WM_S2014_20

This has is not necessarily anything to do with anything that we've been talking about, but uh I was looking through a um a journal basically, and I noticed they had microprocessor transistor counts 1971 to 2011 in Moore's Law. Most people know Moore's Law is that the number of uh transistors on a chip increases or doubles every one and a half to two years, whatever the slope of this line is. It's a log, a semi-log scale.

However, they've redefined Moore's Law, because Moore's Law actually pooped out around 2002 or 2005, because you really can't get much smaller on your feature sizes. We're down to like 0.12 nanometers or something like that now okay, and you know you can't get smaller than an atom. In fact, you can't get smaller than a few thousand atoms. So in any case, what they've done is they've red— redefined the wheel okay.

And um so down here we had a 404 Intel chip, here's an 8086 right here. This is what they put in the first IBM PC, and 8286 was in the early '80s. I remember my son, who just turned forty, was all excited about buying an 8286 computer. He was saving his money when he's a teenager uh to buy an 8286. I told him to save his money, I went out and bought a Mac. Um anyway, um and I've never regretted it since.

Anyway, what they've done is they— well, they didn't really tell you that somewhere here between the Pentium and the Pentium 2, the Pentium was all on one chip. The Pentium 2, because you couldn't join something more than about one centimeter on a side without it cracking due to residual stresses from the soldering or the brazing— or not soldering— operation you're doing, the Pentium 2 was actually two chips on a chip. I mean the package actually had two computer chips in it. I remember walking through the plant and they showed me that um uh back in the late '90s, okay. So here's the Pentium 2 in the mid to late '90s.

Um and now they're basically saying, well, the number of trans— it's not the number of transistors on a chip, is the number of transistors in a computer. Well that's a little different when I got ten cores, right? Why didn't we just start talking about Crays up here, you know, where we have a whole room full of chips, okay? So this is what we call apples and oranges comparisons, okay. They've redefined Moore's Law so it can keep on going. After all, he's a multi-billion— billionaire, um and so we don't want to call him wrong, okay. So we just redefined the uh the wheel. I just, you know, when you read things you have to look at them with a little, uh a little bit of a jaundiced eye sometimes. That's the only point of that.

Does anybody have any questions on what we've been doing? We've been talking about post weld heat treatment, and um Mike had talked about uh PQRs and WPSs and such things, and I wanted to explain a little bit more in my kind of simple-minded lay person's terms.

The— this codes and standards, whether you're talking aluminum or titanium or whatever, you have to have a WPS, which is a welding procedure specification. Mike gave you a handout on these things, and you can think of this as the recipe how to make the weld. That's a welding procedure specification, it specifies what you need to do. Then there's the procedure qualification record, which is bake the cake in my terminology. And then there is a cook.

So you have to have a welder procedure spec— welding procedure specification, which basically says use this preheat, use this contact angle, if you're using this electrode you got to do a post weld heat treat, you got to do, you know, it's the recipe. The procedure qualification record means someone goes out there and they have a qualified welder make a weld on the material you're using, whether it's aluminum, this particular aluminum alloy or whatever. Um you actually make the weld and you look at it visually, you x-ray it, you do magnetic particle, you see if you break it open, put— do tensile tests, there's a whole set of things in the codes, and you test the weld. So it's a procedure that you're going to qualify, and this is the record to show it passed. The cake was well, was properly done, it wasn't burnt, it wasn't undercooked, it didn't fall apart, it works.

And then you have to have— that welder is not going to weld everything on some huge project. You actually have to test a bunch of welders and see if they can reproduce it, and so we have a welder qualification. Most people call that, and have for decades called that— someone would say "I'm a certified welder," and I used to say there is no such thing. It's a qualified welder under the code. However, in the last 10 to 15 years there is such a thing as a certified welder.

