of Easterling. It's out of print, and it's like $180 if you go look on Amazon. You can buy a used copy, but it's pretty pricey. It's actually a pretty good book, but it helps if you're already a metallurgist. Sometimes I feel like I'm talking down to you, particularly to the two of you that are welders, but I guess I'm not really.
How much welding do you actually do? You're actually the supervisor of the divers.
Student: The Navy in general? Like zero welding. Maybe never. Except in very specific cases, and that's only really for practice. We contract.
But you contract it to certified welders. It's too expensive.
One of the ways I learned failure analysis was — when I came back in 1976 from Bethlehem Steel, I had to take a one-third cut in pay to be a faculty member. At that point I had three children, and maybe a fourth one by the time I bought my house in '78. I had to buy it because my mother-in-law couldn't live alone anymore, and we had to move her in. She lived with us for 17 years. But I couldn't afford my house, and I never knew where the mortgage payment was coming from more than two months ahead, and I had to rely on some consulting to do this.
I had one consulting contract with the gold company down south of here in Attleboro, but I was still looking for a little more. A guy down south of here, Bruce Sylvester — his father had worked for Factory Mutual as a metallurgist and then had left Factory Mutual Research to start his own consulting firm, which did non-destructive testing. His chemist was the guy who used to do chemical analyses when the Navy had a Navy shipyard here in Boston, and when they closed the shipyard, that guy went to work for Sylvester, at JG Sylvester and Associates.
JG Sylvester was a pretty good metallurgist, so far as I can tell, but he died when he was about 50, just had a heart attack and died. He left the business to his sons — his wife owned it, so their mother owned the business. I guess the Sylvester sons, the three of them, one of them had a couple of years of college or something, that was probably Bruce. The others may have graduated from high school, but these guys were not rocket scientists.
Bruce wanted to get into the diving business. He loved to dive. They were doing magnetic particle and x-rays. He wanted to get into two areas. He wanted to get into doing nuclear work, failure analysis on nuclear stuff, and he also wanted to get into the diving business. He actually bartered for — Mass Tank, which builds a lot of storage tanks, built him a 10-foot-high diving tank, 20 feet diameter, in the back of their facility. They had all this diving equipment, and they kind of designed their own ultrasonic stuff for underwater, because they didn't want to buy the commercial stuff, it's too expensive, so they kind of made up their own stuff.
Bruce was really trying to push for this. The reason Bruce was looking for a metallurgist — he had moved another guy, Ken Baker, up to office manager, and his metallurgist had been Dave Holt, who had taught me. Dave Holt had been a junior faculty member here, didn't get tenure, and decided to go into the ministry when he didn't get tenure at MIT. He was working for ten bucks an hour doing the metallurgical failure analysis for Sylvester.
None of us knew until about three years later, but Bruce Sylvester was making fake invoices, submitting them to the bank to get loans from the bank based on his accounts receivable, but this was for work that never existed. Finally about 1983 they caught up with Bruce. The bank called. Bruce sort of disappeared. His mother was stuck having to mortgage her home. Bruce had embezzled half a million dollars from his mother. Sylvester still existed.