§1. Reference materials versus standards [00:02]
I have something to hand out. Since it would be very difficult to figure out my outline, I decided to give it to you. This is a beta test course — only the second time I've tried to teach it. I don't know of anyone who teaches a course on codes and standards at a university, so I'm going to want your feedback on whether this was a worthwhile endeavor at the end.
What happened last fall is, I'd been dealing with several problems — I'll actually talk about one of them today — and I realized that a lot of the problems you face out there have nothing to do with the calculus you take. Basically people don't understand how codes and standards work, how purchase orders work, what to do when you have conflicts. That's one of the reasons for doing this. You might say, well, why do we spend the time going through observables and measurements at the beginning?
Does anyone know the difference between a standard and a reference standard? I wouldn't expect you to, but I'm asking.
Student: Precise...
That's close. A standard is something like an ASTM standard — I handed out the one for a seamless carbon pipe, and we're going to go through some of that today. But a standard reference material is actually a NIST term. NIST is the National Institutes of Standards and Technology, down in Gaithersburg, Maryland. They're also located in Boulder, Colorado. These are the folks that run the U.S. atomic clock out in Boulder that keeps time for the United States.
They have a standard reference time machine. You can log onto it, you can buy a watch that will synchronize with it. I have about five clocks at home that synchronize with it. There's one in my office but it doesn't synchronize because Building Two's in the way, so I have to take it home to get it synchronized. After it gets synchronized, that clock doesn't lose more than a few seconds in six months on its quartz movement.
These definitions come off the NIST website. A reference material is material sufficiently homogeneous and stable with certain properties. So a standard reference material could be those Johansson blocks I showed you that are accurate to fifty millionths of an inch in length. They're ground. That's a physical, tangible object that is homogeneous and is traceable back to something. So that's a reference material — not a standard reference material, just a reference material.
A standard is usually a document, a piece of paper, as opposed to a material. Other reference materials are hardness test blocks. I could have gotten one out of the lab — just a little piece of steel that has been manufactured in a very reproducible way, so it has uniform properties. You can take your hardness tester in your laboratory, do the hardness indentation, and check whether your hardness tester is giving you a number within the specified limits of that little hardness test block. That's a reference material.
Down here you have certified reference materials and standard reference material — NIST, registered trademark. NIST has a business of selling these things. Unfortunately, the government will not allow them to sell at a profit, so they have to sell at cost. It's something the Department of Commerce doesn't really like to do, but they have to do it because we need something for mass, length, time, chemistry, hardness. When I looked online this morning, they had standard frozen urine specimens that have certain toxic levels of mercury and other things. If you're going to measure for mercury or lead in someone's urine, you have to have a standard.
People will get that you have to have a standard to measure to. These paper standards essentially call out measurement, tests, and so on. So these two things fit together — standard reference materials and standards. The standards are the documents that tell you the procedures, the tests, the properties, the mass, length, time, magnetic property, whatever. The reason I've gone through properties and measurements is because these standards call them out, and you have to understand where they come from. You also have to have a way to measure them. Remember I started out with Lord Kelvin: if you can't measure it and put numbers on it, you don't know anything about it. You have to have these in order to see if you conform to that standard.
Now the problem is, most people in the world — unless you go to NIST where these people live and breathe standards — will just talk about a "standard," a "hardness standard." We don't say "hardness certified reference standard material." But if you use the word "standard" at NIST and you use it improperly, they'll say, oh no, that's a reference standard, because they live and breathe this. The rest of the world doesn't, and so there's a lot of confusion when people use the terminology. Just use the word "standard." That's one point for today.
§2. History, scope, and fair use [08:09]
I did this outline last fall. I had my secretary type it up yesterday when I was looking at my notes for today. We've gone through some of these things. We went through history yesterday. Who writes the codes? Benjamin Franklin did a test, and then I gave you that paper. The city of Philadelphia started suggesting that people have lightning rods on their houses. Then it moved up through the 1800s, and people learned things. Michael Faraday came along and taught people scientifically about Faraday cages.
Frankly, if you look at a lightning protection system for a home today, it's nothing more than putting a bunch of metal straps around your house with a terminal at the top to try to make a Faraday cage. Does everybody know what a Faraday cage is?
Student: Yes, we had them on the missile handling building in Georgia, but it was to prevent radiation getting out.
Student: For communication purposes — basically can't eavesdrop on the building.
Okay. If you try to use a cell phone inside a galvanized steel building, it's very difficult, because you're inside a Faraday cage. The radiation on the outside can't get in past the metal. Michael Faraday learned that a cage — and it could be like a bird cage — the electromagnetic radiation on the outside can't get to the inside, and anything on the inside can't get to the outside, if the spacing between the wires is small enough. Then depending on the frequency and the distance.
If you want to see a big Faraday cage, go over to the Science Museum where they have the lightning show. They have these big bird cages about two stories tall, and inside they have a Van de Graaff generator to generate lightning. Everybody's safe standing there about a hundred feet from lightning bolts because the bolt is inside the Faraday cage. A Faraday cage is just a metallic enclosure. Many of your laptop computers have a metal case. Even if they don't have a metal case, the plastic case probably has metal powder molded into it to make a Faraday cage, for noise rejection.
Student: [Ours was] for communications protection. It was just four towers, with cables strung between them.
So it's a simple X between them, probably 300 yards from one tower to the other and then one going the other way. There are some good scientific ways to design these things for different frequencies, but it's just a metal enclosure. One of Maxwell's equations describes why it works — basically talks about the charge enclosed inside a circular integral. If you look at the integral form of Maxwell's equations, it will tell you that there is no field inside a volume that doesn't have a charge inside.
Michael Faraday first discovered the principle of a Faraday cage back in the early 1800s, which was after lightning systems and Ben Franklin. There's that little paper that tells you how the lightning protection system standard went through. I mentioned the Iowa State Fire Marshal in 1926, and they had all this data on lightning storms and burning down barns. So there's a history to these things.
