§1. Liability and accommodation in big failures [00:03]
[Tom holds up an asphalt reclaimer tool tip.] This little asphalt reclaimer tool has a carbide center with a diamond on the tip. I didn't tell you why this ever went into commercial production — it's ten times the life. On the break, Adrian asked about liability for some of these screw-ups.
It's sort of like the question of whether you can let a big bank fail. If it's some lower-level person, you're going to send them to jail, bankrupt them, whatever. But if it's a really big failure, you can usually come to some accommodation.
To give you an example: there's a specification going back to the 1970s that you have to quench and temper your high-strength aluminum alloys. It's not the same as martensite — when we go through aluminum welding we'll talk about how you treat aluminum alloys that are quenched and tempered for different metallurgical reasons. You get higher strength, but you get precipitation hardening, not martensite. There's one plant in the United States, the Davenport Iowa works of Alcoa, that has the world's largest rolling mill. That's where they roll all the plate — four-and-a-half-inch-thick, six-inch plate, huge plate — for Boeing and a number of aircraft companies. If you're going to build a huge aircraft, you have to buy from Alcoa at this plant. To quench these big plates, they have a series of water jets. The hot plate goes under this shower of water jets, gets quenched, and is later tempered.
They discovered, a little later than they would have liked, that one of the jets was plugged. By that point Boeing had already done all the machining and put the plates on aircraft that were flying around the world. So you had a big wide plate with a streak that never got quenched and tempered — a local streak. At this point they had to notify the government. They notified NIST, which is an excellent lab in Gaithersburg, Maryland. The head of NIST was Bob Reuven [Schaefer?], the guy who didn't get tenure here at MIT before they hired me.
He was head of the NIST lab, and they set up a partnership between the government, Alcoa, and the airlines. When it's a really big problem there are people who actually want to know how to solve it first, and then tell the boss. They jointly developed an electrical resistance measuring device — a non-destructive probe you put on the surface. I have one that cost about $5,000. You put it on the surface of the metal, measure the electrical resistivity of the aluminum, and you can tell the heat treatment. They could go check the wings at lots of locations to see if they had been properly heat-treated. I was involved in that.
Another example I heard was the Citicorp Tower in New York City. After they built the tower they discovered that some of the welded joints were not good for seismic loading. If you had a really severe earthquake, this building was going to collapse. There's an article in The New Yorker about this whole thing, but it didn't tell the story I heard from another point of view. You had the architects, the engineers who designed the building, the contractor who built it, a bunch of subs, and then the owner, Citicorp. What were they going to do? The story I heard was about John Reed — John Reed went to Sloan School, became chairman of Citicorp, retired four years ago at 75 years old, and is chair of the MIT Corporation. This guy was the real high flyer.
I checked with him last winter, because his brother had a welding problem. His brother is 85 years old and owns a steel tank welding business down in Tampa, Florida — not a small one — and they work with nine percent nickel steel. John Reed had heard from my dean that I knew something about welding, so he wanted his brother to hire me. I talked to one of their engineers, and they didn't want any help. So I talked to John, and he apologized profusely for his brother. But I said, John, can you confirm this story? The story I heard, which John didn't completely confirm — he said he didn't remember the details — he remembered the problem, and he remembered everybody getting together and working it out cooperatively, because it was a huge problem.
The way I heard the story when I was in the Sloan program: they got them all together, and after John Reed heard how big the problem was, he said, okay, I will pay for it. Citicorp will pay for the repair. We're not going to get any attorneys involved. You're going to work at cost, and you're going to work with us, and I will pay for all the materials. And they fixed it. He remembers that he made some decision in the meeting, but he doesn't even remember about the cost, because he was dealing with hundreds of billions of dollars — who cares about a few million.
When the problems are really big you can often go around the attorneys. But when the problems are smaller — say, a hundred-million-dollar problem — the attorneys take control, and then you can't get the engineers together. No one wants to allow their engineering staff to talk to the others. It's pretty dysfunctional as far as solving the problem, because everybody's CYA — who's going to pay for it? When the Seawolf submarine had its two-billion-dollar problems with the welds, they worked it out. What are you going to do? You're going to bankrupt one of the two submarine building facilities? You can't do it.
