CS_F2012_11

Anybody have any questions? I think everybody in here has told me which modules you want to do, but I also like to know what standard you're going to do your presentation on sometime in the next week or so. If you need to come sit down and talk to me about it I'd like to know, because I don't want someone taking a big broad standard that's going to be too broad. I'd like to take a simple standard. I mean the guy who did the bayonet blades, that was fine, that standard was fine, okay. But other things, like I don't want you to do the whole structural welding code or something, or the pressure vessel code. Does it just be in 10 minutes how could you cover something that's thirty-two volumes, right, in terms of you know how it evolved, what it covers, does it have force of law, what other codes does it call out? Well a 32-volume code calls out thousands of other things. A simple little code like a code for making a piece of steel pipe okay, like a 10— this would be a 53, actually this is welded pipe, but that's manageable okay. It might call out another 20 codes. The problem is those other 20, not 20 codes but 20 other standards, those 20 other standards are going to call out another 15 or 20 standards each, and so things start multiplying until they start converging eventually. But anybody— no other questions?

And the things I wanted to talk about, a couple of things that Dr. Belmar talked about. He mentioned just in passing about this trailer that he designed, and the fact that he didn't have to worry about— in other locations he didn't have to worry about the welds because he uses bolts. Well it turns out using bolts is actually required by certain standards if you're actually doing trailers okay, the big semi trailers that go down the highway and stuff. They require that you not weld on the frame. So the frame— if this is your truck okay, not very good truck, and you got the trailer here, there is a frame. There are two frame rails and there's sort of C-shaped pieces of steel like this. If you look at— ever look at the back of the truck? This trailer, it's going to— this tractor that's going to be pulling the trailer, on a tractor trailer there'll be two frame rails looking at it from the back, and these are just formed pieces of steel okay, and they go you know like that, right. Hopefully you're getting some idea of the geometry of the C-sections. They might be half- to three-quarter-inch-thick steel. It's pretty heavy steel.

There are load limits on the highways because they don't want to destroy all the asphalt roads okay. I mean, the one place I'd say really heavy loads is in Michigan. You been to Michigan? You've seen those 48-wheeler tractor trailers? Well no, this isn't tandems. These are— because it's the big automotive area, and they will put on four or five rolls of sheet steel which is probably 150 tons on one trailer, and they will distribute the load because there is no— if you're all the way back on that tracker it's just you know four wheels, and if you count it up it's like 46 or 48 wheels on this tractor trailer okay. Most of the tractor trailers you and I know in Massachusetts or Connecticut or New York or Pennsylvania are 18-wheelers okay. Some of these things in Michigan— I've never seen them anywhere else except Michigan— are 46 or 48 wheelers. I can't remember, but something like that.

Anyway, you want to reduce the weight, and so they actually use a high-strength steel, 100 ksi strength steel. And you can weld 100 ksi strength steel but the fatigue strength is terrible okay. And I think I cover that somewhere in one of the joining courses about— you don't get a significant increase in fatigue strength even though you're using higher strength steel. You tend to get the same fatigue strength as if it was lower strength steel. So this is clearly a fatigue-loaded situation with a trailer bouncing, you know, the weight of the thing bouncing up against the frame rails. And they will— if they have holes they will have actually punched-out holes in these things, and they will bolt to them. And the reason they don't have welds is because historically people have found that fatigue cracks start and they don't get noticed, and all of a sudden the trailer is going down the highway and it breaks in two. It's not a good day.

So in any case there's that particular problem with welding to these things. And why is— it turns out the welds are actually stronger than the bolts on average. If we look at strength and we look at the number density okay of things, the bolts— if we look at a probability plot, and Dr. Belmar was kind of talking about probabilities for safety factors, but this is a strength probability okay. A bolt has a fairly well defined average strength, but a weld may have a higher strength. So the mean strength of the bolt is less than the mean strength of the weld, but the minimum strength— there's going to be some cutoff here— the minimum strength of the weld, because welds tend to have more defects, is lower than the cutoff for the bolt. So in a sense the bolts have less variability even though they have less strength. You can predict statistically the capacity of the bolts better than you can the weld.

So there are codes for designing tractor-trailers that don't allow you to weld, and most of those have been adopted by states that have the force of law. Because, I mean, the states don't care until they actually have some trailer fatigue on the highway, spin out, kill somebody or tie up traffic for four hours or something, and then the state police get involved, they start investigating and figure out the reason, and they will then incorporate that somehow into some legal requirement okay. But that was just sort of an example as he mentioned it. You know, he was designing this trailer for his farm up in Canada or whatever— and I guess that's where he was designing, I don't know— and he kind of knew to use bolts. Because, and it is sort of intuitive okay, that he knew you don't use welds where you really have to— if you have to have a certain level of strength you can use welds, like pressure vessels. You don't want to bolt together a pressure vessel because big thick pressure vessels— if it's thick you couldn't even build them if you didn't weld them. But you're going to spend as much on quality control of non-destructive testing as you do on fabrication making the welds. Whereas on a trailer who's going to spend all that money? You going to take it out and have the welds x-rayed, you know, or magnetic particle tested? No, you're just going to make it.

In fact I always actually call them horse trailer welders. There's a significant number of people in the world who learned to weld growing up on the farm okay, and you learned about how to farm in a rural area and so they know how to weld, but they don't know the science behind the welding. And every now and then they do something stupid, and I always remember one of my favorite stories is a guy brought a weld to me once and it was from a trailer he'd made up in Nova Scotia. And you got the A-frame, you know, this is the hitch right here, and here's the frame, and right here in the corner they made a weld, and they basically cut and formed the angle iron or the steel or the C channel or whatever it was, and they had a gap in there. So they threw a bolt in there to fill the gap, and they welded over the bolt. Well that's actually known as slugging a weld. It was very common in World War Two when they tried to build Liberty ships and stuff. Someone was bored— and they were not necessarily Liberty ships but on battleships and stuff— when you're welding 16-inch-thick armor plate it's really dull putting 600 passes into the same little section of weld okay, when you're welding something 16 inches thick. And so people— so they get a little break, they might take a piece of pipe, or they may take a bunch of welding electrodes themselves and they just throw them in the groove, the big groove, and they weld over them, and no one could find it because they're buried between several inches of steel in this old armor plate. But it did sort of weaken the whole thing. I mean it's like making perforated steel right.

Well so per— slugged welds was always something we used to joke about because it went out with— after World War Two when people started seeing this, you know no one would ever select a weld. I had never seen a slugged weld. They brought me this piece of a trailer, and the story with this trailer is, this person lived on a farm or they bought the trailer in Nova Scotia from a farmer who had built it for him. And it was a trailer for their sailboat, their 40-foot sailboat. And they driven from Nova Scotia with their sailboat all the way down to Newport, Rhode Island for the races in Newport, and they had won the race okay. And they were coming back up in or not— it's not interstate, is route 24 or whatever— coming up from Rhode Island, it's like an interstate highway just south of Boston here, and they're cruising along about 55 or 60 miles an hour with their trailer behind them and they hear a bang okay. And of course they kind of slow down and put their brakes on, and there's a sailboat just passing them on the highway okay. Because the weld broke okay right here, the trailer let go okay. And this all of a sudden you had a freewheeling trailer and a 40-foot sailboat, and it crashed and destroyed the sailboat okay. And they brought me the weld and said what's wrong with this. Well, they wanted to go back to the farmer who had welded up the trailer and complain to him. I said, well I mean it wasn't hard, you saw about a three-quarter-inch bolt that had been put in there and just little welds were put on top of it okay. So it was not a fully welded joint, it's called slugging a weld. That's the only time I've ever seen it okay. But you know, you kind of look at this level of engineering in things and you sometimes wonder about it.

So far as that level of engineering, another bolted thing which we found the other day when we're cleaning out some junk downstairs in the basement, what we found— this is— we call this Tron air in the old days. This little bolt is part of a system to take a jet engine off the airplane wing okay. You have this kind of sort of a roll-up dolly that's about 15 feet tall and you roll it up to the wing. You put this bolt into the hanger for the thing that holds the turbine engine on the wing, and you torque it down. It's, you torque it on this side, then you torque on this side, and then you just pull it out. You release the bolts that held it on the wing and you pull it out, and now you have the jet engine that you can now work on as a mechanic. So they were doing this, and the guy who designed this designed the centerpiece with flats on it. You can see the flats, I'll pass it around, you can look at it. So you could grab this thing with a pair of pliers, and if you grabbed it here with a pair of pliers you could torque this to the proper torque for this size bolt. This is a hardened steel bolt okay. This thing probably is a two-hundred-dollar bolt okay, it's very specialized. You know how many of these do you need? And the other side they had a different size thread, and you were supposed to grab it here and torque here. Well they actually torqued it here, and they didn't bother to grab it here, they just felt well, we talked it here, they held it with the nut here and they torqued there. Well look at them, the two of them are different diameters. It turns out the proper torque for this is just enough to start a crack in this okay.

So you have to worry about your— look, what we called the load path. If the load path was only from here to here and there's no stress on this when you're doing the torque of the large-diameter, what's the problem? However, if your load path just ignores this little shoulder and you don't grab it and keep it restrained, now your load path goes all the way through to the other nut, and all of a sudden you over-torque this thing okay. So in hindsight after they dropped an engine on the tarmac okay— and it costs about five million dollars, engines are not cheap— they looked at it and said, not a good design. And so I mean simple little things, and I'll show you another one a little bit later, of a load path problem okay.

But one of the things I want to talk about today— we've talked about measurement and codes and standards— I wanted to talk about design and the types of design. We often just talk at the university, well you're supposed to have a capstone design course as an undergraduate or something okay, where you can show your creative energy. And one of the things I wanted to do in this course was to point out to you that your creative energy is limited by codes and standards. You're not free to just go out and do whatever you want. Good manufacturing practice basically tells you that you shouldn't just do whatever you want, you should follow the experience of other people who have learned through sad experience with other failures that have caused a lot of property damage or other things.

And so design is actually a complex process. People teach whole courses on it, and it turns out there are different levels of design. And in different businesses you might come up with different lists, but if you were building a bridge or a building, usually start out with something that we often call the conceptual design okay. And the concept is, well, we're going to have a bridge across the Charles River. And if you look at the Mass Ave [Bridge], it was going to have walkways on either side with certain type of railing, and it was going to be low to the ground, it wouldn't have domes or something, and it's going to be made out of steel. And that's the Mass Ave. If you look at the Longfellow Bridge, it was built a few years earlier, it's all made out of stone and has the salt and pepper shakers as the columns and stuff. That's the concept, it's sort of the conceptual design. Could be something as simple as someone just sketching it out on a back of an envelope okay, this is what I want my bridge to look like, or this is what I want my building to look like. It's going to be, you know, pyramidal shaped rather than box shaped building or something okay. That's, you just kind of sketch out a concept.

This is a conceptual design right here. [Tom crumples up four sheets of paper.] I just crumpled up four sheets of paper. Anyone who hasn't heard this story before, does anyone know what that is the conceptual design for? Now, do you know the— you know the story of the Stata Center? You know, the building over here that has all the weird shapes and stuff. So I know this story for a fact because Vicki Sirianni— she's an architect and she was head of MIT physical plant— and they were about to build the Stata Center and they hired Frank Gehry, who's one of the world's renowned architects. And I happened to meet with her the morning after she got back from Los Angeles meeting with Gehry in his office. And she was in the conference room with some of the other people, and Gehry walks in and he starts crumpling up paper, and he throws it on the table and he says to her, that's what your new building's going to look like. So there's the conceptual design of the monstrosity we call the Stata Center okay, crumpled paper okay.

So there's all kinds of levels of conceptual design, and if you're one of the world's renowned architects you can just throw out the biggest piece of crap okay in the world and call it a conceptual design. And you leave it for your underlings to work out the architectural design okay. But the concept was we were going to build this 400-million-dollar building, actually was supposed to be 115 million— that's another story, but it went over budget a little, came in at 430 million. They froze everybody's salary at MIT for two years okay, because they were so far in the hole after going over budget by a factor of nearly four. Anyway, that's another story probably not worth telling right now.

