ML20238A946
| ML20238A946 | |
| Person / Time | |
|---|---|
| Site: | Davis Besse  |
| Issue date: | 08/31/1987 |
| From: | Chon Davis EG&G IDAHO, INC. |
| To: | NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| References | |
| CON-FIN-A-6827 EGG-2510, NUREG-CR-4946, NUDOCS 8709090483 | |
| Download: ML20238A946 (83) | |
Text
N kasy&ame29h W$&WSS&&Q meewMMedh y4 v % upud 4
p h
1%d xq Mes fghw% s y % wp h g@
4grp pmikys d3 y d L
g r pukqmAn m
a$$g g g @g@qiM f$ Qg pammaa-mom w$As%gn@gy$wwww@g ig $w w$g 1os p e9 g
ugsfQ R h0; y 6g$p; s ; y VW$4gMWWNpu@b@k@Nk b bh pdA I@k@MW@p%@ b Nh he#qweewmgamsw s%
uwwe wegh whypAeAsms hhkh ew sug h ?pa g g qqqqg A y g g e h wy,
@6%@M@M$@WWWNWWNW6MR8M Aw?BiWiB2#g$m n pp h kh n w nm : n. wm y m ;w n ene ,e nnx +g pf?
74 , , ,,m,wmw a. .
& qop +. .c.
, c , . , , c
- h ;WerkperYonned,vr&rMMW%4 syn
%c yp,@,...n&a %y J W L $,, We W.. ~
I;dpoicorwacepQEiACQ70GD016W f%pfn0 d&;W M-~9 gn k ' y4 W Y% $U:S.'?N06IeE f[QQmqf4%g&(QMVQ 4 @e c h an h yw%mw.u @ g$ b gHem o: % W ut46 g w %e &g ank g r%j sygg% a p k ~,w y eisy%y0R $ac%uw#%agm y 4 ya s v m ffra m wwm ' mww m+wn mw wQwwwg N'Mp/Q&m%' ' M M AffMA'm%Q W."MdgM r n n w qw npwA.
% gpidgM M$a.g @ M,g
,,b : h ,k bb 4 hhh nu i I
' t ohq sM
- w. p w.g*:g fg g( muww ><c
% w,-g%, qd,/gm, ,o?m y gt .
pc -'.q ;gg<
,j e . q;,4 s 3. ,gs.p f urmu e j:
'4s:,+
4
,r.
s u~ p . ms w
r i
. py .<,1g ,gfa
.,a g
. %;g
.ii<-
?p g o
- 3. 4,,
g im % .
g? g%p
( M 2q;{fg &e.?xgA%
, m g gf,. [ f. , g } f.c(My,;-. n ; m,[j w
q n .s.
!Mhh' S;
v % -
N @w~
- df M '
Y
%{ p h d
JXlQ,Ny$ ? > ' . xv O h ,+l l.; +
Q* % . cT Q . w m , gg .g ey ag a t gj cy >
<# +r weg%g*.
, W ,, Qy
?
y l[y y 1; ' Q' , Q' ' '
.L . q;.?
PN ,
4%:g(? 6^;V < - f,N h mrar. -
W .e pg 3 , ,
m lp s *7 >
a.... m - ,
9g c, p; g
qv ,
( %. ~ r 5,
~
%on fN* n. .'
k 5 l:-
n w~e c
- s[s n [a o nw,. ,c. + ,
,w,.
' L 1 f Ih' ( / , - - 4
-i gy y ,
' 4
['
%. V , ,
hMd.itibNatibhat~boratory; y @n Engineering 4a mg t w s n'a , .
A@
jMa n u g e dlp 9lh e FYl$lD b p a riin'e n t o 'f; E n e r g y/
ww - -
y
' 3,/ N Am ggy
+ m, )!
h%, < 1 g a.,
,mi e
(
-H, M R~-M...
[ ( . f f )( { (
! " j q.4 '(yyL e W, ( [a; ;
- h j m ,,
nu myX , 1
$m$$ HWY ,, + -m .
l
4 - -
4, y 1
h '
4
- g. .
yA
- F bi c y . . ,
o f-F .< ,D w & ) {nag-) ! - Qi at w (1 - e
$y * '
N;p I_
)$ 'ef ' ;W [ ,h (j g(f.
q -
3 4[{
+
A,_ :
? I
- x .hhm._[_ j k$gL,Il,f h ,,
~
s
- j. .
j m ?~ g :y: - -
~
,- n w 4, wran.
, amt sumf> q, %yp a y, a ;
?
s i Y,g%g ; w$w&gph0@%
jw g= s f -~
g N
ahg g
f fQ$QNf 9 i y u
1_ bd@ $ Wd m
mn@h+a$ngMMMN##WMMMMMM k
m 4mmw u.memwem44 m nn h 1
h6 N . nkNh n h~ h k hh h k k- k,h- n _h 1
un
$wdNWqw w w w p m a n n yn p y w ~ ag w wL fM m+gg%gpdn!M my man NknW
$ ,wm&p @k@%w@ggggn gsd7gdm@ MM s MLQ%: @s % PW 2.'o & % et n
@ ya f Jyi mm QQ f x p MW "d"MM gny ,h: Qe?k Mk m
- Ts %wM -
y :: 5v+
Vk~)Yhh m m w4 b Y$m$
- e,h_f
- Y _M ~ if"_k f h$; .
_N N UQ g% v~ . . )
% a [i$ '
3w jdM . > NMh.Qj,(
! MW', ,ay1
- m sQ NN @__0$3
' 4 <i" '
I o s
+ a "e;m r &. - <
w a m o..
s
%<7; sgQpu 4.W gni%e xq m 1 m", < n
- P 7
. &;G yp 4 !
W\' m:cg y y
~ , c : ,
_ , yuqx
- .W 399 4
hhk:d Y. - *>[ h"N x '
% d@b ! ..x-<.3, h!( -hD M '
M Ws 4 hMt$ JWM CA 'R (kQ t
kq w l&m s am /W MRllw w" 3 NR W M":lmn P Wya: M.WNOTICE%[3WWM ~ '&
w~Mp%qY
( QMWJ a%M4hwW:W C 0 W:' ' s &a%m%gi i
pijf M <.n dThI8' r$ Port WO prepared. AE an accoua[of:Workleponsored by[an'a(.:ncy / ,lK 1lJ)dd$fhhof igg NMW@$Qhc% [mXV, tlwWited Sata Gover wnt mor'anyV- r WaWM;M g
s n,&
l M.N8 ,k! m e
M+~the.UnitM gegencylthere~ofl
~ -
morinny Stain..Governmen.
of their emp m-flNstGlo~yees(ma: nty @ Ar
-s g ids
- y '
m
~a~ny~warra
.p yJ ndQ P!DS%
~
4,y gh@
, W JM%cW Fd@r impliddioramenws.any lesal umbility or responalbility for any third party's * +
fT June, or the itsults of such unciof snyinformationtapparaths, product ar proce jpWa hhhqf$
q Lggs@
p Ns(disclosed ls thisiepor6@dpreents thadts use py such third party [would I:
- 4. Jng infelnse,,privsely owned risht a + 'f Nh -
rm u he w:p v ' - .' "m.+mr 1 b
kjN%p%GMir + / ' @';% 9" +" ' '
'<3 h
, pasmh :h ekh .
s N
e
(
m mn ,
T
[,
%@MLMMYYffiq
$g T Sp ,% b.u n ts ' [ h
' i N
f WW sMe .~ l Q Q
? h , i.
k .g
)
RhW 3
b ,M, mh y [w
^
'. v~ .
7
! $y5 .m 7 jbMd E ' , &wn s eggf , hh_ 5Y a $i $
- A 4 .a - :- . L -- =
l NUREG/CR-4346 EGG-2510 Distribution Crtegory: R4 DAVIS-BESSE UNCERTAINTY STUDY Cliff B. Davis Published August 1987 EG&G Idaho inc.
Idaho Falls, Idaho 83415 Prepared for the Division of Reactor and Plant systems Office of Nuclear Regulatory Research U.s. Nuclear Regulatory Commission Washington, D.C. 20555 Under DOE Contract No. DE-AC07-761D01570 FIN No. A6827
l l
l i
ABSTRACT The uncertainties of calculations of loss-of-feedwater transients at Davis-Besse Unit I were determined to address concerns of the U.S. Nuclear Regulatory Commission relative to the effectiveness of feed and biced cooling Davis-Besse Unit 1 is a pressurized water reactor of the raise & loop Babcock & Wilcox design. A detailed, quality-assured RELAP5/ MOD 2 model of Davis-Besse was developed at the Idaho National Engineering Laboratory. The model was used to perform an analysis of the loss-of-feedwater transient that occurred at Davis-Besse on June 9,1985. A loss-of-feedwater transient followed by feed and bleed cooling was also calculated. The evaluation of uncertainty was based on j the comparison.s of calculations and data, comparisons of different calculations of the same ;
transient, sensitivity calculations, and the propagation of the estimated uncertainty in initial and boundary conditions to the final calculated results.
(
l i
1 l
l l
FIN No. A6827-International Cule Assessment: RELAP5 and TRAC-BWR ii r
SUMMARY
On June 9,1985, a loss-of-feedwater (LOFW) tran. R585 model were performed at the Idaho National sient occurred at Unit 1 of the Davis Besse Nuclear Er.gineering Laboratory (INEL). The NRC Office of Power Station. Davis Besse Unit 1, owned and oper- Nuclear Regulatory Research asked that INEL assess ated by the Toledo Edison Company, is a pressurized the uncertainties in the R585 calculations. The INEL water reactor (PWR) of the raised-loop Babcock & developed a detailed, quality-assured RELAP5/ MOD 2 Wilcox (B&W) design with a rated core power of model of Davis-Besse, referred to as the R586 model 2772 MWt. This transient, which was initiated from because it is a RELAP5 model that was developed in 92% power, resulted in a temporary but total loss of 198fp The R586 model was then used to repeat the main and auxiliary feedwater. Auxiliary feedwater was R585 calculation of the Davis-Besse LOFW transient eventually restored, and the plant was taken to a safe and a R585 feed and bleed calculation. The feed and and stable condition. bleed calculation represented a LOFW event initiated Because of the poteniial severity of the event, the from 100% power, with feed and bleed started 20 min-U.S. Nuclear Regulatory Commission (NRC) im- utes after the beginning of the transient. The calcula-mediately began a program to analyze the Davis-Besse tions repeated by the INEL will hereafter be refereed j LOfW transient, including parametric variations. to as the R586 calculations because they were per- ]
These parametric variations were primarily related to formed with the R586 model, the use of feed and bleed cooling. Feed and bleed cool- The R586 calculation of the Davis-Besse LOFW ing, which involves starting the makeup and high- transient was in good qualitative and quantitative agree-pressure injection (HPI) pumps and opening the pilot- ment with the measured data. The trends observed in operated relief valve (PORV) located on the top of the the plart were well represented in the calculation. The pressurizer, would have been used to remove decay maximum deviation between calculated and measured heat from the core if auxiliary feedwater had not been reactor coolant system (RCS) pressure was about restored during the LOFW transient at Davis-Besse. 0.3 MPa (50 psi). The deviations between calculated The NRC pursued a two-pronged thermal-hydraulic and measured RCS temperatures were generally less analysis effort of the LOFW event: an in-house analysis than 3 K (6'F). Even though different thermal-of the event performed through the NRC Office of hydraulic computer codes and input models were used, Nuclear Reactor Regulation (NRR), and an indepen- the R586, R585, and TRAC calculations were similar dent analysis performed at Los Alamos National and showed trends like those observed in the plant. The l Laboratory (LANL). The NRR analysis utilized the differences that were observed between the calculations Nuclear Plant Analyzer (NPA) and the RELAP5/ were primarily due to the assumption of different core MOD 2 thermal-hydraulic computer code. The inter- powers, feedwater flows, and pressurizer spray flows, active features of the NPA abowed 15 calculations to The R586 feed and bleed calculation exhibited the 1 be completed in a short period of time. The LANL phenomena expected in a LOFW event. The secondary f analy3is utilized the TRAC-PFl/ MODI computer side of the once-through steam generators (OTSGs)
)
code. Both analyses were completed quickly to pro- dried out 130 s after reactor trip, resulting in a heatup vide a rapid assessment of the effectiveness of feed and of the RCS. After feed and bleed was initiated at bleed cooling at Davis-Besse. For convenience, the 1200 s, the RCS pressurized until the pressurizer safety NRR calculations will hereafter be referred to as the relief valves (SRVs) opened. The liquid lost through R585 calculations because they were performeo with the PORV and SRVs caused the mixture level in the RELAPS and completed in 1985. The calculations RCS to drop below the pressurizer surge line. The ,
performed by LANL will be referred to as the TRAC resulting flow of 6 team through the PORV caused the l calculations. RCS to depressurize after 3750 s. The makeup pumps Although the R585 and TRAC calculations indicated began refilling the RCS at 5160 s. The RCS pressure that feed and bleed could successfully cool the core dropped below the shutoff head of the HPI pumps at I ifimtiated early enough, the NRC realized that there 5500 s, causing a more rapid refill of the RCS. Feed were several uncertainties in the calculations. One and bleed successfully cooled the core, which was source of uncenainty was due to the code input models covered with liquid throughout the calculation.
used to make the calculations. Both the R585 and The R586, R585, and TRAC calculations all in-TRAC calculations were performed with models based dicated that feed and bleed could successfully remove on Oconee Unit 1, a lowered-hop B&W PWR, that core decay heat if a total LOFW event occurred at were quickly modified to represent Davis-Besse. The Davis-Besse. However, some significant differences modifications to the Oconee model that resulted in the between the R586 calculation and the R585 and TRAC iii
calculations were observed, particularly with respect certainty in the calculated collapsed liquid level in the to event timing and RCS pressure response. These dif- reactor vessel was estimated to be 1 m (3 ft). The ferences affected the course, but not the ultimate out- uncertainty in the RCS temperature when feed and come, of the transient and were attributed to differences bleed was initiated in the R586 calculation was in the boundary conditions. In particular, the core estimated to be 5 K (9 F). This uncertainty corre-decay power wanoo small in the R585 calculation after sponds to 11% uncertainty in the calculated RCS the trip of the reactor coolant pumps because of an er- heatup rate after OTSG dryout. The corresponding ror in the input model. The PORV flow was thought uncertainty in the time required to reach the RCS to be too small in the TRAC calculation. The .:o temperature at which feed and bleed was initiated was parison of calculations indicated that the specific results about 2 min. These uncertainties were caused by were sensitive to the boundary conditions. However, uncertainty in the imtial and boundary conditions, the macroscopic results of all the calculations were primarily the initial OTSG liquid inventory. The similar in that feed and bleed successfully cooled the uncertainty in the initial and boundary conditions did core. not alter the results of the R586 calculation relative to The uncertainty in the R585 and R586 feed and bleed the ability to depressurize the RCS during feed and calculations was evaluated. Several potential sources bleed. l of uncertainty were identified which could contribute The uncertainties in the R585 calculations were to the overall uncertainty in the feed and biced calcula- estimated to be larger than in the R586 calculation tions. These potential soarces of uncertainty included because of the error in core decay power following the thermal-hydraulic computer code, the code input reactor coolan' pump trip as discussed pres sously. This model, the initial conditions and boundary conditions error caused h bias in the R585 calculations in addi-of the calculation, the code user, and the assumed tran- tion to the uncenainty associated with the initial and sient. The important parameters that determine the boundary conditions. The bias in the RCS temperature thermal-hydraulic signature of a feed and bleed tran- at the initiation of feed and bleed ranged from 2 K sient were identified. These parameters included the (4 F) to 11 K (20 F), depending on the time between RCS temperature at the initiation of feed and bleed, the trip of the reactor coolant pumps and the initiation the ability to depre4surize, and the minimum liquid of feed and bleed. The corresponding bias in the time level in the reactor vessel. The uncertainties in the im- required to reach the RCS temperature at which feed portant parameters were estimated based on several and bleed was initiated ranged from I to 6 min.
factors, including subjective judgments. Better esti- The above estimates of uncertainty are valid for the mates of uncertainty would be obtained from more ex- transients analyzed based on the assumed initiating tensive comparisons of feed and bleed calculations and event, equipment performance, and operator actions. ;
experimental data. If different assumptions were made regarding these The uncertainties in the R586 feed and bleed calcula- parameters, the differences in the calculated results tion were thought to be relatively small. The un- could exceed the estimated uncertainties.
1 iv l
l
i
]
[
T ACKNOWLEDGMENTS The author wishes to thank the following people for their contributions to this report:
M. Bolander, for the development of the thermal-hydraulic model of the secondary system; C. Dobbe, for the analysis of the loss-of-feedwater transient at Davis-Besse E. Jenkins,-
for the generation of the report graphies; M. Waterman, for the development of the model of the integrated control system; and P. Wheatley, for technical support and review, The W
author also tnauks personnel from the Babcock and _ ilcox Company and the Toledo Edison ,
Company for their cooperation and support in providing the information used to develop the model of Davis-Besse. The contributions of E. Throm, of the NRC Office of Nuclear Reactor Regulation, and J. Lime, of Los Alamos National laboratory, are also gratefully acknowledged.
v i _.____.___._ _____.___________ _ _ _ ____
)
CONTENTS -
f AB STRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... .... .... . .. ........... il 1
SUMMARY
- ............................................................. iii
- ACKNOWLEDGMENTS . . . . . . ... ......... .. .. . .... . . ... . .... ....... v NOM ENCLATURE .. . . . . . . . . . . . . . . . . . . . . . . . ....... ...... .. ..... ... ... . . x 1
~1. I NTRO D UCTION . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . .. . . . . . . . . . . . . . . . . . . . . . . . 1 '1 l
- 2. CODE ' AND MODEL DESCRIPTION . . . . . . . . . . . . . . . . . . . . .... . . .. ...... .. ... 3.
2.l l Code Description . . . . . . . . ..................... . .. . .. . .. ... ... ... 3
' 2.2 Model' Description . . . . . . . . . . . ,.. .......... ....... . ...... ,, . ... . 3 2.2.1 Thermal-Hydraulic and Control System Model . . . . . . . . . . . .. ....... ..... 3-2.2.2 Initial Conditions . . . . . . . . . . . . . . . . . . . . . . . . . .... .. . .. .. .. 3 2.2.3 Boundary Conditions ...... . .... . . .. .. . . . . . . ... . . 5 ]'
2.2.4 Quality Assurance . . . . . . . . . . . . . . . . . . .. . ............... . . ... 5
- 3. - R ES U LTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . ... .... . . .. 7 1
.. I 3.1 R586 LOFW Transient Calculation . ... . . ..... .. .. . . ......... . .. 7 3.2 Comparison of Davis-Besse LOFW Transient Calculations . ... . .. .. ......... .. 15 .
. 3.3 RJ86 Feed and Bleed Calculation . . . . . . . . . . . .... .. . ..... . ...... .. 19 l
- 3.4 Comparison of Feed and Bleed Calculations . . . . . . l
... . . . ... .... . 26 '
.i 3.4.1 100% Power ...... .............. .. . ... ..... .. .... . .. .. 26 3.4.2 90 % Power . . . . . . . . . . . . . . . . . . . . . . . .... .. . .. . . . . . . . . 30 3.5 - Calculation Uncertainty . . . . . . . . , ... .. . .. ...... . . . . .. .. . .. .. . . . . 37 -
- 4. - CONCLUSIONS ................... . ..... .. ....... .. . . . ... . . 41 i
- 5. REFERENCES........ ... . . .... ... ... ... . .. ... .. . . . ... .. 43 1
APPENDIX A-R$86 MODEL DESCRIPTION . . . ... . ...... .. . ........ . . ..... A-1 l 4
)
' APPENDIX B-COMPUTER RUN TIME STATISTICS . . .. . .. . . . .. . B-1 APPENDIX C-CALCULATION UNCERTAINTY .. ... .... ... . .... . ..... C-1 1
i 1..
vi
-l FIGURES l
' I '. Feedwater flow to OTSG A in the R586 calculation of the Davis-Besse LOFW transient .... 8'
'2. Feedwater fl'ow to OTSG B in the R586 calculation of the Davis-Besse LOFW transient . . . . . . 8
- 3. ^ OTSG A main feedwater control valve area in the R586 calculation of the Davis-Besse LOFW transient .. . . . . . . . ........... .. . ...... ... ........... .......... 9
- 4. OTSG B main feedwater control valve area in the R586 calculation of the Davis-Besse LOFW transient . . . . . ...... . . . . . ... .. ................... . ......... 9
- 9. Reactor power in the R586 calculation of the Davis Besse LOFW trarcient. ..... . . 13
- 10. Reactor coolant pressure in the R586 calculation of the Davis-Besse LOFW transient 13
- 11. . Coolant temperatures in the B loop hot leg in the R586 calculation of the Davis-Besse LOFW transient . . . . . ... .. . ...... . ... ... .. . .... . .. .... . . 14
- 12. Coolant temperatures in the B loop cold leg in the R586 calculation of the Davis-Besse LOFW transient . . . ..... .. . . .. .. . .... . ... ......... .. . . 14 i
- 13. Pressurizer liquid levels in the R$86 calculation of the Davis-Besse LOFW transient . . . . . 15 l
- 14. Mass flow in the A loop hot leg in the R586 calculation of the Davis-Besse LOFW transient . . . . . . .. . ... .... ... ... ..... .. .. ... ...... .... 16
- 15. A comparison of calculated reactre coolant pressures during the Davis-Besse LOFW transient . . . ...... . . . . ....... .......... .. .. . .. .. ...... 16
- 16. A comparison of calculated pressurizer liquid levels during the Davis-Besse LOFW transient . . . . ... ...... .. . .. .. .... . .... . .. .. 17
- 17. - A comparison of calculated coolant temperatures in the B loop hot leg during the Davis-Besse LOFW transient . . .. .. .... ... .... .... .. .. . .. . . 18
- 18. A comparison of calculated startup liquid levels in OTSG B during the Davis-Besse LOFW transient . .... .. . . . . .... ..... .. . .. .... . . 18
- 19. OTSG pressures in the R586 feed and bleed calculation . . .... ... .. . 20
- 20. OTSG startup liquid levels in the R586 feed and bleed calculation . . . . ., 20
- 21. Reactor coolant pressure in the R586 feed and bleed calculation . . 21
- 22. Hot leg coolant temperatures in the R586 feed and bleed calculation . 21 l
vii
l l
1 l
- 23. Pressurizer liquid level in the R586 feed and bleed calculation . . . 22 j
- 24. Hot leg subcooling in the R586 feed and bleed calculation. . . . .. .. .. .. 23
- 25. Mass flow through the pressurizer SRVs in the R586 feed and bleed calculation .. 23 1
- 26. A summary of mass flows into and out of the RCS in the R586 feed and bleed calculation 24 1
- 27. Hot leg collapsed liquid levels in the R586 feed and bleed calculation . . . .. . 24
- 28. Hot leg and PORV void fractions in the R$86 feed and bleed calculation . . 25 l
- 29. Collapsed liquid level in the reactor vessel during the R586 feed and bleed calculation . . 26 j
- 30. A comparison of calculated reactor coolant pressures during feed and bleed . 27
- 31. A comparison of core power from the R586 base and sensitivity feed and bleed calculations . . .. . .. . . .. . .. . . . , 28 )
I
- 32. A comparison of calculated collapsed pressurizer liquid levels during fced and bleed . 28 l I
- 33. A comparison of calculated mass flows out of the RCS during feed and bleed 29 l
- 34. A comparison of calculated mass flows into the RCS during feed and bleed . 29
- 35. A comparison of calculated collapsed liquid levels in the reactor vessel during feed and bleed . .. ,. . . .. . . . 30 i
- 36. A comparison of calculated reactor coolant pressures when feed and bleed was initiated near 2200 s . . . . . . . . . . . . . . . . . 31
- 37. A comparison of calculated mass flow out of the RCS when feed feed and bleed was initiated near 2200 s. . . . . . . . . . . 31
- 38. A comparison of calculated subcooling in the A loop hot leg when feed and bleed was initiated near 2200 s . .. . . . . 32
- 39. A comparison of calculated collapsed liquid levels in the pressurizer when feed and bleed was initiated near 2200 s . . . . . .. . 33
- 40. A comparison of calculated collapsed liquid levels in the reactor vessel when feed and bleed was initiated near 2200 s . .. . . . 33
- 41. A comparison of calculated reactor coolant pressures when feed and bleed was initiated near 1500 s. .... . . . . . . 34 42 A comparison of calculated HPl flow rates when feed and bleed was initiated near 1500 s . 35
- 43. A comparison of calculated fluid subcooling in the A loop hot leg when feed and bleed was initiated near 1500 s . . . . 35
- 44. A comparison of calculated collapsed liquid levels in the pressurizer when feed and bleed was initiated near 1500 s . . . . . . . . 36 <
l viii l i
- 45. A comparison of calculated collapsed liquid levels in the reactor vessel when feed and bleed was initiated near 1500 s . ... . . . . . . . . . .. .... ............ ............ ......... 36
' 46. Results of the volume balance for different times after reactor trip '. . . . . . . . . . . . ........ 38
- 47. The effect of uncertainty in boundary conditions on the results of the volume balance. . . . . . - 39 A-1, RELAP5 Davis-Besse model; loop A . . . . . . , . . -.. ............... .. ............. A4 A-2. RELAP5 Davis-Besse model; loop B. , . . ., . ........ .. .......... . .. . . A-5
- A-3. RELAP5 Davis-Besse model; pressurizer system . . . . . . .... .... . . . . . . . . . . . ' A-6 A4. RELAP5 Davis-Besse model; reactor vessel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . A-7 A-5. RELAPS Davis-Besse model; condensate and feedwater system. . . . . . . . . . . . . . ... . A-8 A16. RELAP5 Davis-Besse model; AFW system .. .. ... ... .. ... .... ........ A-9 A-7. RELAPS Davis-Besse model; OTSG A and steam line .. . . . .. .... .... .. . . . . . . . A- 10 A-8. RELAPS Davis-Besse model; OTSG B and steam line . . . . . . . . . . . . . . . . . . . . . . A-11 '
- A-9. : B&W integrated control system organization . . . . . ... .... ... ... .... .... A-13 C-1. The effect of hot leg nodalization.on collapsed liquid level in the A loop hot leg . . . . . . . . C-5 C 2. The effect of hot leg nodalization on RCS pressure. . . . . . . . . . . . . . .. . ... .. ..... C-6' I
l C-3. Results of the volume balance for different times after reactor trip . . . . . . . . . ... .. C-9 C-4. The effect of uncertainty in boundary conditions on the results of the volume balance . . C-10 i TABLES
- 1. Comparison of desired and calculated initial conditions at 92% power . . . . ..... .. 4 l
l 2.- Comparison of desired and calculated initial conditions at 100% power . . ..... 4'
- 4. Sequence of events for the feed and bleed calculation. . . . . ... . .. .. 19
- 5. Uncertainty in RCS temperature. ,. . . . . .. . .. . .. . 37 B-1. Computer run time statistics . . . . . . . .. . . ... .. .. .... ,. 'B-3 C 1. Uncertainty in initial and boundary conditions . . . . . . . .. . . . . C-7 C-2. Uncertainty in RCS temperature. . . . . . . . . C-7 l
l i
l ix L__-_____-__2__ _ _ _ _ _ _ - _ _ _ _
J' s y ,
, i
.l/ ( , 0 3
J y '
7,
(:
1 i J' H NOMENCQdVRE j
iAEV atmospheric exhaust valve '
i :s NbR. NRC Office of Nuclear Rextor - .
