ML20049J034
| ML20049J034 | |
| Person / Time | |
|---|---|
| Issue date: | 04/30/1978 |
| From: | Hodges M, Israel S, Lauben G, Mcpherson G, Wagner N NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | |
| References | |
| NUREG-0297, NUREG-297, PB-281-133, NUDOCS 8203110377 | |
| Download: ML20049J034 (94) | |
Text
--.
~'
, =.
j
~
i I
f 3
U.S. DEPARTMENT OF COMMERCE National Technicallnformation Service PB-281 133 i
Safety Evaluation Report on Westinghouse Electric Company ECCS Evaluation Model for Plants Equipped with Upper Head injection Nuclear Regulatory Commission, Wcshington, D.C.
Apr 78 Y
)
8203110377 780430 PDR NUREG O297 R PDR
PB 281 133!,l
- =mm_.
l i
-NUREG-0297 a
2, l
1 l
4 l;
k SAFETY EVALUATION REPORT ON WESTINGHOUSE ELECTRIC COMPANY
~
ECCS EVALUATION MODEL FOR PLANTS EO.UIPPED WITH UPPER HEAD INJECTION S. L. Israel G. D. McPherson G. N. Lauben N. H. Wagner M. W. Hodges i
i p ~ a,,,
$d}dh) s,..... /
I Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission n'UF8#A TECHNICAL
""?$.2.,5_1sg3cs mo t
.%w.m,w.-.~=
..e
,,",f,' *"" 335 u.s. NUCLE AR REcuLATORY eomulSs ON BIBLIOGRAPHIC DATA SHEET NUREG-0297 d
- 4. TITLE AND SU8 TIT LE (Add Vo&me No.,,f apprer.sse) t g e, Safety Evaluation Report on Westinghouse Electric Company s U
./
ECCS Evaluation Model for Plants Equipped With Upper 1 RECIPIENT 3 ACCESSION NO.
Head Injection
- 7. AuTHORIS)
- 5. DATE REFORT COMPLE TED Iva^R
=m G.N. Lauben and others
- 9. PE RF ORMING ORGAN 12ATION N AME AND M AILING ADDRESS (tactude I,a CodrJ DATE REPORT ISSUED woNrw lvt,R NRR/ DSS U.S. Nuclear RegJlatory Comission 6 It =
- 'aa
- s Washir.gton, D.C.
20555 8 (Leavt Nank)
- 12. SPONSORING ORGANIZ ATION N AME AND MAILING ADDRESS fractode I,a Codel Same as 9 above
- 11. CONTRACT NO.
- 13. TYPE OF REPORT PE RIOD COVE RE D (tseba,ve deersi Safety Evaluation
- 15. SUPPLEMENTARY NOTES
- 14. (teave uana;
- 16. ABSTR ACT 000 evords or irss)
Westinghouse has planned an additional safety system known as the upper head injection (VHI) system to auament the emergency core cooling system. This system is comprised of additional accumulator tanks and piping arranged to supply cooling water to the top of the core during the blowdown period following a postulated.1:dge-break loss-of-coolant accident (LOCA). The objective of UHI is to add to the core cooling provided by the conventional emergency core cooling system (ECCS). Westinghouse submitted reports describing a proposed I;HI evaluation model to meet the requirements of 10 CFR Part 50. Appendix K, in terms of changes made to the previoush approved Westinghouse ECCS evaluation model. This safety evaluation describcs an acceptable evaluation model for plants equipped with UHl.
l l
l I
- 17. KEY WORDS AND DOCUMENT ANALYSIS 17a DESCRIPTORS f
ECCS Upper Head Injection LOCA l
17tx IDENTIFIERS /OPEN ENDE D TERMS
- 18. AVAILABILITY STATEMENT
- 19. SECURITY CLASS (Tfres report /
- 21. NO O PAGES Unlimited Availability
- 20. Sr.cuRiTY Ct4SS tri,,s,,,i
\\
)2M-/h'/
PRicr.
RM as ii nt g
NUREG-0297 SAFETY EVALUATION REPORT ON WESTINGHOUSE ELECTRIC COMPANY ECCS EVALUATION MODEL FOR PLANTS EQUIPPED WITH UPPER HEAD INJECTION G. N. Lauben N. H. Wagner S. L. Istael G. D. McPherson M. W. Hodges
(
Manuscript Completed: April 1978 Date Publrshed: April 1978 l
Division of Systems Safety l i
>Ifice of Nuclear Re:rtor Regu at on U. S. Nuclear Regalatory Commission Washington, D. C. 20555 I
IL w
6 9
^%
W M M.0+6w&'hQ i
.m___.
.m
-_ _ =._. -.. _ _ _,
n s_.._.__
i ACIUOfLEDGEMENTS Ine review of UNI. Culminating 'n this report, required the efforts of eeny people.
In particular, the authors wish to recognize the following contributors:
G. M. Holahan 4
1 Z. R. Ros:toczy E. D. Throm M. A. McCoy i
I N. Zuber i
Y. Y. Hsu Itaff of Sandia Laboratories
$taff of Irlaho National Engineering Laboratory J
f i
i r
i j
lif
'I
-+ _ _._...
.s
..._.m I
i.
--,,.-y--we-9 p,. * - - -
g-y w
-,gpg
-,g
>-,n w-.-
-re&ww-y.w-%
9 e
-e g
w
-_%wq 9g-s-y caep
,y
,q-g p-gyg-m4,
,.my yw mp g gm.-.
4ec-_-w,m-
.=_:.
~-
r t
TA8tE OF CONTENTS PAGE 1-1 1.0 Introduction...................................................
2.0 UHI System Description...........
................................... 2-1 3.0 General Description of Changes to the ECCS Evaluation Model.......
. 3-1 3.1 Pheno 3enological Changes.......................................... 3-1 3-1 3.2 Programming Changes.......
- 3. 3 Cnangs in the small Break Model....................................... 3-4 4-1 4.0 Acceptability of Model Changes.............................
4.1 Active UHI Injectien.............................................. 4-1 4.1.1 Heat Transfer Correlations................................... 4 1 4.1.2 Que nch..................................................... 4-5
. 4-9 4.1.3 Mimirg in the Upper Head....
. 4 11 4.1.4 Mixing'at the Top of the Core...
.... 4-11 4.1.5 UHI Stop Valves....
4.1.6 Flow Distribution at Core Inlet............................. 4-12
.......... 4-13 4.1.7 Momentum Effect due to phi at Top of Core......
..... 4-13 4.1.8 Effect of Nitrogen...................
4-13 4.2 Upper Head and Upper Plenum Emptying.........
4-14 4.2.1 Upper Head Water Heatup...
4*19 4.2.2 Upper Head Emptying..
4-19
- 4. 2. 3 Heat Transfer...
4-24 4.2.4 Core fluid Behavior......
4-31 4.2.5 Lower Plenum Refill..
4-38 4.3 Reflood.
4.3.1 Modifica. ion of Westinghouse Entrainment Fraction Correlation for Low Ciad Temperatures.........
4-38
.. 4-40 4.3.2 Reflood Heat Transfer.,................
4-41 4.4 Pressure Instabilities.
4.5 Core Subdivision to Accommodate Quenct.........
4-42 y
i f
s
~.
.TAILE OF CONTENTS PAGE 4.6 Slip Effects During 810wdown..................................... 4-42
................ 4-42 4.7 Elevation Pressure Change.......................
4.8 Control Volume Changes in SATAM.,................................. 4-43 4.9 LOCTA-IV Flow and Enthalpy Input........
............. 4-45 4-45 4.10 Bubble Rise Mudel..........
.... 4-46 4.11 0 ift Flux Mode1...............
4-55 4.12 Small Break Model..
i i
.......................... 4-55 4.12.1 Model Changes.
4.12.1 Evaluation................................................. 4-55 4.13 Accumulator Delivery.................
........................ 4-56 5.0 Sensitivity Studies.............................................
5-1 1,
5.1 Large Break Analysis Spectrum........................................ 5-1
......... 5-2 5.2 Systes Parameter Effects.
- 5. 3 Sensitivity of Cl d Temperature Predictions to inccmplete Mi ni ng o f the Uppe r He ad Wate r in the T op Core hode..................
5-3 5-3 5.4 E f f ec t s o f F ini te M! v i ng i n Upper Head...........................
5-4 5.5 Support Column Junctio,eoration...
5.6 Analysis of a Non-UH1 Plant with the UH1 Model........................ 5-5
. 6-1 6.0 5ystem Tests.........
6.1 935A Tests................................................ 6-1 6-4 6.2 Seelscale M000..
7.0 Staff Independent Calculations......................................... 7-1
........ 8-1 i
8.0 Conclusions......
r t
j 9.0 Re f erences.........
9-1 v1
~
r 1.0 INTR 000CT104 For plants which include an ice condenser containmer.t concept. Westinghoase has planned an additions) safety system known as the upper head injection (LHI) system to surrent the emerr,ency core cooling system. This 'ystee is comprised of addi-tional accumulator taaks and piping arranged to supply cooling water to the top of the cc.re during the blowdown period following a postulated large-break loss-of-coolant accident (LOCA). The objective of UHI is to add to the core cooling provided by the ccnventional emergency core cooling systae (LCCS) and so permit operation at linear heat rates comparable to those permitted in plants utilizing the dry containment concept. In this way, plants which include the UHI system would have greater operating flexibility while still meeting t*.e acceptance criteria as defineJ in paragraph 50.46 of 10 CFR Part 50.
In February 1975 Westinghouse submitted WCAP-8479 (Ref. 1) for review by the NRC staff. This report described a proposed UHI evaluation model to meet the require-ments of 10 CFR Part 50, Anpendix r., in terms of changes made to the previously approved Westinghouse (CCS evaluation model. The latter has been reviewed by the staf f and found acceptable as described in References 2 and 3.
On September 26, 1975 and August 13, 1976, the ;,taf f issued Status reports (Refs. 4, 5) on its review of the proposed UHI evalua. ion model and identified several areas of concern.
Westinghouse I as also made changes to their reference LOCA model (non-JHI) that have been reviewed and accepted by the s*:ff as discussed in References 6 and 7.
These enanges are also applicable to the UNI evaluation model. In rcspondir.g *.o the staf f's cencerns noted in References 4 and 5, Westirchouse has made entensive modifications to the proposed UHI model as indicated in the revisions to Ref. 1.
Westinghouse has met with the staff on numerous occasions to discuss data support-ing the proposed evaluation model. These data include esperimental resalts, sensitivity studies, and responses to staff questions pertainina to the model.
The primary documents containing this information are presented in Tabin 1.1.
In addition, varlaus aspects of the proposed UHI model have been di* cussed with the ECc5 subcosmittee of the ACRS on seven separate occasions.
This -* view is concerned with those changes to the Westinghouse [CCS evaluation model that have been proposed for the UHl-LOCA model. Tw objective is to estab-l Ilsh that the UHI-LCCA model is in conformance with 10 CFR Part 50, Appendix K.
This report has incorporated appropriate sections from Peferew es 4 and i in l
l total. Other sections reflect a resolution of pr-Uusly open issues or a discus-sion of subsequent modifications.
- hus, this repcet includes the complete staff 1-1 I
1 l
TA8tt 1.1 WESTINGHOUSE REPORTS ON UH1 t0CA EVALUATION MODEL 1.
Westinghouse Emergency Core Cooling Systes Evaluation Modes.
Appilcation to Plants Equipped with Upper Head Injection Proprietary WCAP-8479 Rev. 2 luov. 1977)
Non pror.rietary WCAP-8480 Rev. 2 (Nov. 1977) 2.
ECCS Heat Transfer Esperiments w th Upper Head Injection i
Vol. 1 Test Facility, Procedures, and Data (October 1974)
Vol. 2 Heat Transfer Data Analysis and Correlations (October IM4)
Vol. 3 k9plementary Information (August 1976)
Proprietary WCAP-8400 Non-proprietary WCAP-8432 3.
810wdcwn Emperisents with Upper Head injection in G-2 17 17 Rod Array Facility Voi. 1 Test Facility, Procedures, and Data (January 1976) vol. 2 Heat Transfer Data Analysis and Correlations (August 1976)
Proprietary WCAP 8582 Non-proprietary WCAP-8642 4.
G 2 Refill Test Proprietary WCAP-8793 Non proprietary WCAP-8794 l
l t
I-2 t
g g
--7, w
ev-,--
+-e,, -,
--- - --,--,-- - - -w,p 4
a n
~. - -. -.. -
3 e
assessment ti date of the UHI-LOCA evaluation model and does not require reference to earlier reports for this information.
At the ACRS meeting of September 9,1976, the stif f identified four major unresolved items and several minor iters On October 4, 1976, the staff outlined methods for resolv;ng these issues in : letter to Westinghouse (Ref. 8).
Tha four major items were:
(1) Core drift flua model, (2) Core flow t4havior, (3) Unquench osring sounter-current flow, and (4) Upper head temperature, iten (1) has been resolve 1 and is discussed in Section 4.11.
Items (2) and (3) were resolved together and the discussion is in Section 4.2.#.
The discussion and resolution of seper head temperature appears at the end of Section 4.1.3.
In crder to comply with paragraph I.C.5.a of Appendia K, a comprehensive reevaler-ation af IIHI heat transfer and quench was undertaken since the last status report with espessis on statistical considerations.
D*
ussion and resolution of t%ese items appears in Sections 4.1.1, 4.1.2 and 4.2.
A number of otter issues identified as having solution procedures will be defined Dased on various 's sitivity studies and coeparisons to data. These sttdies have been performed and are discussed in Sectiens 4.25, 4.4, 4.5, 4.8, 4.13 and 5.0.
The staff a 50 discussed with the Japanese government performance characteristics of thet: Ut! systee tests to better understand both UHI separate ef fects arr*
Integ*a'ed systee ef fects tests perforced in Japan. In addition Westinghouse has arved to anal re these integrated system tests at the FOSA !! f acility to deter-f eine the effect of UHI on the coserved rssults. The staff has also agreed to a pl n with Westinghouse for revie. and analysis of the sealscale HCD3 esperisests.
This test series at INEL will be an integral simulation of a plant
- quipped with UHl.
The upper head injection concept was not proposed as a possible emergency core coolirg subsystem ween 10 CFR Part 50. Appendin K was published. As a result, in areas of UH[ isolementation, the staff has based its standard for Comformance of the UHIAOCA evaluation model with Appendia K requirements of Section 11.5, which states:
"Elenents of evaluation models reviewed will include technical acequacy of the calculational methods, including corpliance with required features of l
Section I of this Appendia K and prevision of a level of safety and margin of 13
\\
conservatise comparable to other acceptable svaluation models, taking into account signific.nt differences in the reacters to which they apply."
The ut*1fration of a UNI systes represents a significant differeiste from the reactor ECC systems which were under consideration durf t.g the ECCS rulemaking prcceedings. As a result, some espects of Appendix K are not directly applicable to the UNI system. For these aspects we have accepted model characteristics on the tests of adequate technical just*fication.
3 6
i 4
k 1-4 a
1
~,.... ~.._ _.._
a l
2.0 THF UNI SYSTfH CE50RIPT!ve The UHI is a passive high pressure fejection system which supplements the accueu-lator cnd pumping systems which constitute the conventional Westinghouse energency core cooling systen (see Figure 2.1).
It consists of two high pressure accsmulatoe tanks which provide a panimum volume of 1800 cu. ft. of Ilssid in onc tant and 1800 cu. f t. of compressed nitrogen gas in the other tank. The two accumulators are conrected at the tops with a 12-inch crossover line. A rupture disc in this line separates the water cnd nitragen gas interfsce and minimizes the amourt of nitrogen gas absorption in the water. This d;s" Jursts to provide the injection path during the LOCA. Borated emer2ency core cooling water is delivered from the water-filled accumulator tSrough two 12-f rch lines which penetrate the containment and branch close to the reactor vessel into four 8 inch lines. The four 8-inch lines reduce to 5-inch lines at the vessel aad dist.iarge the water into the vessel head through 4-ir.ch I.9. vessel head penetrations.
Four hydraulically-operated isolation valves are located outside of the containment close to the accumulators to terminate opper head injection ar.d isolate the nitro-gen gas supply from.he reactor coolant system following the injectior of the borated water. InsiC4 the containment there are reduidant,12-inch and 8-inch, swing disc, check valves in series which prevent the reactor con
- ant from entering the water-filled eccumulator during normal plant operation.
The upper head injection system is passive during normal plant operation. la the event of a LOCA or other severe depressurization accident, tSe Up! system is actuated automatically upon oepressurizatice of tne reactor coolant systta. The UHI system begins to inject horated water (at 70*F-100*F) when the reactor coolant pressure drops below the ca
' pressure setcotr.t in the UH! accumulators (typically, 1500 psla). The asow path of the injected cold water internal to the reactor vessel during the reverse flow portion of the blowd>wn is illustrated schematically in Figure 2.2, for the case of a large cold leg treak. Water from the UHI accumulators enters the upper head and forces water from this region downward, through rollow support columns and control rod guide tubes. While some of this water splies through slots in the guide tubes into the upper plenum, the majority enters the core directly and mises with the flow from the hot legs and steam generator and flovs downward, through the core, to the lower plenum, up the downconer, and potentially out the cold leg break.
As the blowdown progresses, a low UHI accumulator water level sign 11 autaeatically results in the closure of the UH! isolation valves, leaving the upper head filled with water. The isolation valves are automatically locked in the closed positiun to preter.1 an inadvertent discharge of nitrogen gas into the reacter. From this 2-1 l
l I
r
O e4 e
E
=
O is:
=
1.
E I
h 3
b,y t =b "-
I I E $,
h i==
Ig5.s E
[0 34 y
I
\\
31 I
.t g
2g s
O e
n:
,2 s
a-I
- b.
g g
~.
T.
Y 8
s:
- =E I
g_s I
b
{$
=
_._.___.____q p _.j p.!
r Si j
i i%
=
t(
vi a;
h 58[
t m
a "e
5
=
w M
=
se 2-2
Flow FROM UNI ACCUNULATC45 UNI PENETRATICN (TYPICAL GF Fout) g SUPPORT cott'MN
}
yV
.[
- *' "E5
$n N UPPER Cott $UPPCRT (TTFICAL CF Ws\\7'3
,j
'??ff 7T i]
1b EIPANDING FLCw a-7.., \\ -::
,1 wa.ine,
!ff, ;r e
f (y
! y' vr L
it i
E L CATION k ' '* ' ',' t'ihi )
% CCal SittEL 4
FLCW CF t'M!-f t.;EC'FD WATER PLUS EzracinG STEAN WATEt 4:tT;;aE DUtlMG 8tCW"Ces Floure 2.2 UH1 Flow Pottern 3re<,h the Vessel during LOCA Crowdewn 2-3 l
l
polit onw:r1, the UHI systes does not directly influenc3 the progress af the LOCA.
The LOCA proceeds, then, with an upper head filled with water and with a core which is cooler than in the non-UNI case.
The total water delivery of the UHI systes is about 1000 cu. f t. during the first 30 seconos compared to 4000 cv. f t. fer the cold leg accumulators over about 100 seconds.
In the case of a large hot leg break, the core finw is upward during most of the blow h.:; the UHI has a small effect on the core quality and flow. The flow conditions in the hot assembly permit nucleate boiling to continue du "sg most of the blowdown and therefore remain cocl. The refill period differs from that for a cold seg break due to the low resistanse to steam flow from the core to the break.
Consequently, the cere rapidly fills with water, quenching the rods before the upper head empties.
The blowdown behavior of smaller cold leg and bot leg breaks follows the behavior of the large break transfer.t characteristics. Not leg breaks have consistently been shCwn to be less IIaiting than cold leg breaks. with smaller hot leg breaks resultir,in ever. lower peak cladding temperature than those calculated for the limiting hot leg break area.
4 2-4 2
3.0 CENE1AL DESCRIPTION OF C.tANCES TO THE ECC5 EMLUATION M003 The addition of the UHI system requires several significant changes to the accepted Westinghouse ECCS evaluation model. A general description of these changes is given below, while a detailed discussion as to their acceptability to the staff and in accordance with 10 CFq Part 50, Appendix K, fol1cvs in Section 4.0.
3.1 Phenomenoloaftal Chance The flow of water from the upper head into the top of the core results in a higher core flow of lower quality coolant during blowdown irestinghouse has therefore used heat treasfer correlations and quench criteria appropriate to these new conditions. The initial departure frna nucleate colling conditirl has been snaffected by the UHI system because it is predlCted to occur prior to UHI injection.
Once the active UHI injectic.n phase is completed, the injection lines and upper head remain full of subcooled water. During the emptying period this water drains into the core and some of it vaporizes. The resulting backpressure causes a signiffeant delay in the beginning of core reflood. In order to model this situ-atinc., the standard control volume and flow model has been changed to account for the faowpaths between the core and the upper head and the thermal-hydraulic effects which lead to the draining of the upper head.
Twenty LOCTA calculatfor.s are performed up to the beginning of reflood (80CREC).
These calculations represent the different power regions in the core. Ten each use the average core and hot channel transient SATAN hydraulics results as LOCTA boundary conditions. The finst temperatures of these calculations are used to calculate entrainment (carryo;t) rates during reflood. The average cladding temperature at the beginning of reflood is Icwer than for a non-UH1 plant because of quenching during the blowd)wn period. This permits the correlation used to predict the amount of liquid entrained by the st Na flow to be modified to reflect the lower entrainment rates measured in FLECHT (Ref. 9) experiments having lower initial rod temperatures Ite entrainment rates from fuel assemblies both quenched and unquenched (hot assemel bs) are combined to give a weighted core average ligt.id entrainment.
3.2 Proarammina Chances Due to the irjection of su'xooled upper head water into a two phase core flow, an assumption in SATAN which h sogenizes fluid properties throughout each nooe leads 3-1
L2 csiculited pressurs instabilitlIs. $ATAN has besa modified t3 41tlinat2 thes2 instabilities.
When it The quenching phenomenon during blowdown is new to PWit LOCA evaluation.
To occs:es, those fuel rods which quench rapidly release their stored energy.
account for this ef fe:1, the large nuder of fuel rods considered to make up the average core is modeled by dividing them into 10 power-dependent subregions to provide a more realistic release of energy over a longer perled.
With UHI, large amounts of water injected above the core result in low quality flows throughout the reactor vessel and some parts of the reactor coolant system.
Under these conditions, slip, i.e., unequal as well as opposite Ilquid ano gas velocities, becomes important. A more realistic drift flus model has therefore been developed for UHI analysis.
