ML20207J332
| ML20207J332 | |
| Person / Time | |
|---|---|
| Site: | Yankee Rowe |
| Issue date: | 05/31/1988 |
| From: | Fineman C EG&G IDAHO, INC. |
| To: | NRC |
| Shared Package | |
| ML20205M007 | List: |
| References | |
| CON-FIN-D-6022 EGG-TFM-7933, NUDOCS 8809010217 | |
| Download: ML20207J332 (78) | |
Text
a o
-c O f d #b '
f EGG-TFM 7933 TECHNICAL EL\\LUATION REPORT RELAP5YA COMPUTER PROGRAM FOR USE IN PWR SMALL BREAK LOCA ANALYSES C. P. Fineman l
May 1988 i
Idaho National Engineering Laboratory EGAG Idaho, Inc.
i t
i Prepared for the U.S. Nuclear Regulatory Commission 5
Washington D.C. 20555 Un'er 00E Contract No. DE-AC07-761001570 d
FIN No. 06022 l
e 9
e
ABSTRACT A review was completed of the RELAP5YA computer program developed by Yankee Atomic Electric Company (YAEC) for performing pressurized water reactor (PWR) small break loss-of-coolant accident (S8LOCA) licensing f
analyses.
The review consisted of an evaluation of the RELAP5YA computer l
program as wed as the modifications made by YAEC to the RELAP5/M001, cycle 18, computer geogram from which the licensing version of RELAP5YA oeiginated.
Separate effects and integral assessment calculations were reviewed to evaluate the va'iidity and proper implementation of the modifications and ade 4 models.
The review found that RELAP5YA met the requirements of 10CFR50 46, Appendix K, and NUREG-0737, Item !!.K.3.30, and l
it is recommended the code be acc9pted for use in PWR 58LOCA licensing I
analyses on the basis that suggested conditions and requirements are followed.
The largest break size YAEC used to assess RELAP5YA was equivalent to a 0.7 ft2 break in a PWR.
RELAP5YA may be used to analy:e breaks latger than this if the important phenomena for the larger break sizes are similar to that assessed for the 0.7 ft2 and. smaller breaks.
1
(
d e
6 g4
o
SUMMARY
This report documents the review and evaluation of the Ya.1kee Atomic Electric Company (YAEC) computer program RELAP5YA for use in performing pressurized water reactor (PWR) small break loss-of-coolant accident (SBLOCA) licensing analyses.
RELAP5YA is capable of analyzing the steady state and transient tnermal-hydraulic response of a light-water reactor.
The code has features that allow compliance with the requirements'of 10CFR50.16 and Appendix K.
RELAP5YA was based on RELAPS/M001, cycle 18.
developed at the Idaho National Enginaering Laboratory (INEL) under Nuclear Regulatory Connission (NRC) sponsorship.
The code was submitted to the L
Office of Nuclear Reactor Regulation (NRR) for approval as a licensing method.
NRR requested assistance from the INEL in reviewing RELAPSYA.
Assistance in the review was limited to those aspects related to PWR steady state and SBLOCA transient applications.
I The RELAP5YA code was reviewed and assessed as well as the model additions and modifications made by YAEC to RELAP5/M001, cycle 18.
Th i s,.
review was made using the information provided by YAEC in the RELAP5YA code manuals, Volumes 1, 2, and 3, and in the utility's responses to questions submitted by the NRC to YAEC.
The code was also reviewed to ensure that known updates and corrections to RELAP5/M001., cycle 18, were included in RELAPSYA by YAEC or appropriate justification given if excluded.
- Finally, l
the code was reviewed for compliance with NRC requirements.
Based on this review, it is recommended that RELAP5YA be accepted for i
performing PWR SBLOCA licensing analyses, including the calculation of peak l
r clad temperature, provided that suggested conditions and requirements are i
followed.
