§1. Aluminum aircraft crankcase repair industry [00:13]
To finish up aluminum before I go into titanium, I'm going to give you an actual story from January 24th of this year — so it's only six weeks old. A company I've done some work with rebuilds the aluminum crankcase housings for aircraft piston engines. If you've got a little Cessna or a Piper, it'll have a four-cylinder or a six-cylinder engine. It's an air-cooled engine, and the crankcase is not cast iron, it's aluminum, because cast iron is heavy and you're going to fly this thing.
The two big manufacturers are Lycoming engines, in central Pennsylvania, and Teledyne Continental Engines down in Mobile, Alabama. They used to be up in Muskegon, Michigan, but they moved down to Alabama about forty years ago. A little four-cylinder engine could cost you three hundred thousand dollars, and a six-cylinder might cost you half a million. Why are these engines so expensive? Because you're buying a lot of insurance that goes along with them, okay. If the engine fails, people tend to crash, and people who fly airplanes tend to be on the higher end of the income scale, and their families tend to sue. It's not that these things are really hard to make — we're still building them in many ways the way we built them forty or fifty years ago. These designs go back to the 1940s.
You've got an aluminum housing with two clamshell halves, and a steel crankshaft that goes through the middle. The steel crankshaft is very heavily loaded, but the aluminum housing really just holds things together — it doesn't have much stress on it. You've got an aluminum housing, two halves of it, and in the middle some bearings, and the steel crankshaft, and on the end you have your pistons and piston housings. They're usually kind of horizontal. The piston is inside, connected to the crankshaft, and the detonation is in there, you've got your spark plugs out here, and all the stress is on the piston head. But the housing just keeps the oil in. It has the same pressure on the inside as the outside — varying with altitude on the outside, but basically the same — so the aluminum housing isn't pressurized.
The housing keeps the oil in so you can collect it and recirculate it through the pump, but all the high stresses are in the piston head, the piston, and the crankshaft. This is just an envelope to hold everything together. If you went out to buy a new one, it might cost you thirty or forty thousand dollars for these two halves. They do develop cracks, and one of the problems in the industry is no one ever throws a part away because they're so expensive to replace. There's a whole industry out there to repair these things. Say you develop a fatigue crack in the aluminum casting — because these are basically flying fatigue machines — and so you weld-repair it. There are only a few places that can weld-repair it. The Federal Aviation Administration says you must do this at an FAA-approved repair facility that has proven they have the procedures, and there are actually only a few places that do.
The piston engine companies don't particularly want all these mom-and-pop shops out there repairing things. They would much rather sell you a new one, okay. And they don't like the liability of having one out there that's forty years old and has been repaired, because if it ever has a problem they're going to get sued too. So Lycoming and Teledyne — they can't stop the repairs, because if the FAA approves it, the original equipment manufacturers can't stop it. But they're not exactly happy to have a lot of people extending the life to forty years.
Student: [question about altered state defense]
Not to the plaintiff's attorney. The plaintiff's attorney is going to come up with something against everybody, okay. They're going to make sure their expert has a theory that keeps everybody in so he can collect a little bit of money from all of them. You would think in the law, supposedly, if it's been altered from its original state — but just to do a little weld repair on one of these things, they can start arguing a design defect.
In 1992 Congress passed the General Aviation Revitalization Act, GARA, because all the Piper and Beechcraft and Cessna were going bankrupt and the engine companies were having financial problems. Congress basically said if it's more than eighteen years old you can't sue — it's not a manufacturing defect if it lasted eighteen years. Well, some plaintiffs' attorneys say, if you changed your manual, time starts when you changed your manual. Wait a second — you can't update your manual on how to do things? This is still fighting its way through the courts. All it does is keep attorneys busy, keeps experts busy. I put my kids through college, okay, but it's a very dysfunctional system overall.
§2. Jet engine repair industry and counterfeit parts [07:15]
It's not limited to piston engines. This whole repair business applies to jet engines in a much bigger way. On jet engines we're talking about General Electric, Pratt & Whitney, and Rolls-Royce engines that go in 747s and 737s. It's about a 15 or 20 billion dollar worldwide repair industry to rebuild the engine parts. You could have a compressor case that costs fifty thousand dollars brand-new, and you can repair one for ten thousand. It can be a very lucrative business to repair things. But again, you have to be an FAA repair station. Anybody who touches this thing and is not an FAA repair station can go to jail.