It used to be that every company, if you were building a bridge, you'd have to have your company specification for taking A36 steel, which is garden variety steel plate, and you'd have to have your recipe that your company has. And then you'd have to test it, and this testing might cost $2,000 or $3,000. Then you'd have to qualify your welders, and if that welder was working, if you were American Bridge of US Steel and he went to work for, back in the old days, Bethlehem Steel Erectors okay— which Beth's out of business now, but I mean if you went to work for Brown and Root or somebody else, okay, that does construction— he would have to be retested at their, for that company, cuz his qualification only belonged to that company. He was an employee that was qualified when he was working for that company. He goes somewhere else and he's no longer qualified. Well, of course he's still qualified.

It was until about 15 or 20 years ago the American Welding Society decided, "ah, there's money to be made here." Oh no, that's not what they decided. They decided that they would— they decided that they would simplify things and you could be qual— you could be certified. So back in the old days, I don't know, the recipe could cost you anywhere from $1,000 to $2,000 to $10,000, let's say, to have some welding engineer write up the recipe. Procedure qualification record might cost $5,000 to $10,000 to make the weld, do all the test and show that it's good. And this just on a little test plate, okay. And then the welder qualification, every single welder, it typically would cost about $1,000 to have him make the weld and have it tested okay.

Well, welder certification means that the American Welding Society will now administer the test, probably for a couple of thousand, I don't know exactly what it is, prices have gone up uh over the years, but it's probably a couple of thousand. The welder will make the test at an approved testing facility, they'll submit the results to the American Welding Society, and the American Welding Society will keep a record saying, yeah, this guy made a weld 10 years ago and it was a good weld.

Now there are some recertifications, like on some jobs— nuclear work or something— you may have to be recertified every, or you may have to be requalified, retested every three years, or if you haven't welded for a whole year uh you have to be retested, because welding is actually a skill. You know, it takes a little hand-eye coordination, and so if you haven't— it's like playing a video game. You know, you reach a certain level and then you quit playing and you come back two years later and you're not as good as you used to be because you've lost some of the skill, right? It's the same thing. So there are all kinds of rules about when you have to be recertified, but now you can be a certified welder. You're certified by the American Welding Society that you've passed a qualification test, and the qualification test can be carried from company to company. So you no longer have this sort of restriction that it's going to cost a company $1,000 every time they hire a welder to qualify them. Does that make sense?

Okay, so that has been a fairly successful program and the American Welding Society made a lot more money, okay, whether they intended to or not. I don't think they intended to lose money, but anyway. So if I go to the structural welding code, which we've been holding up— and we could go to the boiler and pressure vessel code, but I want to hold it up because it's two feet long, okay— um there will be a blank sheet, and I handed this out a couple of days ago, of the um welding procedure, whether it's pre-qualified or whether it's qualified by testing okay.

There were and have been for at least 40 years, maybe longer, pre-qualified procedures, PQRs. These guys over here in the welding specification, boiler and pressure vessel code, didn't start getting these until about 10 years ago, when they saw that the American Welding Society was starting to make, you know, inroads on qualifications and PQRs by having sort of ones that universally were adopted.

And here are all the things that go into the recipe. Now, not all these things go into every recipe, and you— I'm not, we're not going to take time, but you have to talk about what metal are you welding on, what's the type, the grade of that metal, is it A36 steel, is it 6061 aluminum, um what's the heat treatment, what's the— anyway there's preheat, the be up here somewhere, probably, post heat— well, technique stream, anyway I don't know where postweld heat treatment is over here, okay. Uh stringer bead or weave bead— are you going with a straight line arc, or can you wander back and forth, changes the heat input and the microstructure of the weld. Uh so there's lots of things you might have to— well, you do have to give something about the weld geometry. So that's the blank form.

Someone comes to me, I'll charge them about a thousand bucks to fill out the sheet, if they give me the weld composition— weld metal the weld composition and the process, and uh and things, not the weld composition but the base metal composition. So here's an actual one that's filled out. ASTM [A36] garden variety steel, carbon steel plate, and it will qualify the well— it tells you the geometry of the weld and the stringer. They don't have to fill in everything, sometimes you just put a blank there, which means it's not required. The code tells you in excruciating detail what needs to be filled in and what doesn't.

