Wednesday, October 24, 2007

Feeder fun - part 5

Ok, time to take this baby home!

These feeders were extremely happy, doing their thing, bleeding a little iron, when 2 things happened. First, there was a bad steam explosion in a Japan nuclear plant. If you saw the pictures, you would see that these crazy guys let a steam pipe corrode down to a pop-can. You can see the metal peeled back like an orange! There were only two things to be learned here:
-it's amazing how thin these pipes can get and still work!
-this is a Japanese nuclear plant

The second big thing was that some badly repaired feeder pipe cracked over in Quebec. I actually snuck into a printer room and got some pictures, and showed it around to some smart people. In cross-section, you could see that they joined two pieces of pipe that were not straight, so that there was a kink at the weld. This would result in a tremendous concentration of fatigue stress, and you could see the thousands of micro-fractures at the weld. Here is a good weld, where you can see all the zones.


A pipe butt-weld is sort of the same thing, except it is done from the top, and is 'blind". In other words, you want the bottom to be nice, but you don't know exactly what it is going to turn out like. This is another major 'initial conditions' uncertainty.

Of course, fatigue was not the interpretation of those who scented blood. They went ballistic, saying that all Candu feeders were in danger from flow assisted corrosion (FAC), and they informed the regulator, for good measure. They sent the pictures just to a FAC guy who duly reported it as FAC.

The race was now on for some real evidence, since large departments were being formed, and promotions scattered like sand. Anybody who came up with it, would be highly rewarded! And it came! The technique was familiar to all geophysicists: reflection seismology. But in this case, they shrink it down to ultrasonic size.

Here is their fancy mega-dollar machine that came out of this.



We all know the huge error bars involved with reflections. Yet, when I had my meeting with some top managers, they showed me a result of the pipe thicknesses, presented as single fine dots, spaced over a year. You joined the two together, and there was a line straight down to hell. You couldn't argue with that!

Finally, even that graph was not enough justification for billions. They had to run a crummy modal (seismic) finite element analysis, using 30 year old assumptions, and the 30 year old code. Since the pipes were first designed to that, just shaving off a micro-metre would guarantee failure, and that's what happened.

In reality, modern seismic experience shows that you cannot damage small-bore steel pipes in an earthquake. Sometimes, the piping is the only thing standing! Basically, the steel pipe acts like a rubber hose, and merrily shakes away, not doing anything.

That's how it now stands. I was going to do some more on their bizarre interpretations of weld cross-sections, but blah!

No comments: