ML031110255
| ML031110255 | |
| Person / Time | |
|---|---|
| Site: | Davis Besse  |
| Issue date: | 04/08/2003 |
| From: | - No Known Affiliation |
| To: | - No Known Affiliation, Office of Nuclear Reactor Regulation |
| References | |
| FOIA/PA-2003-0018 | |
| Download: ML031110255 (2) | |
Text
Ellis -
Thanks for the report. I've passed it along to the staff working this issue. I've also attached a section from a draft report we have on RPV (and cladding) flaw distributions that we've put together unrelated to Davis Besse. The report is still under publication review but please feel free to share it with folks 'in house' and, of course, that includes Ian.
One point I want to stress is that we are not arguing there are no flaws in cladding - our own work clearly shows such flaws exist although the great majority of them are small.
This observation is consistent with the size of the defects reported in Ian's paper (0.008 to 0.015 inches deep). Also, Ian's paper shows that many defects tend to have crack tips that are rounded rather than sharp, which is similar to many of our observations.
Mark Kirk put together the attached plot that shows the data we got from inspections of an RPV beltline, including the cladding, with the flaw density adjusted for the wastage area in the Davis Besse head. For flaw depths of about 20% of the cladding thickness, you would expect to have about 0.5 flaws of that size in the wastage area. For flaws more like 40% of the thickness, you would have about 0.04 flaws or a 1-in-25 chance of having such a flaw.
Our approach for Davis Besse stems from the view that the smaller flaws, which have a higher density, have a relatively small effect on the failure stress; they are not all that deep, they tend to be short in length, and both factors are important. The larger flaws, which could have a larger effect on failure pressure, are not very likely. Thus, when faced with modeling a complex geometric configuration and doing so in a relatively short period, we went with the assumption of no flaws in the wasted region. A more robust treatment clearly would treat a distribution of flaws, actual cladding thickness variations, and variations in material properties. However, doing that analysis would/will take considerable time and effort.
We would be interested in comments from you or Ian (or other interested folks) on the flaw data and the report section.
Mike
0
.r4 cN a co
- CO U) a.
00 a)
NO NXw I n CD e - It
$- $$ . O O O . O - O.. aCD O.
- o. ,,* ****
Ct 0 Nl C~ MU i-0m-N 1- (
O~. ..
o 0 0 0 0 0 0 oc 0
w 0L-eojv 9DIe4sem asses S!Aea u! 4unof) Mel paiew!is3