We've talked a little bit about the boiler and pressure vessel code. It came up because they were having serious loss of property and life because steam boilers were blowing up on a fairly regular basis. They still blow up, but not as regularly. A lot of this started out as service, but I brought up last time that in the last fifteen or twenty years we switched to a for-profit, not-for-profit, and for-profit folks.
There are scopes and limitations to these standards. Number one in almost all the ASTM standards is scope. "This specification covers" — those three words are the beginning of almost every standard. So it defines the limitations of the code. There's also fair use, which I'm not sure I could point you to any text that discusses. For example, Section 9 of the boiler and pressure vessel code is a document about this thick — the real document is about three-sixteenths of an inch thick, but all the appendices and addenda make it about an inch and a half thick.
Section 9 of the boiler and pressure vessel code is welder qualifications and certification. The scope of the boiler and pressure vessel code only applies to boilers. It excludes human-occupied pressure vessels. It excludes hot water tanks for your home. I told you about the DOT cement tank trucks — it actually didn't exclude those, because people thought that the DOT covered that.
Under that scope, even though something doesn't come under the code, doesn't mean that Section 9 — which tells you how to do a very good job of certifying and qualifying welders — isn't a good manufacturing process that should be followed when you're trying to do other critical welds that aren't on pressure vessels. I get asked a few times a year by some company to write up a welding procedure for some critical application. It may not be a pressure vessel, but I often use the boiler and pressure vessel code Section 9 for the qualification and certification of welders, because a welder certified and qualified to the boiler and pressure vessel code has a certain minimum level of competency.
It's not what I call a farm tractor welder. A large fraction of the welders in the country actually learned to weld on the farm repairing farm equipment. They never had any real training. Some of them are very good and some of them are very bad — there's no standardized minimum quality. But if someone's passed the requirements of ASME Section 9, they've passed a standardized test, and that level of competence is actually fairly high. If someone tells me they're an ASME code welder, I know they know something about welding. They can hold an electrode and do something of value with it.
If they tell me they've been qualified to the structural welding code — there are two big codes for qualifying welders, and the structural welding code from AWS is the second one. This is for bridges and buildings. Most welders in the country are certified to one or the other. The structural welding code does not have as stringent requirements as ASME Section 9. So if someone comes to me building something new that's not a boiler, a pressure vessel, a bridge or a building — it doesn't fall under any of these codes — does that mean I can't use part of that code and call it out as a useful standard? I don't have to write a new standard, and it doesn't have force of law because it's not under the scope of the code, but it's still a good manufacturing practice. So the better codes actually get used beyond their scope on a regular basis because they're just good practice.
That's what I mean by fair use. I've had people argue with me about fair use a number of times. They say, well it's not required by the code. Just because something's not required by the code doesn't mean the code's not a good minimum-level standard, and you have to decide which codes you want to use.
§3. Conflicts, force of law, and interpretations [17:35]
Since codes do overlap and you can have conflicts, some welders are qualified to both the boiler and pressure vessel code and the structural welding code. There are multiple structural welding codes — there's the one for steel and the one for aluminum. AWS D1.1 and AWS D1.2 — very clever terminology. There are different qualification requirements, because there are different skills for welding aluminum versus steel.
When you run into conflicts, which governs? Some codes are in the Code of Federal Regulations, which means they're law. Does anybody know what the CFRs are? Code of Federal Regulations. Multi-volume set, goes into excruciating detail in certain areas, but it's because something happened once and someone got a congressman to write this into law. Many times the Code of Federal Regulations will call out one of these industrial codes like the boiler and pressure vessel code, and so all of a sudden the boiler and pressure vessel code inherits the force of law because it was written into the Code of Federal Regulations.
When OSHA was formed thirty, forty years ago, they went through and looked at all kinds of codes, and they just wrote them into the law. OSHA is part of the Code of Federal Regulations. If you go through the OSHA documents in the CFR, you'll find references to all kinds of other codes. They all use each other, build on each other. There's a certain collaboration between them. But since they weren't all written by the same person — in fact they're all written by different committees — the codes have problems of conflicts. That's what keeps me in business to a certain extent. There's not a simple answer to what you do when there's conflict in the code. If there was a simple answer, I wouldn't be able to work.
On the problem set: I think I told you on the first day, last fall — actually it was very successful — I said, you're going to have to pick a single code, a simple code (I don't want the Code of Federal Regulations), and you're going to do a ten or fifteen minute presentation on it: the history of the code, the scope, what other codes it calls out, the reference documents. So it could be something like ASTM A106, which I handed out on pipe.
Kathleen, decide what they want to do. Get together with them — talk to them. Consensus. Let's do a committee here. Whether you want to do the problem set, which you can all do at once, or whether you want to do individual presentations. I actually learned quite a bit from the presentations, which is why I'm willing to sit through them, but you all have to sit through them too. We could do probably three a day, so it'd be about three days' worth of work. It's actually more work to put together the presentation and study the code. One person was interested in bayonets — the Army has a code on, if you're going to sell a bayonet to go on the muzzle of a rifle to the Army, you have to meet this code. I didn't even know there was a bayonet code, but I now do know there's a bayonet code.
That's about all I want to say on scopes and limitations. The code will define its scope and the limitations, but there is fair use that goes beyond that. Conflicts we've talked about. Interpretations — even though committees have written their best job, they don't always answer every question, or some people don't know how to read the code. Different codes will come up with interpretations. You can write a letter and ask something like: does qualification as a welder under Part C Section 5 qualify the welder as a welding operator under Part D Section 5? This is an interesting question, I'm sure it's something you've struggled with all your life. The answer comes back probably about six months later, because the committee has to meet, study the question, come to a consensus answer. The answer is no.
Some of the answers — there's one here on ultrasonic testing where the question goes on for a page and a half and the answer goes on for a page. This is the list of all the questions and answers from 1976 to 1999 that were asked of the structural welding code committee. Here's a historical record. Back in the late '90s when the Welding Society and a lot of other societies decided, oh there's a gold mine here of selling these documents — all of this used to be in the records at the American Welding Society. They said, we can dig out all our records, print it up, and sell it. So now you can buy the interpretations. It's another product that can be marketed.
