WM_S2014_12

Nothing really matters too much to just show this, but the guy made a weld that's glycerin, that he just dropped the weld in, and you come back a few minutes later and you can see the hydrogen bubbling out of the weld. So this is sort of like my Sprite okay, where you saw the bubbles, except this is in glycerin so it retains the bubbles, they don't pop to the surface within a fraction of a second. And so you can see, um, they're showing different types of electrodes. This is an E7018 perhaps okay electrode. They have, they're not telling you the strength level. There's 8018s and 9018s and 11018s for different yield stresses. But you can see the hydrogen bubbling out just like the CO2 coming out of the Sprite okay. The whole thing is probably, well, I wouldn't have sent the link to you if I didn't think it was worth looking at. But anybody have any questions? It really does bubble out, and it comes out very quickly, and if you come back a couple hours later there will still be a few bubbles there, okay, because the glycerin traps them. Any questions on that? If not I'm going to go back to my regular mode.

I think I explained last term, to those of you that were in class, that the reason I use this is because I have no idea what I'm going to talk about today. Depends on what questions you ask. So we've talked about faults in fusion welds and how hydrogen can cause cracking, and you get what are called — blow it up a little bit — under-bead cracks, toe cracks. These are under-bead cracks over here. This is a toe crack, that usually the toe cracks right at the toe of the weld okay. Root cracks coming up from the bottom, wherever you have, we have a change in section you have a stress concentration. Remember, the higher the stress, the higher the hardness, the higher the hydrogen level, the greater the propensity. And we use Venn diagrams to show this okay.

Now, I also showed you this before, and this is one of the only places you'll actually — this comes out of the Welding Institute book on welding steels without hydrogen cracking. It talks about the weld hydrogen level in milliliters per hundred grams, which is almost the same as parts per million, within two or three percent I think it works out. And this is the potential hydrogen. Potentially these stick electrodes, these things that we've passed around before, can give you, well, they'll have anywhere from 0.2 to 1 percent moisture or more, and they can give you very high levels of hydrogen potentially in the weld, 100 ppm or more. But in fact much of it goes away very quickly in the welding process, and you may end up with only 4 ppm. And in fact we're going to see that what's called extra low hydrogen welding electrodes tend to be, they will guarantee that when you first take them out of their hermetically sealed can they'll be 4 ppm or less. Cellulosic electrodes or basic electrodes can give you 30 ppm hydrogen in the weld okay.

And what I've done here is, if you actually deconvolute a lot of the data in the literature, you can come up with something like this. If I have a steel of 70 ksi tensile strength, and that's mild steel okay, that's what your little red Radio Flyer wagon that you pulled around as a child, or your typical cheap bicycle, is made out of, 70 ksi steel. You can tolerate 30 ppm and you've got to really use no hydrogen control at all to get enough hydrogen in there to crack it, and you have to have a high enough stress. Well if I'm welding on sheet metal like the Radio Flyer wagon, or the tubes for the bicycle, it's pretty hard to get hydrogen cracking. You don't have enough restraint. You've got to have something that's big and thick, something like a piece of submarine steel that I passed around before. And now I will pass around — this was a piece of 10-inch forging, cast forging. It's a little heavy, I'll pass it around with a rag, the edges are not that sharp. [Tom hands the forging sample around the class.] But there's probably 400 passes in there, and I had put that up on the screen at the last class I think.

That's got a lot of residual stresses, because every one of those passes shrinks about two and a half percent as it solidifies, and then it shrinks some more as it cools down. And when it's restrained, because the rest of this thing weighs 60 tons, there's nowhere for it to shrink to. So all that goes into strain. Some of it's yielded, yielding, but you get yield-level residual stresses. The yield-level residual stresses are the highest stresses you're going to get. In some cases in something like that you can actually get higher than yield because you get triaxial stresses, but we won't get into that okay. But your stress can get as high in thick sections as whatever the tensile strength of the material is. So as your essential stress circle can get bigger and bigger as you go down and strength, or up in strength, you can tolerate less and less hydrogen. Because as this circle gets bigger this one better gets smaller in order to avoid that right. Or you need to control your hardness, it means you start with a steel with a low carbon, because remember hardness is proportional to carbon, right.

