ML13098A250
| ML13098A250 | |
| Person / Time | |
|---|---|
| Site: | Calvert Cliffs  |
| Issue date: | 03/12/2010 |
| From: | Mary Anderson, Cinson A Battelle Memorial Institute, Pacific Northwest National Laboratory |
| To: | Norris W NRC/RES/DE/CIB |
| Norris W | |
| References | |
| Download: ML13098A250 (6) | |
Text
SIA PA Data Summary on Calvert Cliffs PZR Nozzle DMW March 12, 2010 Prepared By:
Anthony Cinson Michael Anderson Phased array data was collected March 7, 2010 by Structural Integrity Associates (SIA) personnel on an area of interest in a pressurizer (PZR) relief nozzle-to-safe end weld at Calvert Cliffs Nuclear Power Plant, Unit 1. This data was submitted to Pacific Northwest National Laboratory (PNNL) via email for an independent review. The data was analyzed using ZETEC UltraVision 3.0 software; a total of 7 data files were transmitted to PNNL. This review was performed to determine if any significant observations could be made that could potentially contradict the on-site evaluation of this region as containing a series of embedded fabrication flaws, i.e., does the data support the on-site evaluation, or can the presence of an ID surface- connected planar flaw (potentially PWSCC) be observed. This was the second set of phased array data to be analyzed; the first set was taken by EPRI, and PNNL reported on the analytic results in a brief letter dated March 11, 2010.
The data provided by SIA is again in the form of undesirable time-encoded sector files, which means that the inspector took data over a certain time period rather than using spatial encoding. Thus, no reference or spatial resolution information can be extracted, nullifying valuable views typically used in data analysis such as volume-corrected side- or end-views. However, although the data was not encoded spatially, it appears that depth calibration was performed, which will appropriately display the through-wall depth of responses observed. The actual scan pattern is not known, but the data appears to be of two types. One type is a simple axial scan towards the direction of the weld. This scan results in a single slice of weld volume, perpendicular to the weld axis, at an unknown circumferential coordinate. Data was independently acquired in this manner from both sides of the weld; three scans are looking upstream, while two scans were collected looking down stream. It is unclear if the scans on either side of the weld were acquired at the same approximate location or at separate circumferential locations.
An additional two data files were supplied that took time data in the form of a circumferential line scan; one clockwise (CW) and one counter clockwise (CCW) around the pipe holding the axial position fixed. The circumferential line scans, although not spatially calibrated, can provide some information in the volume-corrected (VC) end view. Ideally, a spatially encoded raster scan would have provided more valuable volumetric data inside the area in question. The data analyzed only gives a partial view
into the area of interest and may not be representative of the entire weld examination volume.
Axially-oriented, raster-type data were found to have regions that display ID reflectors, potentially caused by geometric weld root responses, together with mid to upper wall responses that are likely embedded weld defects such as slag. Both responses track in to the field of view together as time is progressed forward and the probe nears the weld. A snap shot in time has been captured from data file SIA AX Wedge LKN DS 1 and is displayed in Figure 1 below. In this image there appears to be a disconnection between the upper response and the ID response at this particular frame. Keep in mind this entire scan is only one thin slice in which other regions, if captured, could potentially show ID connection to the embedded flaw response. For example, Figure 2 shows an image from a scan that is looking up stream, in which one can see low level reflections linking the embedded flaw to the ID response. Again, a spatially encoded volumetric raster scan would show more of the region of interest in order to possibly confirm the likelihood that this is an ID connected response.
Figure 1 - ID response with embedded flaw responses observed.
Figure 2 - Embedded flaws shown above with possible ID-connected response.
The circumferentially collected data also provide another perspective view of the region in question. As time is progressed forward on these scans, the VC sectorial view shows mid and upper response move in and out of the image as the ultrasonic energy is reflected from the broad side of the embedded flaw. In one certain period of time, one can see a response from the mid wall with no apparent connection to the ID (Figure 3 -
taken from file: SIA Circ Wedge CCW Circ Scan.rdt). Other images show that possible low intensity responses link the mid wall response to the ID region (Figure 4 also taken from file: SIA Circ Wedge CCW Circ Scan.rdt at a different time position).
Figure 3 - Embedded flaw response with no ID connection visible.
Figure 4 - Stacked indications that could be interpreted as connecting to ID surface at time interval 64.
Creating a merged file from all the data contained in a circumferential scan allows the creation of a VC end view to be displayed. This depth calibrated view, although having no calibration in the length of the scan, can provide some meaningful through-wall localization of the responses. The merged data of file SIA Circ Wedge CCW Circ Scan.rdt in Figure 5 shows an end view with the relative amplitude threshold set to display data above 20%. This image shows mid to upper embedded flaw responses disconnected from the primary ID response seen consistently throughout the data. This strong ID signal would likely mask any corner trap ID response that could potentially link the flaw to the ID. Figure 6 shows the strong ID response gated out of the image revealing possible ID connections. Moreover, time interval 64 in this merged end view, highlighted in red, correlates to Figure 4s unmerged VC sectorial view showing the stacked indications.
Figure 5 - VC end view of CCW circumferential time encoded line scan.
Figure 6 - VC end view of CCW circumferential time encoded line scan with time interval 64 region highlighted.
The limited PA data analyzed here does not provide enough volumetric insight into the region of interest to conclusively evaluate whether that the observed flaws are ID-connected. Plausible arguments could be made to support either of these cases. This type of time-encoded data does not adequately support post-analytical methods of imaging that would be needed to ultimately make this call. Having stated that, we see no reason to question the examiners performing the manual phased array, as they are able to display much more data, albeit real-time, regarding general UT response trends such as weld root, etc., that could be used to disposition these flaws.