ML19317H052
| ML19317H052 | |
| Person / Time | |
|---|---|
| Site: | Rancho Seco |
| Issue date: | 12/30/1977 |
| From: | Walbridgo W SACRAMENTO MUNICIPAL UTILITY DISTRICT |
| To: | |
| Shared Package | |
| ML19317H049 | List: |
| References | |
| REF-PT21-78-028-000 NUDOCS 8004040533 | |
| Download: ML19317H052 (4) | |
Text
01I y
78 -cm.o e c
.' 6
- O sACRAME!dTO t.tUtHCIPAL UTILITY otS1RICT C 6201 s street. Box IM30. Sacramento, california 95813; (9:51 4 December 30, 1977
~Director of Regulatory Operations ATTN: !!r. R. H. Engelken HRC Operations Of fice, Rcgion V Nuclear Regulatory Comission 1990 ?! orth California Boulevard Walnut Creek, California 94595 Re:
Rancho Sece Nuclear Generating Staticrk
~
Operating License DPR-54, Docket Ho. 50-312 10 CFR 21 Report dated December 27, 1977
Dear Mr. Engelken:
In conformance to the requirements of 10 CFR 21, we are sub-citting this, report as the required follow-up to a telephone notification made in your office by Mr. R. J. Rodriguez, Man 0ger of Huclear Operations on December 27, 19i'7.
This notification was made after an evaluation was rendered by the Plaitt Review Committee (PRC) and their findings were concurred irith by fir, f!cdriguez.
Mr. William C. Walbridge, General Mana;er was informed that the 10 CFR 21 regulations were applicable to the situction.
Tne discovery cf the " defect" as defined by 10 CFR 21.3(d) consists of an err.or in the softwar'e for the plant computer. The com-puter, a Bailey 855 model supplied by Babcock and Wilcox Company,(wa calcui.: ting values for the Departure from Nucleate Boiling Ratio DNBR, axtra,.olated to worst case conditions) that were higher than plant conditions warranted. This non-conservative error resulted from the method by which "best guess" reactor coolant flo'.1 data was extracted from a " spurious input" subroutine whenever actual data is, out of range.
The output data from whis subroutine was stored using a multiplication factor different from that assumed in the DNBR calculation.
This led to the flow data being increased by a factor of 10 after which the first significant figure was truncated.
For.the particular case of Rancho Seco, these incorrect flow rates were approximately 136 percent of actual flow. With higher reactor coolant ficw rates than actual as data input, the DNBR calculation produced higher values to reflect the improved heat tiansfer.
A recalculation of data shoucd that at no time did the OMBR drop below the limiting vaue of 1.0.
For this reason, the filing of a report-able occurrence report :as not necessary. U never, this may not be the case for other units which have different reaccor ccolant flow rates.
The results may.be extremely conservative or they tay be non-coriservative 8004040 [ N
Mr. R. H. Engelken Decenber 30, 1977 to "the point where the ONBR limit is vi,olated.
For this reason,- tiabcock and Wilcox was informed of the probleni so they could pursue havin'. this problem corrected at othe; plants with similar software if a similar situation exists at other sites.
The cor;puter programming has been changed to correct the scaling problem, allaving valid DNBRs to be calculated.
Ve truly your, *4 u
's L'
p.
Um. C. Walbridge General Manager e
8 a
e e
9 6
9 9
9 8
O e
e e
l j
~ ~ - - - ~ ~.., _ _.
~
ISTRICT SACRAMEN.O MUNICIPAL UTILITY:
. OFFICC f.tCMortANDtJf.8 6
'oATc:
Decc=ber 28, 1977 4
10 CFR 21 Rcupunne on !!igli D' dR/Flov I
st WcT:
During the utartup testing for Cycle 2 at. Rancho Seco Iuc3 car Generating Station' it vas noticcd that the pricary heat halciace calculated by the Plant process corpulcr.as shoeirip. approximately 120% full power when the recepr was operated.
i Durin0 Cycle J vith sin 11er conditions, it was usual to at 100% Iull power.
observe the prit sry systna heat balance to ind'ieste approximately 90% full power 100% full power.
when t'in setendcry hunt balance was controlling the plant at This dif fercuce v:.a not conuidored to be an operational problent as the prinary In attempting to determin systen he=L balance is nut used "above 15% full power.
what had caused the appart.nt inc cc::e in the primary heat balance, it was found that the reactor caulant systen flow used in the best balance calculation ucs Normally, flov is obtained ohnorually high, approxinately 1367, above actual.
fudepandently f roa each of the four reactor protection cystem channels, then j
In averened end place:1 in t he co: puter nemry for further use and reference.
thi.; et.se the ntored taluc veu unusually high, while the indicated flows from l
. In cach of the p.pS channels ucru endeutially the same as Cycle 1 values.
l researching this dificconce, it van found that a flag had been set in the scit-vere, In111ag the corputer that one of the input flows was bad.
This directed j
for further ust the sof tware to calculate a substitute average flow and store it
~
Jn the norm 1 flo$ averaging routine, the values ebrained carry a Since the l zero pecceding the tuo significant. figuren before the decimal poine.
automatically truncates averaging routino a3 rays expects this leading zero, it and scorcs only the significant digits.
Tn the case at hand, the substitution routine sis: ply calculated the two significant dJgits and passed them back to the averagin;'. rentine which uas expecting the 3-digit input which it irnediatel truncated. As an exanp3c, typical loop flow is 69.5x10 pounds par hour.
The i
0 6 pounds per hour to be effect of truncating then provided the value of 95.cx10 the high stored. Since the priw.any system heat balance relics on stored flow, h%t vns uct known vcs that it is the stored flov which f
power level resulted'.
used in the deterninntion of DNRR in the subsequent thermal-hydtculic calcula-1!cace, wh0n the flag was reset to return the flow to nor=al, a drar.1 Lie This drop amounted to approximately 'l.5 tions.
drup in calculated DER vas obccrved.
1CBR units when operr. ting at. full power with neninal power distributions.
thin had on the results of startup terring doni Of cajor co.:cern was the ef fect.
[
at the begin ing of cyclu in whic!r abnormal power distributions were imposed
!!e vere able to enrrect to simulate design rather then normal configurations.
thi' results of the neartup teer pror. ram by cureciating then with tne before i
Tlic 1:ortinr.houre L'-3 DEM correlation used in the end after flew fix p*:r,P.%.
Sof tware incorporates flev as on a.!ditive z.ud lineal parameter; thes the correcticaa were r.trainht foneard.
In corrceciaci all of cur. t. cst. resul:s we 13 2% fu11' po.vr. the fuund that, there c::trape. lated to the deni.'n evcri:vwr ut
.af nicuis Ichn drupt ed f ro.: 2.99 to 1.04.
No safety violation or consideration 4
~ '
s-u_,.
.--.--.-w,--p.--%-,,.
- 9w*w.
g m-w.
,9e p,.-
,g 4
y-,94
.T9~t8-'"VW'"""*WW V
4-YT Y}
T T
"7*
W"'YW- ' -
. -...=.__
'e Page 2 December 28, 1977 exists at Rancho Seco nint c the minimum acceptalde design ~ DNBR is 1.30 and this value was ucver equtzled or execeded.
The,1cplicacicus of this proble:: are apparsnt to other plants using the sar:e or sirdist so f t.va rc.
If t.he design t.arr. ins for a particular reactor or cycle were not as great as those at R..ncho Suco, they could be deluded into believing that their cininua D:!$'s werc 7,ccater than acceptance criteris, when in f act, this vould not be the esgu uhun t,he correction to actual finu was made. ' Die Rancho Seco compulcr has been eprogra=1cd no that it vfil not truncate v'.icn using the substitution routina.
At thir. point in tice, we have not been abic to establish
'what originally ser the fica in the f.of tusrc which caused the substitution routine to be used. The purpose of this report. is to nahe other unsrs of this software cuare of this idLosyncracy sud to suggest that they revise their software prior to observing the problem.
3 E
p O
g e
s e
m.
, _