I remember working for Bell Helicopter on some things, and the chief counsel said, we've got to quit thinking that we can sue our suppliers. The suppliers will just say, we're not going to supply you anymore, we don't want the liability. So there's lots of business decisions that go into things. The problem is when the trial attorneys get in there — they're just sharks. Welcome to the law profession, which is actually okay because they hire me, and that's how I got grandkids.
But logically, if it's really big — multi-million dollars — you usually find some way to fix it. I've got a problem right now — I'm not going to tell you the product — but it's hundreds of millions of dollars of steel fabrication. They found that they've got a design defect, and the company that fabricated it was supposed to design it. These things are breaking. If they do break at the wrong time, they may kill a few cows, but they won't kill people. But people will be without electricity in the country, and it becomes a big political problem with the regulators.
It turns out that as we've gotten into it, they've got this problem in other parts of the country — same problem. I asked, when are you going to go find out from the other companies in other parts of the country? None of them really want to talk to each other, because the attorneys are all afraid of who goes after the manufacturer first and puts them in receivership. Everybody wants to be first in the trough to get their money out before the company goes bankrupt. In fact, the CEOs of these companies — nobody wants to put the manufacturer in receivership, because if you want to be in business you need a supplier. Ultimately, if it's a really big problem, cooler heads will usually prevail. Does that answer your question?
Student: Yeah, I was curious about the dynamics — whether it's more reputation, or the legal framework, or whether they might be criminal. Just trying to get a sense.
§2. Tylenol, Amtrak, and the Phelps Dodge fraud case [11:20]
These are all kinds of variations. The famous one is when Johnson & Johnson found out that someone was spiking the Tylenol, putting poison in their Tylenol on store shelves. They recalled all of it. They hadn't done anything wrong, but it was their product and their reputation. They have a credo at Johnson & Johnson that basically comes out of the Hippocratic oath: they will not do any harm. Johnson & Johnson actually calculated that their business improved by more than the cost of recalling all the Tylenol. They got great press for doing the right thing, even though it wasn't their fault.
By the same token, I was involved in the northeast extension of Amtrak a few years ago. They had a British firm doing a $300 million Amtrak contract — remember Amtrak is basically covered by the federal government. This British firm, which had all kinds of electrified commuter train tracks elsewhere, was in charge. It was a fixed-cost contract, and they needed a bunch of copper wire for the overhead conductor that these electric trains run on — a very special type of copper wire.
I got a call around the twentieth of December one year: could I go out and look at this product? The British company was going to buy from a little company in Rome, New York. Could I do an inspection and decide whether they could give a waiver on the product? And, by the way, we have to have a decision before Christmas.
I went out, looked at things, and by the 22nd or 23rd I said, no, can't accept it. At that point Amtrak issued an order: don't install any of this product, go find another supplier. The British company said, there is no other supplier, we've searched the world, we can't find anyone who can make this product, and if you don't let us put it up, you're not going to finish the northeast extension.
Through other channels, the government found out that there was a company called Phelps Dodge, one of the world's largest copper companies. They had a product that would meet the spec — actually exceeded the ductility. The ductility was supposed to be more than ten percent, and theirs was more than twenty percent in stretching. For whatever reason, Phelps Dodge had been treated as suspect. I said, use the Phelps Dodge material, it's out of spec but the spec is a silly spec. They ended up using it, and they built it.
But before that, the British company hadn't obeyed Amtrak — they had actually put some of the original product up around Providence. They had a train run by, and it was just the Fourth of July, arcs all along the wire. The wire was wavy, the slider would bounce as it went along and get electric arcs, and the wire would have worn out in three months. Profuse sparking. So I was right in rejecting the other product.
The reason they wanted that other product is they could buy it for $2.5 million from Rome, New York, and they would have to get it from Phelps Dodge for $4.7 million. It was a fixed-cost project, so guess which one they chose. And what did they tell the federal government? There is no one else.