But conceptual design can be— actually today conceptual designs are often done on a CAD system, and you kind of get a nice little rendering in 3D, you know, kind of imaging that you can rotate and everything else and someone does. But it's still sort of a boxy design, and it might show you where the windows are in a building and the entrance and some of the sculpting around doors and things. But it doesn't give you a lot of detail. The architectural design— what's the architectural design involve? You're a civil engineer, Steve.

Well, the architect tells you how many square feet it's going to be, where the rooms are located inside the building. I mean, Gehry didn't have to worry about where the rooms are okay. He didn't have to worry about the details of the design. You walk in the lobby of the Stata Center, most expensive building on campus by far, and what do we have in the lobby? We have concrete block walls and concrete floors okay, right? But everything's sort of— well, I guess the word's catawampus, if you know what catawampus is. Nothing has square corners, you know, everything's— it's crumpled paper, right. But some architects— I'm sure a whole team of architects— were approached by MIT, 40, 50 million dollars for their time worried about how to make— you know what the shapes were, where they're going to put the doors and all the windows, and which windows are on a sloping wall so they have to have an extra frame because they got to be vertical windows on a sloping wall. Don't look at— the thing's a monstrosity. Anyway, but some people love it, it's sort of like politics, you know, you're on one side or the other. Some of us think the Stata Center is a disaster.

Anyway, so the architectural design goes to things about, this is where the doors are going to be, we're going to put the restrooms over here, we're going to put the big lecture hall here, you know. That's the architectural design. It's kind of, these are the pieces of the Lego system that are going to build up the whole three-dimensional object of crumpled paper, right. So what's the detailed design? Come on, you've worked on detailed designs.

Those are the engineering drawings, rather than architectural drawings which kind of show you the layout of the building. I mean, using the architectural layout of a home, it says we have a bedroom, master bedroom here, family room over here, fireplace here. But the detailed design is something that actually starts telling people where you're going to have two-by-sixes, where you're going to have two-by-fours, whether you're going to have a steel beam to hold up some cantilever, whatever. That's the detailed design, where engineers, not architects but engineers, are actually doing calculations on the stresses on beams. But those guys don't necessarily— there's actually often two sets of detailed designs, or you so maybe two sets of architectural designs on a complex building or two sets of detailed designs. Sometimes they call something— rather than detailed designs— in here they call it shop drawings. Anybody know what a shop drawing is?

So the detail design— the engineer says— he's a civil engineer— he says well we need a 2x6 here, we need two-by-fours over here to hold up the floor, we need a joist, you know, here, a fabricated joist to hold up the ceiling or whatever. But he doesn't tell you how to build it okay. He might tell you it's going to be made out of wood because he had to figure out the strength based on wood or steel or whatever, he had to do the properties. But he doesn't worry about things like how do you join it together. If it's a wooden beam and you got the A-frame of a roof, how do they join it? They used to use nails. Well actually if you go back far enough they used to use pegs, right, wooden pegs. Now the finest construction is with wooden pegs, but 400 years ago they used wooden pegs because nails were too expensive okay. But nails got to be cheap with the coming of Andrew Carnegie and steel becoming cheaper, and so they quit using wooden pegs which are very labor-intensive and they started using nails.

They don't use nails nowadays when they make up one of those flat roof trusses. What do they use? Punched metal plates. They take a piece of galvanized steel, they punch it so they make little triangular shape cutouts, so that if this is the plate, you'll put a triangular punch and the pointed triangle is sticking through. You may have 50 of these things or a hundred of them, and just put it on the joint. So you've got— if you've got a two-by-six running up this way and you got a 2x6 this way, you may put one of these galvanized plates with all these punch marks in it that you just lay it on here and then the guy pounds it down with a mallet, and all these little nails if you will— which are just punches of the sheet metal— all go in there, spread out nice and uniformly. Great distribution of stresses compared to toe-nailing at some angle in here. Labor and labor savings. All done in the shop. He laid this out on a great big table flat, put it on a truck and take it to the site, crane drops it in place okay.