Di. ting the upper head emptying period. gravity ef fects become important and density changes are sensitise to a. halpy gradients. To provide a more realistic elevation v
head, this term is based upor. the density of the fluid flowing from the upstream element.
In-plant measurements as well as confirmatory scale model tests inthcate that The staff has enthalpy gradients exist in the upper head in non-UH1 plants.
required that this effect be considered by bestinghouse in their proposed model, for plants equ'pped with UHl.
To better account for the emptying of subcoole1 water from the upper head into the Ibe downComer has been core, the $AI AN ccde is used through the refill phase.
divived into azimuthal r. odes; separate n(des for UH1 columns and ccntrol rod guide tubes have been provided; a containment node has been added; and an additional upgr head node has been included.
As a result of using the $ATAN code through refill, the delay *.ime associated with refilling the intact loop cold legs, free fall in the downcoetr, and the delay due to downcomer hot wall effects are accounted for in a different manner than that used in a non-UHI model. A speClal model for the lower plenum has been developed to account for sweepout during injection. A modification to the static head crossflow term in the annulus has been made to better represent the difference in the liquid isiventory on each side of the annulus.
Core coolant enthalpy input for the LOCTA code hae acen changed to include the upper and lower plenum values in the interpotation at the core end planes.
A summary of the proposed changes to the staMard Westinghouse evaluation model is 4
presented in Table 3.I.
3*2
taste 3.1 MuAil04 IC0f t CHAMCES PROPOSED FOR UNI-t0CA ANALYSI5*
Hydraulics 1.
Moding 2.
Slip model used in flow paths 3.
Drift flum model 4.
Dubble rise model in upper head 5.
Smoothing of the equation nf state 6.
Pseudo-viscosity 7.
Elevation head 8.
Valve closure model 9.
Upper head mixing 10.
LOCTA input 11.
Bypass model/ deficit calculation 12.
Power region LOCIA calculations 13.
Entraiteent correlation 14.
Reflowd calculation 15.
Downconer and Icwer plenum model Heat fransfer I.
Qucnch criteria 2.
Film boiling correlations 3.
Quench time 4.
Heat t.ansfer coefficient af ter quench 5.
Return to film boiling criteria 6.
Reflood heat transfer
- Details of these model changes are described in Ref. 1.
3-3 f
1
.. - ~
_ _ ~ _ - -
. _.. ~.
1,3 Change to the $aall Srn k Model The only change made to th!s model is the addition of a flow path free the UNI accoulator to the upper head. The effect of UH! injection en the small break I
transients is to provide additional coolant inventory to the reactor vessel for the purpose of recovering the core. The model does not consider the enhanced cooling because of UNI injection that is included in L;ee large break model.
I a
i l
i I
i, 3
I s
i J-4 V
4.0 ACCEPTABILITY OF HCDft CHANC15 Cetaffs of the model changes and their acceptability to the staff are discussed in this section. The first three sections (4.1 to 4.3) have been divided according to significant phases during a LOCA in a UHI plant. The first phase corresponds to UHI water delivery from tne accumulators to the upper head which occurs over the time interval from about 3 sect.nds to 25 seconds ir the event of a large break. The second phase is also analyzed with the 5Al.a code and covers the period from 25 seconds to about 100 seconds. This phase includes the emptying of the upper head and depletion of the liquid in the core and upper plenue. Refill of the lower plenum with cold leg actuelator water also occurs during this time period. The f!aal staqe is reflood, which is analyzed with the WREFLOOD co<'e and includes that portlan of the accident until reduced peak clad temperatures have been achieved.
4.1 Active UHI InjeClion During active UHI injection, water is delivered to the upper head as the reactor coolant system pressure falls below the UHI accumuiator pressure. Water in turn is forced dou the support columns and guide tubes to the core and upper plenum where it alzes with flow from the steam generators in the 4dact loops. The condition of low quality downflow through the core dominates during this period.
The principal phenomena of interest during active W injection are mining of the UHI water in the upper head and subsequently in the core; core flow distributions; and the heat transfer in the core. These phenomena and the manner in whlCh they are represented in the evaluation model are discusW 1 in the following subsections.
4.1.1 Heat Transfer Correlations 4.1.1.1 Pressures Above 800 psia The UHI blowdown heat transfer experiments were performed between 800 psia and 100 psia. It was therefore necessary to approve heat transfer correlations above 800 psia where no UH1 heat transfer data exists. At system pressures above 800 psia. Westinghouse originally proposed to use the larger or either the first term of their transitir2n boiling equation or the Dougall-Rohsenow correlation.
l Afte' the clad temperature fl.'st exceeds the fluid saturation temperature by 300'F or more, they proposed not to use the first term of the transition boiling equation. Their prt. posed combination of two heat transfer correlations was not l
acceptable.
I 4-1 I
l
Af t2r Dh8 is first pr:dicud to occur Appaidis K permits the usa sf the Dougall Rohsenow flow film boiling correlatiot or the Westinghouse correlation of steady-state transition boiling with cercain specified restrictions. It is acceptable to the staf f for Westinghouse to une the transition boiling correlation in the same manner as approved for ths. non-OHI model. Westinghouse has agreed to this for pressures above 800 psia.
4.1.1.2 Film Boll (rg_During Active UHI Delf very Westir.ghouse proposed to calculate the film boiling heat transfer coef ficients during active UH! delivery based on modifications to the Borishat. ski Fokin (8-F)
(Ref.10) and Dougall-Rohsenow (D-R) (Ref. II) file 'sof ting corcelattens.
Empirical cunstants were develcred frne film boiling teat tra isfer data cbtaineu' frue the Westinghause UNI blowdoen tests. In the modified form, the B-f correla-tion is applied wh?n the fleid conditions are subcooled. A modified D-R correla-tion is used when the fluid qualities are above a certain transition quality. At qualities between aero and the transition quality, the heat transfer coef ficient is calculated by a 1*near intervolation t'etween the modified B-F and D-2 values.
The heat transfer coef ficient for subcooled conditions derived from the B-F correlation is multiplied by an empirical correction f att9r, o, which is related to a fluid enthalpy rise factor, p.
this factor represerts the ratio of the fluid enthalpy rise f rom the channel inlet to tne enthalpy risa nece55ary to raise tte quality,to the transition value at the point of interest. Ine heat transfer coefficient for flow qualities sbove the transition alue is obtained from the D-R correlation with the original coef'icient (0.023) replace. by the empirical values derived ' rom the UHI c.perioents 4.1.1.3 Requireeents of A,'pendis R Paragraph I.C.S.a of Appendia K states:
" Correlations of heat transfer f rom the f uel cla1dina tu the surroundirg fluid in the post Car regires of traosition ant; film boillag s*all be compared to applicable steady-state and transient state data usity statistical correlation and uncertainty analyses. Such coeparison Sr.all demonstrate that the correlations predict values of heat transfer coefficient equal to cr less than the eean valoe of the applicable esperimental heat te ans f er data throughout the range of parameters for which the corralatiurs art to te used. The comparisons shall quantify the relation of the corre.ations to the statistical uncertainty of the applicable data." These req;irements are satisfied by the experimental data discussed below.
4-2
-o-.
~,
f i
l 4.1. l. 4 bperisertel V<rlficati o Westinghouse has conducted a series of tests (Ref. 12) with a 19:19 array of heater rods (figure 4.1) having rod disseters and spacing typical of a 1717 fuel rod array. The heater rods were stainless steel clad, filled with boron-nitride, and heated by means of electrical current passing through a cer. tral wire. In order to simulate upper head injcction, water was forced into the upper pcrtion of the test section through 13 separate ports or norgles, located a few inches above tha tops of the heater rods. To simulate flos from the unbroken loops, steam was injected into the top of the test sec lon. The test section also included 25 thirbles, siellar in size to those so a ruclear fuel assembly.
(ighty-two of tre 3% heater rods ear contained 6 thermocouples at various eleva-tions over the heated length. The thermocouple loc.ted near the cladding /
boron-nitride Interface was used to determine heat transfer coefficients at the various elevations.
Tests were cciducted on a single-effects basis, i.e., all the parameters except one were neld fined to ascertain its effect. Twenty-two tests were blondowns from appromisa'ely 800 pala to 100 psia to simulate reactor depressurisation during a LOCA. These tests ?namined the effects of variations in the followieq parameter *:
(1) UHI flow rate, (2) UHI flow subcooling, (3) Rod bundle power.
(4) Initial rod temperature, (5) Upper plenus inlet flow rate, (6) Upper plenum inlet flow quality, (7) Buti flow distribution, and (8) System pressure.
Westingiouse predicted heat transfer coefficients using the LOCTA code to calculste flow conditions along the test section based on met.sured test parameters of power, inlet flow and enthalpy. They Compared predicted heat transfer coefficients based upon the average at any one level with the average ceasured coefficlent.
The staff has analyzed the data statistically. It was assumed that tre data are I
distributed binomially. The confidence level was calculated by using the normel approximation to the binomial. The staff required that the correlation provide a
(
confidence level of 90 percent, that at least 50 percent of the measured heat transfer coefficients exceed the calculated value in order to comply with the specifications of paragraph 1.C.5.4 of tcpendia K.
The staff re tewed the data provided by Westinghouse which consisted of compara-tive heat *.ransfer coefficients relating to individu
- values and average values.
4-3 f
l l
l l
I l
27C' e e a t j e r e a s g ie l= gi giisit lc j a i.i i i,i i
VilIll'ii lI 1
e i
i!
t I
i i
ff s
/
- m~p=~I4N$.h.Nb...T_hY 2-
,e W >++t ++4-+ W~
3 Y
siiiiIi k'i: i L!+ + + M escoatt t: stas E h i j ! 5I+1 j ii[if j i ih 6
-g.4 g
e.sies uit j
I (2 3 5 : E !IU si311I: I I:--
- c' i sc* - 's.
- $D i5 i Dii i i1
- SaFFLt Ptatt
'd ? ?444a444111iiiis i.-
l5 -
t,. _ +_ +_ _+_ +2, + t, +V_. i., i i i_ E
..e...
.. nm,w ll -
_ p +++ m ++++t.ii.i!.
- .=
/
m
' f. 'a
^
i
\\
i 90' i s.r. u.v. s..s.
t sr. e so.r. iss..-
atatte e035 natins ists.otov'tts (82 8005) an.r. s.:. so.r.
s e s.i,. r.r e er ouett Stran eents (.)
@ soner=0.atte iniuBLis: SuPfott T..*utLES (8). Cfette inlutit (4)
O *t c'asts tai =stel ti 1 gi une eget tocaticas (es) 1 Fig. 4. )
G.2 Rod Sw.Wie, Baille Ceoss Sect:oa oad Bu dle n
incewmeneofion l
4-4 l
l l
l l
l l
l
Westini,hous2 d owed that the carralation was not biased with regard to clevation, mass flow rate, flow auality and pressure. The staf.' has concluded that the proposed heat transfer correlation is acceptable and satisfies the requirements of paragraph I.c.5.a of a.,ipendin K.
Similar tests were conducted earlier with a sleulation of a 15x15 fuel assetly Predicted heat transf er coef ficients for these tes's were lower than (Ref. 13).
Since all the plant designs which include the UNI systee the average of the data.
have the 17:17 fuel array, the statistical comparison and model appilcability are limited to the 17all array for UNI plants.
4.1.1.5 Hodel timitations limitations must be placed upon the use of the proposed UHI high pressure corre-latian because of its empirical nature and restricted data base. The appilcation is Ilmited to the 17:17 fuel rud array for the following conditions:
(1) (a) Two-Phase Flow:
Cocurrent downflow (b) Single Phase:
Downflow 100 psia to 800 psia (2) fy.. /ressure:
(3) Clad Temperatures:
350*f to 1700*f f
(4) Mass Velocity:
Greater than 30 lb/f t seC.
(51 Ilow Cuality:
up to 90 percent these limitations on the range of parameters have been confirmed by the statistical studies submitted by Westinghouse. For flow conditions outside of these ranges, the Dougall-Rohsenw correlation may be used where appropriate, in If accordance with the specifications of Appendia K to 10 CFR Part W.
counter-current flow occurs in the core curing active UH1 injection, steam cooling heat transfer must te used at that location taking no credit for the liquid present.
4.1.2 Quench 4.1.2.1 Description of Quench Criteria Quench is a rapid cooling process sometimes known as rewet which may occur on surfaces with temperatures significantly above saturation. Westinghouse has established a design quench line based upon the steam flow quality and the wall The wall superheat is the difference between the fuel rod surface superheat.
If the quench criterion is satisfied, the temperature and saturation temperhture.
heat transfer coef ficient is raised to a constant value in one second for the core w.de calculations and instantanecusly for the hot rod analysis.
4-5
ar t,
4.1.2.2 g endin K Aequirements Quench is essentially a rewetting phenomenon wnich can best be described as a change in heat transfer regise, from film boiling to a heat transfer regime in which the heater surf ace is significantly wetted. It is noted that when 10 CFR Part 50. Appendia K, adoresses other heat transfer phenomena, such as cel:Ical heat flus and post-CHF heat transfer during blowdowr, and refill and reflMd heat transfer, it stipulates that the heat transfer correlations must be based upon applicable data. 8ssed upon these analogous situations, the staf f concludes that the quench phenomenon used in the ECCS UHI evaluation model may te considered in conformance with Apper. dis K based on appropriate esperimental data.
Appendu K authorizes the use of post-CHF blowdown heat transfer cerrelations other ti en those set forth in parage.iph 1.C.5.b.
The rule provides that other heat transfer correlations say be used if they are based upon applicable experi-mental data. Westinghuuse has provided suf ficient enerleental basis to peref t use 6f heat transfer characteristic of quenched rod surfaces. The experimental justification is described below.
4.1. 2. 3 E g erimental Verification Data from the heat transfer tests described in Section 4.1.1.4 were used to estab-lish a quench design line. These cata represent average conditions at a given btndle elevation w4.tre all the therancouple locations on the rods were observed to have quenched. These data relate flow quality to wall superheat. lhe design 7
quench line assures, at, the 90 percent confidence le.el, that at least 50 percent ofthetrucdatapopdationliesabovetheline. In reality. 60 percent of the measured points lie above this line.
the data showed a bias with respect to initial te werature. This trend is attributed to the use of a cosire power shape Mich resulted in lower initial rod tenperatures at the top and bottom of the test sectica. Quenching of the ends of the tests sectinn were o'oserved to occur at to.er wall seoerheat. The s'aff concurs with Wes*.inghouse that these initial conditicos artificially influence the quench data but should cot'affett the real quench mecnanise.
Westinghouse presented quench oata relating wall superheat to mass velocity (G).
flow quality (X), and pressure. In order to satisfy a 90 percent confiden.e level that 50 percent of the late be equal or greater thar: the predicter. valur.s. a 5 percent reduction in the criteria was required for pressures fros l'.r0 psia to 400 psia.
Westinghouse des.onstrated the suitability of the one-second quench time by com-parison to a plot of the average heat transfer coefficient as a function of time at a given elevation. Westiaghouse also showed that the constant heat transfer 4-6 w
9
~
i coef ficitt.t used af t:r quench r".flicts the data, frrthermoes, higher coef ficients de not affect the fuel rod temperature response differently than the coefficient voed.
The staff finds the quench criteria are act.eptable because they are conservatively supported by esperimental d4ta.
4.1.2.4 Model tieftati m the staf f has ctncluded that the quench radel is acceptable based on these data and may be applied during the actise LMI Injection phase when the following thereal hydraulic conditfans are satisfied:
(1) (a) Two-Phase flow:
Cocurrent downweed (b) Single phase:
Downflow (2) Minimum fuel assembly 8
mass velocity:
30 lb/ft -sec.
(3) $ystes pressure:
800 pila to 100 psia (4) Clad temperature:
350*f to 1700*F (5) Quality and well superheat as defsred in criteria.
tfestinghouse proposes that the quench design line be entended to flow rates lower (down to $ Ib/sec-f t ) than those used in the tests.
8 Conparison ed he low pressure and high pressure quench data over the full range t
of wall superheats, shows that the low pressure data is clearly tne more conserva-8 tive deta set. The low pressure data set considers flows down tu $ lb/sec ft,
8 whereas the high pressura data onl> inc hies a flow rante down to 30 lbs/sec-et,
The staff considered the cocurrent low pressure data, at flor rates as low as
$ lb/sc* ft8 as being conservative with regar1 to high pressure quench.
Therefore, the low pressure quench desigr. line was modified, conservatively, to permit high pressure quench at flow rates down to 5 lbisec ft'.
4.1.2.5 Return to Film Bollino tiestinghouse has proposed a criterion for retu. nato-film boiling based on the critical hea. flus predicted bj the Macbeth correlation (Ref.14). As shown in Figu
- 4.2, the heat flus at the highest powered quench location drops rrpidly to i
less than 60,000 Btu /hr ft. The correspont'ing critical heat fluses based on the 8
Macbeth correlation are $1 nificantly above the heat fluses espected during the 0
active injection phase.
During active tW1 injection, the mass velocities are high and qualities generally less than 40 percent. Return to film telling af ter rods have quenched is not probable under these flow conditions after the rods have cooled off.
j 41 l
1
-8 R
O s
8 I
~t.
kg 8 3 if I5 e za R
55 ga EE W EB a
~
.15 n y
kh 2
gg
')
- e 8
f I
o e
8 N
o d
i I WA18,08) xnag gygg l
l 48
Westinghouse performed a sensitivity study in which various fractions of the heat transfer area were made to unquench af ter the injection period. The result of this study, discussed in Section 4.2.4.2, Indicated that unquenching large amounts of heat transfer area, did not have a significant effect on the calculated peak clad temperatures. For this reason the staff sccepts the Macbeth correlation as a criterion for determining return to film boiling.
The staff requires that the hot it
- be unquencheJ after the blowdown period. The reasons for this are discussed in Sectiors 4.2.4.3.
o 4.1.3 Mis f ra in the Upper Head A diagram of the upper head slaing s,ojel is presented in figure 4.3.
There are four UNI injection no22Ies located around the periphery of the upu r head. During active injection, cold water from the UHI accumulators mises with hot water initially in the upper head and water at varying temperatures le forced through the support columns and guide tubes.
I%e 5 AIM upper head model, described in Ref. I, is capable of analyting any degree of mining between the fluids during active UHI injection. Two models are
{
considered, namely, perfect mining which assumes that the injected water mines lamediately and homogeneously with the urper head water; and finite mining which allows for a finite amount of entrainment of upper head water by the UHI Jet as indicated in figure 4.3.
Because the estent of mining 11 the upper head is unknown, Westinghouse presented sensitivity studies to demonstrate its effect on i
UHI flow during the active injectiore s.hase.
These sensitivity studies indicated that the supp;rt column flow rate was higher with imperfect upper he.ed mining, but the duratior, of the active injection period was about the same for both models.
- 1. second sensitivity study, discussed in Section 5.4, showed that the imperfect mining model was insensitive to variations of 50 percent in the entrainment ccrrelation. Io assure that reasomble limits are considered for this important phenocenon, the staff will require that buth perfect and finite mining models be cc.nsidered in plant specific calculatlans using the apprb ed model.
Westinghcuse proposes to use a uniform initial upper head temperature for the LOCA/UHI plant calculatinns corresponding to a calculated fraction of the dif fer-ence between the vessel inlet and outlet temperatures. This temperatu e model is based on hydraulic analyses that calculate the recirculation flow rate in the upper head during normal plant operation using press.re vessel hydraulic charac-teristics observed in 1/7-scale model tests.
The staff t'as reviewed the information sepporting the determinattun of the initial upper head temperature aM did not fird it suf ficient to justify the proposeJ 49
?
GUIDE TU8E FLOW METAL HEAT ENTR AINE D
/
,UH j
FLUID
/
\\
/
\\
/
\\
/
\\
/
\\
/
HE AD COOLING j
--1---= \\
1-JET FLOW
% INTERFACE
/
COLO REGION
/
IAETAL HEAT
$UPPORT COLUM*;
FLOW OtAGRAM OF UPPE R HEAD MIXING MODEL Tlqure 4.3 4-10
design procedure. Westinghouse has initiated a program to obtain upper head temperature data in a number of noe9HI plants to verify their assumptions of uniftraupperheadtemperatureandprovidejustificationoftheirupperhead temperature model. Testino on a UHI plant is also included in thi$ program.
However, this information would not be available until the startup tests on the first UH! plant are initiated.
Westinghouse performed I/?-scale model tests with standard upper head internals and 1/5-scale tests with UH] Internals. The results showed that the esisting plant configaration would result In upper Pead temperatures between the inlet and outlet temperatures. It was also determined that modifications to the head cooling jets would return the UH temperatu e uniformly to the inlet temperature.
r The staff These modifications have been performed for all domestic UHI plants.
I ""
IJ 6 eves that this justifies the use of tte inlet temperature (Tcold analysis with the modified design.
The staff has concluded that the initial @per head temperature to be used in the UH!*LOCA analyses is plant specific input and not part of the evalution sodel.
As a result, each applicant must justify the uniformity and magnitude of the initial upper head temperature when submitting h;s LOCA analyses for compliance with 10 CTR 50.46.
If temperature measurement information during startup testing of the first UH]
plant shows significant temperature stratification Wes;inghouse =111 have to modify the generic (CC5 model or an input to account for this behavior.
4.1.4 Mlainq at the Top of the Core One of the limiting f eatures of the sA1 AN code is that it cannot account fr r s
thermal ponequilibrium within a control solume. As a result, the Colder upper head water that reaches the core or t.pper plenues ;s assuned to sia immediately with the fluid in the respective locations. Since the UHI blowrWrn tests (Ref.12) demonstrated an adverse ef f ect of subcooling or heat transf er during blowdown, sensitivity studies were performed to demonstrate that nonalaing which Tre might occur would not lead to significant deterioration in heat transfer.
results, discussed in Section 5.3, demonstrate that thermal equilibrium assumptions are acceptable.
4.l.5 UHI Step Valves The UHI stop valves in the SATAN model are simulated by 11 putting a sharply decreasing sccumulator flow rate at the time of valve closure. The character-istics of this decreasing flow rate are tased en tests conducted on i 12-inch gate The valve to determine the effect of valve petition on overall line resistance.
sensitivity of the UHl vahe setpoint uncertainty has been rcviewed by the staf f i
4 11
relatise to flow behavior in the upper head. Westinghouse reported the results of a sensitivity study in which it was determined that the additional UH! water delivery resulting f ace an incorrect setpoint would have the effect of lowering the uppe* head fluid temperature and delaying the upper head emptying time. The proposed valve closure model is acceptable to the staff because the dynamics of the flow at that point in time are not important to the behavior of the UH1 system. However, the total amount of UH1 water delivered say be important and its effect on peak cl.J temperature should be addressed in the individual plant a
applications.
4.1.6 Flow Distribution at Core Inlet The large number of fuel assemblies with power densities near the core average is represented in the 5 ALAN code using one coolant channel to calcul. ate average flow condition for these assemblies. As a result, the potential for flow ealdistribu-tion at the core inlet is not specifically accounted for in the calculations, for UHI calculations, it is important that this effect be considered. Westinghouse reported the results of flow distribution tests (Ref.1) that were performed with a full size support column and guide tube attached to upper fuel assembly nottles.
No heated fuel rod simulation was included. Therefore, if some thermally induced flow maldistribution could occur, it was not being consite.ed in the test. These tests demonstrated that the upper head flow delivery to any fuel assembly should be close to the average except for 8 peripheral fuel assemblies, and the flow distribution across the inlet of any one assembly can show a larger variation, fuel rod quenching is the most important heat transfer mechanism dur %g active UNI injection. The data base for the quench criteria included tests witn severe flow maldistribution. i.e., no inlet flow ever one half of the test section. The results of these tests showed quenching on the non-flow side of the bundle.
Quench from the conuniform in'et flow distribution tests occurred slightly later than quench under uniform inlet conditiohn (1 to 5 seconds); however, these delay times are small compared to the active UH1 injection period of about 20 seconds, the scatter in these quench data are similar to that observed for the uniform flow tests. In addition, flow mining because of flashing and hydraulic resistance within the core which would smooth out flow differences between channels was not simulated in the tests. The staff concit.ded that while it recognized the short-coming of the flow distribution tests (lack of thermally induced eechanisms for flow maldistributions) the 'soserved behavior of uniform quenching in a very distnrted inlet flow distribution test was a sufficient basis nGt to require any specific penalty to be applied to core inlet flow dif ferences. The staff con-cludes that the use of a single channel to represent the average core during the active UNI injection phase is acceptable.
4-12 l
. - ~
In addition, Westloghouse, by changes to the model representation, esamined the af fect of connecting the guide tut,es f rom the upper head directly to the core with SATAN Calcula-and without slots connecting the guide tubes to the upper pleaus.
tions (Ref.1) indicated that most of the water splashed into the upper plenue, l
but the ef fect on core flow was insignificant. As a result, the UNI model Based on coneetts the guide tube from the upper head to the upper plenus only.
the observed sensitivit, to core flow, the proposed noding scheme is ac eptable to the staff.
4.1.7 Momentum effect Due to UNI at ico of Core In Reference 1 Westinghouse originally proposed to include the momentum ef fect of the discharge jet from the support columns and guides tubes in the momentum equa-inclusion of these effects would tend to promote tion at the top of the core, flow into the core which is benficial for core cooling. The staff espressed concern about the dissipatinn of the support column jet on the guide tube plug and splashout of water from the guide tube slots. Westinghouse has reevaluated the contribution of these jet effects on the momentsas equation at the top of the core and concluded that they are not significant.
Nevertheless, Westl yhouse has deleted this somentim effect from the model and the staff has concluded that the deletion of this somentum flus is conservative and, therefore, conforms to the requirements of 10 Cft Part 50, Appendia K.
4.1.8 (f_fect of Nitrogen The staf f empressed concern to Westinghouse regarding potential Introduction of nitrogen into the reactor vessel from the UH1 system and asked that assurance be provided of compilance with the requirements of Appendix K.
Westinghouse showed the ratio of dissolved gas to water in the UN! system is exceeded by that in the cold leg accinulators citing bCAP-8471, in which the ef fect of nitrogen in the accumulator water was found to be negilgible. Westing-house has provided technical justificaticn that the ef fect of convection and molecular diffusion, to increase the anotnt of nitrogen dissolved in water in the IJHI systee, is neglfgfble.
Based upon the foregoing, the staf f con:1udes that dissolved nitrogen will have negligible ef fect on conditions in the reactor vessel during a LOCA.
4.2 Upper Head and Upper Plenue footying l
This period begins with the isolation of the UH1 accumulators (about 25 seconds)
)
and ends with the depletion of water in the core and upper plenum (about 90 It is seconds for perfect mining, about L0 seconds for taperfect alming).
dominated by emptying of the upper head and cort 6nuous water delivery to the core.
4-13
Core flow dynamics during this period are determined by conditions in the upper head during active UMI injection. If there is finite (imperfect) elair:q between the LMI water and that in the spper head, a discrete volume (for example, a layer of water) of saturated water would emit the ugrer head at the time of UNI isolation. This hotter volume would then flash into steam and produce a continuous blowdown of the upper head at the time of UMI isolation and maintain this continuous blowdown of the upper head at high flow rates as indicated in Figure 4.4 If perfect mining occurs in the upper head during active UMI injection, the fluid in the upper head will be subcooled at the time of UNI isolation (assuming a 1000 cu. ft. UHI delivery). This water must then be heated in order to effec-Lively drain the upper head. During the heatup period, the flow rates down the After the fluid in the support columns are relatively low (about 100 lbs/sec.).
upper head is heated to saturation, the flow down the support columns will increase to high flow rates (greater than 1000 lbs/sec.). This behavior (shown in Figure 4.5) would result in a longer emptying period. If the total UHI delivery is reduced, the heatup portlon would be shortened because of the higher upper head fluid temperature at the time of UN! isolation.
Westinghouse performed plant sensitivity studies to determine which model would produce higher peak clad temperatures. The analyses s'ows that sometime, one a
model gives higher PCI and sometimes the other. For the C,s 0.6 break, the perfect claing case has always been the highest. However, this sensitivity is dependent on other assumptions made in the model, some of which have not been used for an entire break spectum. Therefore, the staff will require that plant specific analyses be performed consideriog both models and the more conservative used to satisfy 10 CFR 50.46.
4.2.1 Upper Head Water Heatup At the time of UNI isolation (about 25 seconds), the upper head is fillcJ with For subse-subcooled water based on th, perfect mining model in the upper head.
quen*. times, Westinghouse has proposed a siphon model to analyze heatup of the water in the upper head. This model is t,ased on the dif ference in density of the fiuld in the support column (water) and guide tube (two-phase) and the relative gvide tube location with respect to the support column. The guide tube inlet is cifectively about 4 feet above the bottom of the support column and it ends about 6 feet above the top of the support column (see Figure 4.5).
The model assumes that steam passes up into the upper head through the guide Water flow tubes, as indicated in Figure 4.5, and condenses in the upper head.
down the support column is based on the steam mass flow rate, density change in the upper head, and flow into or out of the upper head across the cooIIng jets.
If the subcooling in the upper head is greater than 35'F, the energy input to the When the subcooling is less than 35'F, the upper head heats up all of the water.
4-14
~ _,
i 3500.0 e
0.8 DECL BRE'K FINITE MIIINC A
g g
PERFECT F'XING gy g,trER H M ~ ~ ~
3000.0
/
8, es 2500.0 u
p it si
- e.
a I 2000.0 f
,g aan t e D*
$ e 4
e E
8 a 1500.0 g
p 5
g e s.'
'sl 4 e
e n
en/
- ) -
1000a T
l r',
500 0
- f.,,e e
l L
00 00 20.000 40.000 60.000 90.000 100.000 TIME (SEC)
Figure 4.4
$UPPORT COLUMN FLOW RATE 4-15 l
1
W i
l 1
MV 315 E
y s
O#e O I 5
if
[
.a R>
8 r
5 i
s.
om.
O h
g I
l 3
g l
5 Q
O e
=
i W
Z O
3 N
E k
e 8
8 y
8 l
5 5
0 d
t a
/
=
l ll l i
i g
t
- --_ nsus
'l
~
b g
a e
I e; le.
E 8-
-8 l
N l N W AT ER l8 8
a l
I i 4l 4-1G l
The energy input is energy input heats only the water above the guide tubes.
based on the steae flow up the guide tubes and up through the cooling jets from When the fluid above the guide tubes becomes saturated, the downcomer region.
rapid ieper head emptying occurs.
Tests were conducted in Japan (Ref. 15) and at Westinghouse (Ref. 1) to demonstrate the validity of the heatup model. These tests emanined the water heatup in a simulated usper head region when steam is flowing up a guide tube and water is flowing down a support column. The Japanese tests used a half scale
.arter scale simulation of a single guide tube and 2 Support columns and a simulation of 4 guide tubes and 7 support columns at a systes pres'ure of about The water level in the usper head was plotted as a function of time.
30 psia.
During the heatup portion of the test, the level decreased very slowly; while there was a rapid level decrease after the upper head reached an equillbelum The end of upper head heatup was defined by entrapolating backward the coniitton.
slope af the level-time curve (during rapid emptying of the upper head) to the in9rsection with the initial upper head water level. The upper head heatup time,, so calculated, were plotted as a function of inttfal upper head subcooling based on the steam pressure in the lower plenum. This comparison appeared to support the Westinghouse heatup model. However, it was noted that the water temperature above the simulated guide tube did not reach the saturation teepera-ture of the steam in the lower plenum at the end of the heatup period, but rather a lower temperature that was consistent with the reduced pressure in the upper This reduced saturation pressure is reasonable since the head during heatup.
primary pressure drop in the siphon loop during heatup is due to steam flow across the orifice,at the top of the guide tube.
When the heatup times obtained in the Japanese tests are plotted as a function of subcooling based on the reduced pressure in tha upper head during teatup, it is noted that the Westinghouse model overestimates the upper head heatup rate by The explanation of this result is that some of the water in the upper 10 percent.
head below the top of the guide tube is also heating up during this period.
The Westinghouse tests used a fuls scale guide tube and support column and were performed at atmospheric pressure. The heatup times measured at low initial sub-cooling are not conservative. Here agatr. the definition of subcooling based on The staff attributes the the steam pressure appears to confuse the situation.
discrepancies at low subcooling in the Westinghouse tests to two phase flow effects at the support column inlet. Sirce actual reactor conditions would include higher pressures (greater than $5 psla) and initial subcooling above about 20*F, these discrepancies are not considered pertinent to the UH! egaluation model.
In order to assure a conservative drain model, Westinghouse has modified the heatup model to decrease the heatup rate by 10 percent fer subcooling less than 35'F based om the Japanese results noted atove.
4-17 1
o I
L
A single control volume is used in the L>t! model to represent the guide tube.
Because of staff concern about flow conditions in the guide tube during the be.itup period, Westinghouse perf ormed several analyses of guide tube conditions using up to 20 control volumes in the guide tube. these analyses indicated that the fluid density in the guide tutie is conservatively calculated by a single coritrol solume (the difference in elevation heads determine tie flow rate) and substantial amounts of water (low flow quality) are t erried ir.'.o the upper head f ece tt+ guide in a separate anal > sis.
tube during the first 10 seconds of the lettup period, Westinghouse slowed that the energy flow into the upper head during this perind is The minimum er,ergy flow a functlon of flow quality at the gulate tube outlet.
sitturs for a tursogeneous fluid denfity of Ahnut 0.5 lb/f t ; the energy f!w about 3
aloubles for all steam flow and Inu eases tunsidesabi) for high fluw densities.
Based on these analyses, Westinghouse coe.cluded that the assumption of pure steam flow through the guide tubes is a seasonable and valid model for application during the f.eatup period.
In teviewing this supplementary inf ormation the staf f cnncurs with Westinghouse that the controlling pressure drop ouurs at the guide tube outlet; the elevation term in the guide tube is cuaservatively calculated; and the steas velocity across These the guide plates is significantly abose the flooding vel uity threshuld.
results tend to support tne Westinghouse pure steas flow sodet; however, the staff does not toritur that the model consersatively reflects it,e possiDility of reduced energy flow into the upper head t'etauee tf possible liq 9id entrains.ent.
It was noted abuve that the mininum enert,y flow into the apper hesJ occurs wt.en the eisture eh nsity is about 0 's It'/f t which indicates a bivi quality t e phase 2
scoe liquid entraiement in the bioh velocity stere flu in the tpW+ tubes flew can reasonably be espected. 1herefurt, a more conservative nytel shoult Se b.a wl on minimum eneeqf flow into the ug.per bead which would e stend the upper f.ea f In the absence of applicable emperimental datt, the staf f reNuiseif heatup g.eriod.
fhw that tne upper head t.catup period be entefuted by ceploying minier.e energy this mndels up the galde tube ml across tbc couting jets durir.g this period, eestriction is based on sensitivity stud'es which indicate that a lirulunge1 upper j
The cuvient Westinghuae head drain period irvreases peak clad teeperatures.
model wiw includes this requirement.
In an earlier plant antIvsis, it was v oted that water containej in the gueJe tuve Ite staff control volume drained into the upper plenum during the heatup period.
l f
t elieves this to tie unreastistic becaose the steam velocities ecs oss the c9ide plates are significantly above the flooding velocities.
therefore, as J further restriction, any mates corit ained in tt e guide tube contr ol vo?uee must (** h=alJ he ad by there during the heatup period (escept f or watet* cairled intu the uppen This vestriction minimites the water flow to tw r. ore the minieue energy flew).
during the heatup period and increases the possit ility of core unquench.
4-15 f
I
.~..- --,
m
4 The control volumes representing the upper head, support columns, and guide tube-accwnt for tuc phase fl.*w ef fects, structural heat addition, and appropriate hydraulic resistances. Ihis aspect of the model is consistent with other sys'ee noding and is acceptable for the UHl LOCA analysis.
f With the restrictions noted above on increased beatup time, minimim energy flow up the guide tubes and through the couting Jets, and water holdup **e the guide tihe, tie staf f considers the upper head water heatup model to te conservative and ecceptable f;. the inti-10CA evaluat ion ew.edel.
This model is only selevant when sutneoled conditions are calculated to occur at the top of the upper head during active tWil injection.
4.7.2 Upper Head _8mpth the water above the top of Upper head emptying commences when the temperature of the spide tubes reaches saturation an't there is a sharp increase in water ficv dmne the support columns. The plant analyses indicate that sptem blowdown is still occurring at this time. Ih's depressurir3 tion causes flashing in the e.yer head and an espulsion of water and steam through the *Nide tubes as well. After p
I flashing has dissipated the escess energ) in the upper head, steam flow up the guide tube will be initiated and high water flow down the support columns will t,e maintained until the upper head is emg ty.
Du.'ing the (Qper head emptying phase the steam flow up the guide tube is re duced because no condensation otters in the uiter head; only a volume eschange taelween steam and water is reqtaired.
1he two control volumes representing the upper head account for separation of water and stene eiuring Lt is perind ncf the flows in the sgport columns arut *;uede tube are calculated based on byJrJulit insistances and pregeures in the syst(m.
Beuwse of the r educed steam fle,w in t he q,alde tub >$ ducir.1 the reptying phne, tequired that the tesidual Waltr drain out of the gulde (the into ll*
the staff upper glenum.
Because the upper head empt ing calculations are based on appropriate refing atwf seas nable and appropriate 'ydraulic scpresentations, the staf f concludes that these calculations are acre.te' ale lor thc UHI-t0CA analysis.
- 4. 7. 3 Heat fransfer 4,7. 3.1 f ilm isniling at (^
Pressure Westinghouse proposed to use their hiLh 5: essure film teolling correletion (describro in Section 4. l. l.7) at low preuures f or flow ernlities up to 10 percent.
At flow qualities above 10 percent the high pressure film t:oilinq correlation will A s.ndified be n!'fied by multiplicative f ar.ters bated on the sow precure data.
sersion of this correlation is appliec to counterrureent flow. The requirements of 10 CF R part 50, Appendnu K, are described in Section 4.3.1. 3 an 8 are satisfied by tie esperiwntal d et s dMowed hef t.w.
4 1*J
d 4.2.3.2
$perimentalVerificatfor.
Westinghouse conducted a series of tests at pressures from 100 psia to 20 psia to suppo't the use of the UH1 heat transfer correlations and quench criteria at low pressures (Ref, 16). The tests were performed at di*ferent fined pressures except for two " blowdown" tests where the pressure was reduced in a manner siellar t'o that predicted during a LOCA. The test section was the same as that used for the high pressure blowdown tests (Section 4.1.1.4).
In the Tests were conducted with flow ~n both cocurrent and countercurrent modes.
former, flow of steam a'id liquid was limited to the downward direction by closing the hot leg and openir.g only the simulated cold leg side. In the latter case Part of (countercurrent) both legs were open, permitting flow in both directions.
the test program examined the ef fect of flow distribution by using only part of A vertical baffle was in-the injection ports at the top of the test section.
serted in some of the tests to prevent crossflow at the top of the test section.
Emperleental heat transfer coef ficients, heat fium, and fluid conditions were cal-culated in a manner siellar to that used for the high pressure tests. (as dis-cussed in Section 4.1.1.4), for the cocurrent flow low pressure tests with in-jection from all nozzles, f or tests *nvolving flow distributics and/or counter-current flow, the SATAN code was u',ed to Calculate the flow conditions in the test section, and then LOCTA was used to interpolate at intermediate elevations not provided by the Coarsely noded $ATAN ana?ygis.
A confidence level of 90 percent is required, that at least 50 percent of the measured heat transfer ccef ficients emceed the predicted values. Data were pre-sented relating both rieasured and predicted heat traesfer coefficients to the parameters of flow quality, mass flow rate, and pressure.
There-The original correlation =as not conservative at low (20 psis) pressure.
fore, it was necessary to modify the correlation tr a constant f actor in order to provide sufficient conservaties. From 63 psia to 100 psia, another f actor was used to multiply the esk.ated heat trar.sfer coef ficent while matntain8'ig the confidence level at a value above 's0 pe. cent.
With the above podification the staf f concludes that the proposed correlation is in confo.wance with 10 CFR Part 50 Apper+ dix K, and can be used for calculating film bollirq heat transfer coefficients when cocurrent down flow occurs over the full length of the fuel assembly fur which the correlation is being apf ine staf f accepts entrapolation of this correlation from 20 psi to 14.7 psi with the defined penalty f actor t::c:9se this entrapolation is less than that accepted for approved correlations identified in paragraph I.C.S.b.
4 20 i
t l
l
=
4 i
i Application of the proposed correlation is limited to the fallowing range af parameters:
8 (1) Mass se.xity:
? 5 lb/sec-ft (2) Pressure:
!00 psia to 14.7 psia (3) Flow quality:
0 percent to 95 percent (4) Wall teeperature 30C'F to 1700*F for conditionr outside of this range, cooling should be by steam only, neglecting tre preser.ce of any liquid.
Westinghouse also presented '. eat transfer coef fici+nts derivec from countercurrent flow tests wit % uniform f*rm distribution. These tests, which had steam venting free the top and bottom * ( the test section, resulted in heat transfer under local cocurrent as well as 1., cal ccunt'er-currect conditions.
The heat transfer coef ficiat. for test. 749, 750, 760, 761, 762, 7C4 and 767 were predicted fro <e flow conditions calcula.ed with the SATAN code. Thus, the local flow conditions are a function of t'. drif t flus model which is discussed in Section 4.11.
The predicted heat transfer coef ficients were cogared to sensured heat transfsr coefficients, relating both to the major parameters of interest: pressure, flow quality, and mass flow rate. In order to achieve the desired level of confidence It was necessary to amend the basic correlation. This included limiting the minimum coef ficient to a value of 1.0 Stu/hr-f t
- f.
8 The staf f corclu(es that the proposed cos,ntercurrent heat transfer correlation, as modified is in conform.nce wiu, Appendia E of 10 CFR Part 59, paragraph I.C.4.a.
because it is conservatively supporteo by esperimental data.
Entrapolation of the counter cuirent beat transfer correlctica from 20 psia to 14.7 psia is accepted since there is ro observe 1 negative bias at tower pressures and is consistent with entrapolations of correlations identified in paragraph 1.C.5.b.
t ime correlation is limited to the following in range of applicability:
8 (1) Mass selocity 3 0.5 lb/sec-ft (2) Pre.sure 14.7 psia to 100 psia (3) Flow quality 0 percent to 100 percent (no superheat)
(4) Wall temperature, 300*F to 1700*F For flow conditions outside the applicable range, cooling by steam only must be used with the presence of liquid not included in the calculation.
4-21
It should be noted that these tests, both cocurrent and countercurrent are based on uniform flow inlet conditions in the test section. Therefore, the proposed correlations based on these tests do not account for flow maldistribution at the top of a fuel assembly.
Several tests were conducted in the high and Icw pressure test series in which water was introduced over only one-half of the test section inlet. These tests were performet to simulate severe flow maldistribution at the inlet to a single assembly or at the inlet of two adjacent assemblies.
In the case of the high pressure tests. LOCTA was used to calculate average fluid condittoms a:ross the test section, thus smearing out the non uniform inlet condi-tion. Heat transfer coefficients calculated from these analyses were conservative when compared to the data obtained as an average across the test section.
In the case af coeurrent flow at low pressure, a comparison of the uniform and nonuniform flow tests indicated that the heat transfer coefficients for the test with uniform flow exceeded those with non-uniform flow. It was necessary to reduce the calculated heat transfer obtained when assuming uniform flow by the ratio 1:1.13 so that the calculated values for the non uniform cases would provide the desired confidence level (as before) of 90 percent, that at least 50 percent of the data points exceed the calculated ones.
This constant f actor is to tse used in the LOCA analysis only when cocurrent flow esists in the core and there is non uniform trH1 injection at the tcp of the core.
In the case of countercurrent flow, the data from two unifore flow sests was compared wital..ie results of similar tests (similar ir initial conditions with the esteption of uniformity of flow). The ca?culated results for the tests with uniform flow eahibit suitable confiderce levels. The calculated resalts for the uniform tests were then comrared to tre beat transfer coef ficients for the no-flow
$1de of the non-uniform flow tests (which exhibited lower coefficients than the side on which flow was injected--the flow side). Results of this comparison were satisfactory; confidenca levels of 90 percent or greater were determined showing that the measured heat transfer coefficients were as great as or greater than the calculated values obtained from the low pressure, countercurrert heat transfer correlation. ior this reason, the low pressure countercurrent, heat transfer correlation c'oes not require adjustment for non uniform UHI injection.
4.2.3.3 Quench Criteria The fuel rod surface area that is quecebed during active UNI injectICQ #s the only eraa that will be quenched during the blowdown portion of the accident. However, a portion of this quenched area may return to film boiling during the upper head and upper plenum emptying period folicwing tha active llHf injection period if 4 22 O
local flow conditions so indicate. As a result, Westinghosse has proposed a low pressure quench criteria to permit subsegwnt requencning.
fluench data were obtained f rom the low pressure heat tranfer program described in Section 4.2.3.2 and FLECHT SET 8 tests. I,enerally, quench at high well superheats occbrred on only a small percentage of tre rods. The line representing the quench criteria is drawn so that all the data with 100 percent of the rods quenched at an elevation are above the critariin.