,I Regetation o'f r
AFW , auxiliary feedwater ,. P '
9 t.-
W A' $ '
W
.t i' OTSG once-through steam generator.
LARTSl anticipatory reactor trip system N" y PORV: e pilot-operated relief valve B&W. . Babcock and Wilcox
.? PWR- ' pressurized water reactor ,. _
,/ .
DADS data acquisition display system' 1";m Qc volumetric flow due to steam,powhetion- -,
ECCS emergency core cooling system in the ' core
- [
- / .
EPRI - ' Electric Power Research Institute'! ' -Qm volumetric flow of makeup HPI high-pressure injection Q, voluractric flow out the PORV j HPVV- high point vent valve - RCP reactor coolant pump .l INEL Idaho National Engineering Laboratory RCS .rcactor coolant system LANL- Los Alamos National Laboratery ,, u, j^ RELAP Reactor Excursion and Leak Analysis; y;.
' Program LOFW loss of feedwater.
R585 RELAPS completed in 1985/ !
LPI low-pressure injection R586 RELAP5 completed in 1986 k MFP main feed pump SFRCS steam and feed rupture cocirci sys:dm [
MSIV main steam isolation valve '
- t
^
NPA Nuclear Plant Analyzer TBV turbine pypass valve -
.NRC U.S. Nuclear Regulatoly ComSussion TRAC Transivt Reactor Analy*is Code
( ,
I f k
'}(
i
?
'F b
_.~ l. (
ll ,
'l
- 1 1 !
i l
'I I
- c K
. , i jf;, ;f G s 6M .
M -
1 %. a y
. DAVIS-BESSE UNCERTAINTY STUDY l
f
, 1. INTRODUCTION 1
g /
On June 9,1985, a loss bfeedwater (LOFW) event coven . Feed and bleed cooling was initiated at 15,20, occurred at Unit 1 of the Davis Besse Nuclear Power or 35 c.in after the event started in the other three Station.1 Davis-Besse Unit 1, owned and operated by calculations. The calculations performed by LANL are the Toledo Edison Company, is a pressurized water referred to as the TRAC calculations in this report.
reactor (PWR) of the raised-loop Babcock & Wilcox The NRR analysis consisted of 15 calculations, with 7'
(B&W) design with a rated core power of 2772 MWt. j initia\ power of either 90% or 100% of rated core At the time of the LOFW event, Davis-Besse was ; p&r. Parametric calculations investigated the effect operating at 92% power. The transient was initiated - iof. time to initiation of feed and bleed, the effect of by an overspeed of one main fel pump (MFP), which , ' makeup flow prior to feed and bleed, and the effect
' caused rk pump to trip, eventually resulting in a reac- ! cf PORV flow. The NRR calculations are referred to f' " tor trip Subsequent failures then caused the complete is the R585 ca'culations in this report because they loss of all feedwater. Without feedwater, the reactor Twere performed with RELAP5 in 1985.
coolant began to heat up. Auxiliary feedwater was The macroscopic results of the R585 and TRAC restored about 20 min after the MFP trip, and the plant calculations were similar. The TRAC calculations in-was taken to a safe condition. dicated that ked and bleed would successfully cool the 7 Because of the potential severity of the evert, the core ifit was initiated within 20 min of the start of the 4 U.S. Nuclear Regulatory Commission,(NRC) im- transient. LANL also believed that feed and bleed
'4 mediately began a program to analyze the LbBesse would successfully cool the core if it was initiated s
transient, including parametric variations puuadly within 35 min of the start of the transient, although related to the use of feed and bleed ccolig Red and this result lwas not calculated directly but instead was
' bleed coolinF, which involves openinghe pilot- based m an extrapolation. The R585 analysis indicated operated relief valve (PORV) located on the top of the that feed and bleed could successfully cool the core pressurizer andyoartitig makeup pumps and high- if initiard within 37 min of the LOFW.
preisure injectugHPI) pumps, would have been used Althnugh the R585 and TRAC calculations indicated to remove decay heWrom the Davis-Besse core if aux- that fred and bleed could successfully cool the core iliary feedwner ha@t been restored. The NRC pur- if initiated early enough, the NRC realized that there j ', Sed a two-pro'nged ihermal-hydraulic analysis effort were several uncertainties in the calculations. These N ~ of the LOFW event at Davis-Besse. First, the NRC uncertalaties were related to the plant models used in performed an in-house analysis of the event through the calculations, the initial and boundary conditions the NRC Of6ce of Nuclear Reactor Regulation (NRR). assumed in the calculations, and the uncertainty in the Second, an hM:peredent analysis2 was performed at the codes used for the calculations. The uncertainty due Los AlamorNational Laboratory (LANL), Both to the plant model exists because at the time of the analyses used thermal-hydraulic computer codes as Davis-Besse transient the NRC did not have a model their primary calculdral tool. NRR utilized the Reac- 6f Davis Besse Unit 1. The R585 and TRAC calcula-tot Excursion and Leak Analysis Program (RELAPS/ tions were performed with models based on the Oconee j l MOD 2),3 while LANL used the Transient Reactor Unit I that were quickly modified to resemble Davis-Walysis Code (TRAC-PFl/ MODI).4 The NRR .Besse. (Oconee Unit 1 is a B&W lowered-loop PWR).
,( 'nalysis relied heavily on the Nuclear Plant Analyzer The NRC Office of Nuclear Regulatory Research
[ , ; ff.PA)/ The interactive features of the NPA allowed asked that the Idaho National Engineering Laboratory nany calculations to be completed in a short period (INEL) assess the uncertainties in the R585 calcula-r of time following the LOFW transient at Davis-Besse. Mons. The assessment of these uncertainties is the
. A simulation of the LOFW event that occurred at mbject of this report. The uncertainties in the plant Davis-Besse was performed in both analyses model were assessed by developing a quality-assured LANL calculated four parametric variations of the RELAPS/ MOD 2 model of Davis-Besse based on event. These parametric calculations started 3 rem 90% detailed Davis-Besse plant information. This model rated power, close to the power at the start of the was then used to repeat the R585 calculation of the Davis-Besse LOFW transient, ar>d assumed that feed- Davis /Besse LOFW transient, as well as a LOFW water was not restcaed in one calculation, feed and transient from 100% power with feed and bleed ini-bleed cooling wa.s not initiated, causing the core to un- tiated 20 min after the start of the transient. These
1 calculations, called the R$86 calculations because they Davis-Besse LOFW transient and feed and bleed I were performed with RELAP5 in 198f6, were then calculations are described in Section 3 Section 3 also compared to the results of the R585 and TRAC calcula- compares the results of the R585. TRAC, and R586 )
tions to assess the uncertainty in the results. Section 2 calculations and provides an assessment of the of this report provides a description of the uncertainty in the calculations. Conclusions are RELAPS/ MOD 2 computer code and the quality- presented in Section 4. References are provided in assured model of Davis-Besse. The results of the R586 Section 5.
2
M&MML y ll L %,
s q e; r i
,[ y. y%
( {t y as /c (u 3s
/ ,
'i f 'y i ,
A t 2. CODE AND MODEL DESCRIPTION
- /
h
- ha The RU_.A?5/ MOD 2 Wnpn r de and 'he R586 repreerM all the major flow paths of t% primaty and Davisdesse modetlare dmnbed in Sections 2.1 and 2.2, respectives,. b.
.ecoraay coolant systems. The mode' of the primary coolant system,40 called the reactor coo' ant system
} ,
g (RCS), i? presents the two-by-four t.nn0guration of the plant, 4, twu p3, each containing one hot leg and 2Al Code DOScripdOR [i -
two cold qs. The two loops m dNgtuted loop A s
$} ,
' The RELAPS/ MOD 2 etawm ct& was designed and hx,p B. W i ressunzer is connectcd to loop A.
for thctmal-hydraulic analysGf ft sn!<n PWRs At Davb-Besse, the loops are sometimes referred to and related experimental s9 cms. MLMMOD2 as loop 1 and loop 2, with loop I corresponding to was developed at the INEL to ,,smulath wide kariety loop B. The model of the secondary cootmt system of the mal-hydraulic trav6ent im olving steam, inter. < .
includes representations of the conSmat.' and riain and noncondensible finid mixture.t RELAP5 MOD 2 y edwatd systems downstreh of the deaCrJour u.otage has been used p sirri. ate sn.all-break and large-break tanks, the auxiliary feedwater (AFW) system, the race-loss-of-coc'em acd jifis, operational transients mch throur.,n steam generators (OTSGs), and the main steam as lens of feedwater , al even severe accidents to the lines. kcat structures were used to represent di the point of fuel damage. The code can be used to skilate major incud components in the plan. Many oHhe olant tMdwry and reewdary sy stemt of a PWP,, includ. control sygems were also mod 6,'d, iry.Liding a detrRed W M%! axe of plant. "Re version of the cods used raodel of t'he integrated cour6 system (ICS). In addi-
% in :his sph. sa t 6 RELAP5/ MOD 2 Cysle 36.04. don to the ICS, the pressurizer prer, sun ed level con h- '
Mass dent 4 tde code has been pei ormed.6 RELAP5/ MOD 2 iidesenbed in deuilin Reference 3.
f trol systems antiepatory tercor Mr sy> tem, and the steun and feed rupmre control suttm CiFRCS) are
' i' RELAPMMOI8 contains interactive capabilities .tiso mo;1eled. The RELAPS model of Lavis-Best e was wWh, inimjuncion with the NPA, allow an analyst desigmd to allow ft11 interratNe mntrol of the mWor to ecutrolimod dispir2y the resuhs of a calculation as no rNnents, such as pumpa and valves, which an e is running. The ma!yst can interactively contml the optar can carut 1 in the plant The model corrained same nujm couporents. ia r. plant model, such as 101 volumes,232 junctions,188 heat structures, and pumpa and Mva, which an operator nr. control in atmut 147d control varbbles.
the actw.i piant. N NPA display ot'a calculation The Davis Besv model wu based on the experience allou .m analyn traomprehensively view the results pir.ed with the RELAP5 made17of the Oconee-1 of an ir,t.ral tmiel and the interactions between dif- PWR, wrJch was developed and u,ed extensively in
'cr ent compomnts and systems. The combined capabil- the prescrized thennal shock progum. The main dii-tics of DEL AP5/ MOD 2 and the NPA approach the ferencer. between the Oconec and Davis-Besse modeis perfm.. we of an engineering simulator, are duc u geometrical and physical differences between the phts and due to improvements to the code and melsg techniques since the development of the 2.2 MOdel Descdption Oconce podel.
A W @5 model of the Davis-Bece Unit i PWR A derded description of the R586 model of D,cus-e in deve:oped at the INEL to perform the R586 liase is presented in Appendix A.
LO PW ce.lculations described in this report. The trae
< iou alculated insluded the LOFW event that oc- ?. 2.2 initial Conditions. Two steady-state initial-
. nd at Davis-Besse or, kne 9,1985, aad a total .zation calculations were performed with the R586 LOFW followed by primary feed and t*ed. Fec- mcx!ct of DavbBesse. The calculation were perform-ti sn 2.2.1 contms a description of the LtPL AP5 ed at 92% and 100% of full pont. The reactor was model. Desenptions 2 the ' initial and boundary ron- operating at 92% powerjusi prior to 't LOFW event didons for the LOFW <ransients are presentsd in that occurred on June 9,1985. The feed and bleed l
' ~
Sectioni L2.2 and L2.3. A description of the qu #ty calcul# ions were initialized from the 100% power assurstre puformed for the model ayean in steady state. The calenlated and desired initial condi-Section 2.2 4 tions for the 92% and 100% power steady states are r .
t hown m Tables ! and 2, respectively. The desired ini-2.21 Thorcoal Hudraulic and Contro;Sptem tial conditions for the 92W power case were obtained Model. A detaihd RELAP5 model of Davi+Desse from plant data taken at 1:34:00 a.m. on June 9,1935.
Unit I was devdop:d during this tam lhe wdel This time w is about 15 s prior to the ICS failure which 1
a & .Q , l" l l z .
~
l, Table 1 Comparison of desired and calculated initial conditions at 92% power Parameter Desired" RELAP5 Core power, MW 2550 2550 Hot leg pressure, MPa (psia) 14.93 (2166.)b 14.96 (2170.)
Hot leg temperature, K ('F) 591.3 (604.7)b 591.4 (604.8)
Cold leg temperature, K (*F) 566.9 (560.8)b $66.7 (560.4)
Pressurizer level, m (in.) 5.037 (198.3) 5.034 (198.2)
Total reactor coolant flow, kg/s (lbm/s) 18396 (40556) 18111 (39929)
Reactor coolant pump speed, rad /s (rpm) -' 125.7 (1200.)
OTSG outlet pressure, MPa (psia) 6.024 (873.8)b 6.043 (876.5)b Total feedwater flow, kg/s (Ibm /s) 1306 (2878) 1382. (3046.)
Feedwater temperature, K ('F) 505.8 (450.7) 503.9 (447.4)
Steam superheat K ('F) 38.7 (69.6 )b 16.5 (29.7)
OTSG mass (each), kg (Ibm) -2 14751 (32520)
OTSG operating level, % 60.95 b 65.28 b OTSG startup level, m (in.) 3.881 (152.8)b 4.288 (168.8)b
- a. Desired value taken from Davis-Besse data acquisition display system at 1:34:00 a.m. June 9,1985.
- b. Parameter represents an average of both loops.
- c. Desired value unknown.
Table 2. Comparison of desired and calculated initial conditions at 100% power Parameter Desired RELAPS Core power, MW 2772 2772 Ilot leg pressure, MPa (psia) 14.96 (2170.) 14.96 (2170.)
liot leg temperature, K ('F) 592.2 (606.3) 592.6 (607.1)
Cold leg temperature, K (*F) 565.2 (557.7) 565.9 (559.0)*
Pressurizer !cvel, m (in.) 5.080 (200.0) 5.075 (199.8)
Total reactor coolant flow, kg/s (Ibm /s) 18067 (39831) 18143 (39998)
Reactor coolant pump speed, rad /s (rpm) 124.1 (1185.) 125.7 (1200.)
OTSG outlet pressure, MPa (psia) 6.38 (925.) 6.348 (920.7)"
Total feedwater flow, kg/s (Ibm /s) 1477 (3257) 1536. (3386.)
Feedwater temperature, K (*F) 509.9 (458.2)" 511.1 (460.4)
Steam superheat, K (*F) 34.7 (62.5)a 12.2 (21.9)'
OTSG mass (each), kg (Ibm) 17740 (38450) 16696 (36809)*
UTSG operating level, W 69* 67.48 OTSG startup level, m (in.) -b 4.473 (176.1)"
- a. Parameter represents an average of toth loops.
- b. Desired value unknown.
4
caused main feed pump number ! (MFP-1) to over- of both MFPs. The AFW pumps were assumed to fail speed and about 55 s prior to the trip of MFP-1. The to deliver flow to the OTSGs. Letdown was isolated desired initial conditions for the 100% power steady at the start of the transient. No makeup flow was state were taken from a variety of sources, principally allowed until feed and bleed cooling was initiated. Core the Updated Safety Analysis Report.110 wever, the decay heat was input as a table representing the 1979 parameters relating to OTSG performance, including American Nuclear Society standard plus actinide feedwa'er mass flow rate, feedwater temperature, and decay. The reactor coolant pumps (RCPs) were steam superheat, were taken from the Davis-Besse full tripped, representing a manual action,1 min after hot pewe acceptance test data. leg subcooling decreased to 1 i K (20'F). The follow-The calculated initial conditions were generally in ing actions weit taken 20 min after the start of the tun-excellent agreement with the desired initial conditions. sient to represent the manual initiation of feed and bleed For example, the initi:d conditions at 92 % power were cooling. The PORV and hot leg high point vent valves generally within the expected uncertainty of the (IIPVVs) were latched open. The pressurizer heaters measurements. Exceptiens are in the calculated steam were tripped off. Maximum makeup flow was obtained temperature and the feedwater flow rate. The calculated by starting both makeup pumps and locking the makeup phasic steam temperature at the outlet of the OTSG valve at its wide-open position. The HPl pumps were was in good agreement with the data. Ilowever, the staned and "piggybacked" to the discharge of the low-code unrealisticaPy calculated that a small amount of pressure injection (LPI) pumps to increase the HPI liquid was entrained into the steam line. The liquid then shutoff head. The shutoff head was 12.7 MPa evaporated in the steam line, reducing the calculated (1835 psia)in tiie piggyback configuration,1.3 MPa superheat. Because the calculateu equilibrium steam (190 psi) higher than in the normal configuration. . . . _ _
superheat was too low, the calculated fcedwater flow The capacities of some of the key systems and valves was about 5% larger than the actual flow. are presented below to document the values used in the calculations. The sources of these values include 2.2.3 Boundary Conditions. For the R586 calcu- infonnation supplied by B&W and Toledo Edison dur-lation of the Davis-Besse LOFW event of June 9,1985, ing the development of the model. The pressurizer all systems were initially in the automatic mode of heaters were modeled to provide a maximum power control except for MFP-2, which was in the manual of 1.329 MW. The pressurizer spray valve was sized mode of control. The transient was initiated by an to pass 0.012 m3/s (190 ppm) at normal operating overspeed of MFP-1. The ICS controlled the plant conditions. The PORV was modeled to pass 25.2 kg/s response prior to reactor trip except for the core power. (55.5 lbm/s) of saturated steam at 16.10 MPa Core power prior to reactor trip was input based on (2335 psia) and 47.5 kg/s (IN.7 lbm/s) of subcooled data. After reactor trip, the core power included liquid at 16.46 MPa (2387 psia) and 613 K (644*F).
contributions from decay heat and fission power. The The resulting PORV area was 9.48 x 10" m2 decay heat was based on the 1979 American Nuclear (0.01020 ft2), w th a single-phase liquid discharge Society standard," assuming infinite operation, and coefficient of 0.82 and a two-phase discharge coeffi- ..
included actinide decay. The fission power after cient of 1.0. Each hot leg HPVV was modeled with reactor trip was obtained from a separate-effects an area of 1.830 x 10-5 m2 (0.000197 ft2 ), with reactor kinetics calculation. After reactor trip, single-phase and two-phase discharge coefficients of feedwater flow to the OTSGs was specified based on 0.624. The AEV on each steam line was sized to data. Similarly, the atmospheric exhaust valves (AEVs) pass 74.1 kg/s (163.3 lbm/s) of steam at 6.2 MPa were used to limit OTSG pressure to the measured (900 psia).
data. The feedwater and pressure boundsry conditions were specified because the operator actions which 2.2.4 Quality Assurance. Several sources of affected these parameters were not well known. Known information were used in the development of the manual operations taken during the transient were RELAP5/ MOD 2 model of Davis-Besse. These sources modeled. Specifically, the pressurLer spray was included detailed drawings and blueprints, system j
initiated prior to reactor trip, and letdown was, isolated descriptions, including the Updated Safety Analysis j
and both makeup pumps were actuated 1 s after reactor Report, plant acceptance test data, equipment test data, trip. control system calibration data, and com ersations with For the R586 feed and bleed calculation. ali sy tems personnel from B&W and Toledo Edison. B&W and were in the automatic mode of control except as Toledo Edison provided nearly all of the information described below. The transient was initiated by the trip ultimately incorporated into the model.
5
i o The RELAP5 model of Davis-Besse was quality the calculated and desired initial conditions st two dif-assured in several ways. First, the development of each ferent power levels lends confidence to the model, model component was documented on worksheets that Finally, a calculation of the LOFW transient that
~
include references to drevcings and documents de- occurred at Davis-Besse on June 9,1985, was per-scribed above. Second, the worksheets were indepen- formed and compared with plant data. This com-deatly checked by an analyst other than the one who parison, which appears in Section 3 of this report, pro-developed them. Third, the goo <.1 agreement between vides additional confidence in the mcxlel.
'r s'
s w.
Y t
6 s .
. . o
- 3. RESULTS Results of the R586 calculation of the Davis-Besse Table 3. Sequence of events for the LOFW transient are presented in Section 3. . The Davis-Besse LOFW transient '
R586, R585, and TRAC calculations of the Davis- calculation Besse LOFW transient are compared in Section 3.2.
The R586 feed and bleed calculation from 100% power Time is described in Section 3.3. The R586, R585, and (s) Event TRAC feed and bleed calculations are compared in Section 3.4. The uncertainty in the feed and bleed 0.0 Calculation starts at 1:34:00 a.m.
calculations is addressed in Section 3.5. Computer run 15 MFP-1 overspeed begins time statistics for the R586 calculations are presented 55 MFP-1 tripped in Appendix B. 71 Pressurizer spray on 75 Reactor tripped 3.1 R586 LOFW Transient 76 MSIvs begin to close i Calculation 76 Letdown isolated; Second makeup pump The R586 calculation of the LOFW transient that started 1 occuaed at Davis-Besse on June 9,1985, was per- 82 MSIVs fully closed formed with the RELAP5/ MOD 2 model described in 86 Pressurizer heaters on - -
Section 2. The calculation of this transient is sum- 89 Spray off marized in Table 3, which presents a sequence of key events. Figures comparing the response of the calcula- 380 Feedwater terminated tion and the plant follow. The plaat data shown in the 460 Pressurizer heaters off figures were based on computer printouts of the Davis- 480 Spray on Besse data acquisi* ion display system (DADS) as 650 OTSGs dry out digitized by LANL. Time zero in the calculation and 755 PORV opened the figures corresponds to 1:34:00 a.m. Other sources 765 Calculation terminated of data include the plant process computer alarm print-oms. It is believed that the chick times between the DADS data and the alarm printouts varied slightly, with the alarms occurring at a wall clock time 6 s later than the corresponding ever.ts were indicated by the reactor trip by closing the main feedwater control DADS. valves, sharply reducing the feedwater flow. Figures 1 The plant and calculation were steady for 15 s (until through 4 show that the RELAP5/ MOD 2 model of the 1:34:15 a.m.), when MFP-1 began to overspeed. The ICS provided an excellent representation of the exact cause of the overspeed and the subsequent measured response during the early portion of the response of MFP-1 are not known, although MFP-1 transient.
tripped at 55 s. The MFP-1 overspeed was modeled The main steam isolation valves (MSIVs) inad-by linearly increasing MFP-1 speed from its initial vertently closed shortly after the reactor trip. Thus, value to its overspeed trip setpoint between 15 and the only source of steam available to drive the MFP-2
$5 s. The effects of the overspeed on measured and turbine was the steam stored in the steam line and con-calculated feedwater flow are shown in Figures I necting piping. Because the MFP turbines were not and 2. The overspeed of MFP-1 caused an increase modeled, the turbine response during periods of I in feed flow to both OTSGs. The ICS responded to degraded steam flow was not known, and the operator the increase in feed flow by partially closing the main actions taken to regulate MFP-2 speed were not well feedwater control valves (see Figures 3 and 4), thus characterized, the feedwater flow after reactor trip was reducing the flow. The MFP-1 turbine tripped on not calculated but was input based on data. The feed-overspeed at $5 s, causing an immediate reduction in water flows used in the calculation were based on the the flow to both OTSGs. The ICS responded by open- measurements shown in Figures 1 and 2. Zero flow ing the main feedwater control valves, which caused was provided to both OTSGs after 380 s in the calcula-a slight increase in feedwater flow. The reactor tripped tion, even though the flow measurements did not read on high RCS pressure at 75 s in the calculation,9 s zero. The flow was known to be zero after 430 s, earlier than in the plant. The ICS responded to the indicating a bias in the measurements. An operator 7
1000 , , , , , , , i DATA
..... R586 2000 800 - -
' ' .' MFP-1 TRIP ,.-
1500 Q ,
[ 600 - -
v e
$ .$ 3 c: h 1000 S 400 - REACTOR TRIP _
m E )i E 2 3 i
200 -
! -500 0
'. A O
O 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 1. Feedwater flow to OTSG A in the R586 calculation of the Davis-Besse LOFW transient.
l 1000 , , , , , , , i DATA 2000
..... R586 800 - -
. MFP-1 TRIP m Q ,
1500 (
l
.E 600 - -
I v
$ 1
- 1000 S REACTOR TRIP 400 .
E s [! E 3
200 - -500 0 0 0 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 2. Feedwater flow to OTSG B in the R586 calculation of the Davis-Besse LOFW transient.
8
100 . i i e i i i i DATA
....- R586 80 --
80 n
R v
6 60 -- k REAC M MIP -
60 o o -
e o I u
' f l O
O e
$ 40 40 2 o
s !.i >
- i' 20 20 ,
WP-1 TRIP 0 O O 10 0 200 300 400 500 600 700 800 900 Time (s) .
Figure 3. OTSG A main feedwater control valve area in the R586 calculat. ion of the Davis-Besse LOFW transient.
10 0 , i i i i i i i CATA
..... R586
. RCACTOR TRIP 80 -- :j -
80 pl 9h ;
R v
}e 60 --- !j -
60 o
- e
.. L
!3 -
ii O v 'l
- 2 40 -
40 2 o !. O
> i
, l 20 ,j _ ' ~
20 WP-1 TRIP i
0 .
O O 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 4. OTSG b main feedwater control valve area in the R586 calculation of the Davis-Besse LOFW transient.
9
inadvertently actuated SFRCS on low steam pressure The calculated and measured pressures of the OTSGs at 430 s. SFRCS, responding as if a steam line break are illustrated in Figures 7 and 8. A slight pressuriza-had occurred, isolated both OTSGs, terminating feed- tion was calculated and observed after the overspeed water flow. The exact time that feedwater flow de- of MFP-1, followed by a slight depressurization after creased to zero in the plant is not known because of the MFP-1 trip at 55 s. The reactor trip and turbine the possibility that there was insutlicient steam to drive trip at 75 s in the calculation resulted in the immediate the turbine. Flowever, 380 s appears a reasonable closure of the turbine stop valves. Within I s, the estimate basc<1 on the measured flow response and the MSI Vs also began to close. The closure of the turbine known bias. stop valves and the MSIVs resulted in the rapid Calculated and measured startup liquid levels are pressurization of the OTSGs and the opening of the shown in Figures 5 and 6 for the A and B OTSGs, AEVs and the safety relief valves (SRVs). With the respectively. The calculated and measured levels were reactor tripped and the condenser unavailable, the ICS generally in excellent agreement. Both calculated and would normally try to control OTSG pressure at 7.1 measured levels increased slightly after 15 s and then MPa (1030 psia). Ilowever, the measured pressure decreased rapidly after 55 s due to the overspeed and varied significantly from the setpoint, either because -
trip of MFp-l. The levels continued to decrease after of operator actions which are not well quantified or the reactor trip until a nearly constant value was anomalous system behavior. Consequently, the OTSG reached about 120 s. This constant corresponded to the pressure was controlled as a boundary condition after normal posttrip level of 0.89 m (35 in.). The calculated reactor trip in the calculation. The AEVs were opened levels were generally too high after 300 s, indicating if the calculated pressure exceeded the measured that too much feed flow was assumed. The levels began pressure and closed otherwise. The calculated to decrease at 380 s u?n the feedwater flow was pressures diverged from the data near the time of terminated. The OTSGs ure completely dry at about OTSG dryout, in the calculation, the pressure increased 650 s in both the calculation and the transient. Note slowly due to heat transfer to superheated steam. The that although the OTSGs were dry, the calculated and more rapid pressure increases observed in the data in-measured liquid levels did not reach zero because of dicated that a small amount of liquid was left in the the weight of the steam between the differential OTSGs. This liquid then boiled, pressurizing the pressure taps used in the startup level measurement. OTSGs.
5 i i i , , , , i DATA
,s MFP-1 TRIP
.._, R586 4 -
150' n a E .5 E3 e
- b
.$ !. 10 0 $
."2 U
I, rtED STOPPtD g
a l a 50
~
{'
Y..
d
....w
"'.....%.. oisc DRY 0 0 0 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 5. Startup hquid level in OTSG A in the R$86 calculation of the Davis.Besse LOFW transient.
10
5 i i i i i i i DATA
., WP-1 TRIP ....-
R586 4 : -
15 0 i ^
_ 3 - -
e e e 100 o
'O y 2 - - -
3
',. FEED STOPPED s, .i
- 50 1 -
- r w-
. r 'N.,**. OTSG DRY 0 O O 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 6. Startup liquid level in OTSG B in the R586 calculation of the Davis-Besse LOFW transient.
8 i i i i i i e i DATA
R566 110 0
^7 - \ -
g 1 .-
1000 9.;;
0
- i *...... - v u 4 t.,. L 3
m
- o
, m m ,.; : 900 m O OTSO DRY U
,/ ,'
a6 :
a Ii _ _ HEActoR TRIP i
800 5
O 10 0 200 300 400 500 600 700 800 900 Tirne (s)
Figure 7. Pressure in OTSG A in the R566 calculation of the Davis-Besse LOFW transient.
11
8 i , , , , , , ,
110 0 i
i, t ., -- ..... ,
T7 -
'l ..
9 g ;4, eh ,,,,, , , , , , ,
(...- 1000 ;;
v 'y
. ,,,cb
- . v h NYN Ii E :
900 $
c5 6 -
l, DATA _
g
~--- RS66 REACTOR TRIP
%}
800 5
O 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 8. Pressure in OTSG B in the R586 calculation of the Davis-Besse LOFW transient.
Calculated and measured core powers are shown in coolant pressure reached the high pressure setpoint of Figure 9. The measured core power was input as a 15.9 MPa (2300 psis) and the reactor tripped in both boundary condition until the time of reactor trip. It was the calculation and transient. Shortly after reactor trip, necessary to represent core power as a boundary con- the OTSGs removed more heat from the RCS than was dition because a reactor kinetics model of Davis-Besse produced in the core. Consequently, the RCS hac not yet been developed. Consequently, the model temperature and pressure decreased rapidly, along with could not represent mechanistically the effects of the the pressurizer level. The operators acted to maintain MFP 1 overspeed on core power through the influence pressurizer level by isolating letdown and starting the of control rod movements or moderator temperature second makeup pump. As the RCS pressure fell, feedback. The reactor tripped at 75 s in the calcula- pressurizer spray was automatically terminated and the tion on high RCS pressure, slightly before the observed heaters were actuated. These operator and automatic trip at 84 s. The power then decreased rapidly. The actions stabilized RCS pressure and level, but the RCS measurement was lower than the calculation after 100 s temperature continued to decrease. By 300 s, the because the incasured power was based on the detec- pressure, temperature, and level were all lower in the tion of fissions and did not include decay heat. The calculation than in the transient. A small reduction in I measured fission power was signiGeant compared to the feedwater flow between 200 and 300 s would im-decay heat for approximately 100 s after the reactor prove the calculation of these parameters. The termina-trip. The posttrip fission power was associated with tion of feedwater flow at 380 s in the calculation the emission of delayed neutrons. resulted in an increase in the RCS temperature. The Calculated and measured reactor coolant pressures temperatures then increased for the remainder of the are shown in Figure 10. The MFP-1 overspeed at 15 s calculation. The rate of temperature increase was caused a slight depressurization of the RCS. The similar to the data. The calculated and measured rates l depressuritation was caused by the increase in feed- of increase in pressurizer level were also similar.
water flow, which cooled the RCS (see Figures 11 Pressuri7er spray came on near 480 s, briefly halting and 12) and reduced the pressurizer liquid level the pressurization in both the calculation and data.
(Figure 13). The trip of MFP-1 at 55 s reduced the However, the pressure then increased until limited by feedwater llow, causing a rapid increase in reactor the PORV. The PORV first opened at 755 s in the t.oolant temperature, pressurizer level, and reactor calculation and at 880 s in the transient. It was coolant nressure. Pressurizer spray was initiated but discovered through sensitivity calculations that if the was unable to stop the pressure increase. The reactor OTSGs were dry, spray initiation could reduce the rate l
12
100 i i i e i i i i_
DATA
~~~~~
REACTOR TRIP 80 -- -
80 R
v 60 - -
60 R v
u v v E 40 -- ~~40 S 20 -- -
20 t,
Q----..r------r-------r-------r-------r---------r------ '
O O 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 9. Reactor power in the R$86 calculation of the Davis-Besse LOFW transient.
18 i i i i i i i i 2600 REACTOR TRIP 2400
.- n 9 16 -
! SPRAY ON
.' ~
,o I:f S E v .
N .-- . . . . . ..
v u ..
j ,'
,. 2200 ,
u a a: *
.' o y '
, PORV OPEN y e .. v u . .- u o 14 - ', '
n.
2000 DATA
..... R586 l
l 1800 12 O 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 10. Reactor coolant pressure in the R586 calculation of the Davis-Besse LOFW transient.
13
600 , i i i i i i i 620 DATA
..... R586 REACTOR TRIP 590
_600 n s n x
v U p.
- v e 580 -
i -
e a
Is
". ', 580 s -
O u
o u
e .
e c 570 -
a.
g .
[ ,,,,.. 560 b 560 - -
540 550 0 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 11. Coolant teinperatures in the B loop hot leg in the R$86 calculation of the Davis-Besse LOFW transient.
i 590 ' ' ' ' ' ' ' '
600 DATA
..... R586 580 - -
^
5 J, 580 p
- w e I e u , .. u o a
\
Eu 570 -
/ *
%u E
g ..; .....' 560 &g n
a ., . , .. %
............"' n e
560 -
540 550 O 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 12. Coolant temperatures in the B loop cold leg in the R586 calculation of the Davis Besse LOFW transient.
14
10 i , , i , , i i DATA
-.... R586 8 - -
300
.N ^
m E ntActon inie .5 56 -
w , ,,,,.- 200 o
.'?
4 _ ',- _ 2
& , g i
- .~,
, 10 0 2 -
O O i 0 10 0 200 300 400 500 600 700 800 900 t-Time (s)
Figure 13. Pressurizer liquid levels in the R586 calculation of the Davis-Besse LOFW transient.
of pressurization but could not depressurize the RCS, the LOFW transient that occurred at Davis-Besse on as shown in Figure 10. June 9,1985, are compared in this section. The R$86 Calculated and measured hot leg mass flows were calculation was discussed in the pievious section and in excellent agreement, as illustrated in Figure 14. represented the transient from 1:34:00 a.m. The With the RCPs on, the variations in mass 110w were TRAC and R585 calculations started 50 to 55 s later, caused by the changes in coolant density associated at about the time MFP-1 tripped. Different thermal-with the variations in RCS temperature. hydraulic computer codes and input models were used A review of Figures I through 14 indicates that the in the three calculations. The R585 and R586 calcula-calcula>.ed trends were ir. excellent apeement with the tions were performed with the same computer code, data. The magnitudes of the calculated values were also RELAPS/ MOD 2, but wi9 different input models.
generally in good agreement with the data. For exam- The TRAC calculation was actually performed with plc, the maximum deviation between calculated and TRAC-PFl/ MODI .
measured RCS pressure was about 0.3 MPa (50 psi). Calculated and measured RCS pressures are com-The deviation between calculated and Lacasured pared in Figure 15. The R586 and R585 calculations temperatures was generally less than 3 K (6*F), not were generally similar, although two differences were significantly larger than the estimated uncertainty in noted. First, the pressure decreased further after reac-the data. The calculated results were sensitive to the tor trip in the R585 calculation than in the R586 feedwater flow, it would be possible to improve most calculation. This difference was believed to be due to of the calculated results by adjusting the feedwater the core power boundary condition and will be dis-flow. In particular. e 30% reduction in the feedwater cussed later. Second, the RCS pressure did not increase flow after 200 s would improve the calculated to the PORV setpoint in the R585 calculation, but in-resp (mse.110 wever, this improvement would probably stead was limited by the operation of pressurizer spray, not be meaningful, considering the uncertainty in the which was significantly larger than in the R$86 calcula-feed flow and the other data. tion. The spray valve passed 0.012 m3 /s (190 gpm) in the R$86 calculation and about 0.032 m3 /s (500 gpm) 3.2 Comparison of in the R585 calculation. The lower value is appropriate Davis-Besse LOFW for Davis-Besse. The TRAC calculation reached a higher peak pressure prior to reactor trip, then depres-Trans.ient Calculat. ions surized more slowly than the R586 calculation and the The three calculations (R$85, R$86, and TRAC) for data. According to Reference 2, the higher peak 15
10000 , , , , , , , , 22000 DATA
..... p5g6 21500 k 9500'-
ei 21000 kE 5 e n
20500 y
c .- o 8 20000 $ '\
j 9000 -l ..
A g
( 2
{ 19500
)
19000 8500 =
O 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 14. Mass flow in the A loop hot leg in the R586 calculation of the Davis Besse LOFW transient.
18 i , , , , , , ,
-2500 16 -
m i M m e .e 34 1 -
2000 $
, - ~ ,
SPRAY ON
~ ~
0 DATA $
0 o R586 cE a R585 x TRAC -1500 10 - -
8 O 10 0 200 300 400 500 COO 700 800 900 Time (s)
Figure 15. A comparison of calculated reactor coolant pret;sures during the Davis-Besse LOFW transient.
16
pressure was caused by the specification of reactor trip temperature decrease after reactor trip were caused by on time rather than RCS pressure, which resulted in variations in the OTSG liquid levels, as shown in i the addition of too much core power to the RCS. Also, . Figure 18. The level in the TRAC calculation was the TRAC pressure increased more rapidly than the significantly lower after reactor trip than in the other data or other calculations after the OTSGs dried out calculations and the data. Part of this discrepancy is near 500 s. The pressurizer spray was inhibited in the due to the fact that the parameters shown are not direct- l TRAC calculation, and thus the slope of the pressure ly comparable. The TRAC curve represents collapsed curve did not change when the open setpoint of the liquid level, while the R586 and R585 curves mimic spray valve was reached. the output of the plant instrumentation by including the Calculated and measured pressurizer liquid levels are head of steam between the differential pressure taps compared in Figure 16. The R586 and R585 calcula- as part of the indicated level. However, a comparison tions were similar except that the level decreased more of collapsed liquid levels in the R586 and TRAC rapidly after reactor trip in the R585 calculation. A calculations at 300 s revealed that the level was similar difference was observed in the RCS pressure significantly lower in the TRAC calculation. The lower i
comparison discussed previous ly. The initial level was OTSG level contributed to slower rates of decrease in significantly too high in the TRAC calculation, but the RCS pressure, temperature, and pressurizer level, trends were similar to those discussed for the RCS The results shown in Figure 17 showed that the pressure. The rates oflevel increase after the OTSGs calculated results are sensitive to OTSG level, which dried out were similar in all three calculations and the in turn depends on feedwater flow. However, the varia-data, tions in OTSG level between the R586 and TRAC The calculated RCS thermal responses are compared calculations were surprising, considering that the ini-in Figure 17, which shows B loop hot leg tempera- tial OTSG liquid inventories were nearly identical in tures The R586 and R585 calculations were again both calculations and the input feedwater flows were quite similar except that the temperature decreased based on the measured flow rates. The discrepancy in more rapidly after reactor trip in the R585 calculation. OTSG levels may have been a result of the excess core The TRAC calculation decreased slower after reactor power in the TRAC calculation near the time of reac- ;
tor trip. The OTSG level in the R585 calculation was ;
trip. The rates of temperature increase after OTSG dryout were similar. The deviations in the rates of generally lower than in the R586 calculation. This l 10 i i i i i , i i i
8 -
-300
' o n
- 6 i
s %
2 f
./ '200 2 3 4 -
2 o DATA .
D R586 A R585 -10 0 2 x TRAC 0 O O 10 0 200 300 400 500 600 700 800 900 Time (s)
Figure 16. A comparison of calculated pressurizer liquid lesels during the Davis-Besse LOFW transient.
17
i 600- , ,. , , , , , , -620 O DATA - i 4 0' R586 590j. ! EACTOR TRIP .o' R585 X TRAC --600 2
v l, @
e 580' - -
.. 4 5 -
I O .
t580 .sg .
y, s. ~
g570 i $ - -
-560 y 560 - -
)
-540 550 i 0 10 0 200 300 '400 500 600 700 800 900 Time (s) 4 1 ]
- Figure 17. A comparison of calculated coolant temperatures in the B loop hot leg during the Davis-Besse LOFW transient.
I' I
5 ,
REACTOR TRIP O DATA t i / 0 R586 a R585 4 ~
s "
x TRAC -15 0 [
m oc E i v ;,
-3 -
s:
c -100 V-2 - c - .V - .
3
.?r
! ,y .
_a 1
-50 1
u- -
q
,-x -~
E .,
u 5 ~ &'Q Q:. a :. , ,
0 ' ' ' ' '
O ,
0 10 0 200 300 400 500 600 700 800 900 l Time (s)
Figure 18. A comparison of calculated startup liquid levels in OTSG B during the Davis-Besse LOFW transient.
l
'18
indicates that the .R586 calculation had a larger feed Table 4. Sequence of events for the flow, yet the RCS temperatures decreased more rapidly feed and bleed calculation after reactor trip in the R585 calculation. This apparent discrepancy was attributed to the core power boundary Time Lcondition. The power table in the R585 calculation (s) Event
~
went immediately to decay heat after reactor trip. The R586 curve included the effects of the fission power 0.0 Both MFPs trip; turbine trip; reactor following reactor trip. The integrated fission power in trip the R586 calculation was equivalent to 1.5 s of full 0.60 Turbine stop valves fully closed reactor power. The additional power in the_ R586 1.13 Pressurizer spray valve opened calculation increased the RCS temperature by about 2.03 TBVs and AEVs opened 3 K (6'F), equivalent to the difference between the 2.33 OTSG SRVs opened R$86 and R585 curves at 200 s. Different feedwater flows were then used to partially compensate for the 4.58 Pressurizer heaters on different core powers so that the long-term RCS 6.00 Pressurizer spray valve closed responses were similar. 28.5 OTSG SRVs closed Even though different codes and models were used, 130 OTSGs dry out the results of all three calculations were similar 202 Pressurizer heaters off and showed trends like those observed during the transient. The major differences between calculations 217 Pressurizer spray valve opened were primarily due to the assumption of different core 340 PORV first opened powers, feedwater flows, and pressurizer spray 750 Pressurizer liquid solid flows. 1012 Reactor coolant pumps tripped 1103 Reactor vessel vent valves opened t- 3.3 R586 Feed and Bleed ,
1200 Feed and bleed initiated Calculation 1955 Pressurizer SRVs opened
]
The R586 feed and bleed calculation was performed 3090 Last closure of the pressurizer SRVs with the RELAP5/ MOD 2 model described in Sec- 3752 Reactor coolant depressurization began tion 2. The calculation is summarized in Table 4, 4707 Pressurizer spray valve closed which presents a sequence of the key events occurring q' in the transient. The transient was initiated from 100% 5162 Refill of reactor coolant system began rated power by tripping both MFPs at 0.0 s. Tripping $500 HPI flow initiated both MFPs immediately caused a turbine trip and an 5800 Calculation terminated anticipatory reactor trip.
The response of the RCS carly in the transient was controlled by the secondary system. OTSG pressures, shown in Figure 19, increased following the clorure because of the weight of the steam between the dif-of the turbine stop valves at 0.6 s. The turbine bypass ferential pressure taps.
valves (TBVs). AEVs, and SRVs then opened to limit Figure 21 shows calculated pressure in the hot leg the pressurization of the OTSG. The SRVs closed at of the A loop. The turbine trip at 0 s and the closure 28.5 s. The ICS then operated the TBVs and AEVs of the turbine stop valves caused a momentary heatup to control the pressure at 7.24 MPa (1050 psia) for the and pressurization of the reactor coolant. Pressurizer remt.inder of the transient. The OTSG startup levels, spray was initiated at 1.13 s, limiting the peak pressure shown in Figure 20, represented a wide range level to 15.5 MPa (2250 psia). The OTSGs then removed measurement. The MFPs coasted down following the more heat from the reacter coolant than was being pro-MFP trip at 0 s, providing only 0.9 s worth of full flow duced by decay heat in the reactor core, causing a before check valves in the feed lines closed. Because decrease in pressure and hot leg temperature (see the AFW system was assumed to fail, no source of Figure 22). Once the reactor coolant temperature water was available for the OTSGs. Heat transferred stabilized based on the OTSG temperature correspond-from the RCS boiled away the liquid inventory of the ing to 7.24 MPa (1050 psia), the pressure of the reac-OTSGs causing the startup levels to decrease. Both tor coolant reached its minimum value' and then OTSGs were dry by 130 s. Even when the OTSGs increased slowly because of the pressurizer heaters.
were dry, the indicated levels, which were based on Core decay heat caused the reactor coolant temperature calculated differential pressure, did not go to zero and pressure to increase rapidly after the OTSGs dried 19
i l.
i l
l 8 i ' ' '
I
_115 0 A OTSG
-- B OTSG ;
-110 0 I 7.5 - -
?
c.
9
-1050 E
~G o
E 7 -
l 3 -
-1000 a !
E E w u
- o. o, d 6.5 '- TUR8lNE TRIP -
3
--- \
-900 l
6 {
-50 0 50 10 0 150 200 Time (s)
Figure 19. OTSG pressures in the R586 feed and bleed calcul: tion.
i 200 , i i i 200 A OTSG
-- B OTSG 15 0 - - --15 0 .
m ^ .i v
E .5 v
- T>
d
. 100 - - --10 0 e
.E %
- g. 5 a .?
s a
%, OTSG DRYOUT
~-.
0 O !
-50 0 50 10 0 15 0 200 {
Time (s) 1 Figure 20. OTSG startup liquid levels in the R586 feed and bleed calculation. j 20 I
l l
18 - i , , i i i --2600 FLED ANO DLEED
-2400 16 -
n ^
f
. -% SPRAY ON -2200 .
5 a o - e 14 .
t -
-2000 g u
oise onYouT h u
-1800 '
12 -
-1600 HPl ON
' ' ' ' ' ' I 10
-1000 0 1000 2000 3000 4000 5000 6000 Time (s) i Figure 21. Reactor coolant pressure in the R586 feed and bleed calculation.
640 i i i i i i A HOT LEC FEED AND DLEED --
0 HOT LEC 620 -
\ -
-s -
-650 '
x V- 1 E E a a E 600 t
t t
- a. -
a.
E -
-600 E O N 580 - -
s oisc DRYOUT
]
560 - t i ' ' ' ' -550
-1000 0 1000 2000 3000 4000 5000 6000 4 Time (s) )
i Figure 22. Ilot leg coolan temperatures in the R586 feed and bleed calculation.
I l
21 l
_ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 1
out at 130 s. The increase in reactor coolant pressure The reactor coolant pressure (recall Figure 21) caused the heaters to shut off at 202 s and the spray rapidly decreased about 0.7 MPa (100 psi) after the valve to open at 217 s. The initiation of spray reduced PORV was locked open. However, the decrease in the pressurization rate, but the PORV open setpoint pressure eliminated the subcooling in the hot legs and pressure of 16.65 MPa (2415 psia) was reached at allowed saturated boiling in the core. The PORV, 340 s. The POFV then opened and closed repeatedly, which was passing subcooled liquid, was not able to maintaining the pressure between its open and close relieve the volumetric expansion due to the boiling in setpoints. The collapsed liquid level in the pressurizer, the cere; and the reactor coolant repressurized. The shown in Figure 23, increased after the OTSGs dried pressure increased to 17.34 MPa (2515 psia), the set-out because of the heatup and expansion of the reactor point of the pressurizer SRVs, at 1955 s, and the SRVs coolant. The liquid level reached the top of the opened. The open SRVs were able to depressurize the pressurizer at 750 s, and the RCS was liquid solid. The system until the pressure dropped low enough to allow heatup of the reactor coolant continued, and at 952 s the valves to rescat. The SRVs opened and closed the hot leg subcooling (see Figure 24) decreased to four times between 1955 and 3090 s, as shown in 11 K (20'F). The RCPs were tripped, simulating an Figure 25.
operator action,60 s later. The mass balance of the RCS is summarized in Feed and bleed cooling was initiated at 1200 s. Figure 26, which presents PORV, HPVV, makeup, The state of the plant at the initiation of feed and and HPl mass flow rates. The figure reveals that the bleed cooling was as follows: the OTSGs were combined flow through both hot leg HPVVs was in-dry; the RCS was liquid solid; the reactor coolant significant compared to the flow through the PORV.
pressure was near the PORV open setpoint; the hot The PORV flow was also several times larger than the legs were slightly [about 2 K (4*F)] subcooled; and makeup flow at the initiation of feed and bleed. The the RCPs were tripped. Feed and bleed cooling mass loss caused voids to form in the two-phase RCS, war initiated by locking open the PORV and resulting in mixture levels in the vertical components.
HPVVs, starting both makeup pumps, starting the HPI Figure 27 presents collapsed liquid levels in the hot l pumps and aligning them to take suction from the legs. The collapsed liquid level in the A loop hot leg discharge of the LPI pumps, and tripping the (tropped to the elevation of the pressurizer surge line pressurizer heaters. at 2560 s. Thereafter, more steam passed through the 14 i i i , i i m -500 12 - -
v E 10 -- L'ouio m L .-400 C, 7 I> ~
u f 8 ~ -
S
, _300 I F o
- g. '5 36 - -
.}
-200 4 -
p oisc oRYOUT i
-10 0 g i ! I i
-1000 0 1000 2000 3000 4000 5000 6000 Time (s)
Figure 23. Pressurizer hquid level in the R586 feed and bleed calculation.
22
50' i i i e i .
A HOT LEG -80
-- 0 HOT LEG ,
40 - -
3 a
2 -
-60 R
" 30 -
E 'E
=
=
g - - -40 g o 20 - -
o Srn Sm .
j 10 ~- _-20 ,}
.}
C C 0-- W- --O I
-10
-1000 0 1000 2000 3000 4000 5000 6000 Time (s) i Figure 24. Hot leg subcooling in the R586 feed and bleed calculation. l 200 i i i e i i
- -400 !
150 - !
m
?
N
-300 {
E
( i 5
- 10 0 -
-200 y s
E E U O 2 s 50 _- --10 0 l
l 0 O l
-1000 0 1000 2000 3000 4000 5000 6000 )
Time (s)
Figure 25. Mas Dow through the pressurizer SRVs in the R586 feed and bleed calculation.
23
80 , , , , , , l o PORV
~
FCED AFC DLEED o MAKEUP
-15 0 f a HPvv x HPl 60 - -
7 h E 6 -
e -10 0 $,
=
$ 40 -
l y i j i C~ f[ i M
-50 :s 20 - _
C 0 n-k e m - . . -' " 4 S ^ ^' --
A O
-1000 0 1000 2000 3000 4000 5000 6000 Time (c)
Figure 26. A summary of mass flows into and out of the RCS in the R586 feed and bleed calculation.
30 i , , , , ,
o A HOT LEG A 0 HOT LEC SURGE LiNE ELEVATlON
-80 o-eoneee g E 20 -
th -
2 5e -
-60 T
'3
-40 ;
.?' 10 - _
.?
a
-20
% o A 0 ' ^
0
-1000 0 1000 2000 3000 4000 5000 6000 Time (s) 1 Figure 27. Hot leg collapsed liquid levels in the R586 feed and bleed calculation.
l 24
I l
surge line and out the PORV. The increase in steam between the bottom of the lower reactor head, through j at the PORV is illustrated in Figure 28, which presents the core a.nd the upper plenum, to the top of the upper j void fraction in the hot leg volume connected to the reactor head. The upper head began draining about j surge line and in the top volume of the pressurizer 200 s after the initiation of feed and bleed, resulting !
which was connected to the PORV. The void fraction in a decreasing liquid level. The liquid level decreased )
at the connection to the surge line and the PORV in- rapidly until it dropped below the hot leg nozzles. The l creased significantly after the liquid level dropped to level then decreased slowly until makeup began refill-the elevation of the surge line. As expected, there was ing the RCS. The level increased rapidly after HPI was a strong correlation between the void fraction in the initiated. The minimum collapsed liquid level was hot leg and the void fraction at the PORV. about 0.3 m (1 ft) above the top of the core. In fact, The increase in void fraction at the PORV resulted the calculated mixture (froth) level never dropped in an imreased volumetric flow out the PORV. By below the hot legs nozzles. The core was covered with 3752 s, the volumetric flow out the PORV was large liquid or a two-phase mixture throughout the transient.
enough to depressurize the RCS, as shown in Consequently, no core heatup was calculated and the Figure 21. The reactor coolant pressure decreased fuel rod cladding temperatures stayed within a few for the remainder of the calculatiori. The increase in degrees of the fluid temperature.
void fraction and the decrease in pressure caused The calculation was terminated at 5800 s. Hand the flow out the PORV to decrease (see Figure 26). calculations indicated that a quasi-steady state would The decreasing pressure also resulted in an mercasing be achieved at a pressure near 8.6 MPa (1250 psia) makeup flow. The makeup flow exceeded the with subcooled liquid exiting the PORV. At this pres-combined flow out the PORV and the HPVVs after sure, a mass and energy balance could be achieved. ,
$162 s, beginning a gradual refill of the RCS. The The flow out the PORV and HPVVs would balance pressure dropped below the HPI shutoff head of makeup and HPI. The core power would heat the in-12.65 MPa (1835 psia) at 5500 s. The addition of HPI jected water, which would then flow out the open significantly increased the flow into and the refill rate valves. After this quasi-steady state was obtained, the of the RCS. pressure would drop slowly as the decay heat de- .
creased. A source of feedwater would be required to !
The liquid inventory in the reactor vessel is shown in Figure 29. The plot represents the collapsed level ultimately bring the plant to cold shutdown. j 1.5 i i i i i i 1.5 o HOT LEG a PORV 1 - : 1.0 g g a-a-4.or 3-s s E E 3 ,
l 2 O S O
> 0.5 - -- 0.5 >
n 0 ' ' '
0 cbooes O.0
-1000 0 1000 2000 3000 4000 5000 6000 Time (s)
Figure 28. Hot leg and PORV void fractions in the R586 feed and bleed calculation.
l I 25 i
12 i , , , i i VESSEL
-- - TOP OF CORE
-- NO22LE CENTERLINE -35 10 - -
S -
-30 $
e v f 8 - -
S y -___________________ .. __ ______________ ____ _ -25 y
'5 ~5 h ~ ,.
[
6 ~ " " " " " " " " * " " " " " " * * " * * * * * * * * ~ ~ ~ * ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~20
-15 4
-1000 0 1000 2000 3000 4000 5000 6000 Time (s)
Figure 29. Collapsed liquid level in the reactor vessel during the R586 feed and bleed calculation.
3.4 Comparison of Feed and were identified between the R585 and R586 base calculations. First, an earlier dryout of the O'lSGs was Bleed Calculat. ions obtained in the R586 calculation,130 s versus 220 s Feed and bleed calculations have been performed in the R585 calculation. The different dryout times for Davis-Besse at two different power levels. Sec- were primarily caused by differences in the amount of tion 3.4.1 contains a comparison of the R$85 and R$86 feedwater delivered to the OTSGs following MFP trip.
feed and bleed calculations for a LOFW transient ini- In the R586 calculation, an amount of feedwater cor-tiated from 100% power. A comparison of the R585 responding to0.9 sof steady-state flow was delivered and TRAC feed und bleed calculations for a LOFW to the OTSGs as the MFPs coasted down. In the R585 transient initiated from 90% power are compared in calculation,2.3 s of steady-state flow were delivered.
Section 3.4.2. Since the R586 model represented the Davis-Besse feedwater system and not the Oconce sy stem, the R586 3.4.1 100% Power. The R586 feed and bleed ca!- calculation better represented the OTSG dryout time culation described in Section 3.3 was a repeat of one for Davis-Besse.
of the R585 calculations performed shortly after the Second, the reactor coolant pressure increased more June 9,1985, LOFW event at Davis-Besse. The deter- rapidly in the R586 calculation than in the R585
, mination of the uncertainty in the R585 calculations calculation because of the earlier OTSG dryout time l was the major purpose of this study. The following and the smaller pressurizer spray flow rate & scribed comparison of the R586 and R$85 calculations was in Section 3.2. Consequently, the PORV open setpoint used in the determination of the uncertainty in the pressure was reached at 340 s in the R586 calculation calculations. The major difference between calculations versus 620 s in the R585 calculation.
was that the R586 calculation was performed with a Third, the pressure decreased less before the onset quality-assured model of Davis-Besse, while the R585 of boiling in the core after the initiation of feed and calculation was performed with a model based on bleed in the R586 calculation. This was an indication Oconce Unit I that was quickly modified by INEL to that the hot leg temperature was higher in the R586 resemble Davis-Besse. calculation. The higher hot leg temperature was caused Figure 30 presents a comparison of hot leg pressures by the earlier dryout of the OTSG and a slightly higher from the R585 calculation, the R586 base calculation core decay power in the R586 calculation. The core l (described in Section 3.3), and a sensitivity calcula- power in the R586 calculation was augmented to ac-tion w hich will be discussed later. Several differences count for actinide decay.
l 26
18 - i i i i i -2600 FEED AND DLEED
-2400 16 -
m o m y 9 -2200 .y o
O y --2000
{ 14 h 0 E ,
- o !
~ 1800 12 -
O R586 BASE CALCULANON -
O R586 REDUCED POWER o R585 l
-1600 10 O 1000 2000 3000 4000 5000 6000 Time (s)
Figure 30. A comparison of calculated reactor coolant pressures during feed and bleed. I Finally, a major difference in the calculated pres- dryout discussed previously. The pressurizer level sures was observed following the initiation of feed and dropped more rapidly in the R586 base calculation after bleed cooling. The pressure increased following the feed and bleed was initiated because the higher core onset of boiling in the core in the R586 calculation, power generated more steam which flowed to the while the pressure decreased in the R585 calculation. pressurizer. The levels responded similarly in the R585 The deviation between calculations was caused by the and sensitivity calculations.
core power boundary condition. An error was dis- The reactor coolant mass balance is summarized in covered in the R585 model which reduced the core Figures 33 and 34. Figure 33, which shows the com-power 25% after the RCPs tripped. The model, as bined PORV and HPVV flows, reveals that the total developed by INEL, represented the ICS runback of flow out of the RCS was similar in all three calcula- l reactor power following the trip of both RCPs in one tions. The total flow into the system, shown in ;
loop. This runback is applicable only prior to reactor Figure 34, was determined by the reactor coolant trip but was inadvertently applied to the core decay pressure. The R585 calculation had the lowest pressure J power after reactor trip. and thus the highest makeup flow after the inhiation A sensitivity calculation was performed in which the of feed and bleed.
R586 power was reduced 25 % at the initiation of feed A comparison of Figures 33 and 34 shows that the and bleed, as shown in Figure 31. The effect of the RCS contained more mass in the R585 calculation. The reduced power on the reactor coolant pressure is also higher system mass in the R585 calculation resulted shown in Figure 30. The R586 sensitivity calculation in a higher collapsed liquid level in the reactor vessel, depressurized at about the same rate as the R585 as shown in Figure 35. Part of the difference in vessel ;
calculation. The pressure remained higher because of levels was caused by differences in the noding of the ;
3 the difference in hot leg temperature at the initiation upper plenum. The R586 model explicitly represented of feed and bleed, as discussed previously. The sen- the small holes in the plenum cylinder at the hot leg ;
sitivity calculation showed that the major difference nozzle elevation. Modeling these holes allowed a more between calculations was caused by the power bound- accurate representation of the flow paths in the upper ary condition. plenum, the draining from the hot legs, and the mix-Pressurizer collapsed liquid levels from the R585, ture level in the vessel.
R586 base, and R586 sensitivity calculations are shown The comparison of the R585 and R586 feed and in Figure 32. The pressurizer filled with liquid earlier bleed calculations indicates that the transient response, in the R586 calculations because of the earlier OTSG and in particular the reactor coolant pressure, varied 27
I'
, e. .i' 100 a , , i i 1
, o R586 DASE CALCULMION '
O R586 REDUCED POWER 80 -
V -
// - ;t)
E 60 - FEED AND REED y -
3 1.
u o 40 -
,.G s.o.,, .
% N- 0 %,
20 -
c 7'
0 -- -
O 1000 2000 3000 4000 5000 6000 Time (s)
Figure 31. A comparison of core power from the R586 im: and ser.sitivity feed and bleed calculations, 14 , i i i i g-, k @e g _- /} (',
3 -n
,s V s n 30 -
{
i-, -30 ,
8 -
2
.? N .?
.a a- s-
--20 E"
....__. J o PM!i llASE CALCUL AllON O '! R?.% RE DUCED POWER 4 .
a nSas -
-10 2 -
O 1000 2000 3000 4000 5000 6000 Time (s)
Figure 32. A comparison of calculmed eMlapsed pressurizer liqui.i levels during feed and bleed.
\
28
) k k>
'l 0, .
100 -
c- , , i i o R586 BASE CALCULATION -200 FEED AND OLEED o R586 REDUCCD POWER s3a - a R585 -
. -15 0 n 1( (
v as 60 -
7 -
)
o m
-10 0 v
$ 40 -
b" - A .
~
N o
50 8 20
~
o 2 2
'" --O
.O m
-20 O 1000 2000 3000 4000 5000 6000 Time (s)
Ftctr .33.. A comparison of calculated mass flows out of the RCS during feed and bleed, t
i
- w 3. '
10 0 , , , , ,
O RS86 BASE CALCULATION -200 0 R586 REDUCED POWER
.<- h R585 _
7 -
-150 k j 60 -
I
- - -10 0 "
=
2 I -m 40 -
L- jg .
HPl ON y 0
1 2
_-50 20 - -
- e. --e 15:
FEr9 AND DLEED 0 maartmndd O O 1000 2000 3000 4000 5000 6000 Time (s)
Figure .R A cirmparison of calculated mass flows into the RCS during feed and bleed.
29
12 i i i , i o RSBS BASE CALCULATION o R$80 REDUCED POWER a RL85
--. Top OF CORE 35 m ^
E 10 - -
3 e e
-.$ u 30 .$
.?
.?
.k
.a 8 - -
h ##Y'*'*~* -
25
- ~'
b 20 6
0 1000 2000 3000 4000 5000 6000 i Time (c) q J
Figure 35. A comparison of calculated collapsed liquid levels in the reactor vehhel during feed and bleed. ]
substantially. Timing of events also varied significant, rapidly until the subcooling in the hot legs vanished ly. Differences between the calculations wem primarily and saturated boiling began in the core. In the R$85 l attributed to the core power boundary condition, calculation, the pressure then continued to decrease, However, in a macroscopic sense, the two calculations but at a slower rate. HPl was initiated at 4200 s, and provided similar behavior. Both the R586 and R585 the core was successfully cooled by feed and bleed. 1 l calculations indicated that feed and bleed would suc- In the TRAC calculation, however, the reactor coolant I cessfully depressurize the RCS while adequately cool- repressurized slowly. The coolant began to depres- !
ing the core. surize near the end of the TRAC calculation because l of slight voiding at the PORV. Although the calcula-3.4.2 90% Power. The R585 and TRAC feed and tion was not taken to HPI initiation, a LANL extrapola-l bleed calculations from 90% power were parametric tion indicated that feed and bleed would successfully variations of the LOFW transient of June n,1985,at cool the the core. 1 Davis-Desse. These calculations were similar to the The large difference in the pressure response be- l i
calculations descrioed in Section 3.4.1 except that the tween the TRAC and R585 calculations was attributed l initial power was lower and one MFP was not tripped to the boundary conditions. The R585 calculation j 1
immediately, thus resulting in the delivery of more depressurized too quickly because of the erroneous i feedwater to the OTSGs. NRR and LANL performed 25% reduction in con decay power when the RCPs I two comparable feed and bleed calculations. The feed were tripped. This error was discussed in the previous I and bleed calculations were initiated at hot leg temper. section. The TRAC calculation depressurized too slow- l atures near 617 K (650*F) or $89 K (600*F). ly because of a smaller-than-realistic PORV flow. l In the first comparison, feed and bleed was initiated Figure 37 shows the PORV flow from the TRAC near 2400 s in the R585 calculation and near 2000 s calculation and the combined PORV and HPVV flows in the TRAC calculation. The corresponding hot leg from the R585 calculation. The TRAC and R585 l temperatures were near 617 K (650"F) and 609 K curves are comparable because the HPVV Gow was l (636*F), respectively. Hot leg pressures from the R$85 generally insignificant compared to the PORY Cow.
and TRAC calculations are compared in Figure 36. The figure indicates that the PORV flow from the R$85 The pressures were similar prior to feed and bleed ex- calculation was significantly higher than the TRAC cept for the effects of the larger pressurizer spray in Cow, even though the reactor coolant pressure was the R$85 calculation, as described in Section 3.2. After generally higher in the TRAC calculation. A review the initiation of feed and bleed, the pressures dropped of data indicates that the TRAC PORV Ocw was 30 L_ _
'c'i
+ . g
- ' p.; s j .f, : i - . - ./' "w
- V?h )R(. ; - . , ,r . ,
^r f,. /,. '
.(
,.{ n, t .. ,'*-
ig.
r A
[A L
^
'< 5; , ,,,
j . .p . ,
M -
6
/g1 18 .
i .. . i s
- ' N;> . ;) -i. R585 '2500 "7' 4 ', ,,, pwvagupw ..... TRAC'
\; -;13 .
?~,j .,.e -
& E -p '
, _ _,, jy. , , . ,4..,,,,;, , ,, ,_ _,,
Q . i,. J -
. p-. 2000.$
s' .. !
A 'R N(
.S, 4 'r .
f -
.(
- j2 .
N. .% <e >>
. w w
e-e e
.o.
w p
1; .
'c ' . ' \,S \ . 1500 . *
, , %w,10 -
- c. s .
_ g- "
,h- '
Y
.g
- - _*= ,
'8 g
1 -
y i 1 6 e c>' '
e ... ,
1000
,. ( '
' * ,,..i t I () i
< 'O 2000 4000 Time (s)'
6000 8000 10000 lj '
,x
) Fig,uN: 36. A cIorpparison of caiculated reactor coolant pressures v3xa /ced'and bleed was initiated near 220;i s
}<,. r,.
7, e
g p' } l> ,, ,e
. ... s
, 3 c-
,. - i, s
., +.
t , +
) l' 3;
, 10 0 . H--- , , - ,
. R585 200 4
-4
. .... TRAC
'g go ..
+ - .
A k iq /
[ 15 0 ,-., i
, .-~.
p (i
, c N m !
j'
'3 (o. 60
[4 D
E j y f,. -
0 v . >,
,'. v.
+. . t 1 10 0 pg' 'C
$ 90 - (4 ,
g
.t % c
[ >
g }
4, 'l , g k
f ,,,' ~ . ... .kjp,I .... . ... .. ... ... . go ,
,)j g p,o ., f- ..,.............. -
g 1 . ; s f, 4 ;. j I, O-
-j !
' l r
" yl 0
l
..,t. I-' -
( ,a l D .(
-....-- J --.._.1..---~s' L N-20
"' .4000 6000- 8000 10000.-
. O 2000
/ .
Timen(ti) -
t I i j
Fium 37. A canparison of calcClated maw flow u of the RCS when feed and Heed was initiated near 2200 s. l
, 1 l
,.h 4, $
significantly too small. For example, at 5000 $ in the close agreement until subcooling was recovered near TRAC calculation, the PORV was passing liquid at a the end of the R585 calculation. The collapsed liquid i rate of about 24 kg/s (53 lbm/s) at a pressure of about level remained at or near the top of the pressurizer after 14.8 MPa f2150 psia). The Electric Power Research feed and bleed was initiated in both calculations, as Institute (EPRI) conducted critical flow tests through shown in Figure 39. The collapsed liquid level in the
, PORV similar to the one in Davis-Besse. The vessel remained above the top of the core in both E a EPRICrosby dat / indicated that for liquid flow atcalculations, 16.5 MPa as shown in Figure 40. The liquid level (2390 psia), the EPRI PORV would pass about 47 kg/s initially decreased more rapidly in the R585 calcula-(IN lbm/s). It is estimated that at the TRAC pressure tion because of the larger PORV flow. Ilowever, the
. ; of 14.8 MPa (2150 psia), the PORV should have pressure dropped below the ilPI shutoff head after passed about 43 kg/s (95 lbm/s). This estimate was ob- 4200 s in the R585 calculation, allowing flPI flow, and tained by adjusting the EPRI flow rate to accoant for causing an increase in vessel liquid level.
the 9'7c larger bom diameter i/ the Davis-Besse PORY NRR and LANL also performed calculations in compared to the valve tested by EPRI and to account which feed and bleed was initiated earlier and at lower for the different fluid conditions in the TRAC calcula- hot leg temperatures than the calculations dcscribed tion and the EPRI test. Thus, it appears that the TRAC above. In the R585 calculation, feed and bleed was ini-PORV flow rate is about a factor of two low for liquid. tiated near 1600 s at a 'mt leg temperature of $94 K Consequently, the RCS voided too slowly in the TRAC (610 F). In the TRAC calculation, feed and bleed was calculation. Since the depressurization of the reactor initiated near 1100 s at a hot leg temperature of about coolant was coupled to voiding at the PORV, the 586 K (595 F). A comparison of calculated reactor TRAC calculation depressuriicd too slowly. The cor- coolant pressures is presented in Figure 41. The pres-rect pressure response lies between the R585 and sure dropped rapidly in the R585 calculation after the TRAC curves shown in Figure 36. initiation of feed and bleed and then leveled out just The R585 and TRAC calculations generally were in below the ilPI shutoff head. The pressure decreased good agreement for parameters other than the reactor less rapidly in the TRAC calculation because of the coolant pressure and PORV flow. The R585 and smaller PORV flow rate discussed previously.
TRAf' fluid subcoolings for the A loop hot leg are in the previous discussic of Figure 36, in which compared in Figure 38. The two calculations were in feed and bleed was delayed compared to the current 60 i i i i R585 10 0
p 80 p 40 - -
E E
$ 60 %
8 30 -
8 Sm / 9m k
20 \ f h L im 10 -
20 f<h(c i
t1 0- -- " ' ~ '* "
0 0 2000 4000 6000 8000 10000 Time (s)
Figure 38. A comparison of ca!culated sutwohng in the A loop hot leg when feed and bleed was initiated near 2200 s.
32
14 i i e i \
40 12 '- -
m n E 10 -
v t
v
-- 30 -v e >
8 I
.v_ :
.v_
3
- g. l 4 - 20 .?
. .s 6- _J
\
R585
T R AC 4 ,
10 2
O 2000 4000 6000 8000 10000 Time (s)
Figure 39. A comparison of calcitlated collapsed liquid levels in the pressurizer when feed and bleed was initiated near 2200 s.
12 i , , i o R585 O TRAC o ----- TOP OF CORE
. - - NOZZLE CENTERLit:E -35 l
l
^ -
O I v E 10 - -
v
- o. -
. _o _
e v
$ - -30 -
6
.v_ .v_
o s
.? e .
_ .E
._._..._._....._._._._...._..._._._._..__._.%..._,._._c .
m
._._._._._._._._. -25
' " - * **'* ^* -
~20 6~*" *
- 0 2000 4000 6000 8000 10000 Time (s)
Figure 40. A comparison of calculated collapel liquid levels in the reactor vessel when feed and bleed was initiated r. car 2200 s.
33
18 , ,
FEED AND BLEED R565 2500
..a .) h ----- T R AC 16 i j i -
B ! i 9 I t T ',
a, li. .! ',
- 9.