Which gets you to the next thing, is the erection drawings. So I've been driving in and I come in down Broadway and come across on Galileo Galilei, which is over here by the Whitehead Institute. And somewhere behind the Whitehead Institute they're building a new building, and you got some really big steel beams. I mean these things are like three-inch-thick flanges, four-inch high webs— these things are real, I don't know what they're building back there but they could store M1 tanks okay in whatever they're building. They're really heavy stuff, but I can't really see what they're building because it's down behind a bunch of other buildings. There's nowhere to bring all these things all at once to the building site. So you got a logistical problem. If you're building this out in the middle of Kansas in some field, you'd have that they call a laydown yard, and it doesn't matter.

This fabrication shop that does the shop drawings— tell you what welds to put where okay. So the detailed drawings may tell you what shape beam or what shape steel, but the shop drawings will say put a fillet weld here, you know, half-inch fillet weld, and continuous or intermittent or whatever. There's lots of details on a shop drawing that tells the guy out there in the shop where to put the welds and what size pieces are that it's going to put together to make up these beams. When— any way, over here they're building this huge thing in the erection, they have to keep track of every beam, and they can't store all the beams in a field and that's go out to get the beam they want when they want it. When I drive in, there's usually two or three trucks parked there right beside the Whitehead Institute with these huge beams on them ready to unload. And so it's a huge logistical problem of making sure the right beam arrives at the correct time, because you have no storage space. Once it gets there it's just-in-time erection okay. And they have bolting plates and they bolt them together in the field, they may weld in the field whatever, and they're supposed to build in a particular way. But they don't always build it a particular way just because the erection drawing says this is how you're supposed to build it, or the architectural drawing says you're supposed to build it this way. Sometimes for some reason there's a change.

And you don't always see this in a job— that's trying to— and this could be building a nuclear submarine okay as opposed to a building okay, building anything— they will actually go in afterwards and have an architect or engineer do a whole new set of drawings called the as-built drawings. How do we actually build it as opposed to how we thought we are going to build it okay? And that's just in case later you need to get in there and do something. Because people have found, when they're doing repairs on things they have some problem, and they go back to the original drawings and they come up with the correction based on the original drawings, and then they get down in the bottom of the ocean on an oil rig or something and they find out, oh, they didn't build it that way. Just because we have the drawings, that's not the way it was really built. And everything they just spent a million dollars putting together doesn't fit because it wasn't built the way the drawing said. So many times you have a set of as-built drawings.

So there's a whole series in the whole process of design. It's not— a design is not someone sitting down and laying out the final set of drawings. There's a whole hierarchy of these things. And in fact in building something like a nuclear submarine or a Boeing jet, a new Boeing jet, you can spend a billion dollars just doing the drawings. And it used to be in Course 2 men, which used to be Course 13, they had a whole group of people for shipbuilding 15 years ago, 10-15 years ago, when computers and CAD designs were getting more sophisticated, more powerful, that they could have a virtual walkthrough in a building or a nuclear submarine where you're going to put the piping, the ductwork, where the electrical cables are going to go. Because in the past on a nuclear submarine you'd have some guy designing the electrical cables and he puts them right in the middle of the ductwork because the guy who's designing the ductwork is putting the ductwork where he wants it to go and he doesn't worry about the electrical cables. And so you get down in the middle of the ship at the shipyard and the guys say, this is impossible, I can't build it okay. And that's why they would modify it on site, it would double the cost of a ship.

And it turns out where they designed the 777, I think that was the first Boeing aircraft design that was entirely designed on the computer. I know certainly that the new— digital, they called it the digital 737, after the 777— the original 737 design was a huge number of drawings. They had rooms full of draftsmen just sitting there making drawings for this aircraft, the old 737 back in the '70s, and they got updated and they had to have this library of what's the latest drawing. I mean I've been to Bell Helicopter and you know, you ask, can I see the drawing on this? They have to go through a library going back 30 years to figure out what drawing applied at this time when this helicopter was built, because the drawings change over time. It's not a constant thing. In manufacturing the drawings for automobiles are just horrendous in their complexity okay, the contours and shapes. I mean you like the nice aerodynamic design of an automobile? Well, the drawings on that just make that very difficult to do.