Based on this comparison the staf f corc145 that the design cuench line for cocurrent downward flow reflects the emperimental data and is acceptable.
Application of this quench line must te limited to the followirg range of paraseters:
1 (1) Mass velocity:
3 5.0 lb/sec-ft (2) Pressure:
14.7 psia to 100 psia 300*F S, i 170G'F T
(3) Wall temperature Since these criterion are based on cocurrent downflow tests with a uniform flow inlet condition, the application of this criterion is also limited to these conditions.
Westinghouse proposed a criterion for quenching when countercurrent flow occurs in the core bases on data obtained in the Icw pressure refill tests. ho data were obtained where all the instrumented reds were quenched at a given elevation; instead the data represented a fracticn of rods quenched at a glien wall superheat and quality.
The data for low pressure countercurrent cuench was sparse, with a wide variation in the fraction of rods quenched. However, the results were consistent, with the fewer rods quenching at higher superheath and the larger number quentn4g at lower superheats. The criterion line which was established constitutes a conservative quench criterion, allowing only 60 percent of the rods to quench. In addition, a sensitivity study (Section 4.2.4.2) was ccnducted in which large pcrtions of the core were caused to be unquenct.ed af ter the imittal UHI injection period. It was ascertained that unquenching as much as 40 percent of the core issuediately af ter the UNI injection period and as much as 60 percent of the core af ter the drain period had little ef fect upon the calculated peak clad teoperatures. Because of this insensitivity and the experimental data presented, the staff finds the low pressure, countercurrent flow design cuench line acceptable.
This quench criterion is limited to tre fc11owing range of paraeeters:
4-23
(1) Mass velocity:
3 0.5 lb/sec-ft' (2) Pressure:
14.7 psia to 100 psia (3) Wall temperature:
300'F ii.,1 I?00*F (4) 60 percent of the roos quenched Note that, for both cocurrent and countercurrent quench 3t low pressures, the pressure range has been entended from 20 psia to 14.7 psia. The staf f considers that the quench criteria are suf ficieritly conservative and tha's ti., ur.sitivity study dis.ussed above showed that there was little danger in allowing this estrapolation.
4.2.3.4 Retern tc Film Bolling (UnqJenCh)
Westinghouse has proposed to use the MacLeth correlation to determine when the quenched rods would return to film boiling (unquench) for both cocurrent and counter current flow conditions at low pressures. To support this model, Westing-house has preseited data on quenched rods returning to film boiling from the G 2 refill tests. All of these data were obtained daring counter flow tests where steam was vented from both the hot and cold legs in the test section.
In view of the small e*fd of later quench and unquench (up to 80 percent of the core af ter the drain period) on 3,eek clad temperature, as noted in the sensitivity study (Section 4.2.4.2), the staff ac a ts the use of the Macbeth carrelation for determining conditions for return to file an. fling.
4.2.4 fore Fluid Behavior Af ter the llH! accumulators ave isolated, the flow through the core is dependent on conditions in the upper plenum for a s.ajor portion of the upper head and upper plenum emptying period. in all the calculations, homogeneous conditions are assumed in the upper plenue. At the time of the last status report, this assump-tion had not been justified and its ispact on the overall calculations had not been determined.
It is possible that the two-phase misture in the upper plenum will have both temporal and spatial variations in void f raction that must be considered in the core-wide analysis as well as the hot rod analysis. In the core wide anal + sis, it is important to maintain the core in a quenched condition as long as possible, while the concern in the hot rod analysis is a degradation of the heat transfer at the peak clad tempcrature location.
Temporal variations in the two-phase elature are of concern at high void fractions where.ariations may be large under ccnditions of counter current flow. The ef fect core wide could be alternate unquenching and quenching at dif ferent core locations.
444
/
/
!patial variation; in the upper plenum are obvious -the support cglesas are dis-tinctly dif f erent than the guide tubes. The support columns are smaller in diam-eter than the guide tubes and steam up flow occurs over the outside of the support The guide tubes are slotted awl tht steam flow would be up trarough the columns.
In addition, variations in center of the guide tubes and out through the slots.
steam generation across the core durirg counter current flow could produce conunf
- There are dif ferences in formities in the water distribution in ite upper plenum.
the connections of tt.e support coltens and guide tubes to the core hold down plate l
fuel whlo sust af fect the relative liquid fic,w from the upper plenum to the The actual fluid ccnditions (flow and quality) in the whole assemblies below.
The non-UHI Appenjin K model core take on added significance in thc UHI model.
uses only hot pin ccnditions to start the reflood calculation and ts not concerned The 5ATAN model represents the Core with quenching any portion of the f uel reds.
i by 10 fuel ref ons.
This The power of these ten rods is based (n the core power distribution.
enhanced core representation allows a better description of the heat release in Twenty faiAN in ordwr to provide an istprosed Calculation of core fluid behavior.
it.el rod calculations are performed le which two each use the same radial power as ten cf the 10CIAs use the 5ATAN hot assembly was used in the ten 5AIAN pins, fluid conditions as boundary conditiors. The otter 10 LOCT As use tne SATAN As described in average core fluid conditions as bounoary conditions.
Section 3.1. the twenty (CCTA calet.latiors are used to initialise t$e surf ace temperatures for the reflood (alculation. If portions of these twenty rods are quencned or at low temperature just prior to reflood, the flooding rate is i
1 enhanc ed.
To provide an insight to the potential problems, a oescription of flow conditions after UHl Isolation is presented below for a 0.6 OtCL break analyled with the perf ect mining model in the upper heao.
Immediately after active injection, mcst of the core clad surface area is quenched, the core fluid quality is very low (< 10 percent), ami a substantial ine system amount of UNI 11guld has also been detosited in the upper plenum.
The 3P is in the downward pressure is at 150 psia and is depressurittnq slowly.
direction through tte core and remains so throughout the teneat period (12 seconds During most of this time the total dow ward mass flows are a
to 43 seconds).
In addition, rods are not substantial enough to preclude any upflow of steam.
From 43 seconds to 53 seconds the calculated to unquench during this period.
upper head flashes, increasing the down flow through both the guide tubes and The core pressure drop is still substantially downuard during support columns.
A'. the end of this period causing the cocurrent downflow condition to continue.
the flashing period the cold leg accueulator liquid has reached the tottom of the The At this time the dominant mode becomes countercurrent flow.
core barrel.
Water from upper head draining period (55 secc.nds to 75 seconds) also commences.
the upper head is draining directly atreve 112 of tbc 193 assemblies in a cheeserboard fashion.
4-25
,-_._m
.-... - ~ ~...
Ihe period of time which appears to have the greatest potentlal for flow maldistribution is it.e period af ter the upper head has drained. During this period, the upper plenum and the core are drying out and countercurrent core flow the drif t flus and cross flow models are important in describing this
- prevails, Section 4.11 behavior and tend to minimise water flow to the hot assembly.
describes these models in more detail.
Westinghouse conducted cold flow tests to deteralne the ef fect of the ground plate (tte capfer plate hold 6ng the simulateu fuel rods in the test catris) on flow distritntion within the t,undle. flow rates in each quadrant were approsiaately equal but individual channel flow rates as few as 10 percent and as high as The staff considers these differ-250 pe rcent of the normal value were cbserved.
ences to provide additional conservatism in the test data since the enhancement of heat transfer at high finws would t.e ecre than of fset by the ef fects of low flow if such inhalances were to continue f6r the length of the channel.
Data f rom seven countercurrent flow tests showeif the hot spot at four dif ferent locations at the elevation with the highest initial temperature, just above the It should be noted that the heat generation rate is highest at the midplane.
56nce the hot spot moves arour.d the test section, heat tracsfer data midplane.
are not considered spatially dependent and intratundle flow distribution ef fects are not important.
Io resolve the questions of enthalpy and flow distribution effeCis, it was sug-gesteJ that a pherwnenological model f or flow distribution and unquench may not t,e Core flow mal.listr6bution (nuld re ult in water starvation of sections s
necessary.
This in turn would cause these parts of the core surf ace to of the core.
A more restrictive counter-current unquench model would also cause more unquench of the ca.re s uri ne to unquench. lhit would result in a core temperature It distribution dif f erent f rom that pied 6ctcd oy the current Westinghouse model.
is this temperiture distribution at the begir.'iing of reflood which impacts the final peat cladding teererature, Sensitivity studies were therefore prcposed to emplore the effects of a wide The first spectrum of unqin nch characteristics and temperature distributions.
sensitivity study emplored temperatura distribution effeci. (Section 4.2.4.1).
The second study esamined hydraulir feedback ef f ects resulting f rom unquenching The values of various amuunts of the core at discrete times (Section 4.2.4.2).
these parameters were established as tour.Js to a quench data and geometry ef fects, the maalmum amount of core surf ace area assinec t a be artifically unquenched in This these studies at the end of active injection is assumed to be 40 percent.
conservatively assumes that all anembites under guide tubes are unquenched.
Linder counter-current flow conditions af ter the upper head is empty. 2/3 of the This is based on a support column assemblies are assumed to artifically unquench.
This conservative overall assessment of the G-2 counter-current unquench data.
results is a total of up to 80 percent of the assemblies being artificially unquenched by the end of the upper head drain period.
4-26
_~
The ultimate otsjective of this stuffy was to estahllsh, if act$sstry, a ll2iting tenditio,e of these parameters for use a t#tt plant calculations.
4.2.4.8 feeperature Distribution,Sensi,tielty A serisitivity stu.fy using 10 CIA atu' $4tfl000 was designed to eletermine the elf ect of core temperature distribution at tie leglemisvg of reflood (BOCRit) on peak claJJing temperature, the desired retult of this study was that peak cladding temperature showed a simple re'atioriship with average cladding temperature or stored energy, and is relat Ivesly trutpendent of radial and dif al temperature distributinn.
j Westinghouse initially performed the study using the UHI evaluation model, this did out show 4 anwustonic intresse in geak cladling teoperatuse with stored energy.
1his was first attributed to the conservative stees tooling model. The stw1/ was nest pertoemed using fif CHI results cely, with similar results. It was suggested 1
that the calculated time to quench the 6 foot elevation (I/tQf ) used in the flHHi currelation was not suitable for HHI tumlitions with large portions of cold clad-ding surf ace at the treginning of rcflcod-Iberefore Westinghouse proposed to use the hRtil000 calculated 1/lG6 instead of that generated as part of the f L(CHI torrelation. It was sho=n by compariton to applicable data that this setth>d of calculating f/lQ6 and fif(Hi beat trarsfer was Sultable for this study. In order to Isetter assess the real sensitivity to stored energy, it was reasonable to delete the artificial tw.nervations required tiy paragraph 1.D.5 of Appendia it.
Of tourse the appsoved model contains the se coe" vatises.
Appronseately to calculations =cre pes foiwd using various temperatuse distribu-tions 4t (W.RtC.
lhe results stea d a well behaved but not monotonic det.eadence on stated energy. Of intriest was the f act that the minimum core stnred energy _
did not produce the minia.us peak clad temperature, this result was also observed in toc previous studies. Itowever sirwe only fil(HI teat transfer and a realistic I/IQ6 were used for this stud /, it is net attributable only to steam cooling.
Ac t ually, ref loesd entrais. ment has.dnntages and disadvant ages. Il the stored energy is low aval the entrainment is low precursor (noling at the higher elevation is low.
On the othtr ha#wt, the lower entraineer.t allows a higher flooding rate, at least for a while, f or the siestins. house analysis the optimum stored energy allows suf fit icut entrais. ment for early precursor cooling t.ut not 50 much to retard early taucegh front rise, the tease case calculations which wire dove prior to this study were esamined to assuse that they were consistent with this study. By repeating the base cases with reflood heat transfer model used in this stmfy, it was shown that the base case stored energy was very low anti resulted in consistent peak cladding temg era-tures (see Iiguve 4.6).
4 21 J
1
i a
e i
j
' ']
I I
l l
l j
J1
- Pr4PpCT Mlf lN 4
-.j..,i.re..m.en_ N o.
..I f
pW l
. pi j-
-t--2eo*
-- j i
j--
f6 t
t t
,--..g,.__..__._
i 4
J l
! casal :
- Y _-4_ _j a
._. h.-preo 4._._ e ___.
_._.4_..
I s'
j g {
{
)
- ;,.j i
e4
_4.*
i
_4
_+_.__.-
J i
i i l+ t3,_i.
j.!
e i
_ g _. i.
y* t i
E I i
l t
t i
_g-n Iq f'
}
i i
$f a i l
I 1
u 4
l
__Jg._ glw I
t 1
I f
f I
1
$ f I
l I
__ a, _ _
i k
i l
l
-- 400 100 we,
soo eso. Too g too t
_m._ i _4 _. _J. _.. i _. _ _ d.. _ _ _ _.. L.-
l
} breiGHTe o CL. AeolNG f tM PERATU Rf p
AT BtM Rec -?F _ (570se0. rMa 441f t.
i I
i 1
1 1
i 1
i I
i FIGURE 4. 6 Peak Claddirg icoperature vs Stored Energy at BOCREC I
l 4 28 l
l I
e
The results showed that the stored energy would have to be increased by artiff-
,,e' clelly imposing flow maldistributions or poor heat transfer over at least 1/2 of the core before an adverse offett would be observed. It should be noted that this study did not address hot pin effects which will be addressed in Section 4.2.4.3.
The well behaved dependence of PCT on stored energy made it possible to esplore hyde.nlle feedback effects with a limited number of calculatlons. The study also showed that a large fractir,n of the core would have to be at elevated temperature to adversely affect PCT.
4.2.4.2 Hydraulle Feedback Sensitivity The studies in Section 4.2.4.1 did not model any hydraulle effects that would result from artificially unouenching the various groups of rods.' When rods are unquenched, the steam generation is less. Less steam generation should allow more water to penetrate the core and fall to the lower plenum, thus hastening the onset I
of reflood. This postulated water peeetration benefit therefore has the potential of offsettirg the potential detriment of higher temperatures at the beginning of reflood caused by unquenching. Therefore a study using SATAN as we!! as LOCTA and WttFL000 was performed in which the hydraulic effect of unquenching large sections of the core was modeled.
For this study, three large channels were defined in SATAh in addition to the hot assembly. One channel was typical of assembiles below guide tubes and represented about 40 percent of the core. For all of the calculations esca-t the two base cases no water was allowed to penetrate this channel af ter acsive UHI Injection.
This is regarded as an estreme condition of flow maldistribution for guide tube assemblies. The other two channels were typical of support column assemblies, with flow paths to the support column volume. One of these channels was artifl*
clally disconnected from upper plerum liquid af ter the end of upper head draining.
ha cross flow was allowed for the Imperfect mining cases. However, steae crossflow was necessary for the perfect mining cases to maintain hydraulf-stability.
For leperfect mining, tha espected reduction in stese flow was observed. As a consequence, end-of-bypass was reached earlier and the by pass deficit was smaller, resulting in earlier 80CRfC, lower stored energy and lower peat cladding temperature. The results were very lesensitive to the fraction of support column asseabiles unquenched af ter drainirg the upper head. The resulting stored An energies were in a range which resulted in minimum peak cladding temperatures.
additional hydraulic feedback benefit is that the drain rate through the core is reduced since at least 40 percent of the core is not allowed to participate in l
I draining after active injection. This keeps the other 60 percent quenched longer.
For imperfect al Alng the unquenCh time is synonomous with end-of*5ATAN. Thus if unquench n celayed while the downcomer is filling, this 15 a definite benefit in its ef fect on the bypass deficit (sec Sectitn 4.2.5).
4 29 I
I
For perfect slaing the reduced dra's area has a differect benefit. $lnce more water reeains in the core af ter the upper head and oper plenum have drained, it takes longer to unquench and the time between unquench and end-of $ATAN is shorter. Thus reducing the stores energy. Also more water is concentrated in (f.e afsture filled channels because the gt.ide tube channel acts as a natural chiency venting the steae generated in the other channels. Once again, the stored energy levels are within a range which does not result in higher peak cladding temperatures. As would be especteJ from the stored energy study, having 40 percent of the core hot and acting as a steam vent does not have a negative effect on peak cladding temperature.
if.eSe studies show that for imposed severe flow distribution, the resultant stored energy due to hydraulic feedback does not result in peak cladding temperatu es higher than when base case stored energies are used. fherefore the staff concludes that an artificial flow cistribution requirement for the purpose of calculating stored energy at BOCREC ts not needed to assure conservatism.
4.2.4.3 Hot Pin Flow Distribution Effects The stored energy and hydrartic feedback discussed above have their ultimate ef fect on carry-out-rate fraction and flooding rate during reflood. Flow distri-t>ution and unquenching also have a direct effect on the temperature of the het rod. The LOCTA hot rod temperature calculation receives its hydraulic Loundary conditions from the hot assembly in the SATAN calculation. Therefore any arti-ficial flow distribution effects imposed on the hot assesely has a direct penalty on the hot rod and indirect ef fect on 40 percent of the stored energy calculation.
The staf f telieved it would not be possible to assure sufficient 11guld delivery to a given rod such as the hod rod to guarantee tuench throughout the transient.
Therefore, it has been required that if the hot rod is quenched at the end of active upper head injection, it shall be unquenched from that time until reflood.
Westinghouse proposed to isolate the tot asse*bly free the upper plenue after the upper head drain period. This imposes a further heat transfer penalty on the hot rod and the wre stored energy calculatlun. The staff also believes that such isolation should be imposed during the upper head reheat period. During bath of these periods very little water is being delivered from the upper head, and upper plenum ealdistributions must be corsteered. Therefore Westinghouse has also agreed to isolate the hot assembly during the reheat part of the calculation.
lhe one remaining period for ehich there was a question was the post flashing upper head drain period. During this time interval, Ifquid is being delivered from the upper head only thorugh the support columns. That is, delivery is in a checkerboard f ashion throughout the core. This type of situation was simulated with the non unifore low pressure teat transfer tests. Recently these tests ha,e been reanalyzed with the approved 1ATAN drif t flun model assuming the delivery mas 4-30
~.. -. _ - _ _
unifere across the test assembly. Heat transfer and toeperatures were conserva-tively predicted on the no flow side of the,e tests even with the analytical ass e ption of uniform delivery. In this case, the no-flow side can be thought of as a guide tube assembly such as the hot assembly which receives no direct flow An alternate plant calculation was performed with steller during draining.
cnalytical assumptions. Upper read flew to tie hot assembly was determined by dividing the total support column flow du?ing LP.e drain period by the total number of asseabiles.
The results showed that the alternate case calculated higher heat transfer coeffi-tients than the standard case. it was also shown that almost no flow wasThus delivered to the hot assembly from the upper plenum in this alternate case.
the standard case is conservative compared to an alternate analysis which closely slaulates *;.e conservative analysis of an emperteent designed to simulate this Therefore the staff concludes that no additional restrictions need be situation.
applied to the hot assemely to account for uncertate'.ees in flow distribution and e..49enthfng.
4.2.5 Reffil 4.2.5.1 End-of-SAfAN Calculation in the approved standard Wntinghouse ECCS evaluation model, the SATAN The calculations are terminated shortly after end of bypass has bee.i dete ef.ied.
end of-$Af AN (105) la the non-UH1 model is taken as the earliest of the following:
(1) Downconer downflow is larger than the ECC$ Injection flow.
(2) lero vessel side breek flow first occurs.
(3) Botton-of-core recovery (BCCREC).
For large breaks, (1) and (2) occur before (3). In practice, E05 occurs within a second or two of end-of-bypass (E0BY). Appendia K requires a physically based model for the calculation of refill and reflooJ. For this reason certain delay Lines were identified by the staf f to be e;. plied at the beginning of the WREFL000 calculation in the non-UNI model. These delays are:
An amount of water equivalent to that injected into the systen prior to the (1) end of bypass must be subtracted from the total vessel in ventory, and there is a delay asseclated with refilling the intact loop cold legs and the upper downconer annulus.
(2) Delay time associated with free fall of water from the top of the downconer to the lower pisnus.
l (3) Delay in water delivery to the lower plenus due to water held up by steam generated at hot walls.
4-M i
e
In a UMI plant, the $ATAN calculation is estended considerably beyond (08Y so the water in the upper head at that time can be accounted for properly. The duration of the 5ATAN calculation beyond (08f is strongly influenced by the upper head
. taing model and the way in which (CS 95 defined.
Two models for mining injected liquid with the upper head fluid have been examined by Westinghouse (Section 4.1.3).
The first is the perfect misteg model where the injected liquid is homogeneously mised with the upper head fluid at each time step during upper heaf accumulator injecticn. Ine second model, finite mining, does not mis all the injected UH! fluid with the esisting hot upper heaJ liquid but i
deposits a layer of some cold liquid at the bottom of the upper head. Because the hot fluid layer on top in the finite elming mojel can flash and pressurire the upper head, the upper head is emptied socner than is the case for the perfect mining model. This leads to en earlier (0$.
Westinghouse Pas used three definitions for (0$ at various times which impacted the calculations in various ways. The original definition is the time at which the upper head is empty (Method 1). The difficulty with this definition is that a substantial amount of water remains in the upper plenum and Wlifl000 does not have a satisfactory method for handling this water. In a subsequent proposal. 10$ was defined as the time at which the mass inventory filled the lower plenum within the core barrel with water (Method 2).
This led te, much later times for E05. In f act, the times were so much later, tPat it was apparent that reflood had already begun. Calculations performed with Method 2 st. owed that the perfect mining model always resulted in the highest pean cladding temperatures.
To obviate the artificial $ATAN reflood problem another method was tried (Method 3),
This method defined (OS as the tire at w5ich upflow is calculated to occur between the t.o lower plenum volumes. This occuns when sufficient water has b n dissi-d pated from the core and upper plena which in turn removes the major resistance to reflooding and alle=s water to rise within the core barrel. A refill procedure was then propcsed to calculate the tis.e it takes the water to reach the bottom af the core from the bottom of the barrel. This procedure accounted for liquid acceleration, steam volume replacement, and steam generation due to hot metal.
.lowever, it was noted that the rate of liquid rise calculated by SATAN at this time was slower than the rate c.sIcalated by the refill procedure. This indicated that some resistance was still not being accounted for in the refill procedure.
This resistance could not be entirely attributed to artificial 5ATAN reflood water entering the bottom of the core.