So we've spent a lot more time than I thought I might talking about the basics of steel metallurgy, to get to the point where we can now start talking about welding procedures. But before I get there I want to put this up. It turns out we just live in this perfect world at 20 degrees centigrade. And if I look at the tendency to form a crack, which is notch tensile strength — the lower the notched tensile strength, the easier it is to form a crack — and I look at a hydrogen-free specimen, it will have a fairly high strength. If I look at a hydrogen-embrittled specimen, I could have up to 75 percent reduction in strength because of the hydrogen. So a steel that might have 100 ksi strength might crack at 25 ksi okay. However, if I go up in temperature, it takes more and more stress. Why? Because as I go up in temperature that hydrogen comes out faster, it doesn't have time to migrate to that crack tip, it diffuses out before it does its damage.

Um, right, question is — and I have realized by watching some of the tapes, because I've got the mic on, people can't hear your question. So the question is, how do you know what temperature you go to? Well, people have learned that about an hour at 375, or three hours or four hours depending. It turns out the higher the strength of the steel, the lower you have to get in hydrogen, and so that means the longer the time. And the experience is, if you're at this 200 ksi, 700 mega, or 1.4 gigapascal strength level, you're going to need 23 hours, a full day, at 375 okay. Why is that? Well, there are hydrogen traps in the steel. The hydrogen not only will go to crack tips, it will go to inclusions, little dirt particles in the steel. And in fact there's a guy out at Colorado School of Mines who was elected to the National Academy because he showed that if you have slightly dirty steel it's better than if you have very clean steel in terms of hydrogen embrittlement okay. You want to have some inclusions to trap the hydrogen so it doesn't go all to the bad places. You have hydrogen traps, and people have studied what temperature the hydrogen is released from the traps. And so typically 375 Fahrenheit is the number that people use. There's still hydrogen comes out above that, but in general it's a time-temperature diffusion thing. What's the concentration up here? And these types of steels, one hour at 375 is plenty. In these steels you might need a day at 375 to get to these levels at the higher strength levels. There's not a fixed number, but I'm going to show you something that allows you to calculate all this.

Well, any — the question is, does this 375 cause warping? No. Steel to start, relieve residual stresses, or introduce warping in steel, you've got to be above 900 degrees Fahrenheit. Typically we stress-relieve steels in the 1100 to 1200 degree Fahrenheit range, and that's where you're actually starting to see the steel creep, like Silly Putty if you will. You can't see it creep like Silly Putty, but it's creeping enough to relieve the strains that are locked in. We talk about residual stresses, they're really locked-in strains okay. And if I have a piece of steel — let me take this and say this is a piece of steel — and I machine away the top layer, if it's got residual stresses, it will warp because I took away the residual stresses here. Might warp down, it might warp up. But in fact that type of distortion — if I take a heat-treated part, whether it's steel or aluminum, that's been heat-treated, I start machining it and it walks on me. It can move ten, fifteen thousandths. I'm trying to do precision machining on a heat-treated metal part, I got problems okay. It will move on me, and so you have to stress-relieve the part before you go to do some machining.

In fact a number of years ago it was a big problem in the lithography business for making computer chips. They needed to machine things within tolerances of like five microns okay. That's 0.0— 0.2 mils or 0.002 inches okay, five microns. You can't do that even if you stress-relieve the steel or the aluminum, it will still move that much if you machine off a quarter of the thickness out of one-inch plate. It will — you can machine it and you come back and you measure it and it's moved. So you have to machine it again okay, and then it moves a little bit more okay. It does sort of start tapering off, but you have to be careful. You try to grind something flat and you start getting very precise. In fact, if we want to talk precision, does anybody know what the typical manufacturing tolerances are for machining steel or whatever in parts per million? So if I have something one inch long, and I want to machine — I just got a big bar of steel and I want to turn it to one inch diameter — what's the typical tolerance that I could expect a machinist to come back with? Without — a novice machinist?

Plus minus one thou. It might be for you and me it might be plus or minus five, one five thousandth. But a guy who's a machinist, and this is his trade, he ought to be able to hold one thousandth okay. Now a guy who's been a machinist for 20 years, he ought to be able to hold two tenths okay, five microns out of one inch. Or maybe he can hold a tenth. But boy, he's measuring things, he's making sure that there's no stress on it when he clamps it up and everything else. That's per inch. If I've got something 10 inches long, the tolerance is not 5 microns, it's 50 microns, which is 2 thousandths of an inch. So telling you that, what is the most precise manufacturing in the world in parts per million, or parts per thousand, of what tolerance you can hold? Well first of all, you start getting down to 10 to the minus 5, you're now getting into thermal expansion. When you're removing that material or adding that material, you've got to know what temperature it is, because the typical coefficient of thermal expansion is 10 to the minus 5 per degree C. One degree C change in temperature and you're going to have 10 parts per million of movement okay.