When the federal government found out about Phelps Dodge, they said, why did you say there's no one else? You certified that there was no other supplier. We went to a three-day or five-day mediation in Connecticut. We ended up prevailing. The next day, after the decision came down — I was no longer involved — the story was that the FBI came in and confiscated all the computers and the passports of the senior management of British Rail for fraud against the federal government. They had lied. The Justice Department doesn't like that, particularly when you're trying to make a few million dollars off of your lie. I'm sure there was an accommodation, and it probably cost the company quite a few million dollars from their fraud. I doubt anyone went to jail, but nonetheless.
There are other stories — RV brake shoes, for example — where if you get a government Justice Department attorney involved, you can resolve things. Okay. I'm going to show a video, it goes for about ten minutes, on hydrogen embrittlement. Oh — Microsoft has decided it wants to do an auto update right now.
§3. Video: hydrogen in welds and the diffusion demonstration [19:15]
[Video plays.] Defects such as slag inclusions, porosity, and hydrogen cracking depend on the materials and process conditions. Hydrogen cracking — also called hydrogen-induced cracking, hydrogen-assisted cracking, cold cracking, and delayed cracking — typically occurs at stress concentrations and can propagate through the weld metal or heat-affected zone. Hydrogen cracking results when three independent conditions occur: hydrogen in the weld, a crack-susceptible microstructure, and tensile stresses.
Because susceptibility to hydrogen increases with hydrogen content, this demonstration focuses on low-hydrogen welding processes. SMAW using basic-type electrodes, gas metal arc welding, and gas tungsten arc welding are considered low-hydrogen welding processes. Hydrogen content in the weld is measured per 100 grams of deposited weld. A typical low-hydrogen process produces welds with four milliliters of hydrogen per 100 grams or less. In contrast, shielded metal arc welding using cellulosic-type electrodes produces welds with hydrogen levels of 46 milliliters per 100 grams — ten times more hydrogen. If a weld remains at high temperature, through interpass temperature control or postheat, the hydrogen has the ability to diffuse away. The average diffusion rate of hydrogen in steel at 100 degrees Celsius (212 degrees Fahrenheit) is typically 1,000 times faster than at 20 degrees Celsius.
Modeling shows the hydrogen level for welding a quarter-inch and five-eighths-inch thick pipe materials without preheating, as well as five-eighths-inch thick material with a 250 degree Fahrenheit preheat. The hydrogen levels are modeled after a typical cellulosic-coated SMAW electrode. Hydrogen diffuses from the root pass into the heat-affected zone, base material, and surrounding atmosphere as the weld cools. The maximum hydrogen level is present days after welding, at 1.75 milliliters per 100 grams for the quarter-inch thick pipe and 1.5 milliliters for the same with preheat. With preheat for thick-wall pipe, more hydrogen diffuses away, lowering the maximum hydrogen level to 0.75 milliliters per 100 grams. Postheating following the completion of welding can further reduce weld hydrogen.
Hydrogen can come from sources other than the welding process itself. It can be introduced by organic materials or moisture on the electrodes or the base materials. Hydrocarbons and moisture get broken into atomic hydrogen in the intensity of the welding arc. The liquid weld pool has very high solubility for hydrogen, but as the weld solidifies and cools to room temperature, the hydrogen becomes trapped or supersaturated in the solid steel. You may wonder how hydrogen can stay in solution if no bubbles of porosity are evident. Look to the periodic table. Iron has a relatively large atomic size compared to hydrogen. If you imagine the iron atoms as basketballs and the hydrogen atoms as ping-pong balls, hydrogen atoms can fit in the spaces between iron atoms.
The following welding demonstration shows the differences between a high-hydrogen SMAW cellulosic electrode and a basic-coated low-hydrogen SMAW electrode. The low-hydrogen electrode used has the designation H4R, meaning the hydrogen content of the as-received electrode is four milliliters per 100 grams or less, and it is covered with a moisture-resistant coating. Three V-bead-on-plate welds are deposited as part of the shielded metal arc process. Immediately on completion, the welds are placed in mineral oil to observe the diffusion of hydrogen. The first sample is welded with the cellulosic high-hydrogen electrode. A large number of hydrogen bubbles can be seen slowly diffusing out.