I Thus, a fourth method has been pro esed which allows $ATAN to Continue to 80CREC j
without an interim refili procedure. This method imposes phase separation at the core inlet ficw path, which orchibits premature SATAN reflood. Using this condia
(
j tion. [0$ is defined as the time wten the lower plenum volume inside the core barrel reaches a density of 59 lbs/ft. Continuation of SATAN to this condition 8
4-32 l
l t
4 Is virtually synonymous with 80CR[C and assures proper accountieg Cf :11 system resistances during refill. for these reasons, the staff concludes that this fourth definition of E05 is acceptable.
4.2.S.2 Pefill Celay flees It is now necessary to determine if the b1tEFLD00 Initialtration is being done in a manner consistent with the requirements of Appendis ( end if the temperatures at the beginning of reflood are be' wJ deterefned satisfactorily.
Appendis K requires that the refill and reflood calculation take into considera-tion the appropriate thereal and hydraulic characteristics of the core and reactor For non OH! plants some of these characteristics are accounted for as systes.
delay times described at the beginning of this sectien. Since the SATAN CaICule*
tions are extended into the refill period when analyzing a UH! plant, the delay times described above are ace)unted for in the SATAN calculation in the manner described below.
(1) Delay time associated with refilling of intact loop cold legs after bypass deficit has been subtracted.
The " required" bypass deficit is calculated by determining the amount of ECC5 accumulator water which collects in the " bypass region (Intact loop cold legs, upper downcomer annulus) starting from the tie. of accumulator injection up until the time of "end of bypass." Separate mass a.Mi energy inventories are computed for (CC accumulatory injected water and for water which was initially part of the hot reactor coolant system inventory, so that only ECC accumulator water is included in the final deficit.
After the end-of bypass, the SATAN-VI code predicts that a Ce*tain amount of injected water is spilled ce* the break. It was Westinghouse's position that if the post-bypass spillage after filling the vessel was equal to the deficit, that the rule.oald t.e satisfied and no artificial water removal would be required. The staff believed that it should be denonstrated that the final end-of 5ATA% inventory would not be affected if artificial water removal was apolied just after erd-of bypass. The adequacy of the SATAN j
calculation for discarding infected water was demonstrated by comparison of l
the unaltered SATAN calculatten to a calculation for which the deficit was discarded at the end-of-bypass by providing a lower plenue drain. The results indicated that artificial removal of water from the lower plenue did For not materially affect the water inventory later in the UHI transient.
both cases, the same amount of water was lost from the bypass region. Either the water was removed oy draining or it was lost as spillage from the break.
l 1
Also, the rate of vessel refill was not materially affected by the artificial d;scard. Therefore, it is possitie to compare the bypass deficit with the 4 33 l
t
amount of spillage af ter ifw vessel is full in it.e unaltered 5AIAN c alculation. If this amount of spillep is not saf ficient to of fset If.e deficit, the reaalnder of the deficit must tre recoved at the t,eginning of 1stif t000. Suf ficient in<entory it.fs.wation me.st be trovided for each large break plant case, to verify the spillage credit. this agproach assures that the ICC water la.jected during the b) pass period is subtracted either b/
spillage or of assumption af ter it.e end of bypass, ar.cf properly conf orms tu j
Ihe calculattans demonstrate that for perfect alaing paragraph 1.C l.c.
cases, the specific nument of deficit accountability does r.ot af fect the l
water inventory available at 20CitC. It should be noted, however, that for finite mistrw) cases, the spillage is f requently not suf ficient to accwnt for the deficit. Iherefore, the anot,nt of wa'er notess ry to en6e up the deficit for any case will t,e removed at the t.eginnIn.) of WRifl0CD.
(2) helay time associated with free fall, lhls delay time is accounted for in t
aAIAN somentum equation calculation 9.
(3) Not wall delaf ties. Ihe mechani,as. hic h leail to t it wall delay (i.e., heat transfer free twt metal, floodino) are simulated in the SAIA8s calculation.
j In the approved standard Vest 6nqNo.se model the hut well delay time is deter-sIned rne the Cff AR( bat wall test results (Ref. 1).
Io show that t'w time delay calctlatcJ t,y 5AT Af1 is conservative Westinghouse presented the results of calculatier.s cieg.aring the SAf Ah detessined delay tier s.8th the CREAR( dela/. 1he predicted itela' times wese f rom the end-of t ypass to the onset of mass accumulat f on in the In=er plenum f or teo cases of double eemled q.sillotine taeaks. the delay calci.lateJ 14 the SAfAH code inclialed e f fec ts of free f all. 'et wall, snd steam flow f ar.e the co e.
Westinghouse has made 4 siellar t rag arisor, for the large break spectrum with siellar results. This is an acctptat,lc meths.d to assure that these ef fects are felng satisfactorily accounted for in 5AIAN.
4.2.5.3 Down Comer thfet Iwo aspects of (CC cold leg 4 cia.nlatt r 6njei t ton eetated to the downc.mer are treated dif ferently in the LAtt model as cc,mpared to t.ee r.re.PIMI model. F irst, the proposed Westinghouse nadalitation splits the denncsmer into separate floo paths
( 4ction 4.8).
5ecnrut, the tml drif t f!sa model (Sectic.n 4.11) considers a j
separate formulatirm f or the downconer. Preliminary staf f airlit calculatio ns (Section 7.0) have indicated that these no.lels may have a substantial imp. set on ECC cold leg delivery, the staf f believed at was prudent to esasine the proposed tot! dow wome* model mora carefully. The following criterit were estaolished for an acceptabte downcomer model.
4 34 i
I
(1) The erut-of-bypass predictions must be accurate or conservathe relative to small scale data amj tte entrapolation to full scale must be based on the sure conservative of the two presently prevailing theories (i.e.
Kutateladie I
scaling);
(2) The ICC delivery rate ' a the lower pleme must t,e conservative relative to small scale data; i
t (3) the predicted water storaye in tie dowrKt,eer and the lower plenum must be conservative relative to small sc.le data.
In sus. port of the UHi downccecr dr.f t flea model. Westisushouse has sut.eitted calculatic.ns of the Combustion Engine ring 1/5 scale LCC t,ypass tests which show that the model conservatively predicts toth end-of-bypass and ICC delivery to the lower planus. Calculations have also t.een st.taitted in comparison with the Creare Westin9 oisse alto performed additional calculations with I/lS-Scale correlation.
h saturated water and sutgooled water t.Vf subcooled) for toth f ast and slow steam
-ramp rates. These calculations are ascurate or conservative relative to the Creare 1/IS-scale data.
All of the calculstions sutmitted I,y Westinghouse support the model a;, conserva-tive relative to the staf f criteria. Howeven, in each of these calculations, ti:e flow pattern in t*e dowe. comer is mar 6cJ1) asymmetric, with (CC vater flowing down the side of the downcomer near the intact loop ami a steam water min'ure flowlsig up the side of the dowe<omer associated with the broben cold leg. Although this flow pattern may be viewed as a possitle flow pattern in it.e full scale rec tos, Ifie (CC it es v.ot representative of it.e flow g.atteins in the small scale tests.
efetiverv in the small scale tests es g vemeinantly sp.n adic arwl s cametrat. lie f ivw pattera asal the Ju neuwer nuding wat thee rf or e cor.siJued to t e an unconfirmed modeling acumption used to achieve an overall conservative calculation of ICC downtonar flow. $irse the finw pattsrn resulting f rom the thiwrwoner drif t flua model has not t>ee-n verified by ess.eriments, no t,enefit can be talen f rom the possible change-in system ef fects (such as d.r.ncoact rrtial and azimuthal pressure gradients) associated with this modeling. Since all of the information sutialtted t'y b'est inyhouse hat suppried the conse.vatise of it'e medel and since no nonconservative systen ef f et ts have been identified to date, the i,taf f concluded that the f ollowing tra+arisons were the only a fdstional information needed to confirm the saudel:
(1) Comparison (of end of bypass, lo.er pieme delivery, an.1 lower piemm and downtomer amt storage) teetwee.. tfe t>HI strif t flux dowsicomer model calculations anil 10fl fest L1 4; 4-)$
4
(2) Comparison betweco the non-UNI drift flua downconer model and toff Test L1-4; (3) Comparison of the UH! drif t flux downcener model calculation of water st rage in the lower plenus with the Creare measured water storage in the 1/15-scale tests; (4) Comparison of the calculated end-of-bypass and lower plee.us delivery with the Creare transient steam flow tests.
(5) A sensitivity study on a representative plant transient *.c m firm that no algnificant, nonconservative system ef fects are associated =eDh the UHI down-comer atinuthat hoding.
Westi..ghouse performed detaileu compartsons of the LOFT L1-4 esperiment witt their UHI hodel and their non-UNI model for the standard problem progras. Further analysis of this test wss done as requested 'vy the staff for this review. In general the calculations with the (FI model were reasonable. Momentum fluxes calculated in the downtomer appeare3 to t.e within the range of the data, inJicating that the asymmetric tiehavior was not being emaggerated. Lower plenum void fraction predictions were gnerally high reflecting a conservative prediction of that inventory. The annulus void f raction, however, was underpredicted, and the overall inventory in the by pass region was not conservatively predicted beyond 45 seconds. Special phase separation models were utilized to determine the effect of sweepout on the bypass inventory. These models had litt'e influence on the (1-4 predictions. As a result of the analysis of the Creare I/15-scale trantlent tests, a modification 16 the head term in the downcomer crossflow model was developed to more accurately represent the ef fect of a densitf gradient across the dowrcomer. This modification and the Creare tests are discussed beles.
Analysis of LI-4 was performed with this " head-fix" Overall tre results were more conservative. The behavior was less asyrretric than before. The bypass inventory was conservatively under predicted f ee all times la t6e transient.
These results support the inclusion of the head fix in the UHI downComer model.
Westinghouse made comparlslons to the Creare 1/15-scale transient emperients. The initial approach was to mo1el the esperiments with actual control volume size.
However, code stability problems were encountered because of the small size.
The IVR doenComer model was then used with the base llM test paramenters scaled accordingly (i.e., steam flow was scaled by Kutateladze scaling). Westinghouse showed that scaling the experiments to a full slie represcritation was conservative relative to the known parameters. The comparisons showed that the tredictions were not always conservative in te.ms of end-of-bypass and ECC del:very to the lower plenum. Also the correct end-of-bypass trend with transient time was not always predicted. bestinghouse esasined these calculations and observed more asymmetric void fraction behavior than was observed in the tests. Further study 4-36 4
showed that this w s due to a modelling cssumption made with rstpect to calculating the head tire betwern (djacse.t c21muthal dcwnconers volumes. It had the effect of not allowing enough water flow from one volume across to the ither.
An appropriate modification was made to fin the head term in the cross flow calculation. the results and analyses showed that conservative end-of-bypass and transient time were predicted.
For these analyses and the L1-4 calculations it has been shown that the head fix causes the downcomer to behave much more one dimensionally and results in more conservative bypass behavior.
In order to f urther address the problems of lower plenum water storage Westing-house developed a phase separation model. This model is based on the lower plenum voiding tests at Creare and Dartmouth. These experiments measured lower plenum The liquid depression as a function of steam or air flow and system geometry.
Westinghouse model forces the liquid level to depress by artificially lowering the skirt furt".er into the lower plenum. kestinghouse proposed a level depression correlation which was not conservative relative to all the data sets. They argued that the escluded data was very atypical of a PWR geometry. Tne staff and our consultants at Dartmouth agreed that these data could rightfully be excluded since they included distorted geometries which should produce non-representative flow regimes. The correlation was also compared to some very recent Dartmouth data and found to be conservative.
Westinghouse performed a series of plant analyses utilizing the lower plenum separation model and the head fim. The calculations were done for.6 double-ended cold leg break. The results are shown in Table 4.1.
TABLE a.1 DOWNCOMER M00f t PE AR CL A0 TEMPERATURE $[NSITIVITY Perfect Imperfect Mining Mixing Case PCT f *F) SPCT (*F)
PCT (*F) aPCT (*F)
Base 1 2190 1800 Bast 2 2020
-170 1800 0
Head iin 2020 0
1997
+197 L.P. Phase Separation 2130
+110 2150
+153 Each successive case incorporated all changes from the previous case. The clad temperatures increments are from one case to the next to show the effect of cach individual model change. Base Case I was performed ilth the model as it was last August. Since then it was discovered that the flow limitation on low pressure quench was incorrect, it was also decided to perform the perfect mining analysis with a containment backpressure using an approved model rather than using 14.7 psia. These two changes resulted in base case 2 for perfect mining. The 4-37
- Y'
quench II:lt fis was the only change fir bas 3 cIs3 2 imperfect cining. The lower quench limit oily hat an effect on the perfect sizing case when requenChing during the drain period occurs. $1nce the bypass deficit is made up during the drain period of the perfect mining case, the head fin, which affects ECC delivery to the Iower plenum, has no effect. Such is net the case for imperfect slaing and the bead s'la results in a substantial penalty. The lower plenum phase separation model has an effect cm both cases until the Core unquenches just prior to reflood.
As can te seen, the ef fect is in the conservative direction.
From this analysis we believe that the inclusion of both the head fix and the phase separation model is justified. Westinghouse has performed satisfe tory analyses of all the itees required by the staff. The resulting changes have been shown to be conservative and a*e satisfactory.
4.2.5.4 Residual Water The current defintions of end-of-$ATAN allows time for SATAN to disiipate most of the liquid mass on the hot sice of the system. However, for some cases a residual amount of water from the upper t. cad finds its way into the steam generator inlet plenue at E05. This liquid is now accounted for during reflood. Westforhouse has shown that steam flows during reflood are much too low to entrain this liquid into the steam generators. Powever, metal heat can cause the liquid to evaporate during reflood. Therefore, the liquid should be evaporated at a rate appropriate to hsat release from the steam generator plenum walls. This steam must be adoed to the CRF during reflood. Small amounts of Ilquid may also remain in the hot leg and elsewhere on the hot side. Westingoouse now correctly ass wes that this mass is vaporlied at a rate equal to the boil off rate at E05 and is added to the core exit steam during reflood uritil it has been dissipated. This is acceptable.
4.3 Reflood 4.3.1 Modification of Westinghouse Entrainment Fraction Correlation for low Clad Temperatures The Westinghouse mass entrainment correlation described in WCAP-8170 has been modified to account for the effects of Icw initial clad temperature as found j
during FLECHT-SET Phase A testing IRef. 7). These tests provided the data base for the modification. The low clad temperature FLECHT tests show that the point of significant entraient occurs higber up the rod than for high tesperature tests. The entrainment correlat.on, which is a function of the quench front i
elevation, was modified by a multiplier in the quench front elevation ters; the multiplier is a function of the rod ir.itial clad temperature and system pressure.
./
'destinghouse calculates the entrainment fraction as an area weighted entrainment fraction. The core is divided into ten power regions of equal area. During blowdown each of these regions will quench to varying eatents. The lowest power I
4-38 i
I l
e i
m g
regius trill quench along their ettf r3 length whil2 the higher power regions may cuench only at the ends. la the r: flood calculation, the watir I W31 is Compired with the input unquenched elevation for each power region. If the water level is found to equal or exceed the unquenched elevation, calculation of entrainment for that power region proceeds with a T of 1600*F; otherwise the peak initial clad temperature for the region is used. The entrafrment fractions f on the ten r60(ons are then averaged to obtain the weighted entrainment fraction (CRF).
Calcu%tions show a strong sensitivity of the peak clad temperature to quench front elevation. For a typical large cold leg break calculatten, the water level rises quickly c rough the first 3 or 4 feet which were quenched during blowdown and most of upper head and plenum emptying. All the surface temperatures in this region are typically in the 300*F-400*F range at BOCREC which results in very little liquid entrainment during reficod. Temperatures in the next 2 to 3 feet are usually in the 500'F-800*F rarge. at beginning of reflood. This is a sensitive temperature range where the CRF correlation changes repidly. Thus, a few degrees increase can retard the gaench frc,..t progression and result in higher peak cladding temperatures. From *.is it can be seen that the amount of cladding surface quenched and unquenened just below the core hot spot is important.
The modified mass entrainment correlation predicts the quench front elevation well up to a or 5 feet. Westinghouse comparisons of the LOCTA model with cold channel entrainment to hot rod temperatures froa FLECHT-SET tests 1212,1516,1617 and 1715 show that the LOCTA model overpredicts the heater rod temperatures for these tests.
These tests were stepped injection flooding rate tests at approminately 20 psia and low initial clad temperature. Entrainment rates from these te.ts have been ccupared with the modified mass entrainment correlation and agreement was poor, but the clad temperaturas for these tests were predicted conservatively by tSe proposed Westinghouse model. To provide n additional margin of conservatism in the reflood calcualtions, the staff requires that the calculated mass entrainment shall be modified as follows if the clad temperature transient is terminated by the quench front:
(1) The proposed entrainment correlatiur. shall be used for entrainment fractions le a than or equal to 0.5 (2) For :alculated entrainment f ractions, F, in excess of 0.5, a new entrainment f raction, F1, shall be calculated such that:
F1 = 0.5(1+F)
The modified entrainment correlation is acceptable for low initial clad temperature reflood applications. The modified entrainment correlation combined with the Westinghouse steam cooling acdel satisfy the requiremer>ts of Appendia K.D.5 4-39 c-..
for a stian cooling model 13 calculita r$ flood helt transfir fir r0 flood rit;s less than I inch per sscond. The (rza weighting af the sntrainme;t is alss acceptable.
4.3.2 Reflood Heat Transfer The Westinghouse modified It[CHT heat transfer correlation which was approved for This Correlation use in non-Uhl plants is proposed for use in UHI pisnts also.
This quench time is uses the time to quench the 6-foot elevation as a parameter.
much shorter for low ittitial clad temperatures than for the high initial clad temperatures upon which the correlation is based. Over the range for whlCh data are available, the Westinghouse modified "FLECHT heat transfer correlation either models the data or is very conservative with respect to the data.
Cosparisons with FttCHT runs 6850, 7151, 7252, 7354 and 7455 show that the modi-fled Ft[CHT beat transfer correlation, combined with the area weighted Therefore, the entrainment, conservatively predicts hot rod temperatures.
modified Ftf CHT heat transfer correlation is acceptable for use in UH! plants and satisfies the requirements of Appendix K.D.5 for flooding rates in escess of one inch per second, but less than 3 inches per second. for flooding rates in excess of 3 inches per second, Westinghouse has agreed to hold the heat transfer coef ficient constant at 2.5 Btu /hr-f t' *F (Ref.1). The 2.5 Btu /hr-f t* *F value is characteristic of the initial heat transfer coef ficient at the start of Use of the FtECHT correlation for flooding rates in excess of 3 inches reflood.
per second would be nonconservative for UHI plants because of the lower steam The high penetration rate due to cooler fuel rods in the lower part of the core.
reflood rates occur during the time that the lower portion of the core is being flooded and would result in less entrainment than predicted by FtlCHT The Westinghouse steam cooling normalization f actor h* origf rully correlations.
proposed f or the UHI model (Ref.1) is nct acceptable. Westinghouse pro.ded one blocked and one unblocked steam cooling calculation to verify the proposed h*.
This would only allow approval for a containment pressure of 14.7 psia with no A previously approved normalization rcom for deviations from that pressurr factor was presented in (17); this factor is acceptable for both UHI aad non-UHI plants and is not restricted to a single pressure. Westinghouse has agreed to utillie this acceptable factor for UHI plants.
The entrainment and heat transfer correlations presented have been discussed in terms of cosine asial flua shapes. Westinghouse must provide parametric studies to determine the range of asial off set fcr which the cosine shape analysis is The cosine shape analysis may not be applied to any axial shape which applicable.
would yield higher peak clad temperatores than that obtained from the cosine ana)ysis. Westinghouse has agreed to perform these studies on a case-by-case basis as required. If the analysis ce the first UHI plant clearly indicates the axial shape effect, this study may be referenced in future plant analyses.
l 4-40 i
l l
l
4.4 Pr ss-r? Instabilitits The taped parameter method used in SATAN and the assumption of uniform properties within a Control volume leadJ to enthalpy discontinuities at Control volume inter-f aces which are accentuated for a UHI LOCA analysis because of the presence of cold water. These discontinuities result in pressure spikes and numerical instability. Westinghouse has modified its pseudo-viscosity term and a smoothed equatior ;f state in order to eliminate the source of the numerical Instability.
The pseudo-viscosity term provides added impedance to flow at the discontinuity and its gradient is opposite to the velocity gradient. Thus, the shocks are sucared out 50 that the mathematical surfaces of discontinuity are replaced by thin layers in which the thermodynamic prcperties vary rapidly bJt continuously.
This term is used in the $ATAN evaluation model for non-tH19 ants but a dif ferent 1
form of pseudo viscosity is proposed for the UHI model, The UH: pseudo viscosity ters (Ref.1) does not depend as strongly on the speed of sound in each control volume and f or small enthalpy gradients reduces to the previously accepted non-UH!
form.
The inclusion of the pseudo-viscosity term, in the equations of hydrodynamics, permits the stepwise numerical solution of the equations in problems involving shocks. Westinghouse provided an analysis of a non-UHI plant with the approved non-UNI evaluation model having this new definition of pseudo viscosity and com-The UHI pared the results with an analysis using the $A1 AN non-OHI model.
pseudo-viscosity model resulted in a PCI only 20*f higher at BOCRfC. Therefore, the s'aff accepts the new UHI pseudo viscosity model.
Wes'.inghouse uses polynomial functions to fit the steamwater properties in 14 UNI version of SATAN (Ref. 1).
This formulation is used to eliminate discontinuities at the saturation point which would produce local pressure spikes caused by phase Westinghause has demonstrated that t'he combined effects of the changes.
pseudo-viscosity ters and thi continuous equations of state do not significantly change transient behavior when the r3ntrol volume stres are varied (Ref.1),
indicating an overall minor effect resulting from these modifications.
Because the effects of these.90difications were shown to be small, the staff finds use of a continuous equation of state acceptable.
The staf f has noted, however, that methods used to correct pressure instabilities at saturation crossing (water packing) in EELAP4 can have a substantive effect on the (CC behavior. We will continue to p6rsue the suitability of various water packing models in our audit development program (see Section 7.0).
4-41 i
1 L
l t
4.5 tore Subdivision to Accose prits quer.th Westingt.ouse uses the $AIAli code for,':etilcting core cooling behavior during
- 1. lowdown. to account for steam q.:ncration and quenching in various regions of the core, each control volume of the average chantwl contains ten fuel subregions wtere separate fuel rod calculations are made. The subregions use a simple lumped parameter, constant property model wfere within each axial conteel volume the fluid psoperties are assumed constant as in the standa d eajel, fach hot charnel uses a rware detailed f act red calculation identical to that used in t** note-UHI mode 1.