Many of those questions that come into the committee — some are difficult questions that don't have a simple answer — the committee will study them, and they may revise and improve the code. So codes get thicker with time. Well, they don't always get thicker. I actually brought you an example of a code that got smaller with time.
This is the 1997 structural welding code for aluminum. It's this thick. This is the 2003 structural welding code for aluminum. They're not the same thickness — the 2003 is thinner. Why? It wasn't thinner paper. 24-pound paper versus 16-pound paper, no, that's not the answer. Smaller type? No, it's not smaller type. Section, Chapter 2 of these structural welding codes is on design. What happened is in 2000, the Aluminum Association came out with the Aluminum Design Manual. Chapter 2 is much shorter than it used to be, because now they just call out this.
So did it get smaller? No, it got a lot thicker, but the document itself — that code now calls out this code, which didn't exist until 1994. The first version of this came out in 1994. By 2000 they'd come up with an updated version that was a little better. They essentially wrote the design out of the full red one and put it into the red and white, and simplified the red and white.
Another example of how codes develop historically: back in 1989, the US Department of Commerce at NIST, where they have the Fire Protection Center, decided to write a fire investigation handbook. They used to have a fire technology lab, but they closed it down a year or two ago. This is basically 1989 technology. Someone just typed it up into a word processor, a lot of things from a lot of scientists — this was just a guide that NIST came up with. That evolved into something I brought in a couple of times. The NFPA took that document, formed a committee, and now has this thickening code. I have the 2008 at home. I think I have the '99 code here. The '99 code is half as thick as this, which is twice as thick as the '89 fire investigation handbook. Historically, that's how this code came about — all in twenty, twenty-five years.
There are interpretations. You can ask the code committees, if they are standing committees, and they might meet twice a year typically. They'll put it on their agenda and answer the question, give you a long, complicated answer like yes or no.
Student: Are the interpretations considered [code]?
No, not until the code is revised. It may get revised and go into the code, but the interpretations don't necessarily hold the force of law. Some judge might be able to decide whether they could be admitted as an interpretation, but that's beyond me. You're really only responsible for the standard, and if you seek extra knowledge from the interpretation, great. You're not responsible for knowing the interpretations.
To expand on that, this structural welding code — about 60% of it is the code, 40% is annexes and commentaries written by the people who wrote the code. If you read the commentary, it's "this is what we mean." The codes are written sort of tersely on purpose. They're written in legalese. People found that rather than having to respond to a bunch of interpretations, they've actually started including commentaries in the code. This isn't the only code — the boiler and pressure vessel code, I think, has commentaries. The codes are getting more and more complex. But the code itself is just the code, without its commentaries or its interpretations.
§4. Collaboration, competition, and ISO 9000 [29:30]
Collaboration and competition in codes. There's certainly a lot of collaboration. I told you about how NASA, DOD, and the FAA got together to come up with the Aerospace Materials Handbook. It's this thick in 13 volumes. This happens to be Chapter 3 on aluminum alloys — lots of data on fatigue strength, creep strength, tensile strength of a whole range of aluminum alloys. That comes from the aluminum companies — Alcan, Alcoa, Kaiser. Boeing is not an aluminum company, but if a company wants to submit that data for evaluation, then they can include it in the code, and the FAA says, if you're going to build an aircraft, this is the minimum standard for the material properties. If you want to use something else, you'll have to justify it, but if you use this, we'll accept it and we won't make you jump through hoops, because we find these are the standard properties you can expect for this aluminum alloy under these conditions.
That's collaboration. Collaboration in codes got to be a big deal in the early '90s after the first Gulf War. A couple things came out of the first Gulf War. One was rapid prototyping became a big deal. The war didn't last long enough to do a lot of research for the war. What — a week, two weeks or something? But there were things that were needed once they started flying helicopters over in all the dust in Saudi Arabia. They found that all of a sudden the sand was sandblasting the engines, and they weren't lasting hundreds of hours — just a few hours.
So there were a couple of major failures in that first Gulf War and a couple of major successes. Remember they took about six months to move everything over there and prepare for the invasion. They had plenty of time to do all the logistics to get everything over to Saudi Arabia before they actually did the advance. The advance didn't take very long, but they learned things even before that. In about a six-month period, they had to identify a problem and solve a problem. There were some problems they never quite solved that cost a lot of money. There were other problems that they did solve that saved a lot of money.
One of the things that came out of that was ARPA, the Advanced Research Projects Agency, started a whole major program on rapid prototyping — how you could conceive, design, and develop something and get it to market in a very short period of time. One of the great failures of those types of programs was NASA tried to develop the X-33 space plane on a thirty-month schedule, from design through fabrication to test flight. The whole $1.3 billion program was a flop, because we didn't have the ability to do it in thirty months. They gave it to the Lockheed Martin Skunk Works, and they ran into some problems. You don't have a lot of time. If everything goes well, you might be able to do it, but not everything goes well in research.
Rapid prototyping became a big buzzword in the early '90s, in part because of the Gulf War. The other thing that became a buzzword was dual use technology. The Democrats came in, and peace was breaking out with the former Soviet Union. The defense budgets were going down. The way the Democrats decided to balance the budget, as one of the secretaries of defense said, was: we're going to leverage our knowledge of industrial uses and standards and combine it with military needs.
Let me back up. There was a predecessor to that. I ended up getting involved in the mid '80s during Star Wars. Ronald Reagan and the Republicans said, we're going to be able to build this mystical defense system called Star Wars — we'll have fifty-megawatt lasers up in space and we'll shoot down any missile in the boost stage that's coming out of some other country, and we'll protect the United States that way. That's basically how Reagan won the Cold War. He bankrupted the Soviet Union — they could not keep up. Reagan also introduced the idea of deficit spending, which the Republicans never believed in until Ronald Reagan, and Ronald Reagan taught the Republicans that you can buy votes with deficits. Now we have a competition between the Democrats and the Republicans of who can buy the most votes with the deficit. This is the finances of the United States.