Okay, so this is a submerged arc welding process, this is the electrode class, the size of the electrode, the joint detail that was qualified, this is the PQR— or this is the welding procedure, the PQR that qualifies this welding procedure specification. And then I'm not going to put it all up— oh, this is the— these are the pre-qualified welding procedures that you'll find in the welding code that have been there for decades. And they took all that and they reduced it to a half a page okay.

So the detail sort of varies. I gave you a one-page sheet where there's all kinds of detail. Here's a pre-qualified procedure and they simplified everything uh to all kinds of things. Here it's a weld symbol that— uh, Simone, did you talk about weld symbols, or was that Mike? That was Mike, okay. Mike talked about him. Anyway, and then I give you, I think this is about 10 pages. The American Welding Society code committee wrote a welding procedure, but they don't write it on a half a page, they don't write it on one page, it takes them 10 pages to write a procedure. And this is a standard welding procedure and you can buy this from the American Welding Society, probably cost you $500 or something, I don't know. Uh, but they went out and they spent tens of thousands of dollars uh prepared by the committee on welder qual— welding qualification on the direction of the tech, who cares.

Um and this tells you what type of material, what thicknesses, starts out with an E610 cellulosic electrode— this is for pipes— and E718, it says primary pipe applications. And I'm going to let you go through it. Here's their statement of disclaimer, uh here's all the people who served on the committee, there's a forward, and eventually you get into something that starts telling you the same types of things as were on those sheets that I showed you before, but they go on for pages, five or six pages, is talking about all kinds of joint details and things like that. So um it's actually a big business out there, but if you don't do it you end up with bad welds, and even if you do it you could end up with bad welds, or you might be accused of bad welds, okay.

Um and I'm not going to go through this other thing. Anybody have any questions about welding procedures? Like I say, it's a big business and it is the way you maintain quality. Someone asked me in class in the end of class on Friday or whenever it was, and about uh defects in welds, because I think Simone or you or Mike were talking about defects, anyway in the weld[s], and you cannot make a weld that's absolutely perfect.

You want to see— yeah, it doesn't work. All it does, you hit the on button it goes flash at you.

Student: [inaudible]

Yep, it just flashes at me. I mean that doesn't come on. There's no light, there's no bulb that comes on. I got my own projector for right now, but anyway. But okay, fine, but I mean, you know, ordinarily I plug it in, everything's fine, but this morning I don't usually get the flashing on button. Is that— do the smell in—

Student: [inaudible]

What's the smell, I don't know, I'm used to it now.

Student: [inaudible]

Oh, I don't think it's one of these, so it might be— could be that, okay. Well, I'll turn it off then, how about that. It'll turn itself off. Oh okay. Not mercury. Ozone, you got— remember ozone, what they actually sell you, little ozone generators to purify the air. Of course, ozone also gives you a headache, so if anyone's getting a headache, I would advise you to just ignore it. Actually, I mean, so anyway, I'm getting an analyst [?]. Okay, that's what— actually before he came in that's what— and I'm not going to get all upset for him. I actually called him in to let him know that we had— anyway.

Um the quality control— we know we will have defects in the welds, and so what we do, if anomalies— that's right, Mike made the difference between deficiencies or flaws and defects, which most people don't usually— they just kind of throw the words around um any way they want. In the code, in the design, they have a safety factor for the base material. If this was a uh a building that we were— the structural steel for a building— the safety factor on the base material would be 1.67, five-thirds. The weld, the safety factor would be 3.3. If this was a bridge that's fatigue loaded— and Dr. Belmar's talked a little bit about fatigue in this one, or I don't remember when you talked about it— okay, he has talked about fatigue, that was what his thesis was on, uh fatigue and fracture. But in any case, um if you have a fatigue loaded structure that's going to vibrate, like a bridge, the safety factor is two on the base metal, not 1.67. We up it a little bit.