In any case, having a huge database where you can check— and that this was one of the mathematical problems of in these CAD designs— to see if you are putting the electrical cable or the drainage pipe through the air duct, because you don't really want, you know, big holes in your air ducts, they don't work so well. You don't want electrical cables going through your sewer pipes, but that's what they would have. And so one of the mathematical things was for the computer to find where things were intersecting when they should have been intersecting okay. And that's not necessarily the easiest problem. It's sort of standard practice today, but 10 or 15 years ago it was not standard practice. And so today they design most complex things— very detailed designs, the detailed designs are done in a CAD system. And so you still have a whole group of draftsmen, but they are computer draftsmen, IT engineers more so than the guys with pen and paper and whatnot.

Student: [Inaudible — apparently asks about data volumes, "terabytes."]

Yeah, terabytes okay. Well, I— they used to say that the paperwork— now this is not the drawings, this is just the quality control paperwork, and you know that occurred during building of a 747— the paper weighed more than the 747 okay, and had to be stored forever, until that aircraft went out of service 30 years later, 20— for 20 years later. Now it's all stored on some disk somewhere in a digital form okay. So there really has been sort of a revolution in manufacturing from aircraft to nuclear submarines to automobiles to whatever. It's sort of amazing what's happened. And there's a whole set of standards on how to do that, that have developed over the last 15 or 20 years on how to do those types of things okay.

Well, let me give you an example of an as-built that didn't quite do it the way and caused a serious problem. And this is the Hyatt Regency Hotel in Kansas City on July seventeenth of 1981. People been to Hyatt hotels, they know they have these big tall atriums. I lived in Atlanta, Georgia as a teenager, I think when they built the Hyatt Regency Atlanta, which was the first big Hyatt Regency with a big atrium. That building is still there, they built a bigger Hyatt right next to— but the old was only 40 stories tall. I think the new one's like 60 stories tall, but people still go to the old one to commit suicide okay. They go jumping off the balconies, and so you're sitting there eating your breakfast, you know, at the restaurant on the atrium floor, and all of a sudden someone comes splat right next to you. Actually doesn't happen that often, but this was one of the big things. I mean I remember as a teenager when I was just barely a teenager okay— because I moved from Atlanta when I was 13 or 14— but anyway I remember in the newspapers, this was this big 40-story atrium was— people would just come, and I think my parents, we went down to see it, everybody in Atlanta had to go see this atrium. And this was their selling point for the Hyatt Regency. And then about six months later people started using it as a place to kill themselves.

But nonetheless the Hyatt still likes to build atriums. If you go to Hyatt San Francisco, big atrium, you know, most places. And they built one in Kansas City, and the one in Kansas City they decided to have walkways going across the big atrium. So you had this ring of balconies outside the rooms, and actually some of the suite hotels have the same type of design with atriums going ten stories up. In fact they may be a spin-off of the Hyatt chain now. But nonetheless, the— they decided in Kansas City that they were going to have walkways that might take you from the 10th floor down to the 8th floor, and you could cut across 180 degrees across the atrium. And so they had a series of these at different angles and this was sort of artwork up there. And so on July seventeenth 1981, it was a Friday night, they're having a party, and there was a big band and the whole atrium was just a huge party scene. And there were a bunch of people on the cantilevered walkways going across the atrium, 10, 15, 20 stories in the air, kind of dancing to the music, soft on their feet. And then all of a sudden all these atriums— all these walkways started to collapse. And 114 people died and over 200 were injured.

And so the Kansas City Star won a Pulitzer Prize for tracking down what really happened in the design. And so here's the thing from the Kansas City Star, by the way this comes right out of the Petroski book To Engineer is Human, and when he wrote the book in— well in 1982 so— but it's updated 1986-87. Some of this was still in litigation. But what happened was the architect who had designed it decided they were going to have a hanger rod, I don't remember if this is three-quarter inch or 1-inch steel rod. You know, one-inch steel rod might hold 50 tons in theory, and if you put a safety factor on it— and this might have been a— this might have been a building safety factor. If they use the structural welding code, they wouldn't use the welding code because it wasn't welded, but nonetheless they might have used a safety factor of 1.67 for the steel. It should have been a factor of 2 safety because it's dynamically loaded right. This is not just static building, this is something that could be swinging out there in the atrium, so it's fatigue loaded. But anyway, the detail design basically showed a rod that had two C-channels that had been welded together. The rod would go through here and have a nut on the bottom that would hold the C-channel that held the walkway, and there are a number of these rods hanging down.