The second sample was welded with a low-hydrogen electrode. Compared to the first sample, this coincides with the fact that approximately ten times less hydrogen will be introduced to a weld made with a low-hydrogen electrode. The third sample was welded with a poorly maintained low-hydrogen electrode that was allowed to absorb moisture. Despite the moisture-resistant coating, you can see a large number of hydrogen bubbles diffuse — similar to the cellulosic electrode.
To show the beneficial increase in hydrogen diffusion at elevated temperature, a low-hydrogen electrode weld is allowed to sit at room temperature. The hydrogen bubbles slow down. Then it's placed in a bath of oil heated to a higher temperature. The temperature shown allows hydrogen to diffuse 1,000 times faster than room temperature. Diffusion begins again, and a rapid release of hydrogen occurs. Welds that we heat and allow to slow-cool will allow more hydrogen to diffuse, reducing susceptibility to hydrogen embrittlement. Remember, limiting hydrogen in the weld reduces hydrogen cracking susceptibility. A low-hydrogen electrode introduces less hydrogen than a cellulosic electrode. However, improperly maintained low-hydrogen electrodes can introduce approximately as much hydrogen as a cellulosic electrode. If it is not feasible to significantly reduce hydrogen introduced to a weld, careful control of preheat, interpass temperature, or postheat increases hydrogen diffusion and reduces cracking susceptibility.
§4. Diffusion times, helicopter steels, and the Sierras case [27:37]
[Video ends.] Okay, so it looks like much of the hydrogen's gone. It takes about a week or two for most of the hydrogen to diffuse out at room temperature, but if you heat it up to 250 degrees that goes a lot quicker. That's what we do for very high-strength steels. If you're building bridges, the AWS welding code says you must wait, I think it's 72 hours, before you do your non-destructive testing. They call it delayed cracking.
It turns out the ASTM specs, if you're heat-treating, say, a helicopter rotor blade or mast — which is very high-strength steel — can tolerate not four milliliters of hydrogen but less than one milliliter of hydrogen when welded. One of you asked where the hydrogen comes from. For a typical steel coming out of the steel mill, just because of the humidity in the air when they were melting it, you'll have one or two parts per million hydrogen out of the mill. Not a problem for lower-strength steels. When you weld, you can get that up to 30 parts per million. Grams per hundred is about a 1.1 conversion factor, so just take it as the same. That four-milliliter stick electrode is the best you can buy in normal commercial practice. Gas metal arc welding, which we're going to discuss, has even lower hydrogen — that's one of the reasons to go to it. But at very high-strength steels like aircraft landing gear or helicopter rotor mast, when you're at 250 ksi, you tolerate less than one part per million. That can be from the steel-making process alone.
Student: Can hydrogen cracking still occur a year or two years after the weld is completed?
No, not unless it's been reintroduced. It will cure itself if you keep it up. I know one case where it lasted a year, and that was in the High Sierras, where the weld was frozen all winter. If you want to keep hydrogen in steel: I had a student do his thesis 30 years ago on hydrogen cracking in armor steels. He was working for the Army over here at Watertown Arsenal. He would make his welds and within five seconds put them in liquid nitrogen, so he held the hydrogen in there long enough to do his hydrogen analysis. You all saw how fast it comes out. To get the actual hydrogen at the time of the weld, you can only extrapolate back, because you can't weld and do a diffusion analysis at the same time. Although Morris Cohen suggested that. That's where I learned about the impracticality of great academics.
Student: A story about the practical side of preheating: we were welding on a destroyer's rudder steel. When you have to preheat thick steel, that heat goes somewhere — it spreads. So if a welder has to get into a tight spot to weld, they're basically welding in an oven. You can burn yourself just touching the surface you're working on. The practical aspects of preheat can get pretty tricky.