And then I'd have to go back, I'm going to talk about fatigue in welds in a little bit, maybe hopefully today. Uh but we handle the fatigue problem a little bit differently. Rather than just a straight safety factor, we actually have to know what the geometry of the weld is, and so we have different classes uh of welds okay, and I'm going to show you that. But in any case, we hold the welds to a higher standard because we know they're going to be made to a lower standard, I guess, is one way to put it, and it hopefully all averages out.

But there's lots of people who think, well, my weld wasn't perfect, you know, I had a problem, and I went in and I cut the weld up in 14 pieces and I found a little pore in there. Well, first of all, do fracture mechanics and see if the pore makes a difference, and I will tell you that 19 times out of 20 those little imperfections, fracture mechanics will tell you they're insignificant. And that's why the code can allow typically defects an eighth of an inch in size— defects, flaws an eighth of an inch in size, sometimes larger. Sometimes your undercut can be up to one inch out of 12 inch, for example, in the code, because people have not always done fracture mechanics, but through experience they've seen these bridges last 80 years with no problem, no fatigue cracks and whatnot, and they know what size imperfections are permissible at the design stress levels that the codes allow okay.

So the quality control consists of, you have these recipes, qualifications, certifications, and then you, on top of that you overlay a stiffer surf— uh safety factor, and in general the code is a good historical record of what works. If you follow the code—

[Question about paints?] Uh, the paints affect hydrogen cracking, because you don't paint until after you've welded. Uh, the paints, some paints can actually keep the hydrogen in that would ordinarily be diffusing out, you know, like the CO2 in my Sprite okay. Over a few days you can put paints on that will lock the hydrogen in. Uh sometimes— well, uh anyway, I'll just leave it at that. That's not usually a problem um, usually in lower strength materials it's really not a problem. In higher strength materials, like 200 ksi steels, it can be a big problem. But those things, people have got their own design specifications, they do electroplating.

I don't know if I told you the— I think I may have told you the Bell Helicopter mast. They actually electroplate it for corrosion resistance. It doesn't have any welds in it, uh but they electroplate it for corrosion resistance, and when they take it out of the electroplating bath, within five minutes it's in the hydrogen bakeout oven. They don't wait an hour, because it's a really high strength steel and it's a very critical part. You've only got one, there's no redundancy, and if it fails you go down, okay. And most of the time when you go down in a helicopter, you die okay. Um and you can't even auto-rotate. Anybody know what auto-rotation is in a helicopter?

Student: [response about tilting]

Yeah, except you have to have a little bit of forward speed. There's an operating envelope. If you lose your engine in a helicopter and you are— if you are hovering and have no forward speed, you drop like a rock. If you are moving along forward and you lose your engine, if you then pull, you tilt up, which is what you're thinking of, and now the wind from your forward motion is hitting the blades and will cause them to spin. It's called auto-rotation, and you actually go slowly down. If you ever saw one of these children's toys, a little helicopter blade that you shoot up and it kind of goes slowly down— that's auto-rotation okay. Coming down, the wind is going through the blades and causing the blades to spin, so you're using the gravitational falling energy and converting it to rotation, which gives you some lift and you don't fall like a rock okay.

I've auto-rotated a helicopter once. I got to go to flight uh ground school for the Bell 407 and they let me go on a little uh test ride. They had a real pilot, test pilot there, taking me on the trip fortunately. Anyway, but he let me kind of auto— he'd let me do an auto-rotation. Um, you have to do auto-rotations about once a year just to, you know, to requalify as a pilot okay. But anyway, um, it's always sort of fun to auto-rotate. Lots of people go out there and just don't have anything else to do with their helicopter. Let's say your— who was the uh Harrison Ford— he had been there the week before me, he had bought himself a new helicopter, and you know, people like that, "hey, let's go auto-rotate for a while," you know, go up down, go up and down, you know, it's just sort of like riding a carnival ride, right? If you got a few extra million, right, why not. Um so anyway, that was auto-rotation. What was I talking about, I don't remember, but it's another story. Barry?