The Republicans bankrupted the former Soviet Union by doing this Star Wars buildup. The Democrats, when they came in in '92 when Clinton came in, decided they were going to do dual use technology. They decided they didn't need all the military standards. They were going to get rid of them and combine them into civilian standards. Dual use — you build a Jeep, a Jeep Wrangler. What's the difference between a Jeep Wrangler and a regular Jeep the Army uses? It turns out there's a pretty big difference. Nowadays with the MRAP vehicles that weigh 60 tons, there's a big difference between a Jeep Wrangler and a 60-ton MRAP.
So there has been a divergence in military needs and civilian needs. Nonetheless, twenty years ago, the Democrats were going to balance the military budget by doing dual use, and they essentially got rid of an awful lot of the military standards. The standard for statistical quality inspection and manufacturing, which was a MIL standard — I have a copy of the MIL standard in my office — became ASME B1.18.18 almost word for word. ASME said, thank you, military, we'll adopt this, we'll bless it, and we'll maintain it so the military doesn't have to. So it did save some things, but I don't know if there's that many civilian uses of bayonets. So it wasn't a perfect mapping between military and civilian.
There probably are still some military standards in some areas. I haven't been designing with the military in the last few years, but I'm sure they still have some. There are collaborations where people try to combine things — the aluminum design manual and the aluminum structural welding code is one example. The dual-use stuff is another. But there's also competition. The American National Standards Institute basically doesn't write any standards, but they will adopt your standard and help market it around the world and convince people you should be using this standard to build your pipeline in Uruguay. You should use an American Petroleum Institute standard, which is ANSI standard X, written by the American Petroleum Institute, but it's a certified American National Standard — meaning there was a collaboration between API and ANSI, that ANSI would promote and promulgate the standard.
In the meantime, in the early '90s, the European Commission started saying, those wretched Americans, they're using their strength in standards to control the world's commerce. Everybody in the world except the Chinese and a few others builds pressure vessels to the boiler and pressure vessel code. The Chinese just do it on their own because they're not going to pay $11,200 a year for the boiler and pressure vessel code. They'll just do it on their own — they're a sovereign nation, so they don't build pressure vessels to the ASME code unless they have to sell them in the United States, in which case they have to buy that code. I'm sure they copy that code, and they buy one copy for the country.
Student: All the trade has to pass through the Western Hemisphere, you know, from the Pacific to the Atlantic — yeah, that standard...
That's not an American — I didn't say American, but by and large every pipeline built in the world is built by an oil company that will call out the American Petroleum Institute standards. Most utilities, chemical companies — which are those same oil companies — if they're going to build a chemical refinery, they're going to build it to the ASME code. That's what their designers know, that's what they're going to call out. American companies controlled a lot of commerce and would call out things. Other manufacturers around the world would have to use American standards because they wanted to sell in the United States, and therefore a lot of those countries would end up adopting the American standards within their own country.
The Europeans saw this as a competitive disadvantage, and they formed the International Standards Organization in Switzerland, which is not part of the EC. The Germans have the D standards — I don't know what the D stands for, German and Germany in German begin with a D. These D standards, just like our ANSI standards, were used all throughout Europe, but they didn't really spread to Japan and the United States as well as the Americans were exporting our standards.
That's why you can now get a BMW built in America —
Student: Can you? When I was in Germany, you bought a car over there from a German, you had to have it redone for America, could bring it back to America, right?
When I lived in Tokyo, you couldn't buy a Toyota in Tokyo and then bring it back without spending thousands of dollars to bring it up to US standards.
Student: Yes, you're right.
There's a lot of this stuff in standards that's done for the export market. Since the United States was a third of the world's market — now it's only a quarter — we controlled enough of the market that everybody wanted to build to our specs.
To give you another example: when our gasoline was priced at $1.75 or $2 a gallon twenty years ago, you could never justify building a car out of aluminum at that price. The value of a pound saved in an automobile is $2 a pound over the life of the vehicle. So at a price of gasoline of anything less than $4 or $5 a gallon, you had to build it out of steel. You couldn't recoup your investment from the weight savings of using aluminum. At that time, the Norwegians were charging $6 a gallon. Most of the rest of the world was charging similar. If people were building cars for their own market, they would have made them out of aluminum twenty years ago.
The Germans actually did build the Audi, which is all aluminum, but that's a $100,000 vehicle. Any fool can build a $100,000 vehicle out of aluminum, but it takes a little know-how to build a $20,000 vehicle out of aluminum. We built Duesenbergs in the 1930s that were all aluminum. I think Mellon — of Mellon Bank, out of Pittsburgh — they just happened to fund a little company called Alcoa. That's where they got to be a big bank. They funded Charles Martin Hall, who started Alcoa. I have a book on aluminum that shows a picture of Andrew Mellon with his all-aluminum Pierce Arrow automobile, specially built for him out of aluminum rather than steel.
Steel is the cheapest for the initial cost, but it's more expensive when you think of the operating cost — all the fuel, the extra weight. So why was everybody building a steel car? The reason was they wanted to ship their Toyotas or BMWs to the United States, because otherwise they would be too expensive. Americans wouldn't buy them. They were paying a penalty in their own country on gasoline in order to meet the American market. That's a little oversimplified, but it's the same type of thing. We controlled the market, and people do things to be able to get into our market. At least they used to. They're not doing it quite as much now because world trade has increased. Anyway, away from the economics lesson, because I'm not an economics professor. That doesn't mean I won't expound on it though.
Competition. ISO really came about to compete with this American — it wasn't a stranglehold, but it was certainly a dominant position in standards. It wasn't something we set out strategically to do, but because of our market size, we were the 800-pound gorilla on the block in terms of standards.
One of the first things ISO came up with was ISO 9000. It was a big hit. Anybody know what ISO 9000 is? In the early '90s, Ford Motor Company came up with — for their world suppliers — if you were not ISO 9000 certified, Ford would not purchase from you, period, no exceptions, to their first-tier suppliers. Second tier, they wanted you to be ISO 9000, but they couldn't control it, because they were really only controlling the first-tier suppliers.
Student: I think it's broader than just materials. I think it's processes and procedures.
It is — documenting your procedures. It's not even materials. It's not like ASTM. You had to have documented procedures. If you were going to fill out a purchase order, you had to have it documented so that if the person who was your purchase-order guru died tomorrow, someone else could look at the document and see how to fill out a purchase order, so you wouldn't lose all that. It's a way to maintain your corporate knowledge.
Everything you did — going to get a drink of water — was supposed to be documented. Maybe that's a little exaggeration, but almost everything you did had to be documented. They had a standard, ISO 9000, which is not a small standard, that taught you how to document procedures. That was enough of a success — and it came at the time of what was called total quality management, which was coming out of Japan with this guy Deming from the United States who had taught the Japanese. The Japanese were these great world-beater manufacturers in the early '90s in terms of the quality of Toyotas. They had the Toyota Production System, which was a bunch of procedures — not "use this material," but "how do you go about your daily task."
That's ISO 9000. Ford came in and said, we won't buy from anybody that's not ISO 9000. All of a sudden American companies were in a scramble to become ISO 9000 qualified, otherwise you couldn't sell to them. Other suppliers — not just Ford — Boeing adopted it, and a number of companies have adopted ISO 9000 as the standard by which you have to prove you have a quality management program in process that documents all your procedures. It's really a management standard of how do you manage your business. I don't know all that much about ISO 9000, but if you're interested, I have a book on it, which I haven't read. In any case, that was competition in one area between ISO and ANSI, and ISO won that competition.
But they still compete on other standards. There is competition out there on standards, because standards affect your ability to market things. If someone goes out for a request for proposal, they will list certain standards. There's a whole section in that RFP about what standards you have to follow. If they list US standards as opposed to European standards, that gives American companies an advantage, assuming the American companies know the American standards. The European companies had to be bilingual or trilingual in terms of standards in order to compete on a global basis. So there are some people trying to improve collaboration and make international standards, and there are some people trying to hold it to themselves for a competitive advantage. There's that dynamic going on.
§5. Purchase orders and the "what you paid for" principle [48:54]
Cost we talked about yesterday. Today, purchase orders. I'm going to ask you a philosophical question. You go to General Motors or Ford or Toyota, and you buy a new car that's basically stripped down — doesn't have any extras. Might have an air conditioner, but doesn't have the 6 CD changer. Actually they don't even have that anymore. It doesn't have all the bells and whistles. You buy it, and you come back a year later and say, well this doesn't have these other options. Do you think they're going to install those other options at their expense?
Student: Well, you didn't order those options.
Most people would say you don't have a right to it if you give them that example. But I fortunately get involved in controversy several times a year — put another child through school — where someone says, we ordered these welded components to this standard, and for whatever reason we had some problem, or someone went in and did something, and we found that there are defects in our welds. Surprise. I'm not going to talk a lot on safety factors — Dr. Belmir is going to talk next Wednesday on safety factors.
But if you went through the structural welding code: for a building, the safety factor should be 1.67 — five-thirds is the number. For a bridge, which is dynamically loaded, fatigue should be 2.0. For a weld it should be 3.3. So if you're going to calculate the stress, this code in Chapter 2 tells you how to calculate the stress on a weld. It doesn't tell you how to calculate the stress on the plate or the I-beam — the American Institute of Steel Construction will tell you how to do that. But for the welding code, you have essentially double the safety factor.
Why? Because we know welds are going to have defects. Some people define a weld as a continuous defect surrounded by good material. Just by the very nature of putting down a weld, you can end up with lots of different types of flaws. So what they do is just put a bigger safety factor on it to allow for it. I could have a weld that's 50% flaws in a statically loaded structure, and it will probably perform just as well as the base plate, because I have double the safety factor. So I sort of assumed I could have a really lousy welder, or a really good welder who had a bad day — for whatever reason, you put a bigger safety factor on it. That's all I want to say about safety factors right now, but Dr. Belmir will talk more about it later.
Then someone comes in and does some sort of test, and it says, oh, we found a flaw in our weld. Well, duh. I actually wrote a paper once for a conference as a keynote speaker called "In Search of the Perfect Weld." What is a perfect weld? You can't even define a weld very well, much less a perfect weld. In the welding code, they've dealt with this a million times.
Under contractor responsibilities, it says: non-specified NDT other than visual. What's NDT? Non-destructive testing. So non-destructive testing other than visual. The visual inspection requirements are right here in 6.9. All welds shall be visually inspected and shall be acceptable. Pretty simple. You have to inspect every one of them. The guy who makes the weld has to have his sight — you cannot have blind welders. They have to look at the weld and determine whether it looks okay for the criteria in Table Six, which means if it doesn't look like a dog just did his business there, you can probably accept it. The visual inspection requirements are not that rigorous, but there are requirements that someone look at it.
But if you had anything else — magnetic particle, ultrasonic, radiographic, eddy current, a whole host of different specified techniques — if NDT other than visual (because all welds have to be visually inspected) is not specified in the original contract document but is subsequently requested by the owner — this is like going in and saying, I didn't get this feature on my Toyota, but I've decided I want it even though I didn't pay for it. The code says: the contractor shall perform any requested testing, or shall allow any testing to be performed in conformance with 6.14. The owner shall be responsible for all associated costs — including handling, surface preparation, NDT, and repair of discontinuities other than those described — unless there's fraud involved. Unless you can show that the contractor was trying to defraud or gross non-conformance to this code, the owner has to pay for this extra work if they didn't ask for it in the purchase order. If they didn't pay for it and they want a higher level of inspection, you've got to pay for it, folks. It's just like buying that Toyota that didn't have the features and you decide you want the features. The contractor's got to give it to you, but the owner has to pay for it.
How many of you have worked in a SUPSHIP? None of you? Shipyard?
Student: I was the customer.
Student: Yeah, but I worked at the yard.
She's the only one. That's unusual. Usually half the students in one of these classes have worked in SUPSHIP. Let me tell you, it's not the most exciting job in the world. You mostly spend your day either drinking coffee or arguing with the people in the shipyard over: is this weld big enough, or is this defect within the allowable specification. Some of you will probably go to a SUPSHIP, right?