§5. Seawolf, the Cole, and the blue jelly suits [32:07]
In fact, the Seawolf submarine had hydrogen cracking. I actually determined it was hydrogen due to a lubricant in the gas metal arc weld zone — not cleaning the lubricants off well enough. That was my conclusion. Electric Boat didn't like hearing that, because they had it left over from a previous job. When they started to repair it, they first had to dig all the welds out and then start re-welding. Congress was not happy at the time with two-billion-dollar problems. They actually started welding in what they call blue jelly suits. The foundations of the submarine are pretty heavy steel, a lot of it in egg-crate construction. You're crawling into some hole, and they wanted you to preheat to 400 degrees Fahrenheit. How would you like to be the welder in egg-crate construction underneath that? They actually put the welders on little wheeled carts, like a mechanic uses underneath a car. They put them in blue jelly suits, pumped chilled liquid through them, they were breathing air through a mask, and they had like ten minutes of welding time before they had to come out and someone else had to go in.
It's sort of like the USS Cole. They repaired the USS Cole at a cost greater than building a new destroyer, because they didn't want — whoever bombed the Cole — to think they had destroyed a capital ship of the USA. So they spent more money to repair it. That was sort of the Seawolf too, but with blue jelly suits. They don't use blue jelly suits anymore. The reason they had to in this case is that they could get all the up-to-two-inch-thick plate they wanted from the steel mill, but they couldn't get the four-inch plate for more than a year, for various reasons — capacity of the steel mills. So they had to reuse the really heavy steel plate, and that's what they were welding in the blue jelly suits.
When they went to other parts of the submarine, they actually just replaced — it's cheaper to scrap the old stuff. They had tighter chemistry control on the weld metal and on the base metal, so they didn't get to higher hardnesses and greater susceptibility, and they kept the hydrogen down.
§6. Bay Bridge corrosion-hydrogen and the Venn diagram [35:04]
These problems show up all the time. The Bay Bridge in California — it's coming from corrosion. They have big 46-inch-diameter steel tie rods, and they started corroding. The corrosion process creates hydrogen that diffuses into the steel — not as much as in welding — and they said, oh, they cracked within a couple of weeks. Well, that's hydrogen. The hydrogen will usually diffuse out within a few days. For federal highway work, you don't do your inspection for cracks until 72 hours later, because the cracks usually form in the first few days. The US Navy and some shipyards wait seven days. Make sure the cracks have formed if they're going to form. What good does it do to inspect if the cracks haven't formed yet and they form after you inspected?
They had a Venn diagram, the three circles. This is the way the corrosion folks typically talk about stress corrosion cracking, hydrogen embrittlement, and a number of corrosion processes. Frankly, hydrogen embrittlement can be considered a corrosion process. You have to have three things: stress, a microstructure that's susceptible, and in this case hydrogen. If you have no hydrogen, or no high stress, you won't have hydrogen embrittlement. If you don't have hard martensite — martensite is the most susceptible of all the steel structures to hydrogen embrittlement — under low levels of stress and hydrogen, you're fine. The reason they like a Venn diagram is the problem occurs at the intersection of all three. If you have less hydrogen, that becomes a smaller circle and doesn't intersect the other two. If you have less residual stress — that's why we stress-relieve — you don't intersect. If you don't have a martensitic steel — well, how do we get a non-martensitic high-strength steel? HY-80 and HY-100 are martensitic steels. We go to high-strength low-alloy steels, HSLA-80, HSLA-100. We didn't have that technology in the 1960s. We had quench and temper.
§7. HSLA steels and the Japan study [37:54]
The Japanese developed what they call accelerated-cooled steels. Actually they were first developed by Jones and Laughlin Steel in the United States in the 1960s for automotive dies, but we didn't do much with it. The Japanese had a huge shipbuilding industry and they wanted to be more productive to keep the Koreans from catching up with them in the 1980s. So they started developing high-strength low-alloy steels. The reason the US Office of Naval Research sent me to Japan in the mid-1980s is that the Japanese had the best technology in the world for making HSLA steels. We didn't even have a steel company that could make it. The US Navy was considering investing a hundred million dollars in a steel mill under what they call a Title III program, where the government can pay for the capital equipment to produce something the military felt was necessary. They wanted me to go over there and learn how the Japanese did their accelerated cooling to make higher-strength steels with lower carbon, lower hardenability, and therefore better weldability. High hardenability is bad weldability.