Westirghouse per formed a senstivity slicly using 5 instead of 10 subsegions (Ref. 1),
lhe results f ruticated that the 10 region model was mov e conservative.
Because of these resolts, the staf t' cs.ncludes that the 10-region core division is acceptable for tra UHi LOCA analysis.
4.6
$,1ip p ect5 Darf_ng f edwn
.In a 10CA with UHI, trk large amount af water injerted into the core results in low spalities throughout the venel aret some parts of it'e reactor coolant system.
Unjer these conditions, the ef fects el slip due to aneleration and f r iction pressure gradients become important.
The slip calculation is esade in alt flowpaths except cross flow in the core based on the drift flur andel this slip calet lation is then incitefed a the momer.tum equatlon.
Section 1.C.3 of 10 (fR Part 50, Apperdis ll, States:
i "The f ollowirig ef fects shall be tale n into.;ccount in the (onservation of somentum equation: (l) tes4) oral (hang.* of momentias. (21 annenture convection, (3) area change momentum f it.t. (i.) e.imenttwo change dee to compressibility, (5) pressure Inss s esulting ison wall f riction. (6) pressure fou resulting f r om Se es c h arige, amt (1) graviteticnal acceleratiort Any caission of one or spese of these terms under state.f ciectetances shall l'e justifier b) t empara, tive analyses er by esperiment.nl dita."
the application of the slip efte<te. le realistic for flow calculations during the LOCA aruf is in ennfoemarv e with 10 (IP Part 50, Appendia K.
the staff has reviewed the slip asadel in the cas e as discussed in Section 4.12, 4.1 flevatica Preume (hange f or tettl analpls, West ingtayuse uset tie juixtion fluid density from the upstresa element in evatustr the elevatten Isessure chanige. During the initial portion et t-12
_m__ _ _ _
the translint, this fsrs will has a air.or af fIct on the transist becaust the olivition tsris is only a 52311 f raction of the tttal prsssure drop. HowevIr.
towards the end of the transient, the gravitational term becomes leportant.
Westing *iouse has presented data corparing the nen*UHI method with that used for UNI analyses. the data indicates that the forse used for UHI is conservative.
The staff finds that the use of the junction fluid density from the upstream element to calculate th elevation pressure change for UHI analyses is acceptable and is in conformance with 10 Cf R part 50, Appendia K.Section I.C.3.
4.8
[ontenl_. Volume Chae,.ges in %AI Ale for "HI anaivsis. Westinghouse uses the 5ATAN code through the refill period of the LOCA transient. The control volume f(bene used in SATAN was revised (figure 4.7) to provide additional downcteer and upper head no' des, guide tube and support column nodes. UHI accumulator rww.'e. and a contalteent node. The additional downcomer nodes split the downtorer flow into separate paths. One path receives flow f rom two intact loops wh'le the other receives flow f rom the remainirg intact loop and the brolen 1. ;p.
A sensitivity study was performed to compare the ef fects of a single four-elemee-downcocer to the above UHI model. A comparison of pean clad tesiperatores at the end of blowdowre indicated that the UHI model was slightly more conservative. This calculation was done with the upper head flashing model which has since been discarded. Ihat model resulted in very little core steam genesation soon after active injection. Thus the most significant bypass mechanism was not present. In the cuerent model, steam generation and steam flow out the bottom of the core are significant af ter end-of-bypass is predicted. The <plit acding sch e provides a ready vesit path on the broken loop side of the d'ws:ener.
The entire SLbject of the UH1 dowrKomer mmfel is ad< tressed in $ct tlem a..". 5. 3.
the additional control volumn trat rt present tt'e sopport columns and guide tubes are not important dueiq the acti e UHI injection phise when the flow through these elements is con'.solled be the pr ess ure in tt+ UHI accumulator. These control volumn sto it fluence t'i* Ltper tvad beatus period as distuned in Section 4.2..
The additional upper head nofe pernitt tte calculation to account for different fluid conditions entering and leav>ng the support coltans and guide tubes. This amteling also g ermits calculatir.g uppe r tead conditions with finite slaing between the UHI water and the inventcry i's the ut.per head.
the physical layout of the t,ril system includes t-o larqe accumulater tants (one filled with water, the: other with nitrogen) cor.ented t'y a !?-inch diameter pipe.
A small surge tank cor.taining water ard r.itrcen is alsn connected across this 19st deletion of this small sutga 12-inch stiameter pipe.
wet t ingN ne has ;h+w a tank f rom the ev<ttu,etion model vioding is acceptehle 4 43
19 17 I
I at i
COLLHNS j
curet insts 51 52 g
9 11 10 12 4
8 43 3
?
13 49 2
6 14 1
3 50 16 15 SATAN WI REACTOR VESSEL Ftrare 4.14 4 44 O
Figura 2.1 is o scusatic cf the UH! syites showing the small surge tank and the lirge gas and ilquid cccumulttir ecssels. Westinghoes) was (sted 13 justify not A consirvttlee scoping explicitly including the surge tank in their UHI model.
ev61uation showed that only 2 f t8 of Ilquid would be delivered from the surge tank cog ared to 1000 fts from the main tank. Also, since the surge tar.k flow is low, the resulting flow quality in the 12-inch delivery line would be > 99 partent.
The staff (f.erefore agrees The two phase flow effects are therefore regligible.
that the surge tank need not be esplicitly model in the UH! systee.
Westinghouse The (441.gvaluation podel uses four antal nodes in the core region.
The study done has provided a sensittwity study to justify the anlat core noding.
with the 6 Cf ctg perf ect mixing case, showed a PCT of 2130*F for 8 core nodes The staff therefore believes tne compared to 2180*F for a siellar 4-node case.
standard 4 ru e representation can be retained.
Westinghouse performed a time step study by decreasing the allowable mass balance lhe following results were obtained:
- error, Allowable P,CT mass error 2219 I.10 4.10' 2226 2180 8.10 (old base core) e criterion.
Based on these results Westinghouse has agreed to use the 4.10 Westinghouse has psoposed to use the gap area (due to differential thermal espan-The staff slan) between the hot leg pipe and the core barrel as a ve..t path.
However we must reject this agrees in principle that this flow pa.h esists.
because a model for the speciflC gap size has not been proposed.
4.9 10CTA-IV flow and Enthalpy input
$lnce more than one junction entsts at the top of each upper het channel volume in UHI systems, a method was needed to determine for LOCIA suitable inlet flows and the enthalples and thus, to avoid the use of meaningless negative enthalples, method of algebraically summing flows and using the enthalpy f rom the downward flow (s) is acceptable.
4.10 Bubble Rise Model When flashing occurs in the upper head in the latter stages of eeptying, the vapor
(
formation will cause a level swell in the remaining water. Westinghouse has proposed using the Yeh correlation (Ref.18) to determine the vo'd f raction under l
Comparisons of the Yen correlation with available data (Ref. I) f these conditions.
i i
l 4 45 l
l t
i 1
i
Im addition, show good agreemez.t ov;r the pr:sstre range e.f int;rist.
Westinghoun perfirmed sinsitivity studits with a line r and knifIra v$id distribution in the upper head during flashing and showed that the v31d model had As a result, the staff has an insignificant effect on peak clad temperatures.
concluded that the Yeh :orrelation is acceptable for determining void fractions in the upper head.
4.11 Orift Flun Model ite SATAN program includes a drif t flux fic,.# model (Ref. I) in all flow paths which include elevation changes. This model permits the determination of separate The results of flow for each phase in the coCurrent and Counter-current regimes.
this separated flow model are then used in the energy balence equation for all The momentum equation, w.alth remains unchanged by the drift flus flow paths.
ihls total mass flum model, is used to predetermine the total junction mass flua, is related to the separate volumetric flumes as follows:
I + #g I (1)
G=#f f
g where:
G = total mass flux at junction (Ibs/sec-fta) 3 g = liquid density (Ibs/ft )
p 8
8 jf = liquid volumetric flut (ft /ft -sec) 8 p = gas density (Ibs/ft )
g 8
8 j = gas volumetric flum (ft /ft -sec)
The drif t flux equation af ter Zuber (Ref.19) is used to relate the flumes af the
-two phases in both counter-current and cocurrent flow regimes:
V j = a Co j,
- a 93 (2)
- Tut, TGC, 0
where:
a = flow path void fraction C,
flow path void distribution coef ficient a
gy = drift velocity (ft/sec)
V requires the determination of C, v ),
Solution of these equations for j, and j o g g
and a, m the flow path, tow values of C,and V will minimize the separation ef fect of gravity in vertical flow paths.
For non-UH1 plants, downcomer flow paths constitute the most important place for consideration of the vertical slip drift flus model. In these plants very little In such water exists elsewhere by the time gravity effects can beccee doeinant.
and C, to minimize cases the conservative approach is to select low values of Vg downcomer penetration. This was done for non-UNI dealysis by setting C = 1.
The 9
value of V was f rom luber's recoassendation for churn turbulent bubbly flow. If the drift flux line escceded the Vallis flooding curve (Ref. 20) for annular ffow.
4-46
was reduced sa the Vallis curv2 and the drift flux line then the esf ue af Vg in the Wallis flooding cirrelttion era chosen Consttnts far V would coincide.
to be consistant with the Pushkina-Sorokin data (Ref. 21) for the onset of annular gg Although the value of C, a 1 is t sed on turbulent fles and the liquid downflow.
Wallis flooding Ilmit is annular, the conbination was shown to be conservative with respect to downconer penetration data.
Therefore, they Westinghcuse believes that eere gravity sepas ation is justified.
~
proposed that certain general changes be made in the determination of C,for all For downflow and counter fit,w in the downcomer an additional vertical flow paths.
change in the Wallis flooding curve was proposed to provide additinnel gravity separation in those junctions and hatten the end-of-bypass.
The determination of C, is a function of a for UH1 analysis is based on a modified Another Acmand slip correlation (Ref. 22) for cocurrent turbulent bubbly flow. i h d value of C, is evaluated based on the Pushkina-Sorokin (Ref. 21) data wh ch s owe that for high void fractions in the annular regime, the onset of lir.uid downflow This conduit occurred at a constant steam velocity for various conduit sizes.
size variation is regarded as a way of varying e in the annular regime. The C,
formulation is made so the drift flux line is tangent to the Wallis flooding curve The method, amis consistent with the Pushkina-$orokin data.
and intersects the,1 termed the counter flow method by Westinghouse, has proposed to be used when the 9
C,so calculated is less than the C, calculated by the modified Armand The This occurs at void fractions in the range of about C.8 to 1.0.
correlation.
change from the Armand correlation to the counter flow method always results in This method appears to be the drift flum tangency with the Wallis flooding curve.
consistert in this flow regime, since the formulation of the counter flow C, assumes tangency.
Concern was empressed as to whether the part8cular drif t flux formulation based on-a Wallis flooding curve and the Pushktna-Scankin intercept is valid throughout the A'so it is not certain that annular flow cocurrent and counter current regimes.
In certain portions of the is the regime that esists for high void fractions.
reactor system, drop or drop annular flow may exist at high void fraction.
The hig*t void fraction regime is of particular importance, since Westinghouse has shown that this regime predominates when gravity separation is important.
In the low void fraction regime (0..8). the Armand correlation was developed for Westinghouse has statcd that for vertical flow, the horizontal cocurrent flow.
slip ef fects due to friction and gravity can be added, and the C, fm the modified Armand formulation is suitable for both cocurrent and counter current There is disagreement on this point. It has been shown that C, in the flow.
The drift flux vertical direction would not be the same as in the horizontal.
The staff believes that formulations have been shown not to be similar (Ref. 23).
4 47
952 af the Armand ctrrstation is not warranted in the count %r curr' int flow regines. la pirticultr, if velds are concentratid neir the walls Es may be the case in the core when the fuel rods are dry, C, may actually be less than 1.0.
Review of the various plant cases has indicated that the determination of the C, in the low void fraction regime is not important. Vold fractions in the core below 80 percent occur only during the first 30 seconds. DurP ; this time the flow is strongly cocurrent and not greatly affected by drift fluz parameters. For example, in a llHI/non-UHI drif t flus study the fluid behavior was almost identical for both cases during this period of time, even though the drift flux modsis were quite different.
Several approaches have been taken to determine the most appropriate slip formula-tions for the beestinghouse UHI model.
As an integra) verification of the SATAN program with the proposed LD'I drif t flux model, five G-2 refill test results were compared to the 5AIAN predicted fluid Of interest in the ra ntarcurrent tests was the ability to predict calculations.
a the amount of steam and water opflow, Jwnflow, and accumulation in the test section. The primary porpose was to verify the drif t fium model appit=J to the test section. This would provide some assurance that the reactor core fivid behavior could be predicted with some confidence. For these calculations, SATAN used the measured heat fiumes in the experiments as boundary conditions. The refill tests were chosen since this is the time when gravity dominated slip effects are most prevalent. The preoicticn of s'eam flow in the hot and cold legs was fair, and the predicted mass accumulating in the Jowncomer was fairly good in most cases. However, the liquid inventory did not agree well with the predictions. It is not known how much of the error was due to experimental measurement or code preuiction. Six variations of the drift flux model were used to predict test 706. Reductions in the flooding limit generally resultea in poorer predictions. Increasing the Armand C, to redute the transition void f rac-tion improbed the liquid balance but caused the steam flow split to be worse and the downcomer inventory to be too high. This increase in Armand C, towever, is g
much too high and is not Consistent with other known data. The best predicition was obtained with a reduced C but def'ning V 50 the j Intercept is not altered 9
g by the change in C,. However, it is still not known if this agreement is artificial since the emperimental man balance is still not known. Therefore, the emperimental predictions do not form a sufficient basis for selecting drif t flus parameters.
These predictions and certain paraestric plant studies, however, do indicate which parameters are important.
In order to make the final selections of drift flus parameters in the core, several studies were performed based cn a phenomenological understanding of possible 4-43
The base cts 2 la j
flow regimes cnd rod conditigns which tre postuleted ta esist.
In the drift fluz pirametats suasarized la Taels 4.2.
this study (PI) usIs the countercurrent low void fraction regime, C, values are selected not f rom cocurrent
'j At void fractions horizontal data but estimated from countercurrent information.
above 0.8, an annular flow regime is used if the rods in the hot channels are Other-quenched and in the average core region if most of the rods are quenched.
The droplet model allows less water wise, the droplet flow model is selected.
Since the hot channels are at higher power and less likely to be penetration.
quenched, the droplet model is more likely to be used; further isolating the hot assemblies from the coper plenum liquid (see Section 4.2.4).
Idestinghcuse is now IJntil recently, SATAN used a hoeogeneous cross flow moJel.
prop 2 sing that above a transition void frv tlon no liquid be allowed to flow The transition is selected based on the point at which a laterally in the core.
Above the transition substantial phase changewecurgintheirdriftfluxmodel.
Below the transition the flow ($ more nearly homogeneous.
separatisn o: curs.
Westinghouse believes that their cross flow model is more compatible with the Another realistic cross flow model was used for the second case drift flua model.
in the sta dy (P2) in which the horirental slip data of Arnarwi is used in the low void fra: tion regime. That is:
C,= 1.2 (o i.8)
For the droplet regime the high void f raction Armand data is used:
C, = 2-o (o >.8 droplet flow)
A third case (P3) was For the arnular regime no liquid cross flow was alicwed.
the same of P2 but without the droplet drift flux codel in the average core It was suggested that this would allow more liquid into the average core region.
and generate more steam to supply the hot channels.
Preliminary results f rom ti.* drif t flux study showed that certain cha70es to the The Ishii annular flow model model shown in Table 4.2 migh'. be move appropriate.
allows much more water penetration then the Westinghouse Counter flow model (now This is because the Pushkina-$orokin data was tate.
called UNI flim flow model).
The at high gas velocities and represents a more chaotic fluid file interface.
Since staf f believes that this condition skulo general *y be more conservative.
Co < 1 is not realistic at high velocities, it was proposed that this be excluded for Reyncids nuchers greater than 2000. These changes were incorporated in a A summary of these fo.r studies with the old base case is shown fourth case (P4).
below.
l i
4 49 l
i l
l l
\\
\\
~
\\
Table 4.2 DRIr.' FLUX FARAMETERS V
Voie C*
...S).
.R.egi. _r_e_..._.
F r_a.c.t.i_on c
,3,
,3 gjggg)*1.41[egge(nf59)}.25 Bubbly
..8 C = 2. A ~.~ 2
.15-e.5 V
Countercurrent C, a 1.2
.$.5 Z
L
'f i
.I f
Bubbly Downward Co-cu rent
.:.8 C,*1.2 gj(gg)*Vgj(gg) c,,(eg. current)
V C~~Ico'u'nte'r-current) cibiilfuFwYr~d'"'
~ ' ~ ~ '
gj(uS)* gj(BS) I NI' <"*"O Co-current
._.8 C *1.2 V
r.TJcoJter-current)
T -
p
~ _.. - ~.. - - -
Annular
.8 C *2 Ishit et al A.I.Ch C journal (Vol. 22 No. 2} March 1976 p. 283ff.
Equations 22 sad 23 or CQuettons 28 thru 33 Droplet 3o nward C, a 3.6 - 3.
.R ><.9 C (co.curreat)
Co. current
..8 C,=.
n.9 Vgj(DD)*Vgj(Dil ITcounter-curreniT f < ggc(c,-r... _
3.).
(j,,)
M Droolet g[pg)1.41 Countercurrent
...8 C =2-V
\\
-I._
9 Droplet Upward
.8 C,=3.6-3
.8-v.9 Vggg,0)*Tgj(D7) I*' C (cocurr.nt) c Co-current C,=.
e.9 T-7t;Tcounter-current)
m DRIFT flCX StuCY iclad Case _
Drift Flum Crossflow
- BOCRfC PCT Old Jase Old W h
~1600 2010 Pl SER W
~l400 2030 g
- 1100 1930 P3 Modified SER stR
- 1550 1950 P4 New SIR
- 1450 2030 Table 4 4 summafires the P4 drift flua model.
As was especte6. P4 was more conservative than P3. Therefore, it was decided to adopt the new hybrid drif t flus model. In addition, this, model seems to best address all the possible flow regines espected in the reactor.
The FI/P2 comkarl.on h9 wever raised questions aboat the appropriate cross flow Dif ferences were observed in the hot channel fluid behavior but these were model not thought to be too (sportant. The safor cause of the 100*F difference between F1 arJ P2 was postulated to be the dif feience in average channel dryout time and in order to determine if the consequent impact on average channel stored energy, This case was this was indeed 50. Westinghouse performed an alternate case. PIA.
the sesult a repeat of the Pl reflood using the l'2 average channel stored energy, confirmed that 80*F of the 100*F dif ference was due directly to dif f erence in average channel fluid behavior. Since the Westi.'qhouse cross flow model is more restrictive with respect to allowing liquid enchange between channels, and since the P1 results were higher those in P2. it was decided to adopt the Westinghouse cross flow model.
A calculation was performed with finite upper head mining assumptions, the new This case, hyDrid drift flux model, and the Westinghouse cross-flow m< del.
This was substantially denoted as P5. has a peak cladding temperature of 1770*F.
lower than the old base case cf 2050*l. The dif ference was attributed to the new droplet flow model, which inhibits drainage through the core and eate Js the time It was also suggested that some of the difference may be due to dry out the core.
Therefore another to the different methods of computing sitred energy at BOCREC.
calculation was done in whch the old tease Core used the new 20 power region 10CTA Ibe PCI for this method for calculating temperatures at beginning of reflood.
case was 2018'F. tndicating that the r+w P.R. LOCIA method was not a majnr contributor to the dif ference in PCf. ibus, the anomoly or:urred that the new drif t flux model which allowed less water penetration in the core was slightly move conservative for the perfect mining case but less conservative for finite 4-51 T
~
l 1
t Table 4.4 DRIFT FLUX PARAMETER $
Void C
V a
03 Regime Fraction Bubbly C'== 0.5
==0.15 Igges (ej -p ) )
g C
23 + 0.2 eT5<=<0.5
- Vgj(DZ)*I'*I Counter-Current at.8 2
1
\\
- 1 l
Bubbly C.(co. current) gMDB)
- Igj(BZ) C,(counter-current)
Downward 31 8 C, = 1.2 V
Co-current Bubbly 1 - a C,(co-current)
Upward 21 8 C, = 1.2 Vgj(UB)*Igj(BZ) T To C,(counter-carrent)
Film Flow for g
Quenched Core a>.8 C, Film =
1 Vgggg) = (1 - it,)Vgc where 8*
1 M = 0.7 where '[99,(
,,))1/4 K = 3.2 og
.5 V = Kl q
M g
5 k
8
/
Droplet gg ' I'1 ~ #eI c
Z (1 - aC,)
Counter-current a3 C, = 2 = a Vgj(DZ)
- I**
,e proplet Downward For Cocurrent Up or Down Co-current
=>.8 If Reynolds Nweer > 2000.0 Ygj(DD)
- Igj(DZ)
C, = 1.0 0, (counter-current)
I Droplet Downward if Reynolds Number < 2000.0
}
Co-current a8 C, = 3.6 - 34 for 0.8 <a< 0.9 Vgj(UD)
- Ygj(CZ)
C, a
.i = 0.9 1 - ac, (counter.
Current)
two ecs:s until the final sampla csiculation f4r finita cining tras rtcaired.
As a Other changes in the model r75 ult'-d la mor3 quenching fir this final cast.
result the droplet model was not operative during most of the transient.
This allowed the water to drain through the core much faster, drying out the core Ibe PCT for the final finite mixing case was about 1945'F. The influence sooner.
of the droplet model was further verified by Corparirsg Cort flow regines in the final calculation and the P5 calculation. This clearly shownc the predominance of Since the droplet rodel in Pt and the film flow model in the final calculation.
the new calculation is more conservative then PS, we believe it is not necessary to prescribe a separate model for imperfect mining.
As mentioned earlier the determination of e in the flow path is aisc required for successful utillration of the drift flux model. In the past Westinghouse used an averaging technique in the two bounding elements. Calculations showed that during volumes. That is, part of the transient, " layer-caking" was occurring in the coet the upper voi had two phase stature; the next volume was steam; the neat was stature, and the.. tom was steam. This particular combination seemed to be an un'ealistic result, especially since a drift flux model was employed.
Westinghosse propa>eJ lnat the averaging procedure might *e the cause of the The staff agreed and Westinghouse proceeded to devise a more physically problem.
based codel which utilizes continuity wave methoJs. Sucstantial esperience with the new model over the past year, indicates, that the " layer caking" has disappeared and the results are not greatly altered. The staff belle,es that this model is satisfactory.