This comes up all the time, because for some reason someone sees something and they find a defect, and they say, oh, I have a defect. Well, of course. But if that defect's not huge — a factor of two of the size of the weld — it probably is allowed for in the safety factor. People get in fights all the time over whether they get or should get what they paid for. If it's not specified in the contract documents, you don't get it. I'm sorry, that's the way it is. But people will argue this over because they don't understand things like safety factors and how the code has developed over the years a number of different approaches to dealing with the problem that manufacturing doesn't produce a perfect product every time.
§6. Measurement, the Seawolf, and the limits of what to look for [57:43]
When I've been to places like Bath Iron Works or Newport News or Electric Boat, these fights occur all the time, and there's not a great relationship between SUPSHIP and the yard in most cases. I remember I was evaluating a program for the Maritime Administration on an automatic inspector of the size of fillet welds. The fillet weld, let's say, was supposed to be a quarter-inch fillet weld. They developed this little laser-based inspector that could scan along the weld and automatically inspect — it could detect the size of a fillet weld. A fillet weld is, if you have a plate like this, it intersects a plate like this, and you put a weld in the corner — that's a fillet weld. Lots of these on ships.
The size of that fillet weld is defined in the United States as the leg length. The problem is there are two leg lengths. How do you define the size of the fillet when you have two measurements? If one's under and one's over — there are standards to help you define that, because the question has come up before. In Europe they use a more scientific definition, the throat thickness. We measure it as the leg length. The two should be related if it was a perfect triangle, which it's not. The throat thickness, if you have a lot of reinforcement, you could argue you have a bigger throat. You start thinking very hard about this and you have to say, what do I really want to measure? What you really want is some surrogate for the strength.
They had this little laser-based measurer. You'd just take your hand and roll the thing along the side of the fillet weld, and it would pop out a number of the average size, the variation — the thing did all the statistics. Now you have more numbers than you ever wanted. I can remember early in my career working on some quality-control things and going down to Electric Boat and saying, hey, we can measure such-and-such about welding. The guy said, the last thing I want is another thing I can measure, because I'm going to be held to that standard. So there's an upside and downside to measuring things.
This laser-based measurer could measure things within plus or minus about ten thousandths of an inch. They got some real inspectors at the shipyard to go out and showed them the same welds, asked them with their little hand gauges or with their eyes what the size of the weld was — because they now had something automated, non-judgmental, that would give them a number and statistics very easily. They found that the inspectors, even with their measurement tools, couldn't do any better than about 0.020. Here was a machine that had twice the accuracy of the inspectors. But SUPSHIP was arguing over the size of the fillet weld, because they came in with this thing and it was ten thousandths under — even though, before, when that spec was written, no one could have even measured ten thousandths. So as we improve our measurement techniques, we get into more fights.
When I was an undergraduate student working in the lab around here, we were doing superconductivity. We only owned one germanium thermometer. They cost $500, which was a lot of money back then. We decided, well, if that one breaks, we're going to be out of business — we need to buy another one. We spent six months trying to figure out which one was correct. Because you put them both down in liquid helium and start heating things up, and they're off by a couple of tenths of a degree. So which one's right? We knew what the temperature was when we only had one measurement. We had two measurements now, and we didn't have a clue. I was trying to measure 18.1 degrees versus 18.2 degrees, for which one's the better superconductor. We were trying to break the world's record. At that time it was 20.3 or something.
The reason for going through observables is — you can talk about accuracy and precision, but you ultimately have to bring it back to reality. To give you a reality check: a lot of the boiler and pressure vessel code and the structural welding code were built up by 80 to 100 years of experience and committees of people working together in good faith. There are certain things in the welding code — they're never looking for a flaw smaller than an eighth of an inch. Why? The Air Force did a big study, and you can't find flaws with more than 50% accuracy at a sixteenth of an inch. You have like a 95% probability of finding a flaw at an eighth of an inch. Before the Air Force did this study twenty years ago, the code had already decided: don't look for flaws smaller than an eighth of an inch because they're not going to be harmful. They didn't actually say it that way, but no one had ever had a problem with a vessel that might have had flaws of that size.
If you go through fracture mechanics, flaws less than an eighth of an inch with steel with the typical toughness of structural steel don't mean a hill of beans. You have a 3.3 safety factor on top of it. The fundamental equation of fracture mechanics, which you'll get in some of the other lectures, is: the fracture toughness of the material must be greater than the stress times the square root of pi times the crack length. This is a material property. It only gets so bad for steels — steels actually have a fair amount of toughness. This is the flaw size you're looking for — an eighth of an inch. Pi, I know with a lot of precision. Sigma is the stress. If you're a civil engineer, they don't know Greek, so they use S.
The code gave you a design stress for the base material, so you know your limits on stress. If you designed it properly, you plug in the lower-than-maximum yield stress, you plug in an eighth of an inch, and you compare it to the toughness of the steel — and you find an eighth of an inch is fine, no problems.
So what happened to the Seawolf submarine? They were looking for eighth-inch flaws. That's what the codes say. That's what reality says is all you can find with any probability. But what happened is some grinder was grinding the weld smooth. On a submarine you can't have these big humps of a weld when you're going through the water — it's amazing how quiet these things have to be. They have to grind the weld flat. This grinder grinding the weld notices that the swarf — the little particles from grinding, called swarf, good Scrabble word, S-W-A-R-F — the swarf is lining up in little lines, less than a sixteenth of an inch in length. He says, I've never seen that before. He goes and tells someone, and someone says, hm, I've never seen that before. They start investigating it, and the welds had a whole family, a cluster — they were full of these little micro-cracks.
Once they got them and started looking at them by cutting out the weld and doing destructive testing — things you would never do regularly — they realized the whole ship was full of these things. They sampled around. They had completed 18% of the hull. They tracked it down, and — I had to write a report on it — it was the weld wire was on the high side of chemistry. We knew it was on the high side of chemistry, and I thought there were some other things. It's in one of my lectures you're going to see on video — you'll get a longer version of the story.
