ML20056C678

Informs That Several Quality Issues Were Discovered During 930610 Meeting Re Gist Test Program Conducted by GE Which Include Lack of Design of Info for Gist Facility & Discrepancies in Tracg Models Supplied by GE

ML20056C678
Person / Time
Site:	05200004
Issue date:	06/16/1993
From:	Kenneth Jones IDAHO NATIONAL ENGINEERING & ENVIRONMENTAL LABORATORY
To:	Han J NRC
Shared Package
ML20056C679	List: ML20056C678 ML20056H139
References
NUDOCS 9307190167
Download: ML20056C678 (3)

Text

o s .

^

IMEb toona katsanno inprneervng Laboratory June 16, 1993 Dr. J. T. Han '

U. 5. Nuclear Regulatory Comission 5640 Nicholson Lane ,

NL/N-353 Washington, DC 20555 G15i TEST PROGRAM QUALITY RELATED DEFICIENCIES - KRJ-05-93

Dear Dr. Han:

As discussed during our meeting on June 10th with RES and NRR, several cuality assurance issues, related to the GIST Test Program conducted by General Electric Co., were dis' covered during the RELAP5 code assessment work perforced under FIN L2549. These issues include a lack of design information for the GIST facility and discrepancies in the TRACG models suppliec by GE. As detailed below, these problems have a direct impact on the ability to perform an effective assessment of the RELAP5/ MOD 3 code.

There is a very limited amount of design data available for the GIST l facility. We examined the design records file (DRF A00-02917) for the GIST Facility during our data gathering trip to the GIST facility last October. This documentation consisted of several boxes of poorly .

Organized documents, memos and note's collected from individuals who had .

worked on the pro 3ect. Confiouration control over design data appeared to be non-existent. In many cases, the DRF contains several conflicting conceptual designs for a particular component with no indication of the final as-built configuration used. Model development efforts were particularly hampered by the complete lack of isometric drawings of the 1 facility. Due to the lack of facility design data, the RELAP5/M003 models developed for code assessment are based to a large degree on the TRACG mocels supplied by GE. ,

Several discrepancies exist between the TRACG input decks supplied by GE ano the limited amount of available facility design data.

a Several elevation discrepancies exist in the TRACG models. Unlike RELAPS/ MOD 3 the TRACG code does not perform consistency checks for elevations around closed loops. As a result, extra care must be '

taken to prevent artificial " gravity head pumps" from being >

introcuted into the model from elevation differences between two sides of a fle.< loop. The following table summarizes the overall piping elevation changes for several flow paths for each model.

b 3 ')7[E d M 7)J//.XA Enclosure hEGrG.- p.o. sax is2s issha1aus.to s341s

l Dr. J. T. Han

. . June 16, 1993 KRJ-05-93 Page 2 Table 1. Summarv of Elevation Discrecancies in the TRACG inout Decks.

l Design A07 B01 207 C01A D03A Data m (in) m (in) m fin) m (in) m (in) m (in) l GDES Line i 5.5580 5.6997 6.6421 6.6421 5.6997 5.6997

! i (220) (224.40) (261.50) (261.50) (224.40) (224.40)

! Intact l 2.3114 4.6277 2.1586 2.1586 2.1526 4.6277 l Steamline (91) (182.19) (84.984) (84.984) (84.984) (182.19) l Broken 2.3114 3.6277 N/A N/A 0.9450 3.6277 Steamline (91) (142.82) (37.205) (142.82)

Co ncomer ! 7.4422 7.5814 7.5814 7.5814 7.5814 7.5814

, l.nnulas .(293) (298.48) (298.48) (298.48)

(298.48) (298.48)

RPV Elev. 7.4422 7.5620 7.5620 7.5620 7.5520 7.5620 Ad.Jacent (293) (297.72) (297.72) (297.72) (297.72) (297.72) to the Annulus Vent Line 6.9596 5.5790 5.5790 5.5790 5.5790 5.5790 (274) (219.64) (219.64) (219.64) (219.64) (219.64) 1 Most notably, the GDC5 line elevation changes in the two main steam line break cases are over a meter greater than the design data indicates. This error effectively raises the suppression pool elevation, increasing the static pressure head by about 10". for these two cases. The ciscrepancy is reduced to about 0.1 m in the other three TRACG cases.

• The two vent lines were lumped into a single line in the TR.ACG nodel, however, the hydraulic diameter used in the models was incorrectly r based on a c1rcular cross-section of the combined area.

• Wall friction terms are neglected in the steam lines, GDCS lines and break lines. It is not clear from the available information whether or not these losses are accounted for in the local form loss coefficients.

• The moceling of the DPV ta11 pipe cuenchers and the outlet of the vent in the suppression pool does not correspond to the facility design.

TEE components are used in the model to provide separate steam and licuid flow paths into the suppression pool for each of these connections. This moceling approach appears to have been used to correct a limitation of the TRACG code: however, the impact of this moceling approach on the calculations is not discussed in either the TRACC mocel notetooK or the GIST Final Report (GEFR-00350).

_. . 1 l

Dr. J. T. Han June 16, 1993 KRJ-05-93 Page 3 .

.Thermallossestotheenvironmentdreneglectedinthemodels. The -

DRF did not appear to contain any information concerning the  ;

properties or thickness of the insulatic used in the facility. j The TRACG model notebook was of little benefit in resolving these discrepancies. Model documentation contained few references and did not '

appear to conform to even minimal quality assurance practices.

The modeling discrepancies were corrected in the RELAP5/ MOD 3 model where design data was available; however, it should be noted that in many cases, f while information such as overall elevation drops may be corrected in the i mocel, it was not possible to verify other key information such as piping ,

lengths due to the lack of design data. '

As always, piease do not hesitate to contact me at (208)526-0809 if you I have any cuestions.

Sincerely, 5/

K. R. Jones l NRC Thermal Hydraulic  !'

Analysis Programs i

cc: D. E. Bessette. USNRC, MS NL/N-353

• W. H. Rettig, DOE ID MS 1134 -  ;

L. M. Shotkin, USNRC MS NL/N-353 '

i i

i 1

t

?

4 I j

~

L i

-