So when people — I've worked on some helicopter bearings on some of the world's largest helicopters. 42-inch bearing, half-inch raceway, what were they, quarter-inch balls or something like that. And it was 42 inches wide. It was supposed to be 42 inches plus or minus 5 thousandths. Well that's like one part in 10,000. So what do they put them — the mic, which is just a big bar mic to measure this 42 inches, they put it right on the bed that was filled with oil while they were machining, so it'd be at this exact same temperature. Because the measuring tool has to be at the same temperature as the part you're measuring, or you're going to get an inaccurate measurement right. So they bathe the measuring tool in the same cooling bath and lubricating bath they did the other stuff. So again I'm going to ask you, what is the world's most precise manufacturing, in parts per thousand or parts per million of accuracy tolerance?

Well we can do either one okay. The industry — I'll tell you the secret is, it's the biggest part and it's the smallest part okay. If you go to shipbuilding, you've got to hold a quarter of an inch over a hundred yards. Start figuring that out. It's about one part in twenty or thirty thousand okay. It's much finer than some guy trying to mill or machine a bar to one inch plus or minus ten thousandths of an inch okay. Or ten thousandths of an inch um, a tenth, one tenth of an inch okay. It's much better than ten minus four it's — or maybe it's, I can't remember, you can calculate it. Quarter inch over — because you've got to put the steel plates together and they can't be put together like that, they have to go together like that, right. The other one is, it turns out it used to be — ten years ago shipbuilding was by far the most precise. Turns out semiconductors, when I'm getting down to 0.14 microns or, no, 0.14, 14 nanometers over a centimeter, you start figuring that out, and you're getting to just a little bit better than shipbuilding now. You have to go on the extremes, very small or very large, to get the most precise tolerance. In a shipyard, they have to worry about when the sun comes out from behind the clouds, because the heat on that side of the ship will cause the whole ship to bow when they're building the ship, and the distortion will — yes?

No, it's, what tolerance and dimension can you hold. If your spec is one inch exactly, what's your tolerance on that? You ought to be able to give it to a novice machinist and he ought to be able to hold one thousandth of an inch, so it's 1.001 or 0.999 to 1.001. That's one part per thousand okay. If he's holding a tenth of a thousandth, point triple zero one plus or minus, that's one part in ten thousand okay. But you start thinking about a quarter of an inch over a hundred yards, you can calculate what parts per thousand is, and it's going to be — I don't remember, one in twenty thousand, one — it's more than one in twenty thousand, it's one in thirty thousand or one in a hundred thousand. If you start getting down to nanometers of a silicon chip, those dimensions over the length of the chip, which is about a centimeter, you're also getting into the one part in tens of thousands or hundreds of thousands okay.

What — on the chips you don't etch the ships okay. We've got chips on one end, we've got ships on the other end right okay. The chips, they have to — they use photolithography, although now they're using high ultraviolet lithography, because you get into the problem of the wavelength of light okay, which I mean, these dimensions are smaller than the wavelength of just visible light. So they're using ultraviolet, they've talked about using electron beam lithography and things. Some of them, I believe they actually pattern one part of the chip and then they match it to another and match it to another because over a centimeter you can't hold that type of tolerance okay. But there's a lot of things, and I'm not an expert in that area, but at one time this was supposed to be a 20-billion-dollar problem for the semiconductor industry about 10 or 15 years ago. Your dad worked on some of that. Yeah.

So anyway, so the point here, getting back to welding, unless there's other questions, is, you can get rid of the hydrogen at higher temperatures. In room temperature you can lock it so it doesn't diffuse around to do its damage at lower temperatures. The absolutely worst temperature is right around room temperature. Lucky us okay, that's where we live, and that's where hydrogen is the worst. This is just a schematic. I do have a plot that actually gives you hard data. This is not from welding, it's from hydrogen sulfide stress corrosion cracking. Yesterday one of you asked the question about, what's stress corrosion cracking. And I got up here and said, well, it's the same Venn diagram: stress, a susceptible microstructure — in our case it's the strength of the steel, but stress corrosion cracking, do you have carbides along the grain boundaries of the stainless steel that it can etch and eat away because of the difference in composition — and do you have chlorides that are going to eat that away. So at a higher level — this at a lower level is hydrogen stress corrosion cracking, but we don't usually call it that, sometimes people do. But again, hydrogen stress corrosion cracking would occur at the end of the cathode, whereas most corrosion occurs at the anode.