Some grinder happened to notice something out of the ordinary. They were looking for flaws that were smaller than anything the specification had ever asked for before. This was the first generation of that class of submarine. No one was going to take a chance. It was a billion-dollar vessel. I used to hear numbers of half a billion dollars it cost the Navy to fix that. More recently I've heard the ultimate cost was about $2 billion to fix the problem. You could have built a new sub for that.
Student: They built two Seawolves.
And then they canceled the program. The reason — it was built to defeat the Soviet Union, and there is no longer a Soviet Union. If you start firing those boomers off, then it's all over anyway, folks. You have to be willing to do it, but you don't ever want to do it. That's another conflict.
Student: You ever see the movie Seven Days in May? There's only a few...
Student: I grew up in Atlanta. You leave on a submarine, and you come back to Georgia on a submarine, and you never get port calls, and you're wondering about what if you did launch — I always pictured, if we launch, you'd have to think about all those people that died, and then if there is a nuclear war, Georgia and Bangor are going to get bombed.
Okay, well, at least you've thought about it, you've got a plan. Have you written up the procedure for that? Is your ISO up to date?
§7. A106 pipe: a purchase-order dispute in action [69:24]
Let's talk about another example. This is actually the reason I decided to teach this class last fall. I was involved — or I am still involved — in a situation where someone bought some A106 pipe, which is also B31 something. Whether you're buying ASME or ASTM, there are like five different standards that cover this pipe. In fact they were using ASTM A106, which is the standard specification for seamless carbon steel pipe for high-temperature service. I handed this out because we're going to spend a little time talking about it. We talked about scope, and we talked about reference documents. What I made the students do last fall is take any standard — there's going to be a bunch of reference documents — and they actually spread this out to ANSI standards, military standards, federal standards, other standards.
Any standard you call out probably lists another fifteen or twenty standards. This is the burden I talked about yesterday. If you're going to go into the business, you have to know all these standards and how each one references everything. And every one you go to will reference another standard. This is growing exponentially in terms of standards you have to understand and know. It actually does start to converge at a few thousand standards. But it's a burden if you wanted to enter a business that you didn't already understand.
Just to read this standard — I happen to have D97. I'm sure there's a newer one. '97 is 1997. That's the way you read ASTM numbers. This is also an American National Standard — ANSI adopted this. Down at the bottom, ASTM will always tell you: originally published as A106 D26. This standard started in 1926. Going through the standard, it'll tell you how the steel must be made. Actually before I do that — the title says, standard specification for seamless carbon steel pipe for high-temperature service. High-temperature service, I'm sure if it's defined, there's some standard that defines it, but I don't know what it is. It's a few hundred degrees, probably above the boiling point of water. In fact I bet that's what it is.
There are other standards. This is A53 — the way I remember these is 53 is half of 106. They don't always work that way, but in this case it does. This is a standard specification for pipe, steel, black, and hot-dip zinc-coated, welded and seamless. So this specification covers seamless and welded, black and hot-dip. What's the difference between these standards? This one is seamless carbon steel. This one is seamless and welded. This one does not say high temperature; the other one will say high temperature if you look at the title. So one is for seamless high-temperature, the other is for black iron pipe. I can go to the hardware store and buy a piece of welded black iron pipe. That's A53, not A106.
If someone wanted to make the plumbing for my basement out of seamless steel, not only would it cost a lot more, but I could operate it at higher temperatures. It would be equivalent to A106 as well as A53. So it could come under two specifications. You'd better hope they don't have conflicts among A53 and A106, at least on certain levels. It turns out they won't, because it's the same committee responsible for both. The other thing is, this one does not reference A53, and A53 doesn't reference A106. So far as the committee is concerned, they're two different pipes — although the same physical pipe, if it was seamless, would come under both.
How would you know that? If you're a member of the committee, you'd know that. Or if you've been working in the business for twenty years, you might know that. But if you're just someone going to read the book for the first time, you don't have a clue, and there's nothing to tell you where to go to look for the other standards that might also apply. This is one of the problems of how standards are written. There's not a designer's guide to reading standards. I guess because no one would read it — it would be the most boring thing in the world.
The process. The steel shall be killed steel — that means it has to be deoxidized; if you took the casting part of my course, we talk about that — with the primary melting process being open hearth (which no one in the world uses anymore — we gave it up about 1985), basic oxygen (which everybody uses), or electric furnace (which everybody uses), possibly with separate degassing or refining. So you can make it any way you want — that's what it says to me as a metallurgist. If someone comes along with something else, they have to re-qualify it.
There's a heat analysis and a product analysis. Does anyone know what the difference is? This is chemical analysis. Section 8 and 9 talk about the chemical analysis in terms of a heat analysis and a product analysis. This causes confusion to people in materials all the time, because they don't know the difference. It's okay to guess — I don't slap you down that bad when you get it wrong. A heat analysis: when you make the steel, you might have 150 tons or 300 tons in the vessel, and before you cast the steel you dip in and take a little sample on a long rod, you cool it, you send it off to the shop. Within less than five minutes you have an analysis back. That little bulb of metal represents 300 tons or 150 tons of steel. That's the heat analysis. The heat is the big molten bath. That's what the heat is certified to and sold to — the composition of that little sample out of 300 tons.
Generally it's actually pretty good. But if you get it down to a small piece — I don't know where all the brothers and sisters of this 300 tons went, and if this is all I've got left and I do a sample on this product, that chemical analysis might be a little different than the average of all 300 tons, because there is some segregation. It's not perfectly homogeneous in the steel. So there is another specification written by ASTM that tells you how closely the product analysis should equal the heat analysis. There's a specification for the specification. People will often do a product analysis and say, oh, this doesn't match the heat analysis, it must not be the same steel. Well, within certain limits there is a specification that says, yes it is the same steel — it's just that a one-pound piece out of 300 tons may not be exactly the same composition.
There are really exciting tables in here that you'll want to read tonight for hours. How was the steel bought? This was going into a big petroleum chemical refining plant, and someone bought 600 lengths of steel — most of it 6 to 20 inch diameter. This was not small pipe. They bought it 8 or 10 years ago when there was a big shortage of steel pipe in the world. If you were trying to build a plant at that point in time, you just had to find a source. So they went and bought it, and this stuff came from Romania — best Romanian steel, best quality Romanian steel. That's not necessarily the greatest recommendation in the world.