The only problem is, this was supposed to be a solid rod, it wasn't a threaded rod. How do you get the nut on? Well, you could have threaded as 15 feet down to the next walkway, but who wants to sit there and turn the nut for 15 feet to get it at the right position? No one had ever stopped to think about a shop drawing of how you're actually going to— or an erection drawing of how you're actually going to make this. So in the field they said, I can't make this, what idiot designed this okay. So they decided they would just fix the problem, and they have two rods, one rod hanging down with a nut up here and another rod coming through here, the nut on the bottom, solve the problem, they can erect this. As-built— this was design.

Turns out in this, in the management of this whole stream of design, anybody makes a change down here has to submit it to the people up here to have them sign off on it. Well in a big complex building like this there are thousands of changes, and they have to be signed off on them. And frankly if you've ever been involved in one of these things, people will go, you got to get it built, and you make the change that day, you submit the drawing, and it gets approved three weeks later by the architect. And under the law the architect in most states has to be a registered architect, and the engineer— something called the engineer of record who is responsible for the detailed design. You got an architect here for the architectural design, you got an engineer responsible for the detailed design. He could have a hundred engineers underneath him, but he has to be a professional engineer, he has to be licensed by the state, and he puts his seal on there okay, that he has checked it and he knows that this design has been calculated and checked and it's a good design, that it has enough strength.

Well, it turns out it either got done late or people just missed it among the thousands of details that were coming in for changes, or the person had no experience. I'm not sure they ever figured out exactly what happened. Certainly no one was coming forward saying, I did it, you know, after 114 people died. But what happens here, the load path— the original design was going straight through the steel rod here. The load path goes through the rod, takes a dogleg over, and then comes down through the other rod. And as Petroski says, well, think about it. If you had two people hanging from a rope, that's interesting. So you got a rope and you got a person up here hanging from the rope, and you got another person hanging from the rope down here. All you have to have is a rope that can bear the weight of two people right. But if you had a rope coming down with a person on it, and then this person was— you tied a rope to this guy's leg and that person was on there— now it turns out that person's leg is bearing, or this rope is still just bearing the same weight, but this person's leg becomes the weak link okay. And it turns out that's exactly what happened. The weak link was between those two bolts in the staggered system, and that basically sheared rather than a straight tensile pull. You change the entire load system with that small little change. The guys are erecting it, they didn't think it's a big deal, they figured I can make this, I couldn't make the other, you know, whoever designed this, it's impossible.

There's lots of design foibles. I mean there was, I think it's the 1978 Chevy Vega that, you know, they build the whole engine and they put the engine into the frame of the vehicle, and it turns out when you needed to go change the spark plugs on a Chevy Vega you had to remove the engine, because some of the other components were blocking— it was impossible to get a wrench in to remove the spark plug. Well in the plan it was no problem, the spark plugs were already in the engine when they put it in the vehicle, but no one ever bothered to check, could you ever service that engine okay. So there's little details that you have to worry about, the whole system going forward.

Well, the Hyatt Regency was— the as-built was not like the as-designed, and of course there was all kinds of finger-pointing so far as that goes. Another story out of automotive that I ran into, it's '90 or '91, I bought a Ford Taurus. And turns out I bought it in February. Never turn the air conditioner on in February for whatever reason. Come May, I turn the air conditioner on one day, doesn't work. Brand new car— brand new anymore, four months old, but still under warranty. So I take you to the dealership, they said, oh we don't have the parts to fix it. That's— fine, give me a new vehicle, I don't care okay. Because I'm of the Massachusetts lemon [law], all I have to do is bring it in three times to have them fix it, and if they can't fix it within three days then I can get a brand new car.