A question was raised concerning use of a single phase flow resistance formulation for the momentum equation in counter-current flow. Westinghouse elected to show that the friction term is unimportant in counter current flow. They provided a ters by term break down of the sement terms for a case during which counter flow It was sho=n that fcr most of the time during counter-current flow, persisteJ.
the liquid downflow was small. During these tiees the friction term was very small compared to the gravity term. It appeared that for larger liquid downflow in the film or annular flow regine the two phase multiplier appeared to be satis-In factory, since the friction term woutc be closely istuted to the liquid flow.
dreplet flow however the friction term would be more closely related to the steam flow, since that is the phase in centact with the fuel rud. Westinghouse was able that for flow qualities above 20 percent, the two-phase muiltiplier was to show wit' tin a f actor of 2 of the steam flow multiplier. A calculation was then repeated for that part of the transient for which counter current flow existed.
For this calculation t'e friction ters was arbitrarily doubled whenever The countercurrent flow existed in a flow path independert of flow regine.
average channel and hot channe? dr>out tires were virtually ni different from the base case and the PCT at [05 was 2C*F less for the modified cise, Considering all tnese f actors it was uncit.ced that the heitinghouse two phase flow f actcrs are satistifactor, for counter-current ficw.
4 53
D3.2 in the equation af V, used to detirwine the Wallis flooding cxrv) intsrcept This has the affect af incrx sing t e flooding curvs End rslax-uith the j, axis.
Ing, the end-of-bypass criter %n.
1he resultant flochng curve was compared to the CE 1/5-scale data (Ref. 24) and shown to be conserv tive. Since scaling methods are still a question, Westing %use compared the K* and ja methods of scaling to show that the L' method whici is utilized in their flooding curve would show almost no benefit when the CE data is scaltd up to a FVR. Westinghouse also used SATAN to model the CE experiments and showed that the amount of water col' cited was underpredicted. In psrticular, the three tests where partial collection occurred, SATAN underp edlCted collection by factors of 2 or more. Westinghouse also compared the CREARE (Ref. 25) and BCL (Ref. 26) data to e flooding curve scaled by Ja.
The revised curve underpredicted tne data in all cases. The staff believes that the J' scaling method gives too much credit for scaling trends observed between the 1/15-scale and 1/5-scale data. The 1/5-scale data was not s6fficient in the partial by pass region and geouetry may have emaggerated the azimJthal effects. The staff has therefore required that sCallng be determined for bypass using the K* method and the small scale (1/13 and 1/30) esperiments.
The criteria for determination at a suitable dowrcomer model including proper slip parameters are discrssed in Section 4.2.5.3.
Westinghouse performed a sensitivity study to the drift flux model in the steam generator. The amount of slip was reduced by reducing the V in W Ndng 9C limit equation by a factor of three. The study showed the reduced slip case was less than 20*F higher in peak cladding temperature. In the case with nominal steam generator drift flux parameters, five times as such water remained in the steam generator inlet plena. The reflood calculations did not consider vaporiza-tion of this liquid as required in Section 4.9.
The staff believes that the more conservative case will be with the added plenum inventory at the end of-5ATAN.
Therefc-e, the base case drift flux model seculd be used.
4.11.1 Conclusions The The subject of two phase flow in the reactor coolant system is compten.
Westinghouse drift flus model has undergene considerable evaluation.
In general, the use of a two-phase vertical slip model in UHI systems is important and must be used. The use of a drift flex model is a satisfactory approach.
Selection of final acceptable core drift flux parameters was the result of numerous studies. It is recommended that esperiments in the future be designed to f
determine slip flow parameters more realistically. This would obviate the need l
l for performing sensitivity studies and the resultant conservative prescriptions in these models.
i 4-54 l
9
4.12.1 Hodel Changes Small break spectrum analyses for a UHI equipped plant tre perfirmed with the WFLA5H (Ref. 4) program to compute blowdown hydrodynamics, and the LOCTA-IV (Ref. 5) program to compute hat channel thermodynamics. Both programs are iden*
tical to the programs used in the Westinghouse evaluation model (with UNI injec-tion dynamics included in the WitA5H model by the addition of a single flowpath to the upper head volume). UHI high pressure injection is actuated in the same manner as the low pressure accumulators when systea pressure der.reases below the Coolant flow resistance in the flowpath from UHI accumulator gas tank pressure.
the upper head to the upper p enum regime is also modified in the WFtASH model to account for the si.pport columns anr1 control rod guide tube flow areas of the UN!
Since the upper plenum, core, and lower plenum regions are modeled as a system.
Flow from the upper head is $1mply added to the single single volume in WFL ASH.
volume, and when subcooled or in the sa.urated liquid State, is added directly to the water inventory in that volume without contributing to core upper region Core cooling above the coolant misture level is by steam generated below cooling.
the misture level and f ?owing up through the stature and upper core region.
4.12.2 Ivalestion The Computer programs used for the small break spectrum analyses, and the system models used in these programs are identical to the programs and models used for small break analyses in the Westinghouse Evaluation Model which are in conformance with the requirements of Appendia K,10 CFR Part 50. The lack of detailed heat transfer modeling in the core upper region during UH] injections for small break studies is considered to produce conservative results for clad temperature The fluid injected from the calculations in the uncovered portion of the core.
upper head would interact with steam flowing f rom the lower part of the core condensing some of this steam, and previoing a lower quality coolant for clad cooling in this region. Direct additions of relatively cool water from the upper head to the reactor vessel volume during UH! frjeClion will also produce a conservative suppressior,of steam generation in the coolant misture, so that the simplified modeling for UHI in the small break model provides inherent conservatism for clad temperature calculations.
Results of calculations for the small break spectrim of a UHI equipped,,lant have shown peak clad temperatures well within Appendia K limits, with the peak tempera-ture of 1606'F occurring f or the 3 inch clameter break.
from the results obtained with the conservative approach used in this analysis, and the use of previously approved programs and models, it is concluded that the small break analyses reported for the UHI equipped plant has shown acceptable Clad temperatures over the small beeak spectrum.
4-55
Audit c Iculition) perfirmed with errious elrsions af RELAP4 (see $ection 7.0) showed dif ferent dellery ritts f ar UNI and c41d leg Eccumulitirs as Comparid ta stellar $ATAM calculations. These differences were traced to several causes.
RELAP4 M005 wses an Isothermal gas expansion model (y=1.0); $ATAN uses an Isen-tropic model (y=1.4).
The isentropic model allows cooling by espansion, thus reducing the gas pressure more than the isothermal model. This causes prolonged delivery, especially if the accumulator is enhausted as is the Case with the cold leg accumulator. The staf f concern in this case was that if the act' sal delivety rate was greater than predicted, sufficient water may not be available to fill the vessel after the prolonged drain period. In order to verify their svadel, Westing-hcuse provided accumulator delivery data f rom a plant test with UH1 settings.
Analytic comparison showed that a y71.2 provided an accurate (omparison to the This finding is consistent with the INEL analytic esperience with LOFT cata.
tests L1-3A anJ L14. Westinghouse has agreed to make this change to their gas egansion model.
Dif ferer.ces between $AT AN and RELAP in predicting UHI accumulator delivery were directly attributed to modeling delivery line resistance. Westinghouse compared thenr model to a f ull-scale plant test and showed very good agreement. Therefore, no adjustment to the upper head accumulator model will t,e required.
4-56 l
5 4..J M 4 tantial a=w e.
r.
5.0 SENSITIVITY STUDIES Section 4.0 dealt with model changes and, where appropriate, their compliance with Appendia K.
In addition, the rule requires that a variety of sensitivity studies be done to provide a bests for the acceptance of simpilfying assumptions. This section provides a review of these studies provided in response to staff questions.
5.1 Large Break Analysis Spectrum Fcr the LE! system. Westinghcuse originally analyzed a range bf double-ended cold leg pipe breaks with areas frou 1 f t up to the largest pipe in the primary coolant 2
system ft,r a four-loop plant.
le. addition, a double-ended hot leg guillotine rtpture was analyzed to determine the effect of break location. For the selected range, the calculations were perfomed with three discharge Coefficients ranging from 0.4 to 1.0.
A list of pertinent parameters used in the most ter.ent break spectrum study is presented in Table 5.1.
Sensitivity analyses were based on a UHI accumulator pressure of 1200 psia which means that active UH! injection would be initiated before flashing could occur in the upper head. Since the staff's review has relled on sensitivity studies using this assumption, the application of the UHI-LOCA evaluation model must be linited to similar UHI accumulator pressures which preclude early flashing in the upper head.
TABLE 5.1 UHI-LOCA PLANT PARAMETERS NS$ Power 102% of 3570 PWt Peak Linear Power 102: of 12.6 kw/ft Peaking Factor 2.32 Containment Pressure 20 psia Cold Leg Accumulators Pressure 400 psia Water Volune 1050 cu. f t.
Line Resistance 23 UHI Accumulator Pressure 1500 psia Water Delivered 1000 cu. f t.
Line Resistanc+
23 These break spectrum studies were based on calculations with a proposed UHI-LOCA model which needed to be modified to reflect the staff's evaluation. Consequently.
5-1 l
l
\\
l f
a brzk spectrtm study including break locitiin was reanalyzed with a model reflecting the changes rselstrsd in this riport.
This limited break spectrum was intended to include those cases for which the revised sodel might have different icpacts. The follaming results were obtained:
TABLE 5.2 LIMITED BREAK SPECTRUM PESULTS Break UH Mixing Model PCT *F
.6 DECLG Perfect 2180
.6 Imperfect 2010 1 ftr spilt Perfect 1760 Imperfect 2040 I ft?
- A SATAH calculation was also performed for a large (C = 1.0) double-ended hot leg D
break. In that case the core began to reflocd while the upper head was draining.
No deleterious effect on reflood due to draining was observed.
It should be noted that these cases were not done with the nost " final" model.
The study was done prior to the time step study and befose all of the heat transfer and quench requirements were determined. Final calculations were repeated for the two large,(CD =.6) cold leg breaks. The peak clad temperature for the.6 DECLG perfect nialog core was 2210*F. As is apparent from these results. careful break spectrun calculations will be required for each plant. If these plant-specific calculations so indicate, plant operating Ilmits will be revised to assure compli-ance with 10 CFR 50.46.
A previously reported benchmark calculation perforned last year showed a result of 20$0*F for tne.6 break. Westinghouse suggested that no single ' reason Caused the difference between the new and old calculations. However, it appears that the hot pin heat transfer model and requirements for handling SOCRIC residual hot side water were inportant factors.
5.2 Sys, tem Paraneter Ef fec_ts Westinghouse performed sensitivity studies on several system parameters (containment pressure, cold leg accumulator line resistance, and accurulator pressure) using the original analysis model (Ref. 1). Inese studies shoald only be used to indi-cate probable trends ws h regard for reviewing the evaluation model. Plant-specific UdI-LOCA analyses must use an acceptable low containment pressure. For operating Ilcente reviews it must be denonstrated that a;;rcpriate values of cold leg accumulator pressure, water volune, temperature, and line resistance are used in the calculations. In addition, it rust be shown that conservative values of 5-2 W
9
~'
+~s L4 m,,
I A
sawwe.
.e ee.
on use approprlits system paramet3rs during normal operition, which should be used 13 1 Alt 1& lits the trans h nt cciculations.
Sensitivity of Clad Temperature Predictions to incospleto Misino of the 5.3 Upper Head Water in the Top Core Node The staff was concerned that the Westinghouse assumption of complete mining in the In response to this concern top node alght mask important ef fects of subcooling.
Westinghouse did an analysis in which a portion of the upper head water was forced to bypass the top core node and enter the nest lower node before mining with the fluid in the Core.
The result of this snalysis was an increase in peak Clad temperature of about The staf f therefore finds that the assumption of complete olving in the top 25'F.
core node is acceptable.
- 5. 4 fffects of Finite Minino in Upper Head the staf f was concerned about the ef fects of finite mining in the opper head on Water temperature considerations were discussed conditions predicted in the e. ore.
In this section core flow dynamics will be discussed.
in Section 5.3.
Westinghouse performed ;,14%t analyses for a double ended cold leg guillotine break The mining of the UH] water D = 0.6, a* awning finite and perfect mining.
with C and upper bead water was based on jet entrainment correlations, presented in As seen in Figure 4.11. upper bead emptying occurs about 30 seconds Ref. 27.
earlier when finite mining is assumed. Pecause of the 4.f f's concern about the validity of the finite almin2 model Westinghouse demonstrated that the time of upper head emptying is not affected if tt.e jet entrainment constant is varied by In aodition, the time of t,eit isolation is not affected by finite or 50 percent.
perfect mining in the upper head.
Calculations by Westinghouse indicate that there is 100'F-200'f difference in peak The clad temperatures when either the perfect or finite mining model is used.
However, imperfect perfect mining analysis has always resulted in >*gher PCis.
- Also, sizing is more sensitive to the drif t flum model and entent of quenching.
these models represent bounds of uppet head hydraulic behavior. fherefore, the staff will require that inis sensitivity be examined over the range of large break stres considered in plant analyses. The staff will require that the more conserv-ative mining assurotions be used as a function of break size.
5-3 ws
The support calumn cmit flow path in the Westinghouse UH1 model connects to the average upper core vol me.
Actually, the support column in a UHI plant tirsinctes within the fuel assent ly flow hole in the upper core support plate about 7 inches above the top of the fuel rods. F ull-scale cold flow tests showed that a large fraction of the support col o n flow is directed into the fuel assembly, even though there is cosaunication between fuel assembly flow holes. Strictly speaking, a support colon terminates neither in the core nor the upper plenum, but rather in a small region in bet-een. Since this region is not esplicitly modelled ir-SATAN, it was decided to do a sensitivity study by having the support colu n m
junction connect to the upper plenum rather than the upper core node. This study was dota with one of the interie models last year and showed that the standard Westinghouse method yields the nighest peak cladding temoerature. This was attributed to an increased inventory in the upper plenum. The Westinghouse cross flow model restricts water flow between the hot assembly and the average assembly. If, as in the base case, core water is delivered to the average core, it is not readily transferred to the hot assembly via cross f'ow.
If, however, more water is delivered to the upper plenum, it has a ready path to drain into the hot assembly directly. Westinghouse showed that the primary ef fect relates to the stored energy at BOCRfC, When they utilized the 60/40 average channel / hot channel stored energy model discussed in Sections 3.1 and 4.2.4, the PCis were much closer (see Table 5.3).
TA8tf 5.3
- 5. C. JUNCTION PCT SENSITIVIT' 5.C. Junction Stored Energy Core Average Upper Model Channel Plenum Hot Assembly 2225 1930 60/40 1940 1890 Since this sensitivity was performed with an interim model, it was decided to assess the impact of support column junction location with respect to the eew model. Westinghouse was able to do this without performing a separate calculation.
In particular, it was desirable to determine the effect of the new model restric-tions whlCh were not in effect last year. Although a droplet flow drift flus regime has been added and modificatior,$ to the bubbly flow model were made, the file flow regime was the predominant regime in both calculations. Also, both new I
and old models utilized the Westinghouse cross flow. Restriction on upper plenum flow to the hot assembly would bring the sensitivity results even closer together, but still more water would be available to the hot assembly in the case with upper
(
plenum connection. !t was questioned whether the additional upper plenum water in 5-4 l
(
l I
l kb D t% S r
- ^ ' "
ha 5,[ f.%,1
{[
.w.
rsquiremens se cuww...
comparsd f avcntorias for both calculations cod found virtually no dif fersace.
Considering all these f actors, the staf f concludes that this study need not be repeated and the We tinghouse model for support column Junction location at the upper core node is satisf actory.
Analysis of a Non-UNI Plant with the L1HI Model 5.6 The PCI at A typical non-UNI four-loop plant was reanalyzed with the UHI model.
Vesting-BOCR(C was 1520*f using the non-UNI model and 1330*F for the UHI model.
l This house attributes this dif f erence to the ef fects of the UNI phase slip mode.
l during the model has more fluid inertia and keeps more liquid in the hot channe Westinghouse notes that during this time period in a later stages of blowdown.
t heref ore.
UH1 nlant, substantial water delivery f rom the upper head would occur, itself.
any dif ferences in UHI analyses would be primarily dse to the UHI system The staff does not regard this calculation as a factor in determining a suitable UHI model, but it provides some measure of confidence that the UHI model has been implemented in an acceptable manner.
I I
5-!
e
' -- l " s i
i
6.1 Olscussten of ROSA Tests Systes tests with upper head injection were performed in ROSA-II, a small-scale LOCA test f acility operated by the Japanese Atomic Energy Research Institute at The heated section representing the Core he @ ;fon,t heated Tobal, Japar.
length made up of 140 heater rods. The primary system volumetric scale is about I:400.
In the R054-II/UH1 tests, the simulated upper head was Connected to the upper Relative primary plenum region by one guide tube and four UHI support columns.
system elevations were maintained to represent a reactor situation encept for the test section dimensions.
Ten tests were performed in the spring of 1976 with cold leg break simulations and different test paraseters including a non-UHI run. No detailed analyses of the tests were perf ormed at the time; imwever, the f ollowing observations can be made:
There are long CNS delay times (*20 seconds) 'or the non-UN! run and shorter (1) delay times for the UHI tests ($10 seconds).
There is ilmited quenching in the test section during active UNI (2) injection--mostly at the ends of the rods.
(3) Quenching at the bottom of the rudt occurs up to 10 seconds af ter quenching at the top of the rod.
Fluid temperature measurements in the upper head during active UHI injection (4) indicate that there is incomplete mining between the UN! water an the.vper head water.
Flow reversals occurred in the UMI support columns af ter the UHI was isolated.
(5)
All four UHI support columns ressented in unison.
The DNB delay times are not typical of those espected based on LOCA calculations for large PVR plant designs. Although the linear powers in these tests were low (less than 5 kw/ft), the CNB delay tiees do not appear to be associatec with heat Unfortur.ately, there flus, but more probably with flow conditions in the core.
were no flow measurements made in the primary loops to determine core flow direc-tion prior to 0%8 or UH1 injection, so that a cause and ef fect relationship cannot b
6-1
+
4 p*
t 1
l i
1
..w.e wn ewss et is ctypical af a reactir situation.
With regard ts quenching daring cctivs upper head inj:ctirn, the 1:st results show relatively poor quenching during this period which miy be indicative of inadequate water peretration into the test sectio.. The poor quenching during active UH1 injection is a significant result because it is contrary 4 the most important contribut! u of UHI in mitigating the effects of a LOCA. Here again, the lack of flow measurements precludes establisning a cause and effect relationship..If there was core upflow during UH injection, the lack of significant quenching is understandable. If there was core downflow, the results would indicate that UH!
was not as beneficial as anticipated. Another factor is that the top grid in the test section had about three times the hydraulic resistance of the top no221e ln a fuel assembly which could restrict water from entering the test section. In the
{
heater test section, thermocouples were located at 3, 22, 30 38 and 57 inches along the heater rod length. Because of the relative large spacing betweer.
j there> couples, the actual fraction of heater rod surface area that quenched cannot be accurately established.
The lack of coeplete mining in the upper head during active upper head injection is already considered in the croposed evaluation model. The flow reversal occurs in the support columns after UNI isolation. It has been postulated that conder6a*
tion occurs in the upper head during active upper head injection. However, the volumetric t'H] fluw rate is higher than the volumetric condensation rate so that net flow out of the upper head occurs during injection. When I..H! is isolated, the cordensation continues, reducing the pressure in the upper head, and reverse fluw occurs in the support columns. When the upper head pressure pacteds the upper plenum pressure, downf tw is agafi established in the support columns. Since the whole process lasted about four seconds, the staff does not consider this phenomenon to have a significant effe t on the LOCA. If the whole core unquenched for four seconds, there would ha a s sll clad temperature rise =%ich would still be well below the quench critee'a the.
Old when cownflow is reestablished.
It was noted that t-flow revers 46 occurred in all four suppoet columns simultaneously, lh 5 observation is consistent with the assmed J1 form support column behavior in it.s leHI evaluation a.
Because of the v ential questions of UHI viability raised by the poor quen;hing results, the staf' *ecommended that tbest tests be analyzed by Westinghouse in an appropre-a manne' ;) Detter ader*'ar.d the hydraulic behavior in the core and the resultant effect of the Uh! system on the observed results.
Westinghouse chose to analyte two tests in the ROSA-II series, which were meant to isolate the effects of UHl. Test 603 was performed without UHl; test 604 with UNI. Several modifications to SATAN.ere required to reflect specific features of 6-2 4
The actus1 electrical rod nett generation tad thermal proper-the ROSA f acillt s.
2
- lece vapowered sections of beater rods entended into tf.e lower ties were used.
plenum, the rod con 1uction model was entended into that region.
Certain charges were incorporated to reflect measured boundary conditions. These were the UHI injection flow rate, the steas generator secondary side te.4erature, and the pump coastduwn. To t,etter simulate ROSA phenomena, modifications to the leperfect mining model and the metal heat a2 del were also made. The staf f aq-res that these changes are acceptable and allow better ar.alytical esaminatten of the effects of UHl. These modifications were all incorporated prior to the first aralysis.
It is the Westing %use position that the unique ROSA geon,try enhances steam / water separation in the lower plenue, retaining more blowdown liquid in the lower plenum.
This condition, coupled with the high metal heat release, generates the necessary To stese to prohibit the UN! water from reaching and Quenching much of the Core.
account for this postulated separation, bestinghouse socified the drif t flus model in the lower plenue flow paths by multiplying jf by.3 and increasing j to main-g tain continuity. Temperature seasurements in the lower pienum showed that without this modification, tSe enalysis prejicted premature lower plenum dryout. The staff regarded this particular multiplier as some= hat arbitrary. However, the analysis did show that b,v enhancing separation, the resultant steam generation did indeed keep most of the UHI water out of the core and dioinish the amount of rod quenching. The staff believes that a plausible emplanation for the lialted quen:h has been demonstrated. However, the lack of sufficiens flow and density measure-sents, particularly in the lower pienum, argue against unqualified acceptance of this hypothesis.
The delayed "0NB" shown in the esperiments was also observed in the 5ATAN analysis.
The calculations indicated that the tempsrature rise was due more to a dryout than
~
a high heat flux DNB.
Comparison of tne upper head temperature measurements was made with analyses using bo'.h the perfect and imperfect miair.g models. The results showed that for ROSA the behavior fell between the two models.