There are a lot of counterfeit parts out there too. I showed you turbine blades — they're worth six thousand dollars apiece. When you scrap a turbine blade you put a notch in it, you actually grind a notch, so that any mechanic who sees it says, oh I can't install that, to keep people from selling scrap parts. Parts should be destroyed in some way, even if it's just putting a notch in it, so a mechanic wouldn't reinstall it in an engine. You should have a history on some of these things.
§3. The 4043 filler metal problem [08:35]
So this company has been in business for about thirty or forty years. They used to be an authorized repair station for one of the engine manufacturers, and then the engine manufacturer said, we don't want to start repairing our own product, we want to sell new ones. They sold this and the family took it over. Chuck's not part of the family but he's the chief engineer. I scratched out his last name — you don't need to know it. I had worked with him before. One of the problems they had: the housing is an aluminum casting alloy and they were using 4043 filler metal — they'd grind out a crack and fill it up with 4043. As you'll see, 4043 is softer. I had just worked with them on an accident that occurred in Canada, and the Canadian Transportation Safety Board, similar to our NTSB, had investigated and said, your filler metal is softer than your base metal and therefore you have under-matching weld metal, and that's the cause of this fatigue crack.
I had to help them respond to that. For thirty years, every now and then someone would investigate one of these and find their weld metal was softer than their base metal, and say that's the cause. They didn't go through any fracture mechanics and determine what the stresses were. I don't know if the original manufacturer knows what the stresses are, other than that they're low. It's just a container for oil up there. No one's really done a big analysis on it, and they don't need to — these things are sort of proven by about seventy years of history.
A customer had a 3/8-inch stud. You put studs into the aluminum housing to hold the piston heads on — there are probably fifty or sixty studs that go through and hold the thing together. It's actually a stud going all the way through. They had repaired one, and someone was torquing the stud — it's supposed to torque to 204 foot-inch pounds. The mechanic always uses a torque wrench. You never tighten anything on an aircraft engine unless you have a calibrated torque wrench, particularly on something fatigue-loaded. If you don't get the right torque you're going to get a failure. They were torquing it and it stripped the threads — the stud pulled out. They got it back, fixed it, did a warranty repair, moving on. No one got hurt; they had to reject the crankcase.
But they decided to do a test on their own. As he says, we only have a sample size of two, which is not very big. What they found is: with the stud into the base metal, half of them failed at 204 inch-pounds. With the stud into 4043 weld, half failed — with helicoil and with double helicoil.
Do people know what a helicoil is? If you've got a steel stud going into aluminum, you can put a larger thread in the aluminum, and then a helical coil with a diamond-shaped cross-section — rather than a spring with a circular cross-section — and you put it in there, and now you have a steel insert in the aluminum with a bigger thread, and you put a steel screw into this helical. Helicoils have been around for sixty, seventy, eighty years. It's a way to give a larger thread area in the aluminum casting. A double helicoil is probably two helicoils together, so you get even larger area. Aluminum's not as strong as steel, so you're trying to make a stronger unit, and the overload average is higher with helicoils because you have bigger thread area with a larger diameter.
They also made one into 4145 weld metal and got a really big increase in strength. And he said to me, do you have any ideas as an engineer? He didn't really know what to do.
§4. Working through the specifications [13:17]
So I'm going to walk you through what I did when I got this email — and I did all this for free, by the way. Something I ought to point out: if you've taken this course and you graduate and go off somewhere, you're welcome to call me. For all the pain and suffering of having taken the course, you can call me at any time in the future if you run into some sort of problem. I don't mind talking to former students and trying to work you through it. I've said this for years, and I've only had about a dozen students take me up on it — maybe a couple of dozen — and I've usually at least helped head them in the right direction.
First of all he tells me it's AMS 4280 alloy. I don't know what AMS 4280 alloy is. This is the Society of Automotive Engineers, SAE — you could call it the Society of Aerospace Engineers, but SAE is the organization for what they call mobility, whether you're an airplane or a car. SAE essentially writes specifications for the transportation industry, and they have a whole bunch of Aerospace Material Specifications, AMS. So if you're Boeing or you're Lycoming, you're going to specify things under an SAE specification. There are millions of these specifications out there, but many of them are identical. You can go to this little book, which is really just an index of all the SAE specifications, and look up 4280.