I did that to General Motors once in 1985 and got them to buy back my car, because it was a van and had a sliding side door, and the door fell off five times okay. You know, one time I had them come tow it out of my driveway because I didn't want to drive it over to the dealership, the door was just kind of hanging by sheet metal right. And I was not— trying not to go lemon law on them, but the owner of the dealership wrote a nasty letter to me telling me it was my fault. I said okay, I could have done lemon law after three, now it's five. Here's my lemon law demand. And General Motors came in and said, what do you want? Us to fix it, give you a brand new vehicle, or— and I said, I just wanted a van, I wanted to go on vacation with my family that summer, in two months. And they said okay, we have this, you know, executive vehicle, you know the one that they let their people drive for 5,000 miles and give you a discount? We'll give you another one, you know, as soon as this is available. And it was supposed to show up like in three weeks. Well they called me up and they said, sorry, the executive hasn't finished with it, we'd rather give you a check instead. I said, well I really wanted something to go on vacation— was alright, they gave me a check, I went out and I bought a— coughed a lot— and another dealership, okay, not the same vehicle, but anyway.

The problem with the Taurus was that they had a problem with all their air conditioners. The design of the air conditioner— they come up with a wonderful design that had just one sliding shuttle going back and forth, and said, you know, air conditioner you got compressor to compress the freon and then expand it and stuff, and they had this simplified design. But it had very tight tolerances on the shuttle going back and forth, because it didn't have the same types of seals that were in most air conditioners of prior design. So they had very tight tolerances, so the engineer— or the engineers who had conceived this— they decided, well, when we machine this we're going to have the proper fixturing so that when we clamp this thing and then machine it and then unclamp it we don't lose some of our tolerances okay, makes sense right. So they sent this out for prototype development, had the drawings, showed them not only how to machine it and the dimensions, but it also showed them how to build the fixture to clamp it when you're doing your machining. And they get the part back, they test it, works great, decide to build a hundred million-dollar facility to make these air conditioner compressors for the early '90s Ford Taurus.

And I happen to buy one of these early version air conditioners, which turns out it didn't work. And Ford had been learning, because there were probably people who bought it in February in Arizona who now were turning on their commute in March okay, that it wasn't working. So Ford had been learning that it wasn't working, and they went and fixed it. But as they were researching it, they were trying to figure out why it wasn't working, and they found out— they went to this machinist who'd done the prototype clamping and everything and said, well did you clamp it this way? "Oh no, I knew that wouldn't work, I built my own clamping arrangement." So he designed the clamping arrangement that worked, but they didn't know that. They didn't know the as-built of the prototype okay. They built a hundred million-dollar facility with tens of thousands if not hundreds of thousands of dollars of clamping fixtures that were wrong okay, that this machinist knew better okay. And so they had to retool their whole line to put in the proper clamping so they could get the tolerances they needed on the machining to make the air conditioners work.

Well, in my case, by the time I find out in May that my air conditioner is not working, they're way behind, and all their production of these— you know, major rapid improvement of their screwed-up stuff— is going to production models right, because they can't sell a vehicle in May that, in most parts of the country that don't have air— doesn't have air conditioning. Maybe in North Slope of Alaska, but not in the continental United States. Everybody wants an air conditioner right. And so I come in for repair, they say, well we don't have anything, will be it in July. That's it, will give me a new vehicle. And they said, well we can't. I said, well— I said, go give me a whole new air conditioner. And they said, we can't. I said, well give me a whole new vehicle at that point. And I said, I also know some vice presidents at Ford and I'll be happy to call them if you like. They found me an air conditioner okay. And I'm sure it probably slowed down someone's delivery of ordering a new vehicle right, because I got one that would have gone to the production line to sell new vehicles.

I actually didn't know that full story until— the guy who later became dean of Harvard Business School actually told me some of this other story. He had been studying Ford, and he had heard about this story. I knew what had happened to me, that they wouldn't give me an air conditioner and they were telling me I wouldn't have an air conditioner for most of the summer that year. And I said, nope, not acceptable okay. But the real reason was because some machinist knew more than the engineers, which is not uncommon okay. So that's the moral of that story okay. I'll see you tomorrow.