I The SATAN pressure transient Eatched the esperimental results very well for the first 10 seconds. After th.t time, the esperiment showed a more rapid depressuri-2ation. No convincing reason has treen proposed for this discrepancy. bestinghouse has suggested that changes in upper plents entrainmer.*. related tc pressurlier dinharge may be responsible. This hypothesis would be difficult to test, especially in view of the acknewledgee instrument, tion shortcomings.
In summary, the Westinghouse analysis of the ROSA tests prcvided confirmation of some observed behavicr but was not completely coavincing in other areas. 11 l
6-3 r
~ ~ ~ --
S A #sE e w
+Mae-
+
particular, the atypical lower plenum metal heat was shown to be a possibts cruss of limited quenching in the core. However, lack of sufficie..t instrumentation prevents satisfactory validation of this hypothesis. The staff believes that further analysis of ROSA-l! by Westinghouse would not be meaningful.
6.2 Seelocale M003 The semiscale test progree sponsored by RE5 has been reoriented to include a test series simulating a UHI system. Ihls two-loop configuration, designated M003, will actually be fabricated prior to
.e M002 non-UHI system. M003, line M002, J
will have a 12-foot core and esternal downconer simulator. Unlike M002, M003 will have an upper head injection accumulator with a pipe capable of simulating dif fer-ent mixing models. Also included will t>e surport column and guide tube simulato-s. Design and scaling problems are being discussed among Westinghouse.
R(5, D55 and their consultants. M003 will initially perform benchmark tests without UHI to evaluate the external downconer and 12-foot core. UHI results should begin to become available later this year. INEL will perform pre-and post-test predictions for these esperiments.
It is the Westinghouse position that the M003 design has esperimental Itmitations which preclude its use for suitable integral verification of UHl. In particular, they believe that the entension of the unheated portion of the heater rods into the lower plenum creates an atypical heat source. This could lead to rejection of the UNI water, as was the case in ROSA-II. Westinghouse also contends that the f
single pipe dest;n is not a good downconer simulation and will promote rejection
-westinghouse recommands a properly scaled double pipe of the cold leg ECC water.
system or prefer bly direct lower plenum le.jection. Regarding the double pipe design, Westinghouse recosaends a downcomr.r Cesign to minimize metal teat and to model the espected I'W flow behavior. They inve communicated general princfples to be con.idered for this modeling (ref. 28) but have not proposed specific flow modeling standards. The first test configuration will incorporate the single pipe downcomer simulator.
~f he cent Hui configuration will utillie a double pipe design. We believe that neither the single pipe. double pipe o. lower plenum injection concepts are good simulation; of a large PWR system. State-of-the-art understanding of bypass phenomena does not allJw making a convincing choice for either downcomer simulation.
How<. er, the direct injection eliminates almost all considerition of bypass phencuena. Bypass.ilays a uique role in UHI ant should be modelled in these tests. Wir believe that the single pipe design will provide a conservative simulatior, of bypass phsnomena, and the double pipe design is prot bly seme= hat optimistic. Thus, we tielieve that by incorporating both designs, a suitat,'e spectra of bypass t havior will *e emplored in the UHI semiscale tests.
Westinghouse contends thtt because of these esoerimental atypicalities, prediction of these tests would not be meaningful and should not t,e required of them.
6-4
^
r It is the staf f position that prtdiction af appropriate saperiments remains an imporaant aspect of model verification.
The UH] concept introduces additional Complesity in analyging [CCS performance Plant analyses have shown beyond that encountered in Standard PWR (CC systems.
that timing of injection and hydraulle interactions plays e particularly important The ef fects of upper head mining, accumulator line resistance, quench and role.
unnsench, ugper plen.Je and Core hydrodynamlC uniformity, and downComer modeling However, no complete integral tests simulat-have all been emptored analyticelly.
In addition to the ing tHI behavior are available for analytical comparisen.
stated design deficiencies of RO$A !!. those tests were not complete integral tests, having been terminated at about 30 seconds.
We also believe that the sealscale M003 program has an escellent chance of being a
{
We do not believe that the likelihood of UHI good model verification test series.
However, it order to assure water rejection is as great 45 Westinghouse suggests.
that we are not merely requiring Westinghouse to predict esperimental the flest Seelscale test with atypicalities, we suggest the following procedure:
We will determine, by UNI will be chosen as a target test for predlCtion.
comparing test data with analysis = tether significant esperimental atypicalities If they were not, then we would request Westinghouse to perform a were present.
computer simulation of the esperiment.
Therefore, the major phenomena The first UNI test is a blowJown only esperiment.
of interest are potential rejection of DHI water and quenching of the heater rods.
Not until later tests in this series will upper head draining, refill anJ cold leg Therefore, we also propose that the first integral (CC Interaction be included.
test througn reflood be treated in the manner just described for the fir;t blowdown test.
It should be noted that generic approval of the Westinghouse model is not contingent upon successful prediction of these tests, nor is any plant license e
What is required is the commitment to approval based upon such a contingency.
pretest prediction of suitable irtegral tests.
t 65
-c-r r
t 7.0 STAFF INDEPENDENT CAtCULATIONS lhe staf f performed a nuntier of calculations to enamine various aspects of the UHI It was determined from this effort that analytical methods designed transient.
In partICular, the following for non-UNI plants are not suitable for UHI systems.
prot,lems were observed:
The standard heat transfer correlations available for EM calculations (1)
(Dougall-Rohsens and Groeneveld 5.7) with prohibitions on rewet are not suitable for UHI analysis sterc enartments show much better heat transfer at low quality.
When large amounts of cold water are dumped on top the core such as with UNI, (2) a v rtical slip model is needed to SC.ssfactority describe the hydraulic a
$lt.ce the upper head remains subcooleci f er a substantial period of time, the (3)
A current homogeneous models do not adequately describe upper head behavior.
satisfactory gravity deatn model is required.
A large amount of cladding surf ace area is at low temperature at the beginning (4) of reflood because of UHl. In order to take advantage of this condition, a modified carryout rate fraction and reflood model is required.
There are currently two programs under way to analyze the LOCA transient in plants At $andia Laborateries, DS$ and RES are sponsoring a task to equipped with UHI, modify RELAP4-M005 and other existing codes to provide the capability to analyze The the entire transient taking into account the requirements of Appenoix K. Within purpose of this program is to be able to perform audit type calculations.
budgetary limits, our goal is to produce an interin audit calculation by July The effects of various quench models and vertical slip models are being 1978.
Some esplored and the need for a special gravity drain model is being assessed.
dif ferences have been noted between Rf t AP calculatlocs and the Westinghouse SATAN in particular, RELAP4 does not always pre 11ct delivery of cold leg calculations,
- SATAN, ICC water to the lower plenum, while the cold leg accumulators are active.
however, fills the lower plenua prior to exhaustion of the cold leg accumulators.
The RELAP calculation of lower plenum refill can depend upon downcomer nodaliza-Sandia has also emptored the sensitiv-tion, slip medeling, and water pack fines.
ity of using the approved Westinghouse quench logic but not the quenth heat The entent of core quench was showr. to be af fected by the slip model transfer.
7-1 u.
i Difficulties with all af the slip acdels hav) led ta cuanination af ccriala used.
It is aspects of the constitutive equations and solution techniques in RflAP.
espected that progress on resolving these probleet will be suff(clent to be able to produce a reasonable Interim audi'. calcul*ation.
A recent titAP calculation included en improved accumulator and cold leg nodaliza-tion and a spilt downconer model. Ac' umulater resistance and liquid volumes were The code version used contained the H005 updated to reflect current setpoints.
The calculated hydraulic silp model and did not contain the UHI quench logic.
response ccapared very well with the 5ATAN response.
The time to beginning of reflood was nearly identical to that calculated for the Support column and guide tube flows were very close 5*. TAN perfect mining case.
The period of upper head reheat and beginning of the during active injection.
The upper head drain period upper head drain period were also nearly the same.
1his is understandable, since the SATAN was about 10 seconds Innger for RitAP.
drain model allows more rapfd draining than the standard equilibrium model in R[ TAP.
A version of Pf LAP 4-f t000 developed by the staf f utilizes the westing:.ouse UHI
$andla has modeled a UH] plant and made successful prelfsinary carryover model.
As espected, the entent of core quenching at the beginning of calculations.
reflood is entremely leportant in calculating flooding rates.
At PNL, RES is sponsoring a program to analyre various aspects of UH1-LOCA te N vior with advanced versions of COBRA IV. Thus far, a nonequilibrium drain model has been used to simulate the Westinghouse upper bead drain esperiments with very In the near future, the hydraulic and heat transfer behavior encouraging results.
Also planned are simulations of of the low pressure G-2 tests will be analysed.
the entire reactor vessel foi a full-sized plant.
At this time all of the analytical results performed by NRC and their contractors Careful study of these results will cont 8nue as must be considered preliminary.
Results of these studies will be applied to the more calculations are done.
In partic" Westinghouse UHI model for verification or modification if necessary.
utar, we will focus attention on those models which are important in determining satisfactory delivery of c:Id leg injection ECC water to the reactor core and lower plenum and core hydraulle behavior.
12 4
f s.o coactusle's Westinghouse's proposed UHl*LOCA evaluation model contains new and unique features Of in order to analyze a Pvt hJving upper head injection as an Etc subsystem.
these, the most 8mportant are (1) a quench model, to determine wn the core is rapidly cooled and how long it stays cooled. and (2) a prolonged analysis of core flow and heat transfer conditions because of the delay in bottom of core recovery Westinghouse has presented esperimental data to coepared to a non-UHI plant.
support their heat transfer models, upper head heatup model, and drift flus model.
In addition. the staf f has relied on plant studies to determine the importance of various assumptions in the evaluation model.
The area of most concern is the cose behavior while the upper head and upper For a large creat, this period ple e are emptyin; after active UHI injection.
In order for the UHI system application to be could last for 1,0 seconds or more.
successful. A major portion of the Core must remain quenched during this period and hot rod heat transfer must be sufficient to slow the hot spot temperature rise prior to reflood.
During 'his time frame, core conting is dependent upon. In large measure, fluid A basic assumption in cor.ditions in the upper plenum to supply water to the core.
the SATAN Calculations is that uniform Conditions entst in each Control volume.
While the staff recognizes limitations in numerical analysis, we felt that thisIn assumption should be justified with respect to analyzing core flow behavior.
partictlar, the staff was concerned that temporal and spettal fluid variations in the upper plenum and core were not accounted for in the calculations.
Another open issue w e a criteria for return-to-film boiling to be applied to the quenched rods under counter-current flow conditions in the ccre, as discussed in Westing %use presented return-to-film belling data from their Section 4.2.3.
current flow heat transfer tests to justify the use of a Mact,eth CHF correlation.
The staff did not agree that these data cemonstrated that Macbeth was appropriate.
The staf f was concerned that the return-to-film boiling was not weil behaved and be influenced by local flow variations under counter-current ' low conditicas.
pa/
At the tiae of the last status report, it was suggested that ac; urate phenomeno-logical models for core fluid behavior and counter-current unquenching may not be It was postulated that flow maldistributions and pa-tial unquench may necessary.
An extensive sensitivity study was therefore have some beneficial effects.
This study. described in Section 4.2.4. showed that the analysis could defined.
I e-1
i i
i 5% ' ;?*tantial amounts af care unquench and flow aaldistribtion erithout 5
detriment to the.-*sults.
)
On this basis, the staff concisded that the Westinghouse model which treats core and apper plenam fluid behavior uniformly was acceptable, as was the Macbeth correlation for predicting unquench.
The resolution of core flow betavlor, unquench and drif t flua relied heavily e w parametric stadies in lieu of certain esperimental information and probably did not do justice to UHl. Then as now, we strongly encourage Westinghouse to gererate tte necessary esperimental and analytical justification that would ultimately elisinate the need for this app oach.
A related probles was the definition of an acceptable drift flus model that ultimate!y controls the rate of upper plenum emptying under counter-current flow conditions. In the Course of this review, Westinghouse has presented analyses with different d*1ft flux models. They atteopted to justify their proposed model with data from the counter flow heat transfer tests; however, the results were inconclusive because of uncertainties in the interpretation of the data. The staf f reviewed the data from the plant analyses with the alternate drif t flus models to determine the impact on the calculations. An acceptable drift flus model was defined a 4 is discussed in Section 4.12.
Westinghouse has made several leprovements over the evaluation model originally proposed, namely, in the area of apper head dynamics. They have also performed additional heat transfer tests at low pressures.
This review has been based on a UH! system pressure higher t. san the saturation pressure in the upper head. Any alteration of this concept w)uld require a reevalu.tlen by the staff to assess any unforeseen System ef'ects.
Appli ation of this model for plant specific calculations is subject to the following considerations:
(1) Applies to current 1717 fuel only. (WCAP-8185)
(2) Applies to four-loop ice-condenser plants only.
(3) The plant analyses must be performed with both the perfect and finite misieg models.
(4) The initial upper head temperatures selected must be verified by actual acasureeent for magnitude a% uniformity.
8-2
)
g i
-~
)
j (. ( c. ; g, 3
. g ]g, e s
y (5) Singla failure must be eertuatec. on a case-by-ctse basis.
(6) Various plant parameters including UNI and cold leg acetasulator flows char-
!~.
acteristics must be justified on a case-by-case basis.
I t
t
}
Recently, Westinghouse was required to compare their split downconer UHI model to certain small-scale data. This resulted in certain changes required for satisf ac-tory analytical comparison. The effect has been an increase in peak Cladding tempe ature. With the inclusion of these changes, the staff believes that the 5
model descrited in this report is an acceptable evaluation model for plants equipped with upper head injection. Certain aspects of some component models (notably the recent charhaes to the downconer model and correction of a coding error in the metal water reaction model) are not included in the documentation from Westinghouse (WCAP-6479 R2). The staf f is communicating these and other editorial itees to Westinghouse. Final approval of the model is subject to Westinghouse submittal of an addendum or revision to this topical report to reflect the requirements steted in this report, and to describe the eudel Changes and the effect on Calculated 6*
results for selected breats.
8-3 1
e.
G 3,;C,v.s -75 y-
~
T ~ ~ ~
.j Q4]'] [rg s
.91 f tS c
,3 g'4)Ig f
l 9.0 REFERENCH "1
43 u14.. c l
i
. 4 c
1-4, a 1.
M. Yourg and R. Vijuk, " Westinghouse ICCS Evaluation Model Application to Plants
- I Equipped wlth Upper Head Injection," WCAP-8479 (to be publis%d).
r
{ ^ ".
2.
- 5tati.s Eeport by the Olrectorate of Licensing in the Matter of Westing %use Electric Company ECC5 Evaluation Model Conformance to !O CFR 50, Apperdia K,* October 15, 1974.
3.
- $ipplement to the Status Report the Directorate of Licensing in the Matter of
'b Westinpeuse Electric Company ECC5 Evaluation Model Conformarce to 10 CFR 50 Appendia E," hoventier 13, 1974.
4
- 5tatus Report by the Of fice of Nuclear Re ctor Regulation in the Matter Cf Westinghouse E
Electric Company ECCS Upper Head Injection Evaluation Model Conformance to 10 CFR 50, e[;
Appendis A,* September 1975.
'l.
5.
" Status Feport to the ACRS in the Matter of Westinghouse Electric Company ECC5 Upper
< {-
i,.
Head Injection Evaluation Model Conformance to 10 CFR 50, Appendia K,* August 13, 1976.
6.
- hBC Staff Review of the Westinghouse ICC5 Evaluation Model,* report to R. C. Cetowng, Jr., f rom V. Stello, April 22, 1975.
a -
7.
"MRC 5taff Evaluation of Westinghouse ECCS Evaluation Model Cu nges Occiamented in WCAP 8C2* (Jan,;ry 8, 1976).
u 8.
Letter free D. F. Ross to C. E. Eiche1dinger en Resolution of UN! Model Octccer 4,1976.
9.
WCAP-82 M.
10.
Bortsu stil. V. M., aad ' wain, 8. S.,
- Correlation of Heat !ransfer Data in stable Film Boiling c. Vertical Surfaces in the Presence of Free Lloald Convection in targe Volumes
- International Chemical Engineering, Vol. 5, No. 4, pp. 666-668,1965.
- 11. Dougall, 4. S., and Rohsenow, W. M., "F11e toiling on the Inside of Vertical Ti.bes with Upward Flow of the fluid at tow Qualities," MIT Rept. 9070-26, 1 % 3.
- 12. Andreyc*eck T. S., et al., " Blowdown Esperimerts with Upper
- ead Injection in G-2
'1 a 17 Rod Array Facility,* Vol.1. WCAP-i5S2 (Prop), Septencer 1975.
- 13. Cunningaar, J.
P., et al., "ECC5 Heat Transfer Emperleents with Upper Head Injection,"
[
Vol. I, hCA7-8430 (Prep), October 1974.
9-1 i
a
A -%#F.4w &..Q %..t,N. % Q,4fg;_;.(
.s.
~,. m
~
.% J
+.
,1,s, g, fp
' :.g%
w Q y.J g &.ff
-* t*b - v->2 y.g. g y; 3 4 g y, y., y y,.,,. y.. '.h,bc. e. '> C 4
y w,ug,W y..q:
33s. g..e A. ~ c'pn, g r3 4, %,M,.m. ~. *q
- 9 <
m,'M % *
- N.
,4
'2 r
A.'
- a >w. ' u 3.,
- 3r.
- p
.s W,-v - l, e.;;.
1
~ - -
-r+
a u.
..... ~.
_. % f,<;;f,s W,'gK] s.. '
W i
.MA y.p / 14. a e mcheth; ' Burnout Analys..i.s,,P:et 4: ' Appl cat on e, a..ocal. l Co.ndi.tions. Hypothesis,to i i t
. W ;pt u,ng v
m,g.%.w ;...g j[' c.,
World Data for Uniformly Heated need Tubes' and,nectangular_ Channels,* AffW-8167i'j.
4 c%[ ",.I '*, [ %, (. ~, v 1 n. g,, v. u,.s;y-G ', N w w]; w :, w;.c.!r ~ [ ".
/
Y h 4 sl N - [
i-Au'gk19
(;h jh. V.,,7. sp, r y-.,,m v ye [, u m 63.
as
/
. n,v 4~
a w
w,,,.
~
.gy^^ z ; N'l. ~. w;7*"g* ' ~,*
y.,
, %;.+ es. ~
1 c
o n n.
g ps.
q
.+
O Kansal Electric..w h%. Power Co., Test on upper Head Heating and Esotying for UN!-[CCS (Ransal
.,. e. w. ).1~
2-js',-
?
k>.
us.
. J (1 W. f.3
++ ys g
[a -
L,n
~
~
J U C
... $ ;;[ t.,
'/$'
-.y.W+#.h. 4M ' 5 h:ProprNtary) Dec' ember 1975. ? C.,,, r'.mY_
yo.
2 i,
,c
- , c Vl
.;g.
..,.V t, ;%.-
yy m
,- 7
., p..
w_,
-*. T,
- 1.
G7 4,,.;.,
~
..G;'
%..r., % :
+M~.
'. N.b x ' El6'. 'Hochreiter, 't. E. I.,et ' al., 'G-2,17:17 Refill Heat, Transfer Tests' and Analysis,"
.N ';,,
.e, n s a.
e
<*/.*iT.e t WCM 8793-P (Prop), August 1976.,
~
..a, g.~., g. gc 4
~a
,a q f., u,
y m fc W d u 4.
6, @,,, ( 7. ;,, A
,.c..,,
~.,p, ry e
4 s
., c,.
t
./.4,.
< p.p c
' 11 ' Westinghouse ECCS Evaluation model. Octeter 1975, version WCAP,-8672 (November 1975).
,,..~,w..
Qi,
s
.,.,,. j,,
t j
... A;% ;3..w,;-
gr --
- .9,n 6
" 4.;J M I:
- s..r 4
p.
z,,
1a.
^g i
h18.' V,s Camninghae/ and Hc Yeh, "(apeelernts and Void Correlation for P%R Smal? Break LOCA
-,>1
....),,
- y
.y
.vn z.;_
A
'C
" N 'l^
'f I b <w v aa g & ; " p i' Q [g p s." T ans. ANS,~4elime IF (November 1973).
i
@ Yug
-u s
~
f~ a. 'I l g-s y@[h$.cirp> ~.v, 9 a
~
'i
.+
~,.;
- ~s
~
Volumetric Concentration ig Two-Phase Flow
,.'. J' <'i,,.
.19.,,* N. Zihe.r,.and J.' A. F indlay, *Acerage E"+'J cJ-
. ' Journal of Heat Transfer, November 1965.
C
" O Sf,N.,,
,'.O Systees,".
,3
..,t h
.y.
1 5'y 'e,# fg
- ,4
- 3 e.
a + :,'. :.., n. n,....
.a
.,,.~.,
f c
..li:.'s. 20.) G. 8. Wallis, "One-Dimensional Two-Phase flow," McGraw-Hill Book Co.,1969.
e
~
t
_ F
,.c m
~ n ;
+.
, g. arc,
-,qy
., g
+
M IO, k, 21. 'Pushil'na'& Sorokin', {dreakdom of Liquid file Motion in Vertical Tuhes," Heat M..
f' Transfer - Soviet Research, yo1. I, me. 5. pp. 56-64, September 1969.
y-4 L;1.q[ <
t.,.
m ~.gw
(-
uA5
' 22.
B. Cantineau,' *Two-Phase flow 9%dels," Westinghouse Calculation Ncte 5[-5Al-N-249.
v e
.' n ' y n
+
Zuber, et al., " Steady-State and Transient Void Fraction in Tso-Phase Flow 23.
N.
4 F inal Report for ttw Progras of Two-Phase Flow Inves*.lgation," GEAP-5417, Systees:
Janaiary 1967.
24.
Ct hP&-64, Nne 1973.
Crowley, C. J., and Block, J. A., *ECC Delivery 5tudy + Emperimental Results and 25.
Discussion," Creare Tectnical hote IN-2il, OSobe-1975.
flanigan, L. J., Cudnik, R.
A., and Denning, R. S., Topical F.,, ort on faperimental 4
26.
Studies of [CC Delivery in a 1/15-Scale Transparent vessel Model, ehl-1941, Battelle Coltabus laboratories, Noveter 1975.
27.
M. Albertson, et a)., "Offfusloc of Subnerged Jets," ASCE, Vol. 74 (1948).
28.
C. Eiche1dinger,' tetter to D. F. Ross, NSCf-l',99, Nov. 15, 1977.
s l
l I
l 1
9-2 i
n k
- S a
- i.
.y.
.~..
m
..,.i.
.v. :
4 c
4
.k n
($Q
.a