Aluminum alloy casting — at least that sounds right. Permanent mold — that's the way they made the casting. 355-T71. I can recognize that as an aluminum alloy: Aluminum Association 355 is the casting number, T71 is the heat treatment. And I could buy a little ten-page specification from SAE for $59. What a deal — six bucks a page. This is July 2005; I don't have a newer one. Great business, keeps those people at SAE going. So now at least I know it's aluminum alloy 355. He's used to calling it by AMS 4280; Tom Eagar doesn't know what that is, so he has to go look it up. That's not too hard — I own the book.
Then I go to one of these Aluminum Association charts out of the welding handbook. Under aluminum, if I have 355 alloy welded to 355 alloy, the recommended filler metal is ER 4145. But it's got a few footnotes. One says you could use 4047; another says you could use 4043. Well, that makes sense — he's been using 4043, so he's compliant. But why would you use 4043? If you go to other footnotes on the next page, it says you could use 4009, 4010, or 4011 filler metal. So I had to look all that up, which is still in the welding handbook.
Here are the compositions. 4145 is very highly alloyed — four percent copper, ten percent silicon. 4043 is five percent silicon and 0.3 percent copper. So 4043 is a lean alloy — that's why it's weaker. 4145 is highly alloyed. 4009 turns out to be the exact same composition as 355.
Student: [question about ductility]
The aluminum has lots of ductility anyway. The castings only have five to ten percent elongation, but it doesn't really matter in this application — you're fatigue-limited. So it depends on whether it's going to be an overload failure or fatigue failure, and I'm going to walk through some of that. Yes, silicon makes it brittle — it goes to the grain boundaries — but it makes it very fluid, makes the casting easier. The guys in the cast shop will love it because it fills the mold and you don't make defective castings. So a lot of your aluminum casting alloys are based on aluminum-silicon.
§5. Choosing the filler metal: matching, over-matching, under-matching [19:00]
So I've now done a little bit of research, and the recommended is 4145, but you can use 4043, which they've been using since 1976 — so it has a little bit of history. I think they've done about a hundred thousand of these over the last forty-some years. Out of a hundred thousand, they've probably had a dozen or two dozen fail. They're not dropping out of the skies. It's a low failure rate. And there are a lot of other reasons why you can have failures. The most recent one I saw was because some mechanic over-torqued the steel bolt — nothing to do with the weld. If you over-torque the steel bolt you'll get fatigue in the bolt and you can crash. In any case, it's okay to use 4043.
4145 gives the highest strength because it's the highest-alloy, but if it has higher strength you have higher residual stresses. With 4043, you weld it, and if it's the weakest link in the chain it's going to stretch — as the metal shrinks, you get stretching of the 4043, and the maximum stress in the weld is going to be at the level of the strength of 4043. Whereas with 4145, the maximum residual stress is going to be essentially the strength of the base metal, because that's the part that's going to yield. Think of this as a tensile bar made of two different materials: you pull on it, and the stuff that's going to yield in that composite tensile bar is the soft material, and that's going to limit the residual stress to the softer material.
If you read the handbook you can also use 4009, 4010, 4011. 4009 is an exact match for 355, so if someone comes afterwards and does an analysis they won't find a difference in composition. That's what the Canadian Transportation Board had said in his previous problem of last fall — you should have used matching filler metal. Well, if you've been doing it successfully for forty years on a hundred thousand parts, I'm not sure that's a good conclusion, that your design is wrong. When people are doing failure analysis, they're just looking at something that's different, and they'll jump to the conclusion that the difference is what caused the problem. That's not always true.
So what would I recommend? The technical answer for the highest strength is 4145, because of over-matching filler metal. But I'm not sure static strength is the controlling feature here. It's really fatigue strength, and I told you that usually you would like over-matching filler metal. However, in this particular case where the thickness is probably 1/4 to 3/8 of an inch, you don't have super-high residual stresses, and I think you might prefer in many cases to have under-matching filler metal, because this is not an impact-loaded situation, and you can get a little better fatigue strength of softer material. You were talking about coil tubing before class — copper is very good for fatigue resistance because it's so soft. It doesn't have much strength, but it can take lots of fatigue cycles.