The largest break size YAEC used to ass.ss RELAP5YA was equivalent to a 0./ ft2 break in a PWR, RELAP5YA may be used to analyze
' I.
breaks larger than this if the important phenomena for the larger break sizes are similar to that assessed for the 0.7 ft2 and smaller creaks.
e e
lii 9
CONTENTS 1
ABSTRACT..............................................................
11
SUMMARY
............................................................... iii 1.
INTRODUCTION.....................................................
1 2.
RELAP5YA CODE DESCRIPTION AND ASSESSMENT.........................
3 2.1 General Code Overview......................................
3 2.2 Model Description and Assessment...........................
5 2.2.1 Interphase Drag Models for the Vertical Flow Regime............................................
5 2.2.2 Moody Critical Flow Model.........................
7 2.2.3 Accumulator.......................................
9 2.2.4 Forced Convective loiling.........................
9 2.2.5 Critical Heat Flux................................
10 2.2.6 Rewet and Quench Models...........................
12 2.2.7 Multiple Surface Radiation........................
14 2.2.8 Heat Transfer Logic Options.......................
15 2.2.9 Fuel Behavior Models..............................
15 2.3 A s se s smen t Cal cul ati on s..........................,........
17
^
2.3.1 LOFT Small Break Assessment Calculations..........
17 2.3.2 LOFT Intermediate Break Assessment Calculation....
25 2.3.3 Semiscale Assessment Calculation..................
35 2.3.4 THTF Assessment Calculations......................
44 2.3.5 Assessment Summary................................
44 2.4 Phenomena important to PWR $8LOCA.........................
a6 2.4.1 Noncondensible Gas................................
46 2.4.2 Condensation Heat Transfer........................
50 2.4.3 Natural Circulation...............................
53 g
l 2.5 Calculation of S-UT-8 Phenomena...........................
58 l;
i 3.
COMPLIANCE WITH NRC REQUIREMENTS.................................
59 4
R E LA P 5 /M001 C OD E U PDAT E S.........................................
51
?
e 5.
SYSTEM M00'ELING TECHNIQUES.......................................
62 6.
CONCLUSIONS......................................................
66 7.
REF.ERENCES.................,......................................
71 l
1
{
i r
s e
IV
e FIGURES 2.2.5-1 Comparison of RELAP5YA calculated critical qualities to measured critical qualities from assessment cases...........
13 2.3.1-1 Comparison of primary system pressure in LOFT Test L3-1 to RELAP5YA calculation.....................................
19 2.3.1-2 Comparison of break flow in LOFT Test L3-1 to RELAP5YA calculation.................................................
20 2.3.1-3 Comparison of primary system pressure in LOFT Test L3-6 to RELAP5YA calculation.......'.................................
22 l
2.3.1-4 Comparison of break flow in LOFT Test L3-6 to RELAP5YA calculation.................................................
23 2.3.1-5 Comparison of syste.n mass in LOFT Test L3-6 to RELAP5YA calculation.................................................
24 2.3.1-6 Comparison of primary system pressure in LOFT Test L8-1 to original (88.5*.F ECC) and revised (200'F ECC) RELAP5YA calculations................................................
26 f
t 2.3.2-1 Comoarison of primary system pressure in LOFT Test LS-1
'. [
to RELAP5YA sensitivity calculation.........................
28
~
t i
2.3.2-2 Comoarison of break flow in LOFT Test L5-1 to RELAP5YA sensitivity calculation....................................
29 6
2.3.2-3 Comparison of primary system mass in LOFT Test L5-1 to RELAP5YA sensitivity calculation............................
30 2.3.2-4 Comoarison of accumulator mass flew rate in LOFT Test LS-1 to RELAP5YA sensitivity calculation............................
32 r
2.3.2-5 Comparison of clad temperatures in LOFT Test L5-1 to REL4P5YA sensitivity calculation (0.76 m elev.).......:..............
33 i
2.3.2-6 Comparison of core outlet temocratures in LOFT Test L5-1 to RELAPSYA sensitivity calculation............................
34 2.3.3-1 Comearison of core collapsed liquid level in Semiscale Test S-LH-1 to RELAP5YA sensitivity calculation..................
38 e
2.3.3-2 Comoarison of clad temoeratures in Semiscale Test S-LH-1 to I
RELAP5YA sensitivity calculation (228 cm elev.).............
39 2.3.3-3 Comparison of break flow in Semiscale Test S-LH-1 to RELAP5YA sensitivity calculation.....................................
40 2.3.3-4 Comparison of system pressure _in Semiscale Test S-LH-1 to -
RELAP5YA sensitivity calculation............................
41 '
V l
4 2.3.3-5 Comparison of. integrated break flow in Semiscale Test S-LH-1 to RELAP5YA sensitivity calculation.........................
42 2.4.3-1 Comparison of natural circulation flow rates from Semiscale Test S-NC-2 to RELAP5YA calculation - 30 kW case............
55 2.4.3-2 Comparison of natural circulation flow rates from Semiscale Tests S-NC-2 and S-NC-10 to RELAP5YA calculation -
100 kW case.................................................
56 5-1 Proposed Maine Yankee nodalization for SBLOCA EM calculations................................................
63 5-2 Proposed Yankee Rowe nodalization for SBLOCA EM calculations................................................
64 I
I l
i i
[
i I
l
-l
{
c l
I
{
[
e Vi
TECHNICAL EVALUATION REPORT RELAPSYA COMpVTER PROGRAM FOR USE IN pWR SMALL BREAK LOCA ANALYSES
- 1. INTRODUCTION i
RELAPSYAl is a computer program developed by Yankee Atomic Electric Company (YAEC) for light-water reactor (LWR) system thermal-hydraulic j
analysis.
It provides integral analysis capability of the system and core i
response to normal and off-normal events during steady state and l
RELAPSYA was adapted from RELAP5/M001, cycle 18.2 by YAEC
(
for use in pressurized water reactor (PWR) small break loss-of-coolant j
i accident (58LOCA) analyses.
RELAP5YA was submitted to the Nuclear Regulatory Cosmiission (NRC) by YAEC for review and acceptance for licensing applications as a method to analyze PWR $8LOCAs in a manner that conforms f
to NRC requirements contained in 10CFR50.46, Appendix K, and other pertinent NRC regulations.
The Office of Nuclear Reactor Regulation (NRR) is responsible for the
.- a evaluation and review of computer codes and their proposed applications.
l NRR requested the Idaho National Engineering Laboratory (INEL) to provide i
assistance in the review of tt;s RELAP5YA computer code.
Specifically, the
- [
request for assistance included' l
1.
Evaluation of RELAP5YA as a method to analyze the PWR small break j
spectrum.
[
l I
2.
Evaluation of RELAPSYA for compliance with requirements contained' i
in 10CFR50.44, Appendix K, and NUREG-0737, Item II.K.3.30.
i l
3.
Assurance.that corrections to RELAP5/M001, cycle 18, the base
'h j
code for RELAPSYA were incorporated into RELAP5YA.
l e
J i
I Related to the above reviews, NRR also requested INEL review and t
ev'aluate the ut,lity's responsws to NRC questions regarding the l
i I
loss-of-coolant modeling applications.
The responses re' viewed were for tne f
I I
1 l
}
- 197 general PWR and SWR questions 3 and the additional PWR and BWR 4
questions generated by this review.
YAEC's responses to these questions are contained in References 5 to 11.
This technical evaluation report contains the results of the RELAP5YA review and assessment for PWR SBLOCA analyses.
Section 2 provides a brief overview of the history of RELAP5YA and its development from RELAP5/M001, cycle 18.
This section also discusses YAEC's modifications to the RELAP5 base code and YAEC's code assessment work.
Section 3 reviews the code for compliance with NRC requirements defined in Appendix K to 10CFR50.
Section 4 reviews the implementation status in RELAP5YA of all documented updates to RELAP5/M001, cycle 18, generated by the code developers at the
!NEL.
The integral system modeling techniques proposed for use in YAEC's plant licensing analyses are discussed in Section 5.