Student: [reminds him]

Oh, the part. I was talking about the mast. If your mast fails, you lose your blades and you can't auto-rotate, you just drop like a rock okay. And it doesn't matter if you got forward momentum okay. Um I've had a couple of situations where people were hovering and lost their engine and they just dropped and they died, and there's lawsuits and everything because they didn't have forward momentum. Sometimes it was because you're supposed to get out of that area of, you know, going straight up or hovering. You're not supposed to hang in those areas where you can't auto-rotate safely uh for any length of time. So hovering is a more dangerous thing in a helicopter than flying in a helicopter, because you can use your forward momentum to give you some lift on the way down. So anyway, um, other any questions?

Um, let me talk a little bit about— there's certain things about welds that make them, particularly steel welds and titanium welds. They actually transform. Remember, you go from, in steel, you go from BCC to FCC as you heat up your crystal structure, somewhere around between 723 and 910 you change your crystal structure, and when you cool back down you change your crystal structure again, and that helps you maintain very fine grain size okay. You're going from one crystal structure to another, you nucleate new grains, they take time to grow, and if you do this at the right cooling rates according to your welding procedure, you can get fine grain, high strength, high toughness welds.

But there's more to it than just that. It turns out the weld metal in steel— and the same thing occurs in titanium alloys, because you have hexagonal close packed at low temperatures, and at higher temperatures you have uh BCC, but you have the same type of thing, you have a crystal structure change which allows you to have um fine grain size in your weld. So titanium and steel, because of this allotropic crystal change, um can give you weld metals that have as much strength as the base metal. If you pull a tensile test you'll find it'll fail in the base metal, even though the weld metal is sort of an as-cast structure.

This is a picture of a submerged arc weld in steel, two passes. This is the interface between two of these passes. This big ferrite blob here is a transformed structure, it's in the heat affected zone. You see the weld metal is generally finer structure. And the reason for that, it turns out, is in steels when you weld you usually get some oxygen from the air or from the slag flux that you're welding with, and the base metal might have 20 or 30 parts per million oxygen, the weld could have 300 to 2,000 parts per million oxygen. If you look at that weld metal, it's just full of little bitty spherical oxide inclusions. And you might say that's bad, except it turns out it's good, because it causes nucleation. When you have this transformation, as it's cooling down from FCC to BCC, those little oxide inclusions will nucleate what we call acicular ferrite. Right over here.

So a regular steel grain cooling down, if you look at it in a microscope, will look like a bunch of grains of different sizes. These should be 120-degree angles at the intercepts and stuff, but you have grains like that. But the weld metal will be a bunch of needles, and at the center of each one of these needles, if you look at it in a transmission electron microscope, will be an oxide inclusion okay. I can't draw them all, but it's a fine grain structure with needles— that's what the acicular means, it's needle-like structure— and these are high-angle grain boundaries. When the crack tries to go through there, it's blunted, it changes its direction, and it gives you very high toughness, very high strength for the composition of the weld metal okay.

So it turns out, even if I'm welding A36 [0.3] carbon base metal, I might use a 0.15 carbon weld metal, because I have this finer structure. I don't need as much carbon to give me strength, and I can match the strength of the base metal to the weld metal, and get slightly overmatching strength. I want slightly over— slightly higher strength in the weld metal than the base metal. You mechanical engineers out there, or you materials engineers, do you know why I would rather have a structure where the weld is slightly stronger than the base material?

Student: [response about stress concentrations]

You're welding at stress concentrations, and because you're at a stress concentration— a corner or something— if the thing gets loaded in any way, whether it's bending or if it's tension, if it's undermatching, you're going to pull the weld metal and it's— you're going to, it's going to yield before the base material. That means all the strain in your structure is concentrated right on that weld. Even though your structure may only undergo 1% strain, if the weld metal is only 5% of your overall structure, that 1% gets multiplied by 20, by a factor of 20, because it's only 5% of the structure, and all of a sudden I got 20% strain in that weld, and the weld is over-strained. So you can't think of it in terms of stress as much as you should think of it in terms of localized strains.

The Navy and other people have spent millions and millions of dollars over the years trying to see if they could use undermatching weld metal, because if you could use a lower strength weld metal you can tolerate a lot more hydrogen okay. You could save millions of dollars on your preheating cost. But every time they've done it, they've done the fracture— and you have to weld up whole structures and shoot at them with ballistics if you're the Navy, put them in great big testing machines at Lehigh University, or out in Boulder, Colorado at the National Institute of Standards and Technology, where they have these multi-million pound test machines, things you can put a huge I-beam in and just tear it in two okay. You can take a whole structure panel from a bridge and you could bend it and tear it apart.

With undermatching weld metal, you concentrate the strain at the weld. With overmatching weld metal, you force the base metal to strain, and that's 90, 95, 98% of your structure, and so now your structure behaves like you would like it to. A 1% strain can be absorbed by the structure as a whole. But if the metals, under— weld metals undermatching— a 1% strain gets concentrated in the weld. Now you start seeing 20, 25, 30% strain and the thing fractures okay. And it doesn't really matter whether it's um slow tensile failure, slow bending, or whether it's impact fracture. The impact fracture, also the deforming area or the deforming volume decreases if the weld's weaker, and it will absorb less energy in an impact test. So whether it's dynamic fracture or whether it's slow fracture, undermatching is not good okay.

And fortunately, I haven't seen another millions of dollars worth of testing for about 25 years, but before that, about every 15 years, someone would go out and spend millions of dollars trying to find out if they could use undermatching filler metal. The advantage is in steels, you get this acicular ferrite that will give you strengths up to— the acicular ferrite uh can give you [120] ksi, which means you can weld steels, base metal of 100 ksi, and you'll be overmatching. You'll have a 20% overmatch. You'd like to only have 10% overmatch, but usually you shoot for 10 to 20% overmatch.

If you go above that, uh you're going to get uh martensite. You're going to have a weld metal composition that will give you martensite, and you can go up to 180 ksi without too much difficulty, but you're going to have to do post-weld heat treatment, and that might be 100K— 160 ksi with post-weld heat treatment to temper the martensite.

And you're not Navy folks, but this summer when the Navy guys have to watch some of these lectures— that's what happened to the Seawolf submarine. Seawolf submarine was supposed to be welding HY 100 [ksi], but they had a weld metal that was a little too rich in chemistry, had too much car— it is, had— they said high side chemistry, the carbon was at the maximum of the range, the manganese was the maximum of the range, the chromium, everything was at the maximum of the range, and they were getting 130 ksi weld metal. It was more of a martensitic rather than acicular ferrite, and all of a sudden they ran into all kinds of problems with hydrogen cracking. Cost an additional [?] $2 billion okay for one submarine. Congress was not happy. The submarine was supposed to cost $2 billion, not repairing it okay. So it doubled the price of that sub. And I think we only built one Seawolf, or one of the Seawolf class, but that was also because peace had broken out with former Soviet Union and we didn't need a uh submarine that would be able to shoot down Soviet uh submarines and a lot of other things.

Anyway, with the weld metal you get uh um martensite, and you try to keep your hardness in your weld metal below Rockwell C 30, and the petroleum guys like to keep it below Rockwell C 22, but they can't use the higher strength. It turns out the bridge guys, when you're building a bridge, we can use steels up to 100 ksi, but we don't use steels any stronger than that, because you can't just— you'd have to do post-weld heat treatment. You can't post-weld heat treat a bridge, not economically anyway. If you go above uh 180 ksi— so this is up to 120, this is like 120 to 180, if you're greater than 180, it's hard martensite. And now we're talking preheats, um— I mean, yeah, you can have higher strength base materials but preheats— and post— preheats of 600 degrees Fahrenheit and post-heats, post weld heat treatment required, it's very expensive. This is the problem when they came to me and said, oh we want to use 4340 in our America's Cup yachts—