Student: [question about machining 7075]
You can machine both of them. Aluminum is soft enough that sometimes it gets gummy, but usually the higher the strength, the better — the chips come off. Maybe you're right about 7075; I haven't ever experienced or read that. Usually 1100 aluminum, the really soft stuff, is what gets gummy when you machine, because it just smears, bonds to the tool. Actually, I can imagine in 7075 you might think something's gummy because at the speeds people are machining today, they're going so fast they could actually get a diffusion bond of the aluminum to the steel cutting tool. Aluminum is very reactive with steel — you generate enough heat and you start coating your cutting edges with aluminum, and now you've got dull cutting edges. If you're getting gummy on 7075, it may be that you're cutting too fast.
My first comment would be: slow down. But you really need to see it and look at the tool. I've gotten 1100 aluminum gummy and ended up stopping and looking at the cutting tool — boy, I just made a diffusion bond of aluminum to the cutting edge. No wonder it's dull. Back when I was your age we might machine at spindle speeds of thousands of RPM. Today we're going at tens of thousands of RPM literally. The spindle speeds have shot through the roof in machine shops over the last thirty years. That's a result of new cutting tools and better machine performance — when you're going to those speeds you've got to have a lot of precision in how you put the whole thing together. In fact, that's one of the things your dad worked on. It's all in the design of that cutting tool, better materials, and better design of the spindles and bearings.
So one thing you should do is stick with success. Once every ten years someone criticizes your under-matching strength — deal with it, which is what they've done. However, 4009 is matching composition, doesn't have too much strength that would give you poor fatigue properties like 4145, and you could use 4010 or 4011. People would have a hard time telling the difference between 4009 if they analyzed it afterwards, and this gives you some intermediate strength. You've been successful with low strength; you should be successful with intermediate strength. What would I do? I'd stick with what you've got, or if you want to change to get rid of this criticism — because people say, oh, you don't have the exact same composition, which is a stupid argument, means they don't understand welding science and stresses — if you want to come up with something matching, I wouldn't go to 4145, because they're going to give you the same argument on the other side of the strength. Go with 4009: it's permitted under the recommended materials and it will probably work just fine.
He wanted advice on what he should do. First of all, I'm not going to tell him what he should do. I can tell him what his options are. Don't tell people what to do, because then you're the designer.
§6. Johnson & Johnson anodized aluminum laser instruments [27:27]
Which reminds me of a story. Back when I was an associate professor without tenure — so the early '80s — I was working with a division of Johnson & Johnson. They were looking for some laser tools, instruments. They made stainless steel instruments for surgeons — neurosurgeons, general surgeons — scissors, drills, reamers. They made 5,000 different types of hand tools for surgeons, mostly out of stainless steel. In the early '80s, laser surgery was coming along, and they really wanted something a little bit better.
What would happen is, you're cutting some tissue with a laser, and the laser could have too much power and go through and burn the tissue underneath where you're cutting — that's a problem. Or the surgeon doesn't aim it right and it hits the tool and reflects off, and now the laser cuts the patient somewhere you didn't want, because you're using the surface of the tool as a mirror. These are small little cuts, okay — a thousand cuts will kill you but one won't. It's not that bad.
MIT has this Industrial Liaison Program, Johnson & Johnson's a member of it, so this guy comes by, and he was dropping off a sample for me — I was doing some testing on something else. I said, what are you here on campus for? He says, oh I'm going to the physics department to talk to the people about lasers. I said, you are? He says, yeah, what's wrong with that? I said, oh no problem, go talk to the people in physics. He came back about twenty minutes later just shaking, and said, now I understand. He'd explained that he wanted a material that wouldn't reflect and burn the patient in the wrong place. And the answer from the head of the spectroscopy lab — the world's top spectroscopy lab, knows all about lasers — was: well, all materials absorb energy from lasers. Yeah, well, some absorb a hundred percent of the energy, some absorb one percent of the energy, so there's a wide range in between. From an engineering point of view that might make a difference, but you couldn't tell that to a physicist. One percent, a hundred percent — in engineering that could be a difference.