E '4 \l ~ ..,
2000 $
- -' v
\'
a ........................ ........ ............ 8 8 12 -
- E 6.
O-(U 0
10 - --
8 ' '
0 2000 4000 6000 Time (s) l Figure 41. A comparison of calculated reactor coolant pressures when feed and bleed uas initiated near 1500 s.
calculations, the rapid initial depressurization was between the R585 and TRAC calculations was attrib-limited by Cashing and boiling in the core, in the cur- uted to the upper head model. In the R585 model, the rent calculations, the depressurization was halted by upper head was modeled as a stagnant volume, with the liPI, which nearly balarwed the flow out the PORV. one flow connection to the upper pier.um. Consequent-As discussed previously, the core power was too low ly, the temperature in the upper head remained nearly after the RCP trip in the R$85 calculation and the constant untii the pressure dropped low enough for it l'ORV flow was too low in the TRAC calculation, to flash and void. In the TRAC model, the upper head Consequently, the R585 calculation depressurized too was modeled as a flow-through volume with two flow rapidly, and the TRAC calculation depressurized too connections, one to the upper plenum and one to the slowly. The actual pressure response should lie be. control rod guide tube brazements which were also tween the two calculated results. connected to the upper plenum. The flow circulated Figure 42 shows the llPI flow rates in the TRAC between the upper head and the apper plenum, cou.
and R$85 calculations. IIPI flow was initiated at 1700 pling the temperatures of the two regions. The upper and 1900 s in the R585 and TRAC calculations, head thus remained subcooled in the TRAC calculation, respectively. The HPI flow was much higher in the Nc,te that the R586 model also ur,ed a flow-through R585 calculation because of the lower reactor coolant upper head similar to the TRAC model. The actual pressure shown in Figure 41. The llPI helped main- mixing process that would occur in the upper head for tain subcooling in the RCS, as illustrated by Figure 43, this transient is not known. The lower 40% of the up-which shows subcooling in the A loop hot leg. As per head is below the top of the guide tube brarements shown in the figure, the A loop hot leg remained sub- and should mix well with the upper plenum fluid, cooled throughout both calculations. Hov ever, the mixing that would occur in the upper in the TRAC calculation, the RCS remained liquid 60% of the upper head is governed by a relatively com-solid, as illustrated by Figures 44 and 45, w hich show plicated multidimensional natural circulation process, i collapsed liquid levels in the pressurizer and the reac. which was not mechanistically represented in any of l tor vessel. In the R$85 calculation, the core and the the models. Although the actual mixing process is not i A hiop stayed subcooled, while some voiding occurred well understood, the stagnant and flow-through models in the pressurizer, the B loop, and the upper head after bound the possible responses, it is felt that the flow-the initiation of feed and bleed. The voiding in the through modct provides the best representation of upper head caused the collapsed level in the vessel to the upper head for most transients. The possible decrease. The difference in the vessel level response formation of a steam bubble in the upper head leads 34
)
l 1
1 80 i i R585
..... TRAC 15 0 60 -
- T M N N E v
f 10 ' o
.o 8 40 -
3 C S M
'S
- o 2 50 $
20 -
- ---..... ..............--..j~.-
- I' ' O 0
O 2000 4000 6000 Time (s)
Figure 42. A comparison of calculated HPl flow rates when feed and bleed was initiated near 1500 s.
80 i i R585
..... TRAC 60 -
^
m 6 10 0 D cn /
cn C
- C
.:= l '
. *. :=
o 40 - :/
0 '
. FEED AND BLEED 8 n ,
/ -
o aa 3 /
,, / I
\' 50
.? \ .2 a -l -
o
' \. '
20 HPl ON s
u
' ' O 0
O 2000 4000 6000 Time (s)
Figure 43. A comparison of calculated fluid subcooling in the A kmp hot leg when feed and bleed was initiated near 1500 s.
35
l l
l
(-
14 , ,
12 '- -
40 m : ^
v E 10 -
0 v
e 30 _v
- 8 -
- n v_ j, ,
.t_
g . v
.. J 6 d R585 - 20 .?
1 J
- ----- T R AC
- i. .
4 -
10 2 -
O 2000 4000 6000 Time (s)
Figure 44. A comparison of calculated collapsed liquid levels in the pressurizer when feed and bleed was initiated near 1500 s.
12 i i n ; _............................................
UPPER HEAD VotDS -35 m ^
v E 10 -
N %
v e o R585 -e TRAC >
3 39
- TOP OF CORE .
y NOZZLE CENTERLINE y a
.t_r 8 - - .er
_a _J
-.._..._.._._..._...._.__._.._.._._.._._._._._._._._.._._._._._.-25 6~~-~'----------'----~~--------'-~~-----------20 0 2000 4000 6000 3 Time (s) '
Figure 45. A comparison of calculated collapsed liquid levels in the reactor vessel when feed and bleed was initiated near 1500 s.
36
l i
I to uncertainty in the minimum vessel level for these portant measures of the success of feed and Need were feed and bleed calculations. However, this uncertain- the ability to depressurize and the minimum collapsed ty is not thought to be significant, since the minimum liquid level in the reactor vessel. The uncertainties in level would be at the bottom of the upper head, well the calculation of the significant thermal-hydraulic q above the top of the core. phenomena due to the potential sources of uncertainty j The R585 and TRAC calculations showed that ifini- were based on several factors, including subjective tiated when the hot leg temperature was about 589 K judgments. Details of this assessment are presented in (600'F), feed and bleed would successfully depres. Appendix C. A summary of the assessment of the surize the RCS and cool the core. No significant void- uncertainty follows. Unless otherwise stated, the uncer-ing occurred in the RCS because the core remained tainties apply to the R586 feed and bleed calculation subcooled. HPI flow was initiated relatively early in described in Section 3.3.
these calculations because the saturation pressure cor. The uncertainty in the RCS temperature when feed responding to the hot leg temperature at the time of and bleed was initiated in the R586 calculation is esti-feed and bleed was below the shutoff head of the HPl. mated to be 5 K (9*F). This uncertainty was caused Recall that the pressure will quickly drop to the satura- by uncertainty in the initial and boundary conditions, tion pressure corresponding to the hot leg temperature Table 5 shows the contribution of several different in the absence of HPI flow. Thus, an extension of the parameters to the uncertainty in the RCS temperature.
calculations indicates that if feed and bleed is initiated The largest contributor to this uncertainty was the when the hot leg temperature is below 602 K (624*F), OTSG dryout time. The uncertainty in the OTSG corresponding to the saturation temperature of the HPI dryout time was primarily due to uncertainty in the shutoff head, then the reactor coolant pressure will initial hquid inventory in the OTSGs. The stated uncer-quickly drop below the HPI shutoff head. If the com- tainty bounds the deviations between the calculated and bined HPI and makeup flows are sufficient to prevent measured temperatures for the Davis-Besse LOFW boiling in the core, no significant voiding in the RCS transient of June 9,1985 (see Figures !I and 12). The is expected. If the combined flows are not sufficient stated uncertainty is also representative of the devia-to prevent boiling in the core, the RCS could pres- tion observed between an assessment calculation and surize, temporarily shutting off HPI flow and resulting data from a feed and bleed test
- in an experimental in more extensive voiding of the RCS. Hand calcula- facility scaled to a B&W PWR. The stated uncertain-tions indicate that if feed and bleed is initiated before ty in the RCS temperature corresponds to Ii % uncer-the hot leg temperature reaches 597 K (615'F) and tainty in the calculated heatup rate after OTSG dryout.
more than 10 minutes have elapsed since reactor trip, The corresponding uncertainty in the time required to the reactor coolant pressure will remain below the HPI reach the RCS temperature at which feed and bleed shutoff head and no significant voiding in the RCS will was initiated was about 2 min.
occur. The HPI can then cool the core, assuring the The uncertainties in the R586 calculation due to ini-success of the feed and bleed operation. tial and boundary conditions are also applicable to the R585 calculations. However, the uncertainties in the R585 calculations are larger because of the erroneous 3.5 Calculation Uncertainty Several potential sources of uncertainty were iden.
tified which contributed to the overall uncertainty in Table 5. Uncertainty in RCS temperature the feed and bleed calculations described in this report.
The potential sources of uncertainty include the Uncertainty thermal-hydraulic computer code used to make the Parameter (K) ( F) calculation, the code mput model, the imtial conditions at the start of the calculation, the boundary conditions Initial core power 0.8 1.4 applied during the calculation, the code user, and the assumed transient. A review of the calculations and Decay heat 1.9 3.5 data presented previously in this report hciped iden- OTSG dryout time 4.4 8.0 tify important phenomena relative to the feed and bleed process. These phenomena included the thermal- RCP power 0.9 1.6 hydraulic conditions at the start of feed and bleed 0.8 1.4 RCS heat structures (principally the RCS temperature) and the boundary conditions during feed and bleed. The important boundary conditions included core power, makeup and Total 5.1 9.1 HPl flow, and flow through the PORV. The most im-37 L.
25 % reduction in core power when the RCPs tripped, because the small holes in the plenum cylinder at the as discussed in Section 3.4.1. The reduction in core hot leg nozzle elevation were not explicitly modeled.
power resulted in a calculated heatup rate that was However, the uncertainty due to the upper head and about 25% too low after RCP trip. This error caused upper plenum models should not affect the calculated a bias in the R585 calculations in addition to the uncer. result that the minimum collapsed liquid level remains ;
tainty associated with the initial and boundary condi- above the top of the core.
tions. For the R585 calculations described in this A simple, quasi. steady volume balance was used to report, the error caused the calculated RCS temperature estimate the uncertainty in the calculation of whether to be about 2 K (4'F) too low at the initiation of feed or not the RCS should depressurize. The parameters and bleed. The time required to reach the RCS which compress, and thus pressurize, the RCS are the temperature at which feed and bleed was initiated was volumetric flow due to makeup, Q,, and the vol-about I min too long. The bias was larger in those umetric expansion due to boiling in the core, Qe. The j R585 calculations, not described in this report, in volumetric flow through the PORV, Qp , acts to )
which feed and bleed was initiated at 37 min. For these depressurize the RCS. Figure 46 shows the volumetric i calculations, the errors in RCS temperature at the start flow through the PORY versus quality and the sum of l of feed and bleed and the time to reach this temperature volumetric flows due to makeup and core boiling, Qm were about 11 K (20"F) and 6 min, respectively. + Qc. The core power was varied parametrically as
)
1 The uncertainty in the calculated collapsed liquid a function of time after reactor trip from 100% power.
{
level in the reector vessel was estimated to be i m The results shown in the figure were obtained at a (3 ft). This r_sult was based on the results of assess- pressure of 17.2 MPa (2500 psia), which is close to ment calculations. In the R586 feed and bleed calcula- the pressurizer SRV setpoint pressure, and assumed tion, an uncertainty was identified relative to the RCP both makeup pumps were available. When the 501 m>dalization which could cause the calculated level to umetric flow through the PORV exceeds the volumetric be 0.3 m (1 ft) too high. The uncertainty in the level flow due to the combination of makeup and core boil.
could also be larger because of the upper head model. ing, the RCS will depressurize. Figure 46 shows that ing for those transients in which feed and bleed was the PORV cannot depressurize the RCS for times less initiated at relatively low RCS temperatures. The than 30 min after reactor trip, regardless of the fluid uncertainty in the R585 calculation may be larger state at the PORV. However, as time increases, decay I
l J
! 0.5 , , , , , l l
i l
x PORV (09) '
- -- MAKEUP + CORE BOlU NG (Om +O)c 15 l
, , 0.4 -
, I m a p T = 10 minutes E -- ~ ~~--- - - - - - - - - - - - - - - ~ ~ - - - - - - - - - - - - - ~ ~ - t
, 0.3 -
O 10 o C ...................T..=.20..m..i.nu.t.e.s..................................................o.-
g j
f = 30 minutes
,3 .................................... . .................................................. ,g 2 0.2 -
3 ID T = 60 minutes o E - - . - . - - - . - - - - ~ ~ - - . - - - . - - - . . - - - - - -
-- ~ ~- - -- -- ----
E 2 5 2 o N T = 120 minut es o
0.1 ,..................... . ................................................:-
0.0 O
-0.50 -0.25 0 0.25 0.50 0.75 1 Quo Ht y. x Figure 46. Results of the volume balance for different times after reactor trip.
38
l l
1 heat and steam production in the core decrease. At the depressurization could begin. The uncertainty in about 37 min after reactor trip, the RCS will depres- the volume balance due to the PORV now was larger surize if dry, saturated steam flows through the PORV. than that due to the uncertainty in the other two j The RCS will depressurize with almost any fluid at the parameters. However, the PORV could stdl depres- i PORV at 120 min after reactor trip. surize the RCS once the quality at the PORV ap-The uncertainty in the boundary condit;ons of core proached unity. Thus, the uncertainties in the bound-decay power, PORV flow, and makeup flow does not ary conditions of the R586 feed and bleed calculation significantly alter the results of the R586 calculation do not have a large effect on the calculated ability to relative to the depressurization of the RCS. The sen. depressurize. The calculated depressurization appears sitivity of the calculated depressurization to uncertainty valid and would not be expected to vary significantly in the feed and bleed boundary conditions is illustrated because of uncertainty in the boundary conditions. The I in Figure 47. The figure shows the best-estimate estimated uncertainty in the time of depressurization PORV volumetric flow and the volumetric flow due due to the combined uncertainty in the three boundary to makeup plus core boiling at 60 min after reactor trip conditions is 5 rain. Furthermore, sensitivity calcula-from Figure 46. This time was selected because it was tions described in Appendix C indicated that the approximately the time that depressurization occurred calculated time of depressurization was not sensitive in the R586 feed and bleed calculation. The figure also to hot leg nodalization.
shows the best-estimate PORV flow reduced by 20% The uncertainty in the depressurization in the R585 and the makeup and coreboiling terms resulting from calculations was thought to be much larger than i a 5 % increase in core decay power and a 10% decrease described above for those feed and bleed calculations in makeup flow. These variations correspond to the in which the RCPs tripped. As discussed previou ly, estimeted uncertainties in these parameters. the error in core power when the RCPs tryped The effects of the uncertainty in decay power and significantly affected the depressurization in the R585 makeup flow were relatively small. The volume calculations.
balance indicated that the PORV could still .depres- The above estimates of uncertainty in the important surize the RCS at 60 min with either the higher core feed and bleed parameters are valid for the transients power or the lower makeup flow. However, a slightly analyzed based on the assumed initiating event, equip-higher quahty at the PORV would be required before ment performance, and operator actions. Significantly 0.5 i i i i i l x BEST ESTIMATE PORV FLOW
~
BEST ESTlMATE MAKEUP + CORE BOILING -15 a INCREASED DECAY POWER -
0.4 ~ ^
n o REDUCED MAKEUP
$ 0 REDUCEO PORV FLOW $
m n
O E
y 0.3 -
-10 o m
.o._
o o
{g 0.2
{g
- .. -5 -
0.1 -
O 0.0
-0.50 -0.25 0 0.25 0.50 0.75 1 Qualit y, x figure 47. The effect of uncertamty in boundary conditions on the results of the solume balance.
39
different results could be calculated if different in the time of re,ctor trip thus could result in i assumptions were made regarding these parameters. significantly earlier dryout of the OTSGs and higher }
Two examples that illustrate the effect of different reactor coolant temperatures at the initiation of feed 4 assumptions on the calculated results are discussed and bleed.
below. Second, in the R585 and R586 calculations, no First, in the R585 and R586 feed and bleed ca!cu-j makeup flow was assumed prior to the initiation of feed l lations, tripping both MFPs resulted in an immediate and bleed. If a more realistic scenario had been j anticipatory reactor trip. If the anticipatory reactor modeled, the operators would have provided maximum l trip failed or the main feedwater control valves closed makeup flow until pressurizer level was recovered.
while the MFPs continued running, the reactor trip With maximum makeup, the RCS temperature at the would be delayed until another parameter, such as initiation of feed and bleed would have been about 5 K reactor coolant pressure, reached its trip setpoint. A (9'F) lower, Thus, the sensitivity of the calculated )
delay in reactor trip would cause a rapid reduction results to assumptions regarding operator actions and in steam generator liquid inventory because of the equipment performance can be as large or larger than absence of main feedwater. Relatively minor delays the uncertainties described above.
i l
i a
__o
- 4. CONCLUSIONS I
Feed and bleed can be successfully used to cool the tion exhibited the phenomena expected. The uncertain- {
core at Davis-Besse in the event that all feedwater is ty in the RCS temperature at the initiation of feed and lost. The analysis indicated that il feed and bleed is bleed was estimated to be 5 K (9'F). The correspond-initiated within 20 min and full makeup flow is avail- ing uncertainty in the t'ime required to reach the RCS able, feed and bleed could successfully depressurize temperature at which feed and bleed was initiated was the RCS while cooling the core. The effect veness of 2 min. The uncertainty in the calculated collapsed feed and bleed in transients in which the makeup flow liquid level in the reactor vessel was estimated to be was degraded or the initiation of feed and bleed was 1 m (3 ft).
delayed was not investigated. The uncertainties in the R585 calculations were The important parameters in a feed and bleed tran- larger than in the R586 calculation, primarily because sient are the RCS temperature at the time feed and of an error in the core decay power following RCP bleed was initiated, the depressurization during feed trip. In addition to the uncertainties stated above, the and bleed, and the minimum liquid level in the reac- R585 calculations contained a bias m the RCS tor vessel. The RCS temperature at the start of feed temperature that ranged from 2 to 11 K (4 to 20*F),
and bleed can have a large effect on the transient depending on the time after RCP trip that feed and response, as discussed in Section 3.4.2. Without bleed began. The corresponding errors in the time to depressurization, the core could uncover and heatup reach a given RCS temperature ranged from 1 to while the RCS pressure remained above the HP1 6 min.
shutoff head. The minimum liquid level in the reactor Event tim'mg and RCS pressure response are sen-vessel determines whether or not the core is adequate- sitive to ihe assumed boundary conditions of core ly cooled. power, PORY flow, and makeup flow. The RCS did The R586 calculation of the Davis-Besse LOFW not depressurize in the R586 calculation until about transient was in good qualitative and quantitative agree- 40 min after feed and bleed began. However, as ment with the measured data. As described in Sec- described in Section 3.4, the RCS depressurized im-tion 3.1, the trends observed in the plant were well mediately when the core power was reduced by 25%.
represented in the calculation. Prior to reactor trip, the When the PORV flow was reduced by a factor of two, trends of the calculated feedwater flow, main feedwater the depressurization was delayed by over an hour.
control valve arca, OTSG pressure and level, and Thus, feed and bleed transients in the plant would also RCS response were in excellent agreement with the be expected to be sensitive to variations in equipment measured data, indicating that the ICS behavior was performance and operator actions.
well modeled. After reactor trip, the phenemena of The feed and bleed transient is significantly affected OTSG dryout and RCS heatup were also well repre- by the RCS temperature when feed and bleed is ini-sented. The maximum deviation between calculated tiated. R585 and TRAC calculations showed that when and measured reactor coolant pressure was 0.3 MPa feed and bleed was initiated at a hot leg temperature
! (50 psi). The deviations between calculated and near 589 K (600'F), the RCS easily depressurized measured reactor coolant temperatures were general- below the shutoff head of the HP1 pumps; feed and ly less than 3 K (6'F). bleed then successfully cooled the core without signifi-l The differences observed between the R586, R585, cant voiding of the RCS, Hand calculations indicated and TRAC calculations of the Davis-Besse LOFW that no significant voiding of the RCS would occur if transient were primarily due to the use of different feed and bleed was initiated before the hot leg temper-boundary conditions, including core power, feedwater ature reached 597 K (615'F). If feed and bleed is ini-flow, and pressurizer spray flow. The calculated results tiated at a higher RCS temperature, there is the poten-were sensitive to the feedwater flow. The effect of tial for the RCS to pressurize and for significant variations of the feedwater flow based on its likely voiding to occur. In the R586 calculation, the hot leg uncertainty would probably be as large or larger than temperature was about 621 K (657'F) when feed and the observed differences between the calculations and bleed was initiated. Although feed and bleed was data. Even though different thermal-hydraulic com- calculated to be successful, the depressuri2.ation was puter codes and input models were used, the R586, delayed and significant voiding occurred in the RCS.
R585, and TRAC calculations were similar and showed The uncertainty in the boundary conditions of core trends like those observed in the plant. decay power, PORV flow, and makeup flow did not The R586 feed and bleed calculation provided a significantly alter the results of the R586 calculation reasonable representation of the transient. The calcula- relative to the ability to depressurize the RCS during 41
feed and bleed. However, hand calculations (see Ap- as a stagnant or flow-through region. The best model-pendix C) indicated that feed and bleed would not be ing technique for the upper head is not known, successful without any makeup flow, in the absence However, the uncertainty caused by the upper head of makeup, the depressurization of the RCS would nodalization is not significant relative to core uncover-probably be delayed until after the core heatup began. ing for the feed and bleed transients analyzed.
The initial ano,ivundary conditions were thought to The RCPs should be nodalized such that the correct )
be the source of most of the uncertainty in the feed volume ofliquid will remain trapped in the loop seals l and bleed calculations performed. It was recognized, during a feed and bleed transient. The liquid which re-however, that significantly different results could be mains in the loop seals is not available to cool the core.
obtained if different assumptions were made concern- As described in Appendix C, an uncertainty was iden-ing the initiating event, equipment performance, and tified in the R586 feed and bleed calculation due to RCP operator actions during the transient. In particular, the nodalization which could cause the calculated reactor results were thought to be sensitive to assumptions con- vessel liquid level to be about 0.3 m (1 ft) too high.
cerning feedwater flow, makeup flow, and reactor trip. The results of the R586 feed and bleed calculation The small holes present in the reactor vessel plenum were not sensitive to hot leg nodalization. As described ;
cylinder, located at the elevation of the hot leg nozzles, in Appendix C, the nodalization of the hot leg affected should be explicitly modeled. Modeling these holes the timing of events by less than 80 s. j a! lows a more accurate presentation of the flow paths The fission power produced following reaaor trip in the upper plenum, the draining from the hot legs, should be modeled. The integrated, posttrip fission and the mixture level in the reactor vessel. power in the R586 calculation of the Davis-Besse The calculated mixture level in the reactor vessel can LOFW transient was equivalent to 1.5 s of full reac-be sensitive to the modeling of the upper head in certain tor power. The effect of the posttrip fission power on transients. The upper head may remain subcooled or RCS temperature was about 3 K (6'F), as described flash and drain, depending on whether it is modeled in Section 3.2.
l l
42
- 5. REFERENCES
- 1. U.S. Nuclear Regulatory Commission, Loss ofMain and A.uillary Feedwter Event at the Davis-Besse Plant on June 9,1985, NUREG-1154, July 1985.
- 2. J. F. Lime et al., " Rapid-Response Analysis of the Davis-Besse less-of-Feedwater Event on June 9,1985,"
Topical Meeting on Reactor Physics and Safety, Saratoga Springs, NY, September 17-19, 1986; LA-UR-86-1782.
- 3. V. H. Ransom et al., RELAPS/ MOD 2 Code Manual, Volumes 1 and 2, NUREGICR-4312. EGG-2396, August 1985.
- 4. Safety Code Development Group, TRAC-PF1/ MODI: An Advanced Best-Estimate Computer Programfor Pressurized Water Reactor Thermal-Hydraulic Analysis, NUREGICR-3858, LA-10157-MS, July 1986.
- 5. E. T. Laats et al., User's Manualfor the Nuclear Regulatory Commission 's Nuclear Plant Analyzer, EGG- l RST-7044, September 1985.
- 6. P. D. Wheatley et al., RELAPS/ MOD 2 Code Assessment at the Idaho National Engineering Laboratory, NUREG/CR-4454, EGG-2428, March 1986.
- 7. C. D. Fletcher et al., RELAPS Thermal-Hydraulic Analyses of Pressurized Thermal Shock Sequencesfor the Oconee-1 Pressurized Water Reactor, NUREGICR-3761, EGG-2310, June 1984.
- 8. " Decay Heat Power in Light Water Reactors," American Nuclear Society, ANSilANS-5.1-1979, Lagrange Park, Illinois, August 1979.
- 9. Electric Power Research Institute, EPRI PWR Safety and Relief Test Program Safety and Relief Valve Test Report, EPRI NP-2628-SR, December 1982.
- 10. J. R. Gloudemans, "OTIS Test Results." Thineenth Water Reactor Safety Research lnformation Meeting, Gaithersburg, MD, October 22-25, 1985, NUREGICP-0072, Vol. 4.
