Restarting a Broken Nuke

The Nuclear Regulatory Commission has determined that (in their opinion) Southern California Edison did NOT mislead the NRC about design changes to their failed replacement steam generators for San Onofre Nuclear (Waste) Generating Station.

The billion-dollar mistake was done according to the rules.  They admit that the rules might need to be changed, but the rules were followed, according to the NRC.

Actually, though, SCE arrogantly described their efforts to avoid a proper “10CFR50.59 review process” which would have entailed not just heightened NRC scrutiny, but even some citizen’s public scrutiny as well!  But the NRC doesn’t care what SCE said in those industry publications, where SCE bragged specifically about circumventing 10CFR50.59, which requires an open regulatory review process for significant design changes at a nuclear power plant.

It should be noted that the followed string of changes actually occurred, all without need for a “10CFR50.59” review: The tube material was changed to a newer, more durable, but significantly less thermally conductive alloy, requiring 10% more surface area for the same thermal output.  Hence more tubes, and hence they eliminated the stay cylinder in the center (to pack in even more tubes), and hence ALL the tubes were packed in a bit tighter than before.  The tighter-packed tubes were too tight around the U-bends, near where the anti-vibration bar goes along for the ride (it’s not connected to anything but the tubes themselves).  (See section 8.b.1)

The whole thing was basically redesigned to flutter!  Ooops!  And yet it passed 10CFR50.59 after-the-fact “scrutiny”!  It feels like a foregone conclusion.

So much for finding the real “root cause” of why San Onofre has been inoperable since January.  One key phrase turned up: “The replacement steam generators were designed and fabricated in accordance with quality assurance requirements, and 10CFR50.59 does not require the licensee to presume deficiencies in the design or fabrication.” (p. 36)

Rushing to make a shipping schedule probably isn’t considered a deficiency the licensee needed to presume might occur, either.   But in fact a decision was made: „not to control the positive pressure, the dew point of nitrogen, and the oxygen content on the primary and secondary side of the Unit 3 replacement steam generators [during transport from Japan] to accelerate delivery schedule.‰ (p. 45)  The NRC is still reviewing whether “corrective action” will be taken on that and several other issues.

Also presumed not germane to 10CFR50.59 was whether shipping the entire steam generator in other than a “gravity-neutral position” (in fact 45 degrees OFF from that position) would have any impact.  It seems the lifting points made rotation to a gravity-neutral shipping position difficult.  Shucks.

One of Unit 3’s SGs was apparently damaged in handling: “Unit 3 replacement steam generator 3E0-88 accelerometers indicated up to a 1.23 g spike with a simultaneous recording on all three of the attached accelerometers.  Mitsubishi provided an evaluation of the forces which showed loads were within allowable stress limits but exceeded stress for an operating basis earthquake.  The team was not able to determine if this was properly considered.” (p. 46)  A steam generator weighs well over a million pounds.  Don’t drop it.  Don’t bump it.

Additional fabrication problems required extensive rewelding, nearly doubling the number of rotations of the Unit 3 SGs during fabrication. (p. 63)

As for computer modeling, they have a nice phrase for an error: “nonconservative results”.  Such “nonconservative results” (p. 47) from one computer program which calculates flow rates were diligently fed into another computer program to determine resultant vibrations, which, of course, were way below what actually happened.  Garbage In, Garbage Out.

That’s why it’s good to use real-world modeling, not just computer modeling.  Or, as the Augmented Inspection Team put it: “The accuracy of calculating fluid-elastic instability is limited based on inputs that are best determined by design-specific mockup test data.  Mitsubishi did not perform design-specific mockup tests.” (p. 49)

In one attempt to use “conservatisms” in their vibration analysis, a multiplier of 1.5 was inserted… but the multiplier was actually required to match real-world testing results! (p. 50)

It should be noted that in trying to isolate which software model failed, it was admitted (p. 56) that the models only predict “bulk fluid behavior based on first principles and empirical correlations”. So the models don’t actually handle fabrication differences between steam generators, or structural differences, etc..  In fact, someone with a slide rule did some calculations to check on these wonderful “computer programs”.  They’re that basic.

Back in the real world, here’s what happens: “If operating velocities reach [a] critical value, vibration amplitudes can increase rapidly and fluid-elastic instability forces can lead to rapid pulsation and damaging of tubes.” (p. 49)

The U-bend region of the tubes is most susceptible, but the industry believed everything would hold together — until San Onofre proved otherwise: “This event at SONGS is the first US operating fleet experience of in-plane fluid-elastic instability, sufficient to cause tube-to-tube contact and wear in the U-bend region.”

Shocked, are we?  Well, then I should add that there is no standard for what margin of safety should actually be designed into steam generators to avoid fluid-elastic instability.  Each manufacturer has a different value.

So, what about restarting San Onofre?

The AIT report indicates the only way SCE might be able to restart the reactors is by changing the operational parameters, because: “SONGS replacement steam generators were not designed with adequate margin to preclude the onset of fluid-elastic instability” which can cause “tube leakage and/or tube rupture.” (p 56)

Replacing the steam generators will take several years and cost about a billion dollars.  And what design would they use? And who would fabricate them?  The old SGs weren’t much better:

“The original steam generators installed throughout the domestic fleet of pressurized water reactors, including SONGS, experienced widespread corrosion of the tubes and tube support plates, stress corrosion cracking of the tubes, and wear at tube supports.  These problems led to the replacement of nearly all of the original steam generators, in most cases well before the end of their design lifetime.” (p. 41)

And they can’t simply make the old design out of the new alloy, because the heat transfer characteristics are so different.

They’ve already plugged well over a thousand tubes, many because they were already severely worn, others in the hope of reducing the amount of steam compared to water in certain areas of the steam generators.

Restarting Unit III seems out of the question at this point, and to restart Unit II, SCE has already determined that it would have to be at a lower power setting. (p. 56)

But the only way to know for sure that a lower power setting will work is to try it.  And that might result in a tube rupture, possibly even followed by a cascading failure of one tube after another, resulting in a massive radiological release.

Zirconium cladding, hot fuel, loss of coolant… these are a dangerous combination!  Add in a multitude of companies and agencies each laying blame anywhere but on themselves, confusing and inadequate regulations and regulatory oversight, language barriers, proprietary manufacturing operations, whistleblower intimidation here (and undoubtedly in Japan too) and you have a recipe for … exactly what happened.

Or worse.  Let’s not restart San Onofre.  Ever.

Russell D. Hoffman lives in Carlsbad, California. He is an educational software developer and bladder cancer survivor, as well as a collector of military and nuclear historical documents and books. He is the author and programmer of the award-winning Animated Periodic Table of the Elements. He can be reached at: rhoffman@animatedsoftware.com