So I said, okay, John, sit down, tell me what your problem is. He told me: they wanted to cut the tissue they wanted to cut, didn't want to burn through beneath, and didn't want to reflect. I said, you're posing a problem that is sort of betwixt and between. To have an absorptive material, you want something with no free electrons, because it's the free electrons in a material that re-radiate and cause it to reflect. What are the best mirrors? Silver, gold, copper — they have lots of free electrons. But if you want something with good thermal conductivity among the metals, you're talking about something that has lots of free electrons — silver, gold, copper. That's what he was looking for too. I said, what you need is a composite. It's too bad you can't use aluminum for medical instruments. He says, why not? Well, I thought you might have some problems. He says, we make lots of instruments out of aluminum. I said, then you ought to use anodized aluminum. Anodized aluminum gives you a top surface that's an insulator with no free electrons — aluminum oxide, absorbs lots of energy — and then the aluminum underneath, which has high thermal conductivity, will diffuse the heat away.
He says, how do we test that? I said, you make some little discs of aluminum, go get them anodized to different thicknesses. I talked to him about how many microns thick. He goes out and gets them anodized four, six, eight, and ten microns. We come back, I find a laser somewhere here at MIT, and we basically hit each one with a pulse of laser energy from a YAG laser, and we measured the temperature rise of each little disc, and you can figure out how much was absorbed — it's a little calorimeter. We showed that if you didn't have any layer you didn't absorb any energy, because aluminum is a great reflector, but with an anodized layer — I don't remember exactly, you'd have to look at the patent — somewhere between six and ten microns was the best. I actually calculated that from the wavelength of the light. That made sense in terms of the wavelength of the laser, when you get thick enough to have good absorption and thin enough that you've got the high-conductivity aluminum close to the surface.
§7. Patent indemnification and consulting liability [32:54]
So the reason I started thinking about this story: they patented it, and I got two patents out of this. Good for my tenure case, but I got nothing out of it. I've never made a dime on any of my patents. The ones I patented through MIT, no one is using, and the ones I patented through companies they're all using, but I did it as a consultant, so they had all the rights. I've never gotten any royalties out of any patents. But when the patent issued, about a year and a half later, all of a sudden I thought: if anyone is ever using this and someone gets burned or injured, Johnson & Johnson could be sued, but my name is on the patent, and I could be sued.
So I wrote a letter to Johnson, said I want to be indemnified — that if anyone ever sues, you will treat me just like one of your employees, and your attorneys defending your other employees on the patent will take care of me. Because I don't want to see hundreds of thousands of dollars in legal costs because Johnson & Johnson gets sued and my name's on it, and I'm an independent consultant. This is one of the reasons for telling you this story. If you're ever an independent consultant, be careful about designing things, or design through an LLC company, because if you're the designer you could get sued. A big corporation will protect their employees, but if you get sued, you can be sued personally.
If you don't have the insurance — and you can go out and get professional liability insurance — you have to have the insurance in force when you design it and when the failure occurs. For five thousand dollars a year back in the mid-'80s, I could have spent $5,000 on professional liability insurance of a million dollars or two million, and then I would have to carry that for the next forty years until I die. That didn't sound like such a great deal for something I got no royalties on. So the point of all this is: be careful when you actually design. The only way someone could have tracked that I designed it was that my name showed up on the patent. It took Johnson six months and I finally got a letter from their corporate counsel saying they would indemnify me. I made about ten copies of the letter, placed them all over the office, so if I ever did get involved I could find a copy somewhere. Now the patent's expired, and to my knowledge we've never been sued. Just be careful about designing things. It might be good to ask your dad what he does, because he designs lots of things. Even if you do it right, you can still be sued, and you still have to defend yourself.
§8. Transition to titanium [36:13]
So I want to talk about titanium alloys. Anybody have any questions? Okay. Titanium alloys come in a number of — well, there's actually not that many titanium alloys. Remember, steel is a billion tons a year, aluminum is 45 million tons a year, titanium is 165 thousand tons a year, of which about 26,000 tons actually go into structural materials. Anybody know what the bulk of the titanium is used for? Paint on the walls. Titanium dioxide is in the paint, and ceiling tiles. Titanium dioxide has a higher index of refraction than diamond, so it sparkles. It replaced lead in lead paint for giving you a nice reflective, very white surface — appears very white because it reflects all the visible wavelengths.