43 l.
l l l
l l
1 1
l
\
APPENDIX A R586 MODEL DESCRIPT!ON 1
I A-1 1
i
\
APPENOlX A R586 MODEL DESCRIPTION le8, one OTSO, two pump suction legs, two RCPs, A '!. Introduction and two cold legs (refer to Figures A-1 and A-2). The The R586 model of Davis-Besse Unit I is described cold legs in loop A are designated as Al and A2, cor-in this appendix. The R586 model was developed for responding to the RCP number, either 1 or 2, in ,
use with the RELAPS/ MOD 2^'l computer code and ' loop A. The B cold legs are designated similarly. The i was developed in 1986. The model of the reactor pressurizer and pressurizer surge line (Figure A-3) are coolant system (RCS) is described in Section A-2. The attached to the hot leg in loop A. The RELAP5 vessel model of the secondary coolant system is described in model, shown in Figure A-4, represents major com-Section A-3. Control systems are described in Sec- ponents of the vessel, including an inlet annulus, tion A-4. References are presented in Section A-5. downcomer, lower plenum, core, core bypass, upper The RELAP5 model of Davis-Besse is based on a plenum, upper head, reactor vessel vent valves, and modelA-2 of the Oconee-1 pressurized water reactor. the control rod guide tube brazements. The major flow The Davis-Besse model is based on the experience path through the downcomer, lower plenum, core, and gained with the Oconee model, which was developed upper plenum is represented. The minor and leakage and used extensively in the pressurized thermal shock flow paths are also represented. These minor flow program. The main differences between the Oconee paths include the core bypass between the core barrel and Davis-Besse models are due to geometrical and and core former plates, the leakage path between the physical differences between the plants and improve- downcomer and upper plenum around the hot leg ments to the code and modeling techniques since the nozzles, the leakage path though the control rod guide development of the Oconee model. The modeled tube brazements to the upper head, and the small holes geometrical differences include differences in reactor in the plenum cylinder at the hot leg elevation. The coolant piping due to the raised loop configuration of model of the RCS includes representations of the Davis-Besse; the different condensate, feedwater, and pressurizer pilot-operated relief valve (PORV) and main steam systems; differences in equipment, such safety relief valves (SRVs), hot leg high point vent as reactor coolant pumps (RCPs), high-pressure injec- valves (HPVVs), emergency core cooling system tion (HPI), auxiliary feedwater (AFW), etc; and the (ECCS), makeup and letdown, and pressurizer heaters reduced number of reactor vessel vent valves (four in- and spray. The ECCS includes high-pressure injection stead of eight). The Davis-Besse vessel model is noded (HPI), low-pressure injection (LPI), and core flood differently than Oconee to allow the use of a crossflow tanks. LPI and the core flood tanks are connected to model at the connections between the hot and cold legs the inlet annulus of the reactor vessel. HPl is connected and the reactor vessel and the junction between to each of the four cold legs, downstream of the reac-the pressurizer and surge line. The assessment of tor coolant pumps. Heat structures were used to repre-RELAP5/ MOD 2^4 indicated that improved represen- sent heat transfer from and stored energy in the fuel tation of loop draining could be obtained using the rods. OTSG tubes and tube sheets, loop piping, reactor crossflow model at the connections between the loops vessel wall and internals, pressurizer wall, pressurizer and reactor vessel. An additional difference between surge and spray lines, and pressurizer heaters.
models is that the secondary side of the Davis-Besse The pressurizer heaters provide a maximum power once-through steam generator (OTSG) was modeled of 1.329 MW. The pressurizer sprav valve is sized to two-dimensionally to prmide a more mechanistic pass 0.012 m3 /s (190 gpm) at normal operating representation of AFW wetting, conditions. The PORV is sized to pass 25.2 kg/s (55.5 lbm/s) of saturated steam at 16.10 MPa f2335 psiaj and 47.5 kg/s (104.71bm/s> of subcooled A-2. Reactor Coolant System hquid at 16.46 MPa (2387 psia) and 613 K (644*F).
The model of the RCS is shown schematically in The resulting PORV area was 9.48 x 10-4 m 2 Figures A-1 through A-4. The Davis-Besse plant has (0.01020 ft2), with a single-phase liquid discharge a two by four configuration, i.e., two loops, each con- coefficient of 0.82 and a two-phase discharge taining one hot leg and two cold legs. The loops are coefficient of 1.0. Each hot leg HPVV is modeled with designated loop A and loop B. Sometimes the loops an area of 1.830 x 10 4 m' (0.000197 ft2 ), w th are also referred to as loop 1 and loop 2, with loop 1 single-phase and two-phase discharge coefficients of corresponding to loop B. Each loop contains one hot 0.624.
A3 E_____________________
l l
l O :
5 h
w i I l 1l li. 1 1
$ b i
s } a a e , e
... mnss, m .
}g} Z j
0 T
I 2
y (E) y .g 2s -
e V
(e q; e
\-
5 g e e
- - __ $ j P i a '
i - 4
~ ? l - e
- ' s q
- H H
( l > e qE-
) in I x t' p D B l
E e E,
e T 5 s i
A-4
6 4
1 O
- O N
E R
E F
L P"
u V
5 5 4 6 5 5 gl l 1 1
. 8 5 2 2 l l 0
0 2 B' 0 0 8 5 p 2 2 o l
o l;
e d
l l o
m e
b7 0 s _
l l 2 s 5 e ii 7 5 1 B -
2 4 l
2 2 7 m is v
a 1 _ 9 5 D 0
2 [
0 7
[
0 4 8 2
O27 S P
A 2 2 7 L l l l E
R I
2-1 A 0
1 B e
' 5 r u 2 2 5 3 P u I
a I 1 B 6 2 C ig R
P [2 F _
0
= X 5 C R
3 2
l v 0 2
w s
3 1
2 -
5 5 8
1 2
2 2
1 2
C _
S T _
O 0 6
2 -
- >6 -
ll .
7 0
1 O .
O N
E .
J 0
0 8
Y A
N -
& 0 2
6 .
m e
V t s
y R s O 6 r P 1 iz e _
1 6 r u 2 0 0 s s
8 7 e r
1 p l; -
\ - _
- _ - M d e -
- _ - O o
- R m 1 .- . _ . F e m5 s 6 -
. _ , s
_ e I
B-
/ . _
- i s
v a
0 D 0 5 6 P V 2 f A R 0 L E
S 5 R
<l .
3-l, A
e r
u ig F
3 l 0
8 -
5 -
0 1
M O
R F ._
ll IjI{l 8
0 1
O O
\
N E
J S
T N
E h
E Z
A R
B E
B N U L N
T E
m6 -
5 4 3 2 1
E L
P U 5 5 1 5 51 5 R G 1 1 1 E
5 5 5 5 5 W l 1 O e 1 _ 8 5
L s s
e v
//
j r
' l l
{
l 0:
l t
o c
0 5 a 3 2 I e r
5 5 l l
r e
~ 5 d
-- 0 o 0
N i 5
3 5
0 5
4 5
5 4
2 5
k ::. :. # :.+7i;.. .7 2
0 2
o
- l.
1 0
- I 5
5 7
B-i m
e s
s e
s v
a D 0 l 1 t 1 5 D A 5 5 s 5 5 - . 5 K _ P k +: .. :e
.. 'It' ;- ; A
~i. .*::.M;;::. a::;;;.
R f
!" !!e
[j !! f::- . : #[.:::
L E E P R P
(
g .
1 f
i:
- i:
)"
5 4-5 0 5 i 2 3 4 ! - l A 5 6 , 6 1 - - - - 0 e 5 5 5 0 0 0 0 7 r l 7 7 7 7 5 u l 5 5 5 5 ig
- ~ ) - 9" F
l [. l R
E M
O 1 1 C l
0 l
85 N 0
7 7
2 7
m0 i 2 0
22 W O
l l sc D s usc a
s g u mr rt 57 Dt o
c d
T
A-3. Secondary Coolant System The AFW system is illustrated in Figure A-6. The turbine-driven AFW pumps are modeled explicitly to The model of the secondary coolant system includes represent the variation in flow to the OTSGs with pump representations of the condensate and main feedwater speed. Each pump is normally dedicated to a single systems downstream of the deaerator storage tanks, the OTSG. However, crossover piping downstream of the AFW system, the OTSGs, and the main steam lines, pumps allows either pump to supply either OTSG. The The model of the main f:edwater system (see crossover valve and the admission valves to the OTSGs Figure A-5) iacludes repre:,entations of the deaerator are controlled by the steam and feed rupture control storage tanks, turbine-driven booster and main sy stem (SFRCS).
feedwater pumps, the high-pressure feedwater heaters. The Davis-Besse steam generators are once-through the main feedwater control and block valves, the and are oriented vertically. Between the outer shell and startup control valves, the main stop valves, check the heat exchanger tube bundle is a cylindrical baffle, valves, and connecting piping. Common headers forming a downcomer section. A gap in the baffle i connect the two feedwater trains upstream and allows steam to be drawn from the boiler region into l downstream of the high-pressure heaters. The the downcomer to heat the incoming feedwater. After !
deacrator storage tanks are modeled as time-dependent falling through the downcomer, the feedwater enters i volumes, with pressure and temperature specified as the tube bundle and flows upwards, vaporizing to l a function of plant load, The power added to the feed- saturated steam in the nucleate boiling region. Dry water by the high-pressure heaters is also specified as saturated steam is produced in the film boiling region a function of plant load. The geometry of the high- and raised to the exit steam temperature in the super- )
pressure heaters and the piping upstream of the main heat regica The steam flow then enters the steam an- '
feedwater pumps was not available during the develop- nulus section, which is between the outer shell and the ment of the model. The piping length upstream of the cylindrical baffle and above the feedwater inlet port.
pumps is assumed, while the geometry of the high- The superheated steam then exits the OTSG via the pressure heaters is based on Oconee. Heat structures main steam line.
are used to represent the high-pressure heaters and the The RELAP5 models of the A and B OTSGs are l piping walls. shown in Figures A-7 and A-8. The major components 748 750 752 754 756 DEAERATOR y
r EOOSTER PLM3 i i i MAIN FEEDWATER STORACE 749 "5 Tams 751 753 760 761 f n , ,
r n , ,
i i i i 762Z 2 757 H]CH rwEsstnE 763
>{ ATERS 781 779 777 775 773 OTSG A 400 -+ 1 d F-M l l MAIN FEEDWATER j
!5CLATION VALVE 5 777 l
700 778 776 774 77I OTSG G 300 -%4-{ i-- + 0 l l [
770 u4g Am STARitP FEEDWATER CONTROL VALVES JEN00113 Figure A 5. RELAPS Davis-Besse model; condensate and feedwater system.
A-8
805 820 800 810 X 323 OTSG A CONDENSA1E STORAGE TANKS k RWS 830 CROSSOVER VALVE 855 850 860 -N 423 OTSG B 870 I
END0112 Figure A-6. RELAPS Davis-Besse model; AFW system.
1 A-9 !
i
- n. _ _ _ _ - - . . - -
1 1
l
- l 1
'. . -' 115
- r
- ' :4 :
FROM AFW q-ll 324 LL _L lP 326 O
u d w q d O
- s O
~
u u i e in 357 328 u u u E ....
I 7eo L- N E I I
MAIN g FEEDWATER 356 AEV L%ET u ? 120 G
- J51 _ J47 350 345 -
M5fV ,...,,......:,.m STEAM LlrE mv u v g l v T [ O u
ll...... ,
.- s W ,
u (1
352 .
m it
-l'.Nl
}.
- l . .l=
125 353 COVERNOR 360 tg %
VALVES '
M2 361 i
Figuir A 7. RELAP5 Davis-Besse model; OTSG A and steam line.
A-10
c _-i j .,
't.. ;
l
-215
- .:.U:
(h :-
FROM AFW l 570 "
L
- I ll . 424 l ,
426 , ,.
a A
- n n >
+ W N a
- n y "..
d: '
- iii iii .
428
- a 457
- 781 G ;;;.
2:- E ' :::
d win W
FEEDWATER . _
456 AEV PLET a 6, . 220
- 8" ' G
- 451 __
447 445 -e---- . ';
450 ,
MSiv ..
STEAM LPC ,,
454 gy ,
V : s n .
ll . . . . . .,
> a
- y -
S 8 ',
m .
452 ,.
406 >.... . .
_- Ed:
.s ii
., . .:+,:-
.::t::
1 225 460 m 453 COWHNCR vu.vts
'82 461 i i
1 i
Figure A-8. RELAP5 Davis-Besse model; OTSG B and steam line.
A-11
l of the OTSG are modeled, including the downcomer, anticipatory reactor trip system (ARTS), and steam and tube bundle region, and steam annulus. AFW enters feed rupture control system (SFRCS). These control near the top of the OTSG through eight injection ports systems are described in greater detail below.
which are azimuthally located around the outer edge The RELAP5/ MOD 2 model of the Davis-Besse ICS of the tube bundle. Experiments indicate that the AFW represents the following subsystems: unit load demand wets only the outer rows of tubes. As the AFW falls, development subsystem, integrated master subsystem, it penetrates radially inwards, but wetting at most 10% steam generator feedwater control subsystem, and the of the tubem The bundle region below the AFW injec. reactor control subsystem. Figure A-9 is a schematic tion ports is divided into two radial regions to approx. of the ICS organization and presents an overview of imate -
"o-dimensional wetting behavior of the the ICS functions. The borate control subsystem and AFW. The outer radial region contains 10% of the the non-nuclear instrumentation system are not tubes, while the inner region contains the other 90%. represented. The ICS model is based on information Each radial region is divided axially into eight control obtained from Babcock and Wilcox (B&W) and Davis-volumes. The AFW is connected to the outer radial Besse personnel, plant calibration data, detailed region. The two regions are connected radially with schematics of the subsystems, analog and digital logic crossflow junctiens, allowing AFW to penetrate radial- drawings, and Bailey Meter Company detailed descrip- .
ly if calculated by the code. The OTSG tubes are also tions of the individual modules. 2 divided into two separate heat structures, representing The ICS modules and relays are modeled individual-10 and 90% of the tubes. The primary side of the ly to provide the greatest amount of flexibility for future OTSG is modeled with a single channel. Separate- analysis requirements. Additional control variables are effects calculations performed during the development included in the model to allow the analyst the ability j of the model indicated that two primary channels were to impose false signals during a calculation. For ex- I not needed for most applications, including natural cir- ample, a steam generator level signal can be failed to culation and small-break transients. Although the zero interactively to simulate a failed level transducer.
{
multidimensional mixing of the AFW on the secondary Display parameters and display options available to the '
side caused significant variations in the behavior of the operator are also available to the analyst during interac-wetted and unwettcJ tubes on the primary side, the tive execution, overall OTSG performance was not significantly dif. The RELAP5/ MOD 2 kinetics package is not used ferent when two primary channels were modeled in. in the Davis-Besse model. Consequently, reactor con-stead of one. Heat structures are used to represent the trol rod positioning is not directly coupled to the reac- l stored energy in the secondary shell wall, tube bundle, tor power. Instead, the reactor power is controlled by and cylindrical baffle. The heat transfer hydraulic general table reference. Reactor kinetics will be incor-diameter on the secondary side of the tube bundle is parated at a later date, as the need arises.
based on the minimum tube-to-tube spacing. This The pressurizer pressure control system is modeled hydraulic diameter improved the thermal performance through the representation of pressurizer heaters and of the OTSG and was recommended in the assessment spray. Design data on the pressurizer level control of RELAP5/ MODI.^4 system were not available during the development of The main steam lines models are also shown in the model. Instead, a simple model which controlled Figures A-7 and A 8. The model represents the region the net makeup into the reactor coolant system based from the outlet of the OTSG to the turbine governor on the pressurizer level was developed. The net make- l valve. The safety relief valves (SRVs), atmospheric up represented the combination of makeup and let- l exhaust valves (AEVs), turbine bypass valves (TBVs), down, with the net flow added to the A I cold leg pump l main stcam isolation vahes (MSIVs), and check valves discharge. In the plant, letdown is taken from the B1 1 are modeled. 'Ihe turbine stop valve and the turbine cold leg pump suction, but the model approximation governor valve are combined into a single valve in the i3 thought to be adequate for most applications. The model. Each AEV and TBV is sized to pass 74.1 kg/s capability to model zero, one, or two makeup pumps (163.3 lbm/s) and 185.2 kg/s (408.3 lbm/s), respec. and minimum, normal, or maximum letdown, in any tively, of steam at 6.2 MPa (900 psia). Heat structures combir.ation, was developed, are used to represent the steam line piping. The model represents the ARTS and SFRCS. Reac-tor trip is modeled based on high power, high reactor A-4. Controf Systems P"""" ""*"1Peatum, pown-to40w ratio, mactor pressure versus temperature, RCP trip. turbine trip, Many of the plant control systems are represented. SFRCS actuation, or manual trip. SFRCS is actuated These control systems include the integrated control based on low steam pressure, low feedline differen-system (ICS), pressurizer pressure control system, tial pressure, low or high OTSG level, or reactor A-12
_ E B C S E A BE F GVF F V SL R S GL 4
WVTAE SA F V L N RO I OR TT AN F RO EC S N R E
E E GT S A C S MA P D P AW F f F ED W MR E E M TE i F UE S FU PT P SE F N I
- F 1 E A E C S A AEV GV F F M SL R GL E R A E S SA T D E WV F T V S N T N Y A S I S M A H E M L C D O T V D V DE F R A i A S T A S T P O N IS L R O D O C E T E C S D
I R S A N T DEV F F DE A U N O F OIV O T I R 5 RR S
I L C R E D A D T N
E R I E
SM C ES A NS S E I BA F 'AT F F R P L 4 PS R S U Yy O
VY E T R
BS TN T I N
O F C A E R
NNA NN R F I
BR P F I B R F RE 4 R VE ER A UV U TO TOT GIN G
_ l
>b
.9 f )
x !
i coolant pump trip. The model determines the correct alignment of AFW Jascd on the the type of SFVCS A-2. C. D. Fletcher et al., RELAPS Therinal- ,q,l Hydraulic Analyses of Pressurized D:ermal ;
actuation. In event of a rupture of the steam or feed Shock Sequencesfor the Oconee-1 Pres wi:ed F'>,
lines, SFRCS isolates the OTSGs and aligns AFW inte WaterReactor, NUREG/CR-3761, FGOD310, 'lt the unaffected OTSG. June 1984. ? '
A-3. P. D Wheatley et al., REL4PS/ MOD 2 Code I Assessment at the Idaho National Engineering A-5. References Laboratory, NUREGICR-4454, EGG-2428, March 1986.
A-1. V. IL Ransom et al., RELAPS/ MOD 2 Cole A-4. S. L. Thompson and L.N. Kmetyk, RELAPS 1 Manual, Volumes 1 and 2 NUREGICRh,312, Assessment:LOFTLarge Break L2-5, NUREGI , }
EGG-2396, August 1985. CR-3608, SAND 83-2549, February 1984. j l
l 1
l l
l l
t i
A i
\
l i
i i
7
- h
'k
, A-14 t 1
i i ) l A , 4 i'
i f j f' a . .
) ' l
,F' }! I i
I 4
, t {
it l APPENDIX B COMPUTER RUN TIME STATISTICS c
a
\
?
'/
h' 5 l
1 f
'\
B-1 "L .. - - . .
APPENDlX B COMPUTER RUN TIME STATISTICS
' Table B-1 presents a summary of the computer run performed with version 36.04 of the RELAPS/ MOD 2 time statistics of the RELAPS calculations performed computer code.
at the Idaho National Engineering Laboratory (INEL) The feed and bleed calculation executed' more during this task. These calculations were described in quickly than the Davis Besse LOFW transient Sections 3.1 and 3.3 of the main body of this report calculation. The code was able to take larger time i and represented the loss-of-feedwater (LOFW) steps after the reactor coolant pumps were tripped in
~
transient that occurred at Davis-Besse on . lune 9,1985, the feed and bleed calculation. The code per-and a total IDFW followed by feed and bleed. The formed reliatiiy for both calculations. No code computer used to perform the calculations was the execution failures were encountered in either Cyber 176 at the INEL. The calculations were calculation.
. Table B-1. Computer run time statistics LOFW Feed and Bleed Parameter Calculation Calculation Number of volumes (C) 201 201 Number of heat transfer surfaces 291 291 Transient time, s (RT) 765 5800 Total CPU time, s (CPU) 3469 17290 Number of time steps (DT) 10263 50937 CPU /RT 4.53 2.98 (CPU x 10)/(RT x C) 0.23 0.15 (CPU x 1.E6)/(RT x C x DT) 2.20 0.29 (CPU x 1000)/(C x DT) 1.68 1.69 B.3 ,
i .- ,!
g..-7. ._-.__ _ _ . _ - - -
.,. ;. 3 i
Ik l
L /
I r'
APPENDIX C 1
CALCULATION UNCERTAINTY >
)
i I
>; 1 I
l i I .
i t
C-1
. - - - . . .~- . - _ _ _ _ _ _ . - _ _ - _ . -
1-APPENDIX C L
CALCULATION UNCERTAINTY l
form ine R586 feed and bleed calculation. In general, C-1. Introduction the uncertainty of a computer code for calculadng feed Several potential sources of uncertainty were iden- and bleed transients should ideally be determined by tined which contributed to the overall uncertainty in an assessment pregram which involves comparisons l the feed and bleed calculations described in this report. of calculations and data from a wide ranFe of scale and The ' potential sources of uncertainty include the test parameters. Unfortunately, however, only limited thermal-hydraulic computer code used to make the data and assessment calculations exist for feed and calculation, the code input model, the initial conditions bleed in Babcock and Wilcox (B&W) geometry. One at the start of the calculation, the boundary conditions directly applicable assessment calculation, utilizing applied during the calculation, the code user, and the feed and bleed data from Once-Through Integral assumed transient. A review of the calculations and System Test 220899,c2 indicated that RELAP5/
data presented in the main body of this report helped MOD 2 provided an excellent representation c.' the identify important phenomena relative to the feed and observed trends. The deviations between calculated and bleed process. These phenomena included the thermai- measured RCS pressures, RCS temperatures, and reac-hydraulic conditions at the start of feed and bleed, prin- tor vessel liquid levels were generally less than cipally the reactor coolant system (RCS) temperature. 0.7 MPa (100 psi),5 K (9'F), and 1 m (3 fti, respec-
. and the boundary conditions during feed and bleed. The tively. Although this test was not scaled to tha Davis-Besse feed and bleed transient described in Section 3.3, important boundary conditions during feed and bleed
. includea core power, makeup and high-pressure injec- the assessment results described above are thought to tion (HPI) flow, and Gow through the pilot-operated be representative, relief valve (PORV). The most impertant measures of Additional data comparisons have been performed the success of feed and bleed were the ability to which provide an assessment of the ability of RELAP5/
depressurize and the minimum collapsed liquid level MOD 2 to calculate some of the important phenomena in the reactor vessel, occurring in a feed and bleed transient. Feed and bleed The uncertainties in the calculation of the significant transients exhibit relatively simple thermal-hydraulic thermal-hydraulic phenomena due to the potential processes, involving dryout of the once-through steam sources of uncertainty were assessed, quantitatively generators (OTSGs), heatup of the RCS, initiation of when possible, as described in Section C-2, The feed and bleed, dreining of the RCS dewn to a estimated uncertainties in the important parameters are minimum level, and refill. The ability of the code to summarized in Section C-3. References are presented predict the phenomena of OTSG dryout and RCS in Seelion C-4. Unless otherwise stated, the uncertain- heatup was demonstrated in the calculation of the l
I ties apply to the R586 calculation performed at the Davis-Besse LOFW transient of June 9,1985, as Idaho National Engineering Laboratory (INEL), as described in Section 3.1 of the main body of this described in Section 3.3 of the main body of this report. The ability of the code to predict RCS report. draining and refill in pressurized water reactors (PWRs) with U-tube steam generators has been c
C-2. Evaluation of Potential demonstrated in assessment calculations -3 of smali-
- "k i 55**""' ""'idents. Calculated liquid Sources of Uncertainty levels were generally withm I m (3 ft) of the The potential sources of uncertainty in the feed and corresponding data. Assessment calculations of feed j;
bleed calculations include the thermal-hydraulic com- and bleed experirnents applicable to U-tube steam
(- generator PWRs have also been performed.c-4,5 These l puter code, the code input model, the initial conditiom and boundary conditions, the code usn, and the feed and bleed calculations, although not directly assumed transient, The evaluation of the uncertainty applicable because early versions of the code were in the feed and bleed calculations due to each of these used, indicated that RELAP5 was generally able to potential sources of uncertainty appears in Sec- predict the trends observed in the experiments.
tions C-2.1 through C 2.5. Consequently, RELAPS/ MOD 2 should adequately represent feed and bleed transicats in D&W PWRs. The C-2,1 Thermal-Hydraulic Computer Code The uncertainty inherent in the code for calculating feed c
RELAP5/ MOD 2 computer code " was used to per- and bleed transients, although difficult to quantify, is C-3
not thought to be large compared with other sources Part of the uncertainty associated with the input of uncenainty. model is caused by the nodalization used. The nodaliza-Another method for estimating the uncertainty of a tion of several different components has the potential computer code is to compare the calculated results from to cause uncertainty in feed and bleed calculations.
one code with those from another code. The variation These components include the OTSGs, reactor vessel between the calculated results provides an indication upper plenum and upper head, heat structures, hot legs, of code uncertainty. Sections 3.2 and 3.4.2 present and reactor coolant pumps (RCPs) and RCP suction comparisons of calculated results from the legs. A discussion of the possible uncertainty due to RELAP5/ MOD 2 and TRAC-PFl/MODic-6 computer the nodalization of these components follows.
codes. The comparisons showed that the codes obtained The OTSG nodalization is not thought to significantly similar macroscopic results in that the feed and bleed contribute to the uncertainty of the R$86 feed and bleed operations successfully cooled the core. However, the calculation. The most important OTSG parameter is codes calculated significantly different trends, par- the initial liquid inventory. Since the OTSGs dry out ticularly with respect to RCS pressure. The differences relatively quickly, and since the total liquid inventory between the calculations were attributed to errors in must boil away eventually, the OTSG nodalization is the boundary conditions. The differences in the bound- not significant for feed and bleed calculations. Of ary conditions were thought to overwhelm differences course, the OTSG nodalization might significantly af-between the codes. Consequently, the differences in fcct those calculations in which the OTSGs actively the RELAP5 and TRAC results were not indicative of remove heat throughout the transient.
the uncertainty in either code. Both codes are thought Calculated reactor vessel liquid levels can be sen-to have the capability to adequately represent feed and sitive to the nodalization of the upper plenum and up- !
biced transients. per head. As shown in Figure 35 of the main tuly of this report, the R585 model predicted liquid levels that C-2.2. Input Model. The uncertainty of the cal- were about 1.5 m (5 ft) higher than the R586 model. !
culations due to the input model primarily arises from About half of this difference was attributed to the dif-two different concerns. These concerns include the ac- ferent upper plenum models used in the calculations.
curacy with which the model reflects the geometry of The R586 model explicitly represented the small holes 1 the plant and the adequacy of the nodalization for the in the plenum cylinder at the hot leg nozzle elevation, !
feed and bleed transient. while the R585 model did not. Modeling these holes In general, the R586 RELAP5/ MOD 2 model is allowed a more accurate representation of the flow believed to accurately represent Davis-Besse. The paths in the upper plenum, the draining from the hot detailed information used in the development of the legs, and the mixture level in the vessel. The upper model and the quality-assurance procedures yield a plenum model used in the R$85 calculations thus high degree of confidence that the model adequately caused the calculated liquid level to be about I m (3 ft) represents Davis-Besse. The agreement between data too high for certain transients. Section 3.4.2 explained from the Davis-Besse LOFW transient ofJune 9,1985, that the reactor vessel liquid levels could be significant- t and the corresponding calculation (see Section 3.1) fur- ly different, depending on whether the upper head was ther increases the confidence in the basic geometry, modeled as a flow-through or a stagnant region. It is the response of the integrated control system, and the not known which model generally best represents the applied boundary conditions. Thus, the uncertainty in upper head. Although the uncertainty in level caused the feed and bleed calculations due to uncertainty in f
by t.he upper head model can be as large as the upper the basic input model is thought to be negligible com- head height, about 2 m (6 ft), the uncertainty appears pared to other sources of uncertainty. However, even only for these transients in which feed and bleed was with the overall confidence in the model, the model initiated at relatively low RCS temperatures. Since the does have limitations (see Appendix A), primarily minimum li'luid level remains at or above the bottom related to the lack of information during the develop- of the upper head for these transients, the uncertainty ment of the model. The uncertainty in the R585 calcula- is not significant relative to cooling the core during feed tions performed by the Nuclear Regulatory Commis- and bleed.
sions', Office of Nuclear Reactor Regulation (NRR) Some of the uncertainty in a feed and bleed calcula-due to the geometry of the input model is also thought tion is due to the model of the RCS heat structures.
to be small. As described in Sections 3.2 and 3.4.1, The RCS heats up after the OTSGs dry out. The the R585 calculations were generally in good agree- coolant heatup rate is slowed by the heat structures, ment with the R586 calculations except for differences which absorb energy and heat up along with the associated with boundary conditions of core pcmcr, coolant. For example, after the OTSGs dried out in pirssuriier spray, and feedwater flow. the R586 calculation of the Davis-Herse LOFW C-4
transient, the heat t;tructures absorbed energy at a rate (Figure C-2). The change in hot leg nodalization af-equivalent 20% of the core decay power. The im- fccted the timing of events by less than 80 s. Thus, portant parameters in the heatup of the heat structures the uncertainty in the calculated feed and bleed results are mass and specific heat capacuy. If the heat struc- due to hot leg nodalization is insignificant.
tures are neglected, the calculated heatup rate will be The purpose of the feed and bleed operation is to too large. Since all the major heat structures were cool the core in the absence of a secondary heat sink.
represented in the R586 model, the uncertainty in the Keeping the core covered with liquid assures that the heatup rate due to the heat structures was thought to core will be cooled. The core is not in da .ger of un-be relatively small. covering when the mixture level is above the The R586 feed and bleed calculation showed thet the pressurizer surge line, flowever, in the R586 feed and final RCS depressurization was coupled to the increase bleed calculation, the RCS was losing mass and could in void fraction at the PORV. This increase in void not depressurire with liquid Howing through the fraction occurred when the mixture level dropped PORV. After the hot leg mixture level dropped below helow the pressurizer surge line, allowing steam to pass the surge line and steam passed through the PORV, out through the PORV. The hot leg nodalization af- the RCS depressurized but continued to lose mass, and fccts the hot leg mixture level and the void fraction the mixture level approached the top of the core. The passed to the surge line. Consequently, a sensitivity mixture level decreased until the makeup Dow ex-calculation was performed to investigate the effect of ceeded the flow out the PORV. For the feed and bleed hot leg mxialization. The sensitivity calculation was operatien to be successr ut, the reactor vessel mixture restarted from the base calculation, described in Sec- level should remain abw the top of the core until the tion 3.3, at 1600 s. There was no significant voiding RCS is depressurized far enough so that the makeup in the hot legs at this time. The A loop hot leg was and IIPI systems can maintain liquid inventory.
renoded by reduci.ig the length of the volume con- The minimum liquia level in the vessel depends on nected to the surge line from 6 m (20 ft) m 1.5 m (5 ft) the minimum liquid levels reached in the OTSGs and while keeping the total number of volumes constant. the loop seals of the RCP suction piping. The Davis-The calculated results were msensitive to the hot leg Ilesse RCPs contain a weir which determines the mxlalization. The nodalization had only a small effect minimum level in the loop seals. The liquid i i the RCPS on hot leg liquid level (Figure C-1) ad RCS pressure and OTSGs above the weir should drain into the vessel 25 - ' ' -80 o GASE CALCUL ATION N OOOOGvo,gn .... SEN ;lTlVITY CALCULATION
-- SURCE LINE ELEVAflON 20 - + -
- -60 m n
5 D 5-15 -
E I h -40 -
.T 10 -
.f ___________________
f 20 5 -
c c-0 O O 2000 4000 6000 Time (s)
FiFure C.I. The effect of hot leg nodalization on collapsed hquid level in tne A hop hot leg.
C-5
18 i , 2600 !
1 I i
'2400 j l 16 -
m n 2 2200 .
S $
- 14 -
o g , -2000 g E V E j 1800 12 BASE CALCUL AllON _
--- SENSITIVI TY CALCUL ATION j 1
1600 I I
10 j 0 2000 4000 6000 l Tirne (s) l l
Figure C-2. The effect of hot leg nodalization on RCS pressure. I and cool the core. The liquid below the weir should the important parameters were identiEcd. Second, the rem in trapped in the loop seal and be unavailable to uncertainty in each of these initial and boundary con-cool the core. The elevation of the weir was not ditions war, estimated. Third, the uncertainty in the ,
available during the development of the model and thus initial and boundary conditions was propagated linearly was not incorporated into the model. The RELAP5
)
to determine the total uncertainty in the important '
model has the potential to drain the loop seals down parameters. The method for determining the uncertain-to the bottom of the RCPs, allowing too much water ty and the results obtained are described in more detail j
to drain into the vessel. Thus, the calculated minimum below. ;
vessel liquid level could be too high when the level The initial and boundary conditions which had the approaches tt e top of the core. The uncertainty in the potential to affect the RCS temperature and the minimum reactor vessel liquid level due to the RCP ]
depressurization are listed in Table C-i, along with !
nodalization was estimated to be about 0.3 m (1 ft) in estimates of uncertainty. The uncertainties represent the R586 feed and bleed calculation. estimates of two standard deviations. The probability was thought to be about 95% that the true value of the C-2.3 initial and Boundary Cor Jitions. The initial or boundary condition was within the stated RCS temperature at the start of feed and bleed can have uncertainty of the calculated or input value. The j u large effect on the transient response, as discussed estimates of uncertainty were generally subjective and '
in Section 3.4.2. Operator guidelines may also direct were based on engineering judgment. The uncertainty that feed and bleed cooling be started when the hot leg in the core power was divided into three compo-reaches a certain temperature. Thus, the RCS temper- nents: those associated with the initial power, the ature at the start of feed and bleed is an important decay power (including actinidesh and the fission l parameter. The ability to depressurize is also an im- power produced after reactor trip. The uncertainty in i portant parameter because without depressurization the the fission power was thought to be larger than the core could uncover and heat up while the RCS pressure uncertainty in the other two components. The uncer-remains at the open setpoint of the pressurizer safety tainty in the initial OTSG liquid mass was based on relief valves (SRVs). The effect of uncertainty in the a calculation of the variation in mass due to tube foul-initial and boundary conditions on these two impor- ing.. The uncertainty in the PORV flow was probably tant parameters was determined in the R586 feed and less than 10% for single-phase thermodynamic states bleed calculation as follows. First, the in;tial and upstream of the valve because of the available test data.
tmundary conditions which had the potential to affect The estimated uncertainty was increased to 20% to C-6
Table C-1. Uncertainty in initial and ' the OTSG dryout time, its ultimate effect on RCS boundary conditions temperature was minimal. The uncertainty in the calculated OTSG dryout time due to uncertainty in the Uncertainty . initial and boundary conditions was calculated as Parameter (%) described above. The OTSG dryout time varied be-tween about 50 and 240 s, while the OTSGs dried out Core power at 130 s in the R586 calculation. Thus, the uncertain-Initial power 2 ty in the calculated OTSG dryout time was almut 100 s.
Post trip fission power 35 The major contributor to the uncertainty in the OTSG uccay heat 5 dryout time was the initial OTSG liquid mass.
The analysis of the R586 calculation indicated that PORV flow 20 the parameters which most significantly affected the Makeup flow 10 RCS temperature at 20 min, the time when feed and initial OTSG liquid mass 14 bleed was initiated, were the time of OTSG dryout and Post-trip feedwa:er flow 100 the net power added to the reactor coolant after OTSG dryout. The net power included core decay power, initial stored energy in fuel 10 RCP power, and the power absorbed by heat struc-RCP power 10 tures during the RCS heatup. The uncertainty in the RCS heat structures 10 calculated RCS temperature at 20 min due to the uncer-tainty in each individual contributor (Table C-1) is shown in Table C-2. The largest contributions to the uncertainty in RCS temperature were associated with account for the effects of two. phase flow through the the OTSG dryout time and the core decay heat. The PORV. combined (total) uncertainty in the calculated RCS The total uncertainty in the important parameters (the temperature due to the individual uncertainties shown RCS temperature and the ability to depressurize) was in Table C-2 was 5 K (9 F). The corresponding uncer-obtained t- combining the effects due to the uncer- teinty in the calculated heatup rate was Il%. Based tainty in ti.e individual contributors (the initial and on the average heatup rate, the uncertainty in the time boundary conditions). The effect of a variation in each at which a given temperature was reached was about individual contributor by its uncertainty on an impor- 2 min. The above uncertainties are also thought to be tant parameter was estimated with a hand calculation. applicable to the R585 calculations. As described The change in the important parameter from the code- previously, the R585 and R586 calculations were calculated result was assumed to be a linear function generally similar prior to the initiation of feed and of the individual contributors. The individual con- bleed.
tributors were assumed to be independent, normally The depressurization of the RCS is an important distributed random variables. The total uncertainty in parameter in feed and bleed. The R586 feed and bleed the code calculated result was then determined with a standard statistical formula: the square root of the sum of the squares of the uncertainty due to each individual Table C-2. Uncertainty in RCS contributor. Although the above method was not en- temp'erature tirely rigorous or statistically justified, the method is thought to provide a reasonable, but rough, estimate Uncertainty of the actual uncertainty. Parammr S) ('F)
The RCS temperature at the initiation of feed and - -
biced was primarily determined by the time of OTS9 initial core power 0.8 1.4 dryout end the heatup rate after dryout. The analysis Decay heat 1.9 3.5 of the R586 feed and bleed calculation indicated that the parameters which primarily determined OTSG OTSG dryout time 4.4 8.0 dryout were the initial liquid mass in the OTSGs, the RCP power 0.9 1.6 amount of feedwater delivered to the OTSGs after reae-tor trip, the core and RCP power, the stored energy RCS heat structures 0.8 1.4 in the fuel, and the OTSG pressure and temperature response as controlled by the atmospheric exhaust valve (AEVL The effect of the AEV was neglected in the Total 5.1 9.1 estimate of the uncertainty because, although it affects I
C-7 ;
1 l
calculation indicated that the RCS would depressurite Wp= PORV mass flow rate, and when the void fraction at the PORV reached a high enough value. Hand calculations were performed to V p = specific volume of the fluid at the PORV.
estimate the uncertainty in the calcdated depressunza-tion. The analysis of the feed and bleed calculations The above equations are valid for quasi-steady flow described in Section 3.4 indicated that the key bound- in which all the makeup flows past tt.e core. Some of ary conditions affecting the depressurization were core the core power goes into heating the cold makeup to power, PORV flow, and makeup flow, the saturation temperature. The remainder of the power l A simple, quasi-steady volume balance was used to boils liquid to steam. Note that the subcooling of the j estimate the uncertainty in the calculation of whether makeup is assumed to reduce steam productiou in the l or not the RCS should depressurize. The parameters core rather than condense steam. However, the net I which act to compress, and thus pressurize, the RCS steam production is identical if the makeup is assumed I include makeup How and steam production due to boil- to condense steam rather than reduce steam produc-ing in the core. The parameters which act to depres- tion in the core. Also note that the expansion of makeup surize the RCS are the How through the PORV and due to heating is accounted for in Equation (C-2) by the condensation of steam due to the cold makeup. The multiplying by the specific volume of saturated liquid RCS volume balance can be summarized mathe- rather than of the cold makeup. The volumetric flow matically as out the PORV is based on the PORV How area and )
the fluid state at the PORV. For the volume balance, ,
Q = Qm + Oc ~ O p (C-1) the PORV area was increased by the effective area of l the HPVVs. The PORV flow was calculated with the where Henry-Fauske and homogeneous equilibrium critical
""***'# # "" # "" I* "" *'
Q = net volumetric flow' !
respectively. The fluid state at the PORV was handled Q, = volumetric flow of makeup into the RCS, parametrically rather than calculated directly. )
A va atbn oW dmpk dume balance was pm Q = volumetric production of steam in the core, nna comparison with the results of the R$86 feed and and bleed calculation. The RCS pressure, core power, Qp= volumetric How out the PORV. and makeup flow from the R586 calculation at the time that the RCS depressurization began were used to The sign of Q detennines if the RCS will depressurize evaluate Equations (C-2) and (C-3). Equation (C-4) or pressurire, and the magnitude of Q is roughly pro- was evaluated as a function of void fraction at the portional to the rate of pressure change. The individual PORV. The volume bJance, Equation (C-1), predicted terms in the volume balance are computed as that the RCS should depressurize when the void frac.
tion at the PORV exceeded 0.82. The RCS actually l Qm
- V/mVI (C-2) began to depressurize in the R586 feed and bleed calculation when the void fraction reached 0.78. The Oc = [P - Wm (h r - hm )] vg/hg (C-3) agreement between the volume balance and the cale calculation was considered excellent given the simple Qp =Wvp p (C-4) assumptions of the volume balance. The small dif-ference between the volume balance and the code where calculation was probably caused by the use of different critical flov' mmlels, which provided similar but not W, = makeup mass How rate, identical results. The volume balance was also able to vr = . predict the key results of the R$85 and TRAC calcula-specih.e volume of saturated h. quid, t ons. In particular, the volume balance correctly P = core power, predicted that the R585 calculation described in Sec-hr = specific enthalpy of saturated liquid, themmation
. b '. *""f
"" """" b I""
feed and bleed, while the
- " "E R586 calcul o
hm = specific enthalpy of makeup, tion, which had 25% higher power, would not. The volume balance also correctly predicted that the TRAC vg = difference in specific volume between !
saturated h, quid and gas, calculation illustrated in Figure 36 of the main body of this repon would not continuously depressurize after I hg a difference in specific enthalpy between the initiation of feed and bleed because of the smaller saturated liquid and gas, PORV How in this calculation (see Section 3.4.2). j C-8
The volume balance was used to determine the sen- volumetric flow occurs at a quality near zero. The max.
sitivity of the calculated depressurization to variations imum volumetric flow occurs when the quality is unity in core power. mgure C.3, which wmmarizes the and dry, saturated steam flows through the PORV.
results of the vv.ume bal- vs the volumetric Superheated fluid states are not shown in the figure, flow through the PORV, Q,,(Equation (C-4)), versus since they imply core uncovering in a quasi-steady quality. Quality, x, was evaluated as analysis.
Figure C-3 shows that the PORV cannot depressur-x = (h - h i )/h g (C-5) ire the RCS for times less than 30 min after reactor trip, regardless of the fluid state at the PORV. The where h is the specific enthalpy of the fluid at the compressive term decreases with time after reactor trip PORV. Equation (C-5) yields negative values of as the decay power and the steam production in the qualities for subcooled thermodynetic fluid states, core decrease. At about 37 min after reactor trip, the Figure C-3 also show s the comEm%ompressive term, RCS will depressurize if dry, saturated steam flows which is the sum of the maaup, Qm, and core boil- through the PORV. At 60 min, about the time depres-ing, Qy, terms [ Equations (C.2) and (C.3)). The core surization occurred in the R586 feed and bleed calcula-power was varied parametrically as a function of time tion, the RCS can depressurize if the quality exceeds after reactor uip from 100% rated power. The results 0.65. The RCS will depressurize with almost any fluid shown on the figure were obtained at a pressure of at the PORV at 120 min after reactor trip.
17.2 MPa (2500 psia), which is close to the pressurizer The sensitivity of the calculated depressurization to SRV setpoint pressure, and assumed both makeup uncertainty in the feed and bleed boundary conditions pumps were available. is illustrated in Figure C-4. The figure shows the best-When tN volumetric flow through the PORV ex- estimate PORV flow and makeup plus core boiling ceeds the flow due to the combined effects of makeup terms at 60 min after reactor trip from Figure C-3.
and core boiling, the PORV can depressurize the RCS. This time was selected because it was approximately When the PORV flow is less than the compressive the time that depressurization occurred in the R586 feed term, the RCS will pressurize until the SRVs open. and bleed calculation. The figure also shows the best-The volumetric flow through the PORV is a strong estimate PORV flow reduced by 20% and the makeup function of the quality at the PORV. The minimum and core boiling terms resulting from a 5% increase 0.5 i , i , i x PORV (Op)
-- - MAKEUP + CORE DOILING (Om +O) c 15 n 0.4 - n m
m p T - 10 minutes )
C
{
, 0.3 -
0 5
o O
C .....................T..=.20..m..inu.t.e.s.....................................................
. .. =
T = 00 minutes
,o ...................................... ....................... ......................... ,o b 0.2 -
b c) T = 60 minutes tu
........................................ ................... ....................... E g .?
5
$ T = 120 minutes [
0.i ,................. ... ..
4
' ' ' ' ' 0 0.0
-0.50 -0.25 0 0.25 0.50 0.75 1 i Quolit y, x i Figure C.3. Results of the volume tulance for different imies after reactor trip.
1 C-9 I l
l 0.5 , i , , i x BEST CSTIMATE PORV FLOW BEST ESTIMATE MAKEUP + CORE BOILING
~
a INCREASED DECAY POWER -15 1 m 0.4 ~
o REDUCED MAKEUP
^ j n( 0 REDUCED PORV FLOW $
n I 3
E, O I v I a 0.3 - -
o -
10 o c c U U {
{e 0.2
,/'
j e
j
-5 2 O o
> o,j _ ts-.~ .
,,,,g 0.0 0
-0.50 -0.25 0 0.25 0.50 0.75 1 Quolit y, x Figure C-4. The effect of uncertainty in boundary conditions on the results of the volume balance.
in core decay power and a 10% decreac in makeup duction in the core. Thus, the figures shown, which flow. These variations, which correspond to the were developed for the max; mum possible pressure, estimated uncertainties shown in Table C-1, were corresponding to t.he SRV setpoint pressure, represent selected to make depressurization more difficult. a worst case for depressurization. It should also be The effects of the uncertainty in decay power and pointed out that makeup flow is crucial for the sue-makeup flow were relatively small. The volume cess of feed and bleed in Davis-Besse. Without any balance indicated that the PORV could still depres- makeup flow, the volume balance indicates that PORV i surize the RCS at 60 min with either the higher core would not be able to depressurize the RCS until ap-power or the lower makeup flow. However, a slightly proximately 140 min after reactor trip. By this time higher quality at the PORV would be required before in the R586 feed and bleed calculation, the mixture the depressurization could begin. It was estimated that level probably would have dropped below the top of if the core power had been increased by 5 % in the R586 the core and a core heatup would have begun.
feed and bleed calculation, the depressurization would Although the volume balance is useful for under-have been delayed by about 100 s. The uncertainty in standing the results of code calculations and determin-the volume balance due to the PORV flow was larger ing the uncertainty in the calculated depressurization, than due to the uncertainty in the other two parameters. the method has limitations. In particular, the depres-However, the PORV could still depressurize the RCS surization time required until RCS refill begins and the once the quality at the PORV approached unity. Thus, minimum liquid level in the vessel are not obtained.
the uncertainties in the boundary conditions of the R586 Thus, even though the volume balance predicts that feed and b%ed calculation do not have a large effect depressurization will occur, there is no guarantee that on the calculated ability to depressurize. The calculated the core will remain covered as the RCS depressurizes.
depressurization appears valid and would not be ex- The prediction of depressurization by the volume pected to vary significantly because of uncertainty in balance is thus a necessary, but not a sufficient condi-the boundary conditions. The estimated uncertainty in tion, to guarantee the success of feed and bleed. !
the time of depressurization due to the combined uncer-tainty in the three boundary conditions is 5 min. C-2.4 Code User. The code user is also a poten-Although not shown in the figures, the RCS general- tial contributor to the uncertainty in a calcu ation.
ly depressurizes more easily at lower pressures. This However, the user affects the calculation primarily is primarily because as the RCS pressure decreases, through the selection of the thermal-hydraulic computer the makeup flow increases, which reduces steam pro- code, the code input model, initial conditions, and C-10
boundary conditions. Since the uncertainty related to regarding the initiating event and operator actions can these potential contributors was described previously, be much larger than the uncertainty in the calculations no significant, additional uncertainty was thought to due to the sources described in this report. The uncer-be related to the code user. tainties in this report only apply to the calculation as performed, given the assumed transient, equipment C-2.5 Assumed Transient. The assumed transi- performance, and operator actions.
ent can have a large effect on the calculated results.
The initiating event, equipment performance, and C-3. Uncertainty in Important operator actions strongly influence the course of a tran- Parameters sient. Thus, sigmficantly different results could be calculated if dif ferent assumptions were made relative The important parameters for a feed and bleed tran-to these parameters. Examples of parameters which sient were identified as the RCS temperature at the time significantly affect a feed and bleed transient are the feed and bleed was initiated, the depressurization dur-feedwater How and the makeup flow. If some feed- ing feed and bleed, and the minimum liquid level in water flow continues after reactor trip, sech as oc- the reactor vessel. The estimated uncertainty in the curred in the Davis-Besse LOFW transient of June 9, calculation of the important parameters is discussed 1985, the OTSG dryout can occur much later than was below.
obtained in the R586 feed and bleed calculation. A later The uncertainty in the RCS temperature when feed OTSG dryout delays the time when the RCS reaches .md bleed was initiated in the R586 calculation is a certain temperature and possibly the initiation of feed estimated to be 5 K (9"F). This uncertainty was caused and bleed. Conversely, if less feedwater is delivered by uncertainty in the initial and boundary conditions.
to the OTSGs than in the R586 calculation, highc RCS The largest contributor to this uncertainty was the initial temperatures result at the initiation of feed and bleed, liquid inventory in the OTSGs. The stated uncertainty For example, in the R586 feed and bleed calculation, bounds the deviations between the calculated and the initiating event, the main feed pump (MFP) trip, measured temperatures for the Davis-Besse LOFW was assumed to result in immediate turbine and reac- transient of June 9,1985 (see Figures 11 and 12 from tor trips. If reactor trip on MFP trip failed, the reac- the main body of this report). The stated uncertainty tor trip would be delayed until high RCS pressure or is also representative of the deviation observed between temperature conditions occurred. The reactor trip the assessment calculation and data from a feed and would then occur at a much lower OTSG liquid inven- bleed test in an experimental facility scaled to a B&W tory than in the R586 calculation. In this new transient PWR (see Reference C-2).
scenario, OTSG dryout would occur earlier, the RCS The uncertainties in the R586 calculation due to would be at a higher temperature, and would possibly initial and boundary conditions are also applicable to esen be two. phase when feed and bleed was initiated, the R585 calculations. However, the uncertainties in and RCS voiding and depressurization would occur the R585 calculations are larger because of the er-l more rapidly than was obtained in the R$86 feed and roneous 25% reduction in core power when the RCPs bleed calculation. tripped, as discussed in Section 3.4.1. The reduction The assumed behavior of the makeup system can also in core power resulted in a calculated heatup rate that hase a large effect on the calculateu temperature of the was about 25% too low after RCP trip. This error RCS at the initiation of feed and bleed. In the R586 caused a bias in the R585 calculations in addition to l calculation, no makeup Dow was assumed prior to the the uncertainty associated with the initial and boundary initiation of feed and bleed. If a more realistic scenario conditions. For the R585 calculations described in this had twen modeled. the operators would have manual- report, the error caused the calculated RCS temperature J ly started the second makeup pump, providing max- to be about 2 K (4*F) too low at the initiation of feed
( imum How until the pressurizer level was recovered. and bleed. The time required to reach the RCS In this scenario, the RCS temperatures at 2J min would temperature at which feed and bleed was initiated was have been about 5 K (9"F) lower than in the R586 about 1 min too long. The bias was larger in those calculation. Note that the variation due to the assumed R585 calculations, not described in this report, in makeup response was as large as the total estimated which feed and bleed was initiated at 37 min. For these uncertainty in RCS temperature due to initial and calculations, the errors in RCS temperature at the start i boundary conditions. If maximum makeup for the en- of feed and bleed and the time to reach this temperature j tire transient was assumed, the RCS temperature would were about 1I K (20*F) and 6 min, respectively.
hase been about 17 K (30*F) lower at the initiation The unecrtainty in the boundary conditions of core of feed and bleed than in the R586 calculation. Thus, decay power, PORV flow, and makeup How does not the sensitivity of the calculated results to assumptions significantly alter the results of the R586 calculation C-11
I i
relative to the depressurization of the RCS. The uncer- C-4. References ,
tainty in the time at which the RCS depressurization 1 began due to uncertainty in the boundary condition',
C-1. V. H. Ransom et al., RELAPS/Af0D2 Code was estimated to be 5 min. The calculated time of Afanual, Volumes 1 and 2, NUREG/CR-4312, depressurization did not appear to be sensitive to hot EGG-2396, August 1985.
leg nodalizatmn. The uncertainty m the depressuriza-tion in the R585 calculations was thought to be much C-2. J. R. Gloudemans, "OTIS Test Results," Dur--
larger for those feed and bleed calculations in which teenth li'ater Reactor Safety Research Informa-the RCPs tripped. tion Afecting, Gaithersburg, AfD,0ctober22-25, The uncertainty in the calculated collapsed liquid 1985, NUREG/CP-0072, Vol. 4.
level in the reactor vessel was estimated to be 1 m (3 ft). This result was based on the results of assess-C-3. P. D. Wheatley et al., RELAPS/Af0D2 Code ment calculations. In the R586 feed and bleed calcula- Assessnunt at the Idaho National Engineering tion, an uncertainty was identified relative to the RCP Ldioratory, NUREG/CR-4454, EGG-2428, j nodalization which could cause the calculated level to March 1986. !
be about 0.3 m (1 ft) too high. The uncertainty in the C-4. R. K. Byers, L. N. Kmetyk, RELAPS Asses.e level could also be larger because of the upper head ment: LOFT L9-1/l3-3 Anticipated Transient modeling for th(we transients in which feed and bleed with Afultiple Failures, NUREG/CR-3337, was initiated at relatively low RCS temperatures. The SAND 83-1245 R4, August 1983. j uncertainty in the R585 calculation may be slightly l larger because the small holes in the plenum cylinder C-5. D. J. Shimeck et al., Analysis of Primary Feed l at the hot leg nozzle elevation were not explicitly and Bleed Cooling in PIVR Systems, EGG- l modeled. However, the uncertainty due to the upper SEMI-6022, September 1982. I head and upper plenum models shoult' not affect the C-6. Safety Code Development Group, TRAC-PF// :
calculated result that the minimum collapsed liquid Af0Dit An Advanced Best-Estimare Computer 4 le\el remains abase the top of the core.
p,gg,g, fg, p,,,,u,;.ed IVater Reactor The above estimates of uncertainty in the important Usermal-llydraulic Analysis, NUREG/CR-3858, feed and bleed parameten, are valid for the transients LA-10157-MS, July 1986. ,
analyzed based on the assumed initiating event, equip- i ment performance, and operator actions. Significant. C-7. Aerojet Nuclear Company, RELAP4/Af0D5 A ly different results could be calculated if different Computer Programfor the Transient Thermal-assumptions were made regarding these parameters. Hydraulic Analysis of Nuclear Reactors and in particular, different assumptions regarding feed. Related Systems, Users Afanual, l'olume 1, water flow, makeup flow, and reactor trip could REL4P4/Af0D5 Description, Interim Report significantly alter the calculated results. SRD-113-76, June 1976.
f 1
1 l
C-12
..._.........,___o.
=,,_ NUREG/CR-49 6
BISU RAPHIC DATA SHEET
'"3"3g, 2i EGG-2510
- u. . .,.ucv.o . o . . .. ....
. ..a n...
. ,,,s....u.,,,s.
Davis-Besse Uncertainty S dy ........p... -[ v...
f
.Wo s. s Auaust p 1987
....o... . O g .,0.,i u.0
.. p n Cliff B. Davis Augustg I 1987
- i. c , . a cm .=_=. . i . u .
, ,. o. o.c.. .a . , ,o ... ..o ... ... .oo. . u ,,
EG&G Idaho. Inc. . . . o. 2 ..' ..
P.O. Box 1625 N No. A6827 Idaho Falls. ID 83415 j
...a.,,o...... .... c..oo.. u 4 ,c , p. ...o,. .o.,
.. o .o..
Division of Reactor and Plant Systems Technical Office of Nuclear Regulatory Research '""'oco""'*"-'"-->
U.S. Nuclear Regulatory Commission ;
Washington D.C. 20555 ,
, , u s. .. ... .o, o
,2 ... . . .c , ,m ,
The uncertainties of calculations of I of-feedwater transic s at Davis-Besse Unit I were determined to address concerns te U.S. Nuclear Regulat Commission eclative to the effectiveness of feed and ble ooling. Davis-Besse Unit S a pressurized water reactor of the raised 4oop Ba x & Wilcox design. A de d. quality-assured REl.AP5/ MOD 2 model of Dav' rse was developed at the Idaho ional Engineering Laboratory. The model was to perform an analysis of the loss-of- dwater transient that occurred at Davis-Bes n June 9,1985. A loss-of-feedwater tran. nt followed by feed and bleed cooling also calculated. The evaluation of uncertain was based on the comparisons of cal tions and data, comparisons of different calculatx of the same transient, sensiuvity, culations, and the propagation of the estimated unce ty in irutial and boundary co , sons to the final calculated results. .
FIN . . A6327-International Code Assessment: RELAP5 and TRAC-B v.
. .o.a. o. .c. ... oc ....,.,,
. . ooc o.. . . .. . . . ..
Unlimited
.. new.,e, cs.ui,ic.,,o.
f re.s . sped
...... ,,on. ..o. o .... Unel aMsi fied
, r. ,
unclassified
,, ... o ..cn
_ . _ ._o
b -
mn 4 3,.c @L:4 r.pb ru s , r mWlM'h i. ,: eE -^ . '; c Wl1;" s# ..-- :,
Glyn s;d %'e At ; n
.. - tqVm.. l.. 'a n, .i - . .yy a r. . &.*',
Q.,[u% s 4! u , zc ,.pf' " ,
k h., % r '.y.n MM y TQq m n. Jm& V s.73W+ ' .
'y ;M&f Q A- %; v oW YO th;.-
r-
', ,s;Q 961 (8'VR '% , . & QV.?"%g%:%
'qhD$l'?qf:y b vf9 QWpenf f'NJ Vqp Wyn & m!.y . ^", e yy i &f@.,@
W.:<- W.j?QN?. ~-r :M:&
C Qll? %y' ' A.Q.hw,,ypy ? H t W; m % % q%'> u, ,'wpq,.v'%54a" * ' QM814
- y:s _' .
- m % %
^a y.Q - g Q;, g g. , - ,
b e ,,.
' ' ~ ~ ' I y ; e m 7 .p s
k p ;;, g ( , L Q)um
. $$ y ; -
, Nygy}'j,Oh(,.1 ..
" WM k,d q.$W $,. c.$:lr j. i.. m, , f . .j 'i t. '[ p.
M 'chy( .'; . - 3 96:::3 %NJ $ps...J~i I A "C @f & . .h x; ;M. c ',' , 2.' 2,7'h,k$ h $$.'UN r . t T.; y,A. .& 9 ,, ~; At :.w .km,3. & -.y%, , ,h, v y,. v~.7 ,Qf ' H. <"w
.$.,h. .
w<, J; % d)a ' a p + ~ r- ' ) ;
.n ,.
.,w., s Ah .
s ;,,-
r: >v
/
i.u... , ..t 9
V y ee v.i 1
- .4 E x .. :q . Pj
(!N .a d u a i) M ,7 4hh ,) gk,,h5 f;.{[,.}f$. .
[(!yYH Ndc m:
i q O i a-3 e ( ('< 9m e . < -h. ,[i. y yap 9
. W ' l' Q4 .h .4 %Q. *8-bW'h '.! ' W. O .N d [?~Fb _
f ' l> lr,z ? ;; ? k lW.? '
, wh .
w
)
u -* M. T w 9.v: '.) gi f. & \' tv -i7
' [ '
- s g
s
' [. 5 : .s<
',j 1
, c, - ,j ' Qg wy,, a.t f 9 =M;ck 6.,J v e%D x - ,A, , N.n, N' et' M. t ;t ,
t ,, 1 - - 2 f' &" :A gW e.3 1+
- f. s;; !.ym, o
,e 1 y -
- rt r sv:w
+
7 e% e 'm -
\
- ~
,;v
, , fh, ; Y:vvt c' -
g y k ua...: e g y.y ve, d,V 'r k.e.Rm (, L;b m t su"t.
m i. ):;m':.a p JW .Q y y'y n .p ^ >: %w v4 e s n "s. W\ A /t'"t' (Q.,y
, 1 1 :i
<m m.W , a ...(;,f,F.
- c., }-ll- ) r;.y gTf -> 7 g
m g[Ttd['c 'I ;q ..y W ,i N N .
o i ;3 N 'y;L9 & i"' $,W q 4 ' f.j; -
m'v$, . .F. 4.M,.
M ci, w L,.u.s..< (3A p.t,}f'Nl- ggw/WW' %.
mu-
,' '$4%f ~ r*) ? -.h y
'W'..j%
.'q-g:m y 4,'
- siJ'4 M k..v .~?
3c i s' li, N,n.c;i? n* f (
4 j
64n n
= , - - s m p.a & a,:,c,k y $. h.gi qr c M.e,/yL.?# 3, ..
. p' O y ::w t : v y
't s ~ , . .
.c:<g,M, v v.. -~.Y f .,;p 4 ,Mg t iV.fL:. Sl.W . V y's },4 l h:M %g <p,T !
i ~ Ak rA..,, ,l,
'b , > Wg , M:f: , ,
. p p ;;- . ,
y jr o,pg g-egv J, . 3 pt 3 . -; . g f .N &; f% p ,.q gp,#.v ,A Lg,,g. y:t ,1ge.,g .. ,3_
c# , , . , g .,.j., a, , z ,2_..
(
,y
- Q< *e o m ..g y m p p.,j fte3 i .
- c. y a p.t y i p .,s,3.y, pm ..
%g. pp b u e yjg;i. w.s. . ^... r:.,, o .n . , . ,: j . ,.7, WH e y. 3 37 y .9 t
i, . f.
- L. sg '.,\y '
t > } b~ -%p a (q.Vv 1 4,,p '
M v 4,%ym%.4 g@M Q}f;c.}@,g&y' ts ..., )
n- >
v;q:;q *% ; R_ :y ..w.g;;< M y:, ! g M;\ !u r ..
4 +g:
e m ' O.W '
t .,;y* ' )
., y1 M ;c "Q, g"h :. .M. u 3.y?w ;' q, N;:
M wp qM.cp hi Wlp nm n,p%w,q.$-b' k , r . .e. %'W, M 1W D G l;' 4(g, Mp.Np a ho o;' i
+ >' y.. -y.. i , W}w;.
"~w, a 4 /n .,-)e-ypm,bu y c _y,; m L . + - ,
ymsmg&.4,%
49 q .+>.JK T. 4 2 -1x ,3 . . 6 wy, m.k, . y- ' l":.::h.?y!J % ' 'i"' o . 1>%-AD fn 1 4 -
s 3 i t
'm. .L':
y s i s
t a313 nM:M W-i iimq ,. ' t: ; '
L.D' #"m4g'4O7 f:k 5 ^hQ Nf;p i f; g,. W.y:% /4 w[.'!' q Qv g , &q - e , g' f i t t t
v""5- i . .: , ft m > n 7 <
1 ,-Qc MW G a r ,, a i w tg
.gX. ,s.p
.. o s q g- ( (.'. C ' : r, P
,p' a
,a.y.,7h. >t 3 ,4 7. ~4.t ,.c. t : .,- s.
"c~*: -wS r.
. c f%r.,i 1
1 e N : s c a. n, s e ,m f
,. - im ;' }w< n. :
,w , i .>yL
,; q a.w% u: .. wy
,u g? ;,;M,3 & qcg' aa.f v)c4 u4 ci. ,'n;, ?', ;,e a s Asm
'S F . n. m , . , ',,~, s tv m
!sv 1 r
, ;%gc.- r 5
3.3 g, y t e .
pn - ;
.:4 $ p.-. : ,, g;. . a %' g y .,c 'i ! >
' m to'. : y: - , ,- < .,
.f ' e -"p ' W,b: g ;9 3, m.;ig .,d
'm' e'; y pQ.
, , s' r '8r,.-.a.; M ~, j ,. 2nAf ;'M-,4+
w q,'i .O. . + ;; r T m.c gy' i
"i' +, ig 1 . : gwM
~v ,pc . d F g'
Q ,:' - ' 3,Mw;g&s t s r n: q ; i'i .,, C tQ. ;. .QM 4
.w ] , , +'M 3
- t. 7e n.
O.L . + e b:' ,' e e , g 4
.U'
>a6 !gs' '
'.f ~ - i -
t' 2 Q' ~fi b ,i . :. \ n 'i !'t9%
- fi%ihcy% hh. LA*hki[ - '
i.h N,' b, s.; F y.h (F 64 3 *t v ,.
'% E1;')i;
(,1 f s i , ,
,'~i f.[;t , . ';O M' hkkb i.:a.
.-QJ ' ,
4 JR M
)y : Y' I i
t ,
'.V h Q* ]7, ,
}g;glyp. n-* Of. v ,
'Y",.&' 4 h J q , ,..:.'
Q V ' O,f . i . Y, e'"
g &m e gh ( kUhqj ,;,s54.8 y 9l'/ r,
4
/, ^ j t .q y> X ::{ i l +
!Lp y .cj V 8 gK7 * ' ". ;:3,g :w, g
f.'. E ' N:r ,r.de,h'
- c. x 1, . : t
.q ,
L t
1 i ) n i '
s .7 '
, ,9 t
U,w &. ,d ,di
,Y " '
..g".';, W,s. . ., h 3 ,, .., s
.. d Of'i[cR@M. , ,t'.' 7J.];{ s n 2;' '
.r "f :s ' M' [d~}' ?dY;M (.[ddf!
yiQ;.9. -u~y . y. . ml{S.u., a QR' . @C,si, . . j* l;lM 'p,,-:f !,%
- G
..o<
f.a ,, ,
e &w. 4,,;msm, ,-
%,i, %' , y. ,
,% , ,~, ,.;
- M~
, . ' , ; ,. g,.
(
s e
f,s,4 !;
' ' A
. rg 4 [G s 2ij&
'~
- g. 1; ,.g4 ; m.d,rt R3 q.1, q a w
.; qf ., ,1tqo ... c n ,v rz 2 ys y; e .
4 4 .. . ,, ,. 2 t.. ,. , ~. , 3
.h n ~ :,4k,o '.[',-.
n G
,N; i Q'y Y. an, t, k
.: ,- c
.,4,.-
. , , ... ii: .
am ; s t ,
v s .- s .. .> y ;op 4a \. , , L N f pO Q 3 o Q c. i ~M, ,5 A tI Q / *'
1 r ,/, - i r J. D . I.~( .l g f ' l V ', '* ' cad ' d, .b. } d'f', . Q, , a"W -. 4.%.m 7'f b 3 , a ,
'h ; .,, h G ,, 5 w M, ,d %,I%%,u . .N.W
, , , 1 s
?. L M,. W W N Mr M,W " ' *
%,,nhc MV.s;A,n ' A < -
O, ','; , ,
' ' 't
+
%., is,* i s b~ t 2 ,
r
., , ,. , tw -
u, t' ' .V m /.-wh ;. W 31.
,:.q :. gW,~ ,
. n v.a 'y' , n ,) 1 L
s'f c ? ? ' , : ,.\," 9' 0 6 ; ; ,
.' y &, s Y, V, ;{ $\s&. ,%'
A'
. ? LY{ &+r -lf, A;:
> 3y;, . . s
-(%. s g,.g;Q" g 4",g-Q g,: ;- ) {;;.. t , '1 m m,
- c,
., a c .p r f g ;;{; i ,, ... pgqi 3 ]P. , , 3 yg* v' Qp 'gt r .3, gl i
m,,,, wgy".(s) tn M. #, ;, 4E. 4 ,1 , - .. ' , ' ' ' - * ;. vsa 1 % wa s yt t W -i y'.. o,0 eig;..
e . . r s. e . /, - ' $-s 1' 9-c pl ',, fi 1
. 4 ,
.=q :
f 6
, .Y { L.'
k'd'c, ' ,
\ i y,I
}, s,, a
, .f ? !- ,k ]
L
> t ,,s
, 3, us en9 .. pf . . .
- e , a n, 3,.
ypa et-
,, 3 , 3
..rt i .
..a .> rq ;,,, a,w , , . ,, . , + ,,a, 4
=
,3 3
~
3>m 4
' s'7 a
?.. d, '
~
s l"
' ' ' ' N' &,,qlp %
y'::< -
l.. d ; " ' $ ': e e#
&,q{ G ,,1A+. - HI s . $ ,,1hNLhk y' g C ,y: -
{- t. , ,, . ,
, ( , l . t"-
, 0' 8 '(. '., 1.~
t-
..g'. 29
, g? ,q-
- I?,' l }.t{
r 4 f ,4
- i. .y'.)
..q
.y3;%.. .: -1 i,
- j m,I. -, , .r . m ..tp" \
"w 8
r4 -
T y'1.l' 'gl A+
.s. x f
%' ,9-) e 3 -
1 . . r.f '( , r .".
. ,4 ),,i .2 t
. .i -, an +s - r.e." p ., 4 r -.n
. -s
- r. i. ,7 a
- ,m
- i <
M, .. .Mv3 ..a . l[ N 4'tm d.p. .,M,chs,I g ; q' d. 'E,.
D,". 4; s. s' a c7 .. W '. ( jij l
- r. ' k,wtg A,r M,g' 1g.,.'9 ;y . ..3 -
/ + F
- 4. [ i i q%:g'pC$ a 5; L t3. b}3 .. .g: ,~y[h
. .r a n .v , g! /t;-pt; -%,l ,
ggy !'f. - . ,i $
4ec
'i, r ?if p Aq q
.a .
- e. " N 3 4. ,
( 4s s1 .g g,a y)A
.J hrs >
..., r. ; wb ', .. , o < t -r t < .e i t n < <y
,.i q;,..,.1e,ue 6 13c g,.i ~. d i '; re t %- f (~ . l ?:i ~ s ~
. .. 'e q;,'9 .i. ,!j, . I 6, , , c , , , , . . p es 1 w
,W: ,a.,A. r ( j :J, c.e
'g.< <:17 v, f;i. >. t s s.> 3.1 t..
. W. p -
- 3. . ~e , s
+H',.-
- s. s -
.m
.,:3,,):q e ,p -y -- r ,,a'.,, n q,. ..; y J 6 , ,, / .g ; c . -
m . i og.
j gon; y.3. 4
/
i i ;s, 3, mi_
, , , , , a f.
. - i i im
,~,&: L p.
.#3 t .a'a.., i 5 y .a. -
. 2y- ,
s b +4 cp, ( g, r p.1. y .q;>
' ; py .
u .- ; , . r i
i - - 1 j:# t a' ' ~
p . y .1 0- g 1
]s"gm 1
l ; ' < > ,
r , }= . i,. 4. . + w A .c r .., '
r i
7- ,.:... f.ps. , <., , "
a i(q
' t p .,sa io
, gi y t .+g g y - - ,
x 7'.,: s.".y. v tp '- '- , f
- x A- ,.;.,
4 o 3 s g i x 5.,
6.- 3
"-;y a ; i i' ,.' ;l $ ' ; 3
.Q
Q
\'p y[. ? / W,; v' i w: , ; t j 1W g- Q, ,'
i #
' . l 'i j'c . r, 4
t
, m %,_
QQ t
.E;@':,S
^
r v
,, s 3
O!
3 6
- )
- w
- \; *
., y' ' h (J b A E
?
- m, y' g .. y 1
mQed ' n:.n 9 8 .;, :; b , '
s' ..,4r
);. g' ;;' .' f, g(y a,j i My'1 3 . ,. : M'_%
TA thgl ..
}Y
- i
? ?,YQ 8 ',- M" )
!\ ii L,f.+t h , .,
i 3, v ; g (' ,
- M::. G n; ; ; ,
v y i ., i '
\
- r. .
n:L tyA , .q, .e<w. r*' . . ., ,
viu s
, c e <
r y i
~ ",
. l t
.)
,7 l", ? ' . , .p.
t 8 f0 e i
- t t g j .d ! f' d' : 1 e sG 3 i
o .:t
.2 '. s.>q,. a$) ;cku ' T , . '
N, e_ :b .y ,,
,h;&.
(fy n,
.y d ,j , J i l
f
, R- 4 n.
3i IW.j . *; \ t.
' hs',,.4.Ve.
, s s
- g 3.g .p . {: eg1*.. ,
i' Q:v&;m wp., '. nO S.' m y :-
Fif , .
. v,';
'f.
R h$ . $.
< . .e 'p} 5, s
.a -').' - ' 3*
- qZO G .,o e 1sa 5 I C'? i t Oj ['y$Q Q
- ; - ? b-s ysk&m,e*, mx n' q-m'eM t .- c' < -
b:5 s ,
' g @.
y pnq t , . . 'l9 U.D..
W
' ;N 1 ' l.
. { [. ,
pS. Q a h. y[M,dj lQ Y ' ' . ,'
s g 4
m# , n:w.u. <
l _
- h,, . , t , I' n
r}'&'tM
,,'! M M C! A^ pi(s a
- , 3' t '
ym. 4 ;
, . .t fb)g n>,,-Yj . a,g ! :l,g ,
i l
r,, (,
m%Q? ,. 3 4 I
%'m[ q d' m(. 3'- ..4. c -
,g
(
W;frEG6Gjdahol i p 6 , , . . ,i , ,
^
n .
&,nQ P;O_ /. B.ox1. ,825n. .
&m_Uda.h. o Falls lldsho . .
,e , ms v .
v, :n
" .d@fhm.p~ b .m6 !!@ 5 ;,> 9@+'7 ,rs . . ~. 4 L,p 6%m ,hh y' 5I ' '
umaue m ze.