By ASTM A106 — each length of pipe shall withstand without leakage through the pipe wall a hydrostatic test. You have to pump it up with water and pressurize it. There are different hydrostatic pressures, which are listed on the purchase order. If you want it tested to 1500 PSI or 2500 PSI, you have to tell the steel company what pressure you want it tested to. A lot of these were tested to about 1500 PSI for the smaller diameters. That's like 40% of yield strength for these pipes — they would have been able to take 4 or 5,000 PSI. But they only had to be hydro-tested to 1500.
How big can the flaws be? It's seamless, but how big can the flaws be in that piece of pipe in order to support 1500 PSI but have imperfections? They can be pretty big. I had a case down in the Gulf of Mexico where someone had a hydro test of a pipe made in the United States in a seamless mill, and they were going to put it in as an underwater pipeline in the Gulf to take oil from the drilling rig onshore to the refinery. It was only a 4-inch pipe, but it had been hydro-tested at the mill, magnetic particle inspected and eddy current inspected at the mill, and passed all those tests. They sent it down to Louisiana, sandblasted off all the information identifying the steel maker, coated it with this yellow coating for corrosion resistance because they were going to bury it in the ocean. They welded it, laid it down on the bottom of the ocean, and then had to do one last hydro test on the final assembly — and it leaked. It had already been hydro-tested and passed twice, and the hydro test at the bottom of the ocean was actually a lower pressure than the ones that had passed already. And there was a leak. Cost them $4 million to get it out of the ocean, bring it back, plus the delay in the project. Not everyone was happy.
So we went to look at it. This was about a half-inch-thick pipe, only about 4 inch — pretty high-strength pipe, maybe a centimeter wall, in about a 4 to 6 inch pipe as I remember. One of the tools they use to make seamless pipe — they basically take a hardened steel piece and ram it through this hot piece of 2200-degree-Fahrenheit steel, and just like sticking your finger through a piece of clay, you make a pipe. Only they're doing it with a piece of steel. A piece of that tool broke off and wedged itself into the wall of the pipe. It was probably 3 or 4 inches long, and it was 80% of the wall thickness. It had stayed in there, sort of wedged in.
It passed the first hydro test because it had a plug in it — a V-shaped plug. It passed the hydro test at the mill, it passed the inspection because it was in a critical area that was not very well inspected — I don't want to get into all that — but it passed because it was located in just the wrong spot. It passed the hydro test, it passed another hydro test, and it got down at the bottom of the ocean — I don't know if the plug came loose because of the previous hydro tests starting to give a little plastic deformation, opening it up so the wedge could fall out. When the wedge was no longer there, it wouldn't pass the hydro — it leaked. Couldn't put it into service, had to take it out. It's expensive to go down there at the bottom of the ocean to take it up, repair it, and everything.
Things happen, defects occur. That one had gone through non-destructive testing, hydrostatic testing, and still got to the field. That doesn't happen very often — that's a little unusual. But in any case, these guys for this refinery wanted to purchase the pipe to the hydro test, because that's what's required. But the hydro test says, "except as provided in these other three paragraphs" right beneath it: "when specified by the purchaser, it shall be permissible for pipe to be tested by non-destructive electric test" — you don't know what that is, and neither did I when I first read it, you have to read a little further — or "when specified in the order, the pipe shall be furnished without hydrostatic test, in which case you have to note it as NH for non-hydrostatic."
Someone might be using this for some structural application in a building. They're not going to pressurize it internally. There are some exceptions. "Non-destructive electric test" — it turns out non-destructive electric is how these code guys decided to define ultrasonic, eddy current, or magnetic particle, which they call flux leakage. It warns you that the hydrostatic test in Section 13 has the capability of finding defects of a size permitting the test fluid to leak. Whatever that means — it's a little fuzzy, but it should be able to find flaws, is what it's telling you. But then it's talking about the other tests like the flux leakage, and it basically says, if you find any flaw greater than 12 and a half percent of the thickness of the wall, you're going to have to reject it.
Now we have a bit of conflict within the same spec, because the hydro test is only done at about 40% of yield, but the non-destructive test is looking for something that's 12 and a half percent loss. This is never-never land between 87.5% and 40%, where you could fail one but pass the other. You could pass the hydro even if it had a 50% wall defect. They purchased it to the hydro. The company found one leaking pipe when they were doing some tests, and they decided all 600 were bad. They had a metallurgist do some metallography — said oh, it had a defect in it. It was in a little end, only a couple inches long. I wouldn't want to put it in service — it was definitely a defect, and somehow it had gotten through the whole system of hydro.
They had someone then do the non-destructive electric tests that were not called out in the purchase order, which are more than twice as stringent as the hydrostatic tests they paid for. Remember — this thing in my outline is: you should only get what you pay for. Since they found one defect out of 600 pipes, in only one little 3-inch section of pipes that are 40 feet long — you can multiply 600 times 40 and get how many feet that is, and you have a 3-inch length out of how many feet — that's the only defect they found. They claimed that some other things exceeded the 12 and a half percent limit.
Several of us went on site a year ago August, and we looked at all 600 lengths of pipe. Let me tell you, it's a lot of fun looking at 600 lengths of pipe. But it would pay for one year of one student in college, so it's not so bad. We couldn't find a single defect that exceeded the 12 and a half percent. So now we're in an argument about whether there was a 12-and-a-half-percent defect, and even if there was, you only ordered it to 40% strength, so now you want 87.5% strength. What's the price? This was originally a couple million dollars worth of pipe or less, and they want $60 million in return, because half the pipe had already been fabricated and put into the refinery. They decided, well, we can't tolerate a failure in the refinery, so we're going to rip all that out and put new stuff in. That escalates the cost from a few million dollars to tens of millions of dollars.
There's controversy, and it's because various engineers can't read and interpret that single code — forget all the other codes it references. They can't even interpret the English language of the code. So you get into lots of interesting fights. I'll see you on Tuesday, and we'll talk about some more of these things.