§1. The Miami Art Museum hanging garden [00:03]
There was a six-story building in Miami, with concrete beams forming an overhang at the top — these beams came out almost half as wide as the building, maybe 20 or 30 yards. Underneath, you could have people sitting outdoors in the good weather, in the winter in Miami, at little tables, having their chardonnay and eating their hors d'oeuvres. And this was going to be a hanging garden, because it doesn't freeze in Miami.
And this architect — Swiss architect — decided to take these beams and do a tension connection. These are 20-ton concrete beams. They're reinforced concrete, but rather than sticking it on top so it's in compression, with the joints in compression, he decides to make a tension joint. Stupid. And then they were going to have plantings on top, so you'd be like in a jungle. I'm told it's a very nice building.
In August of 2012 they actually hung the first couple of beams. They were using a martensitic stainless steel — they wanted high strength and some corrosion resistance, so they used a 410 stainless. These 20-ton beams were held by two three-quarter-inch threaded rods. You're going to hold 20 tons with two three-quarter-inch rods. Stupid. And they were supposed to tack weld the threaded rod up above where it came through. They hung these things up there, and one night a couple of them fell down. It was nighttime, no construction workers got hurt — they just dropped a couple of 20-ton beams on the veranda.
Afterwards they looked, and they had put these little tack welds to keep the rods from un-threading. The engineer for the architect says to the contractor, why did you weld them? He says, have you looked at your drawing? On the drawing there's a weld detail: weld it. That was why the contractor welded it. It turns out the prior February, the people designing the concrete beams — who are concrete specialists, not metal people — had sent a little note to the engineer saying we don't think this is the right stainless, but we're not stainless steel experts, you should ask the owner if this is the stainless steel he really wants to use. And the engineer sent back a note saying we know all about this steel, don't bother us. So they didn't bother them. They hung it in August, and the thing came down a couple of days later. Delayed cracking. Rings a bell, right? High-hardness martensitic stainless.
So they repaired some of those, decided to put nuts on or something to get rid of some of the welding. In November they hang some more. And they come down about four weeks later. Everybody's all up in arms, and somebody decided they shouldn't have used martensitic stainless steel — so they decided we're going to use 316 stainless steel, because it has very good chloride corrosion resistance. They order something like seven million dollars worth of steel to do this. They're going to get behind on their contract, and they're starting to fight over whose fault it was, who picked the 410, who welded it, why these other things came down during construction.
§2. Consulting the case: ethical dilemma and the 316 decision [04:44]
Eventually the local general contractor calls me up and says, can we meet with you tomorrow? I said no. I said I can meet with you Friday. Friday morning at seven a.m. they're in my office, explaining this whole story to me. They'd sent me some documents in between that I'd read, and I'd sent the guy back an email saying stupid stuff is stupid — no one should have ever used martensitic stainless steel here. Basic fundamental mistake. No one should have a hanging tension connection — these were supposed to be a 120 ksi or 140 ksi steel. Just dumb. And then welding it in the field.
They had hired me — they wanted to fight over who was responsible for choosing the martensitic stainless. I said, so are they building the building? They said yeah, they switched to 316. I said, how long's your warranty? He says a year. I said, well 316 will probably last a year. This building's supposed to last for 50 years, right? He says, what about 10 years? I said well I don't know, 316 will last 10 years. I said, what's 10 years? He says, statute of repose — if it lasts 10 years you can't sue the original builder after 10 years. I said well I don't know, you may not make 10 years with 316. So this is actually how the conversation went.
A couple weeks later I have to fly to Miami to meet with the owner, the architect, the contractor — big meeting, 40 people there. I get up and I say as a professional engineer I could not sign off on this design. I'm talking about the 316 now — they just ordered seven million dollars worth of 316 to finish building the building, and they're starting to hang it. The engineer who's responsible for having selected this material is defending it. Oh there's nothing wrong with this, you don't know anything about metallurgy. Oh thank you sir, I think I do know something about metallurgy. They had hired a civil engineering firm out here in Waltham — I used to be their metallurgist for about 30 years. They were actually coming in to advise people, only those people knew me well enough that they weren't going to go against me, because I'd been telling them what to do for 30 years.
We have this meeting and the owner is just sitting there. Who's gonna pay for this 10-million-dollar problem on my 70-million-dollar building? But he's sort of aligned with the architect and the engineer, because he's closer to them. The general contractor — the guy who owns the general contractor flies around in his own private jet, he's fairly well-to-do — but he said, we're going to do the right thing here. If Professor Eagar says this is going to fail, we're not going to hang any more steel. When he says that, they could turn around and sue him for the whole delay of the project — tens of millions of dollars. So all these people are positioning themselves and I'm sitting there saying, well what's going on here.
