Thursday, March 8, 2007

What I Should Have Said in Less Than Two Minutes

On February 20-21, 2007, the DOE's Nuclear Energy Research Advisory Committee held an open meeting in Idaho Falls, Idaho. They discussed several activities related to nuclear power developments worldwide. Time was allotted for public comments and I discussed Regulation by Myth for about eight minutes. However, following is what I should have said in less than two minutes.

Gentlemen, you are likely eager to bail out of here after the foregoing endless sets of slide presentations. Today I have only virtual slides. Slide one lists eight nuclear power plants that have applied ultrasonic fuel cleaning. Slide two shows the ultrasonic fuel cleaning equipment. Slide three has two photographs of fouling on nuclear power plant fuel rods. Slide four shows that with fouling, light water reactors have operated with fuel cladding surface temperatures in the range of 1200 degrees Fahrenheit and above (instead of the range of 550 to 600 degrees Fahrenheit that is the design basis for long term operation). Slide five is fantastic and non-existent; it shows the growth of Zirconium dioxide scale and also the increase in dissolved oxygen concentration in the cladding with time at six operating temperatures: 600, 800, 1100, 1500, 2000 and 2500 degrees Fahrenheit. To produce slide five, more experiments are needed with Zircaloy cladding, as called for by the AEC Commissioners decades ago.

Wednesday, March 7, 2007

A Real Small Break LOCA that could not be isolated

It was a good thing that I was at the operating station about 53 years ago at National Reactor Testing Station (NRTS) in the desert near Idaho Falls, Idaho. The Materials Testing Reactor (MTR) was about a year or so in operation and I had the job of installing and starting up the first pressurized water loop in the MTR for prototype Nautilus fuel. With about three years of solid engineering experience in the nuances of in-pile testing, I inherited this job somewhat by default. The Nautilus was running as well as prototype reactors in the Idaho desert, so why test prototype fuel? However, the project was underway when I went to work at Argonne National Laboratory near Chicago during November 1952 and testing of the prototype fuel proceeded by inertia.

We completed fabrication and crude shakedown testing of the loop at Argonne, Chicago, and shipped the gear to NRTS. With expert assistance and design improvement by key personnel from Phillips Petroleum Company, the operating contractor for MTR, installation and startup of the equipment proceeded in record time. So the loop was operating with the prototype Nautilus fuel in place, when I had my baptism in managing an accident. A 1/4 inch diameter sensing line at the main flow metering orifice severed and blow down from 2500 PSI and 570 degrees F. began. I was at the control panel and the first indication of trouble was a loss of flow signal that let to switching of the primary pump to a standby pump, followed immediately by a scram (fast shutdown) of the MTR.

I very quickly knew where the break was, and I knew that it could not be isolated. I decided that the best approach would be to allow the primary pumps to automatically shift from A to B to C and back to A, etc., as the blowdown proceeded. An option would have been to turn off the pumps, but that would have led to a real loss of cooling for the prototype Nautilus fuel while it still had a relatively high level of decay heat. Of course, this option was not an option forever. At some point, the pressurizer would be empty and voiding and real loss of flow would quickly follow.

My first step was to turn on a low capacity piston pump that delivered about one gallon per hour to the loop. I valved this flow to the liquid level reference sensing standpipe for the pressurizer. I knew that the loop was operating with a concentration of dissolved hydrogen for corrosion control. I also knew that the cold reference sensing line acted as a cold thumb and the the hydrogen concentration on the standpipe would be at saturation for 2500 PSI and its temperature of under 100 degrees F. And I knew that as pressure was reduced during blowdown the hydrogen would begin to outgas in the standpipe and that water would thus be bubbled out of the standpipe. The liquid level signal would then be false and the pressurizer would be empty even though the instrumentation would indicate otherwise. I also knew that the modest flow into the standpipe would not upset the accuracy of the calibration of the level sensing system.

So, I allowed the loop to blow down and depressurize, and as the level approached 10 percent of full, in quick order I proceeded as follows: I turned off the primary pumps. Next, with assistance from Fred McMillan of Phillips Petroleum Company, I isolated the in-pile pressure tube assembly from the main loop (two valves) and then opened the cooling of the in-pile assembly to once-through cooling by process water (two more valves).

Of course, it is not always great for one's career to be on the scene of an accident. And it makes little difference if one's moves were somewhat lifesaving. A few months later I was back at the home office at Argonne, Chicago. At one point I was asked what good I thought a gallon per hour of injected flow would do in the blowdown situation. Well, I did not answer.

Decades later, the value of a degassed reference standpipe (or the safety problem with a hydrogen saturated standpipe) was not recognized by some very highly paid consultants as well as the equipment suppliers of huge nuclear power plants.

Saturday, March 3, 2007

Blind faith in single tube tests in the production of TRACE

The following text in italics is copied from the transcript of the full ACRS meeting on February 1, 2007. It is a very small part of the part of the transcript that covers TRACE, however, it reveals more than the previous lengthy discussion of the TRACE activities.

MEMBER ABDEL-KHALIK: But philosophically, if you had a perfect code, and you understand the physics, then it doesn't matter what the scale is because you're verifying phenomena. And therefore, by this process, you're essentially saying the code is nothing more than an empirical fitting tool for the experimental data. Is that true?

MEMBER BANERJEE: It cannot predict new phenomena.

MEMBER ABDEL-KHALIK: Because you are limiting the range of applicability of the code, essentially, to a rather narrow range around where the experiment is. So the code, you philosophically by doing this, you're viewing the code as nothing more than an empirical fitting tool.

MR. BAJOREK: I think that's an accurate statement.

MEMBER POWERS: Do you really want to say that though? I think that's what he was getting at.

MEMBER BANERJEE: It's not predictive of new phenomena.

MR. BAJOREK: That's the -- these codes are not based on first principles. They are based on and held together by closure relations which are based on sub-scale experiments. A lot of those correlations come from single tube tests and you are using that at faith when you start to look at larger and larger scales. Assessment helps to benchmark and let you know whether those correlations are truly applicable with those other conditions but going back to the experiments, we all in integral tests in particular, you want to try to establish a basis for that system global-wide behavior and is it going to behave much like you'd expect in something with much larger scale. But the smaller scale test, that's all you have to run the full test.

MEMBER BANERJEE: As we come to full scale tests.

MR. BAJOREK: If we had full scale tests the --

MEMBER BANERJEE: The assemble system, we can do it in components.

MR. BAJOREK: Components, yes. That's all


The complete transcript from which the above was extracted may be found at:

http://www.nrc.gov/reading-rm/doc-collections/acrs/tr/fullcommittee/2007/ac020107.pdf

However, all of the single tube tests, and also the larger scale tests were conducted with clean (unfouled) heat transfer surfaces. Faith in those tests is blind faith. More later on this.