Student: The reason the high-strength steels need lower hydrogen — is it that they have more martensite?
The important thing about HSLA — that's why we chose LA, low alloy — is that you have less martensite. High-strength steels can have high hardenability, or they require very low hydrogen. Why? High-strength steels have more residual stress locked in. Remember — someone was talking back there about pulling pipes together, jacking pipes together, and when you let go they spring. Those are the locked-in stresses. The higher the strength of the steel, the higher the locked-in stress, the bigger this stress circle gets. The more martensite, like HY-80, the worse the microstructure circle. So you make this HSLA, that's better. You make a lower-strength steel, or you stress-relieve to get rid of the residual stresses, or you lower the hydrogen. We actually do all three because we don't want that intersection in the center. You can shrink each one of those circles in size if you're clever, but it costs money.
§8. Flaws, inclusions, and the limits of perfect materials [40:45]
Student: You know, GD nuclear power — they had brittle fracture very early. So our task is to make sure I'm not — brittle fracture is pretty distinct. In a hypothetical scenario, if you just create a material with no pre-existing flaws, then you can prevent brittle fracture. In practice that's impossible. But that's different than what we're saying here. We're saying the atomic structure itself permits hydrogen to diffuse in.
You always have flaws in the steel — microscopic, smaller than a human hair. Seriously, microscopic flaws. The hydrogen will diffuse to those. If you go on YouTube you can probably find somebody from the 1950s or 1960s at Rensselaer, Apollo Tiffany [?]. They took a piece of steel that they welded and polished — or they may have introduced the hydrogen not by welding but electrochemically, through a rapid corrosion process. They put a notch in it, just like a torn piece of paper, and they put tension on it. They put some blistering on top, and under a microscope they could see the bubbles coming right out of the crack tip, because the crack tip is larger in volume — you're stressing in tension. The hydrogen wants to diffuse there, and as it concentrates at the crack tip, it allows the crack to progress. You can see in the microscope the crack growing, and you can see the bubbles coming up right ahead of it, because the hydrogen wants to diffuse to that crack tip — unless you can diffuse it out to the air faster. It will start even at these little inclusions.
The Naval Research Lab found in the early 1980s that they could weld HY-80 with lasers and have better toughness and more resistance to hydrogen in the weld pool, because the laser actually vaporized away the inclusions in the steel. They ended up with a super-clean steel from the welding process. A scientific curiosity that proves the point: if you're ultra clean, you can resist this — but you can't make steel that clean on a tonnage basis. Maybe a hundred years from now we'll have some way to do it. Today we make steel thirty times cleaner than it was 50 years ago. That's why we don't have laminations, why we don't get crap when you put it in a stress-relief furnace.
But we still can't make perfect materials. A lot of what you hear about new materials — the nanowires, nanotubes, graphene, all this stuff — is based on some physicist calculating on a piece of paper what he thinks the properties of an absolutely atomically perfect material are. Bob Sprague, the head of GE Aircraft Engines 30 years ago, said: whatever you first hear about properties of a new material, write it down, because those are the best properties of the material that will ever be reported. And Jim Williams, who replaced him, used to say, his corollary is: whenever you first hear about the cost of a new material, write it down — that's the cheapest the cost is going to be.
Bob Sprague had another quote. He said physicists think that structure controls properties — that's the big thing from the DMSE materials science department here. Materials scientists, who are wannabe physicists, also think that structure controls properties. Metallurgists know that defects control properties. How big is that little notch? Even microscopic notches in some cases are what limit your properties.
I passed around the other day a sample that was electron beam welding overcladding a nickel alloy on top of a piece of steel. That produced some of the cleanest microstructures, because you're electron-beam-refining every little weld pass. They're kind of pricey, but you can make very clean material. So when we're making steel in space with lasers — that's how the higher quality is going to come. You want to take your submarines and build them up in space. The Air Force might actually believe that. Let's start building submarines in space.