In any case, this is time to failure versus temperature. And this is real data points in hydrogen sulfide, and it shows that the minimum time to failure is right around 25 degrees centigrade, and lo and behold it's just over an hour. So why do you have to get into the furnace within an hour? Well if you keep it hot you can wait five hours. If you keep it cold you could wait five hours. Room temperature is where you've got to be the fastest and get it in that furnace to bake it out. And I showed you the bubbling — those of you who weren't here at the beginning of class, I showed the video that Jerry sent you a link to. It shows hydrogen coming out of a steel weld by just dunking it in glycerin, and you can see the bubbles coming out just like the CO2 in Sprite that I showed you before okay. Any questions on that stuff?

Okay, I gave you a handout a few days ago — last week was a few days ago, I think it was last Thursday, a week ago. This is out of Stout and Doty's book. And let's say that someone — and I gave you the example that I used to work with this Simpson Gumpertz and Heger firm. These were three MIT civil engineers back in the 1950s. They started their own company, it's now got five or six hundred engineers, several offices around the —

He's a welding engineer, and he's walking through the shop on a Monday morning. And you can tell if something's supposed to be preheated to 100 degrees or 150 degrees. You walk past it, big — you know, several tons of steel and it's 150 degrees, you can feel it. In the winter, kind of sit there like the winter, you've got to go up there, because these shops aren't heated. And in the summer, you kind of walk away from it because it can get pretty hot in these shops in the summer when you start preheating tons and tons of steel. In any case, he goes up and he knew the procedure, he knew it was supposed to be preheated. I don't remember the temperature, and you put your hand up — if it's 200 degrees don't put your hand on it okay. He just, you know, he put his hand on it, it was cold. And he asked the welder, he says, I thought you were supposed to preheat this. Oh, we did that on Friday okay. Oh good. This guy obviously didn't understand the principle of preheat. But anyway, so you have to — I have to specify a preheat. I have to tell them what type of electrode, because they will probably in the field be using the stick electrode because it's so easy. Now more and more they're using gas metal arc, which is a semi-automatic process. But whatever it is, I have to tell them what types of electrodes they can use. Then it's going to tell me, do they have to do a post-weld heat treatment, and it will tell me for that steel at what temperature peening may be necessary. That great big thick thing — you could, I think it took them three days to weld that okay, 24 hours a day okay, to put in 400 passes. Yeah?

If you do too much — that's the interesting thing about peening. A little bit of peening is good, it creates surface residual stresses, compressive stresses on the surface. If you over-peen, you're going to start getting like the peened head of a ball-peen hammer, that starts introducing cracks. And so over-peening can reduce fatigue strength. Underpeening will not — well, optimum peening will give you much better fatigue strength. Underpeening can give you low fatigue strength, over-peening can give you low fatigue strength.

To tell you a story on peening — this was 1969, I was between my freshman and sophomore year. I was working at the Naval Air Rework Facility in Norfolk, Virginia rebuilding jet engines. I wasn't rebuilding them, I was an engineering intern okay. And the Vietnam War was going on, and we were getting engines back from Vietnam, and we had — out in the shop they had to rebuild them. Well one of them was all ready to be shipped, all reassembled, ready to go back to Southeast Asia. And they did a final inspection. In one of the inlet fans, right there — you're looking at the front of the engine — had a little crack in it, in the titanium case.

Well to take that off you've got to disassemble half the engine and put a new case on, and they were in a hurry for these engines. And so my boss says, okay, we're going to peen it. Well you can't shot-peen in the engine, because you get little steel balls all over, little BBs all over everywhere, that gets in the engine, it'll destroy the engine, called foreign object damage. So he designed a little tool which the shop made up. It's just a little finger that, on a little pneumatic tool, that go bam, okay, just pound the surface. So I had to go out as the person from the engineering department and witness the testing of this. So I tell the guy okay, bear down like you're comfortable, you know, pushing down on this, some downward force. And he would take a little strip of steel, and it would be like one inch by four inches, and it's called an Almen gauge, A-L-M-E-N. And he would peen the top surface and then we'd measure the bow. The amount of bow tells you the amount of compressive stress you introduced on the surface okay. So I was there. He would peen it, I would time him, he'd try to cover the whole area, and then I would measure the bow. And it was just scatter all over the map okay. We couldn't get any reproducibility by and large on this peening operation.

Bring the data back into the chief engineer and I show it to him. He says oh, okay, use these parameters and go out there on that engine and have him peening that crack. Or actually they first had to weld the crack and then peen it to stress-relieve it. You had to stress-relieve this titanium. So I go out, I witness, I time the peening. The guy does — of course it's not as simple as the Almen gauge which is nice and flat. This is a weld that's got a little hump on it, so he couldn't even peen it the same way. Nonetheless we did it. I come back and I report okay we've done it. He says sign this. I said what's this — and he had already signed it, but he needed two signatures, two engineering signatures. He says, this is certification that we've done this repair on this. I said well, and that it's a good repair? I said, well, what if I — wha— what happens if I sign it? He says, if this plane goes down, you'll be in jail within 24 hours. I said oh, okay. So I signed it okay. I mean, I'm signing — I have professional responsibility that we had done our best job of peening that, and we had done our best job. Whether it was any good or not, well, we don't know. But you know, after 40 years later, 43 years later, I don't think that engine's still running. I think I probably won't go to jail for that. But part of the theme of that story is, sometimes you just have to take risk okay.

But that was thin material, I didn't understand all this at that time but my boss probably did. It's relatively thin material, the residual stresses weren't that bad. The Pratt & Whitney spec said we were supposed to do x, y and z. Well, I forget Pratt Whitney, a bunch of theoretical, you know, physicists sitting up there in Harvard okay. So you just do what you have to do sometimes. But you don't do it without some knowledge. Well, I did at that time anyway. It's a true story. I always think back on, should I have signed that. I actually have come to the conclusion, yeah, I probably should have signed it. I know enough now that I figured it was okay.

So we now want to figure out, well, we've been doing tests for 70, 80 years on trying to understand stress, hydrogen and hardness. One of the tests — we need to reproduce residual stresses. And one of the tests that people use is called the cruciform test. If you look in Stout Doty, which was written back in the 1940s, they would take a plate, they would put another plate on top, and they would weld, fillet weld around the edges. They'd bolt them together initially and weld the fillet weld. You bolt it together so it wouldn't bow up on you and turn into a V-shaped joint on you. So they take a high-strength bolt, bolted together, and they weld it with their electrodes, and they then look at it and see if you get a crack. Very sophisticated test. Another one was the cruciform test. These are kind of like 50, 60 year old tests, where you basically take three plates and you weld them together on the ends, and then you put a fillet weld here, a fillet weld here, fillet weld on the other side. And the fourth one is the one with the highest restraint, where the plates can't move, and that's the one you look for the cracks okay. So you can get with relatively small pieces that are not two inches thick — you can use a half-inch cruciform plate, half-inch thick cruciform plate, you can get yield-level stresses on the type of steel you're interested with whatever your electrode is okay.

Pretty crude tests. Now, when I was a welding engineer at Bethlehem Steel in the mid-70s, we might use — the Japanese liked to use the Tekken test. Was nothing more, you take like a one-inch plate or a half-inch plate, and you cut one of them in cross-section like that. The other one you bevel straight across. So you do a double bevel on one and you do a single bevel on the other. And you put a little weld right in here, and you see if it cracks. Because the Tekken test, there's tremendous residual stress here even though you're only using small plates, small test. Tekken — anybody know what Tekken means in Japanese? Railroad. Okay, this was a railroad test. The Japanese were building lots of railroads, still in the 70s and 80s, for transportation, subways and things like that. And that was the Tekken test. But there are dozens of these types of tests where you try to use a small amount of material and test the weldability of your steel, whatever strength level it is, whatever electrodes you're going to use. You do a quick test and see if you get a crack okay.

I think I told you the other day that if I'm welding 100 ksi steel or 120 ksi steel for a bridge, if I'm welding a 70 ksi steel for a bridge, I don't really have to do much of anything. I'll show you what we have to do. 120 ksi I have to test that steel afterwards to see if I developed any cracks. I can't just weld it to my recipe and assume everything's good. I'm going to have to go inspect it afterwards at the higher strength level. Why? Because I can use almost any electrode with this steel and it won't, you know, with a little bit of care I won't get enough hydrogen. This one I could get cracking. But it requires that I wait 48 hours before I do the test. Because if I tested one hour later I may not have gotten a crack yet okay, and then I could pass it and say no cracks, and three hours later there would be a crack. It's delayed cracking.

So far as that goes, you know, can't tell you how many times people say, wow, we had a crack — we have a crack here, but we know it wasn't hydrogen cracking because we tested it right afterwards. And I say well, how soon did you test? Oh, right afterwards. I said well, how do you know it's not hydrogen? Well, we tested, there's no crack. You know, well, it could have formed afterwards okay, after your one hour that you did your — you have to wait. And the US Navy waits seven days before testing for cracks. So one of the things you do to prevent cracking is you preheat or post-heat the steel. So what I want to do now is hand out an appendix from the Structural Welding Code okay. We're going to spend a little bit of time on this.