I remember the Hyatt Regency collapse in Kansas City back around 1980. They had walkways — it was an indoor atrium, and they had tension threaded rods. They weren't stainless steel. They actually had a shear connection. The designer had designed it as a straight-through connection, which may have been strong enough, but you couldn't run a threaded rod for 60 feet. So the erection detail had offset, with one threaded rod with a nut and another one with a nut on the top — it was a shear connection. And all of a sudden that shear connection couldn't take the load, and the big party, a couple hundred people died. Came crashing down four or five stories on the inside. So I'm sitting there thinking, this is sort of like the Hyatt Regency — except here you've got people sipping chardonnay down below rather than up on the walkway, so they wouldn't fall, they'd just get crushed. Hyatt Regency, but still a tension connection. Stupid. Super stupid. Why do you make tension connections?
§3. The letter, the duplex solution, and the open litigation [09:15]
I had the same sort of ethical dilemma. What am I going to do? I called up the attorney the next day when I got back — or maybe we said it on the way to the airport — and said, I've got a problem. If those guys are going to keep building this thing out of 316, I have to notify — I have a public responsibility as a PE, as a professional engineer. I can't let them continue to do that without warning somebody. I can't stop them from doing it, but I've got to tell them they have a serious problem. He says, well just write me a letter. Next morning I get back to my office and write the letter, and by 10 o'clock he's got a copy. He says, oh don't write it to me — write it to the general contractor, rather than to the attorney. I thought they were going to take this and read it to the other people. No, they mail it to the owners, which means they're now on notice. And they really have to show that to the City of Miami building authorities, who can shut the project down, because they've been notified.
It was a fairly firm letter, that this was a poor choice of material. In the meantime, the engineer goes back and talks to his metallurgist in Sweden — they were using a Swedish steel company. They also asked me what material could you use? I said, there are some nickel-based alloys you could use that have very good corrosion resistance in seawater. You might be able to use a duplex stainless, but I'd have to look it up and research it, see if it'll be good enough at these stress levels in this environment. I said, I don't know off the top of my head. That's when I found out — before I thought there were just a few of these — I didn't know there were seven million dollars worth of metal here.
The first response from the engineer was, well, we're 50 yards from Biscayne Bay and we're up six stories in the air, so it's not a marine environment. Pretty good, huh? I thought that's cute. I said, did you ever hear of hurricanes? They kind of come through Miami every now and then, and they have a salt spray that comes with them. It'll get up six stories in the air. I think that's a marine environment. I'd given them some references of how 316 in tension can cause stress corrosion cracking, particularly if it's been welded.
It goes on for about two or three months. The engineers out in Waltham — we were sort of on opposite sides, but sort of trying to work together. I'm not going to design this building for them. I didn't design the stupid tension connection to begin with, and I would have changed it to a compression connection where I don't have tensile stresses to cause cracks. I'm not going to start telling them how to solve their problem — they've got to make the decision, otherwise I'm the designer, right? And if anything happens, I've got problems, and no one was paying me enough money to take on a 100-million-dollar liability. These other engineers were working with the guys in Sweden, and they probably did a million dollars worth of tests. I'm just sitting there sending them paper saying this is a bad choice.
They finally did go to a duplex stainless steel. They got some of the duplex in a big rush. By June or July they were putting up duplex stainless steel, and they had their opening in May. But they're still going to have a fight over who chose the martensitic stainless. The engineer is still trying to defend their choice — that there was nothing wrong with martensitic stainless. I wouldn't accept that from a first-term sophomore. And now that I put it on, it'll go on to YouTube, and they can probably play this for me at the trial. I don't really care, because those are idiots, pure idiots. I couldn't say that at trial, but if they want to play that part of this video at trial, it's okay with me. I can't get up and call them idiots live, but when I'm talking to my students, I can say how stupid this mistake was. And this is one of the world's larger engineering firms — they've got hundreds of engineers all around the world. The problem is, when you're trying to use a sophisticated material — stainless steel in a marine environment susceptible to corrosion cracking — you've got problems.
§4. Ferritic stainless steels and the ductile-brittle transition [14:19]
Let's go to ferritic stainless steels. Some of the medical instruments are actually ferritic stainless, like a 430 — they don't turn to martensite, because of the ratio of carbon and nickel and chrome. The problem with the ferritic stainless steels is the same thing we had with the Liberty ships, which were not stainless steels obviously, but carbon steels. Ferritic steels have something called a ductile-brittle transition temperature. So it's impact energy versus temperature. Remember I told you with the Liberty ships, we learned it's not just the force of fracture, it's the energy of fracture. This is what Pellini at Naval Research Lab and Morris Cohen at MIT and other people had known about for years, but they didn't really learn how to design with it until the 1950s.
The steel will change its resistance to fracture. It'll be brittle fracture down low — is this in joules? So 20 joules is 15 foot-pounds. Typically the Coast Guard uses 30 joules, or 20 foot-pounds. The U.S. Navy often uses the same thing. This is brittle, very brittle down here, barely acceptable here, and a good piece of steel is up here at 50 or 80. This is an as-received steel, this is a shielded metal arc weld — and you can see we shifted the ductile-brittle transition temperature from around 30 or 40 degrees. The midpoint is often called — there are lots of different areas on this curve that people pick, but let's say this was good for 40 degrees Fahrenheit, or 40 degrees centigrade, for these ferritic stainlesses. If you weld it, you increase the temperature to well above room temperature, or anything you're going to use. If you do gas metal arc welding with a higher heat input, it makes it even worse. If you do a simulated heat-affected zone, you get grain growth from the welding, and you can destroy your toughness in these things if you weld them.
That's one of the problems with ferritic stainlesses. We do sometimes weld ferritic stainless, but usually only if it's sheet material. You don't have to worry so much about brittle fracture — we have a different type of stress state called plane stress rather than plane strain. So I'm not going to talk a lot about ferritic stainlesses, because you're not going to use them in heavy sections. Even anything above an eighth of an inch is a heavy section for ferritic stainless. They are used, but you have to be really careful about how you use them.
§5. Sensitization in austenitic stainless steels [17:12]
The problem with austenitic stainless steels, things like the 316 — if it's straight 316 and you make a weld, in the heat-affected zone there's a particular temperature range, between let's say 600 and a thousand degrees Fahrenheit, where you can precipitate chromium carbides if you have more than 300 parts per million carbon in your steel. Here are your grain boundaries — you actually precipitate chromium carbides. The carbon can diffuse from large distances, because it's a nice light small element. The chromium can only diffuse from a few — maybe 10 microns away — during the time you have, and so you get a chromium-depleted region right in here. There's nothing wrong with the chromium carbides; the problem is the chromium-depleted region.
Another view of it: here's your carbide, which is very high in chromium, greater than 70 percent chromium. This is chromium versus distance across a grain boundary. You have this little chromium carbide very high in chromium, but the carbon has diffused from far away, the chromium diffused from a short distance, and you end up with a chromium-depleted region. You get down to six or eight percent chromium in that depleted region, which is less than the 12 percent that usually gives you your stainlessness. So now you have a region at the grain boundary that's depleted in chromium, and this becomes an anode compared to — this is the cathode. I have a huge area, and I just stress corrosion crack right through this.
If I use the right type of etchant — a 10 percent oxalic acid etch, there's an ASTM procedure A262, which has a bunch of different procedures for testing for what we call sensitization. This is called sensitization. You can take this steel — it'll be fine as it comes from the steel mill, everything cooled down at a uniform rate. You weld it, and now at the grain boundaries the chromium has formed carbides if you have a little bit of carbon in your steel. And you run into very sensitive, easily etched grain boundaries. This picture actually came out of the ASTM specification — they didn't reference it, but they stole it.
§6. Argon-oxygen decarburization and the 304L revolution [19:55]
So what General Electric and other people did, because they didn't think they would run into this problem — they basically went from 304 stainless to 304L. In fact, they got the carbon down to less than 100 parts per million. 304 stainless has a specification — the carbon can be between 300 and 800 parts per million. To be in spec, that will cause chromium carbide precipitation during welding. If you go to 304L, which we've known about since the 1940s — but until about 1960 it was very expensive to produce. You had to melt your stainless, take off the slag, put another slag on to try to draw the carbon out.
I mentioned this — there was a guy who did his doctoral thesis in the late 50s here, in the basement of Building 8, who found a way to bubble argon through the bath of molten stainless and remove the carbon to very low levels. It used to double the cost of stainless steel back in the 1940s to go from 304 to 304L. All of a sudden it costs almost nothing to go through this process, which is called argon-oxygen decarburization. You take the carbon out of the stainless steel — you don't have to do a double slag practice. Almost all of the stainless steel today, 304 is 304L. Certainly anything other than sheet material. Some sheet material is still straight 304, some rod that's never going to be welded is 304, but most of it is 304L.
What General Electric did for the reactors — they actually went to an ultra-low carbon, with a specification that it had to be less than 100 parts per million. But then they found the carbon was adding strength, and their stainless steel didn't have enough strength when they got below 300 parts per million carbon. So now General Electric uses LN steel, where they go to very low carbon and add some nitrogen back to replace the carbon, to bring the strength back up. That's an extremely easily welded steel in terms of carbide precipitation and stress corrosion cracking. So they got around a lot of stress corrosion cracking problems that way.
Here's the time scale for stress corrosion cracking — I think I said degrees F, so it's between 600 and 900 degrees centigrade. This is the regular 304, .08 carbon max. This is the bottom range for the 304. So 304 is in here. Within less than a second you can form these chromium carbide precipitates that will lower your chrome at the grain boundaries and cause stress corrosion cracking. We get intergranular cracking in the grain boundaries, which looks sort of like hydrogen cracking in steels — but remember I told you stress corrosion cracking occurs at the anode and hydrogen cracking occurs at the cathode.
If I go to less than .03, which is the 304L, then I can change this to many minutes. Certainly my welding is going to be here, and the cooling of even a big heavy plate is going to be in 1 to 10 or 20 or 30 minutes, depending on what I'm welding. So if I get to very low carbon, essentially less than 100 parts per million, I can get rid of the chromium carbide precipitation.
§7. Chloride cracking, nickel content, and the superferritic chimney liner [23:55]
I'll show you what a cracked material might look like — a little bit scary. This actually is caustic cracking, but it's transgranular stress corrosion cracking. That's quite a bit of cracking in that 308L weld metal. This is carbon steel shell, 308 weld metal, but 316L tube sheet in some heat exchanger, and they had some caustic sodium hydroxide solution, and these things just crack. That's basically what happened to my stuff — my hot dog cooker — it was basically stress corrosion cracking, although that was in the base material.
Just to show you one last thing on chloride cracking and stainless steels before we take our break. If you look at nickel content versus time to failure in hours, this shaded region represents the nickel range of many austenitic stainless steels. We tend to use austenitic stainless steels in this range. If you test in boiling magnesium chloride — which is a standard ASTM test for looking at chloride cracking resistance of stainless steels — we have things that will fail within an hour. It is a chemical reaction. It's not necessarily hydrogen exactly, but it's similar — you're eating things away.
If you have much higher nickel, like the Inconels with 60 percent, you won't have this type of problem at all. That's one of the reasons for going to the Inconels: generally you don't have problems with chlorides. Or if you go to the ferritics. But the problem with ferritics and martensitics is, the martensitics will hydrogen crack, and the ferritics have grain growth and impact toughness problems. But you can go to some of the ferritics when you want chloride cracking resistance.
Student: [asking about high-efficiency boilers]
Has anybody got a very low temperature boiler in their home that actually exhausts wet steam out the side of the house rather than hot steam going up the stack? A high-efficiency boiler? You do, okay. I've got one too. They exhaust out the side, and you see steam coming out. For me it's five yards from my front stoop, and when the boiler goes on, you see this steam coming out behind my rhododendrons, because that's where it exits.
I had a fireplace and I said, well can't you send it up the stack? They said only if we line the stack with this special stainless steel. I asked what type, or I looked it up and found out — it was one of the superferritics. When you're burning gas or oil, they're worried about the carbon monoxide and there might be some chlorine if you're an oil burner. I'm burning gas, but there shouldn't be chlorine in my gas. But some people are burning oil and there'll be some chlorides in your oil. They don't want cracks in your liner for your chimney. They want a superferritic that can take moist chlorides and not crack.
They were going to charge me — this was in my poorer days — two or three thousand dollars to line my stack. Back in the 1970s, I could have paid my mortgage for six months for whatever they wanted. So I didn't do it, and I just blew the steam out the side, which is what I'm still doing all these years later. The superferritics can take moist chlorine, hot chlorides, and not crack. Austenitics will be gone in an hour. So there are regulations, and I still steam-heat my rhododendrons during the winter, or whenever we're using a lot of hot water, because my boiler runs my hot water. Okay, let's take a break for seven minutes.