Section 6 sumarizes the conclusions reached from this review and the references are listed in l
Section 7.
l
- I 1
I i
i 4
=
d e
2
~
g
___9.__
)
s e
- 2. RELAP5YA CODE DESCRIPTf0N AND ASSESSMENT -
This section presents a brief description of RELAP5YA and discusses its relationship to the RELAP5/M001, cycle 18, code from which it was developed.
Then each of the PWR-related modifications and assessments made by YAEC will be discussed.
The results of RELAP5YA integral assessment calculations will then be reviewed followed by a discussion of the code's ability to simulate the phenomena important to PWR SBLOCAs, Finally, the code's ability to simulate the type of phenomena observed in Semiscale Test S-UT-8 is discussed.
2.1 General Code Overview RELAP5YA is a light-water reactor system transient simulation code based on a nonequilibrium and nonhomogeneous model for two-phase conditions.
To minimize the number of constitutive relations required for mathematical model closure, the least massivs chase (steam when the static quality is less than 0.5 and liquid when the static quality is greater than or equal to 0.5) was assumed to be at the local saturation condition.
The code formulation of the hydrodynamic components, power sources, heated structures, trips, and control systems provides a flexible method for modeling LWR systems.
RELAP5YA includes many general component models from which general systems can be simulated.
The component models include i
pumps, valves, pipes, heat structures, reactor point kinetics, l
accumulators, and control system comconents.
Special process models are included to account for form losses, abrupt area changes, branches, choked
\\
flow, boron tracking, and noncondensible gases.
i RELAP5YA is based on RELAP5/M001 which was developed under USNRC i
sponsorship.
YAEC decided to maintain the same modeling philosophy in 3
-t RELAPSYA as in the base code.
The same formulation of the oifferential equations for the thermal-hydraulic models, the same basic constitutive relations, code architecture, principle solution techniques, and user convenient features were retained in RELAPSYA.
As the RELAP5/M001 coce develoement effort continued, the cycle of the M001 code on snich RELAP5YA i
i was based changed until cycle 18 was rel' eased.
After cycle 18 was g
.y
)
released, further changes in RELAP5/M001 development where incorporated into RELAP5YA by using selected updates obtained from the RELAP5 code devel:pment group at the 1NEL.
This is discussed more fully in Section 4.
i RELAP5/H001 was specifically developed to include the capability to simulate SBLOCAs and operational transients.
The application of RELAP5/M001 to a wide variety of thermal-hydraulic problems, including many small break LOCAs, indicated the formulation of the differential equations and solution techniques can provide numerically stable solutions.
With RELAP5YA maintaining the same basic approach, it can be concluded that it
)
will also provide numerically stable solutions and, theref ore, the general code models and structure are adequate for PWR SBLOCA analyses.
l l
YAEC's assessment of RELAP5YA identified several areas that required new or improved models to meet its LOCA ar.alysis needs.
In addition, several changes to RELAP5/M001 were needed to meet the Evaluation Model l
requirements of Appendix K to 10CFR50.
As a result YAEC began a code development program in the following areas:
l 1.
New interphase drag models for the vertical flow regime map.
2.
Addition of the Moody two-phase critical flow model.
3.
Addition of a new accumulator model.
i 4.* Modify the forced convective boiling algorithm.
i i
5.
Addition of a new critical heat flux option.
y e
6.
Addition of a rowet and quench model for reflood and spray cooling.
7.
Addition of a multiple surface radiation heat transfer model.
9 4
4 O
s 8.
Addition of return to nucleate boiling and transition boiling lockout logic.
t 9.
Addition of new fuel rod behavior <aodels to represent fuel rod fission gases, rod deformation and rupture, gap conductance, and zircaloy-water reaction.
The modifications made by YAEC and their assessment are presented in more detail in Sectjon 2.2.
2.2 Model Description and Assessment,,
2.2.1 Interchase Orac Models for the Vertical Flow tecime Ouring an assessment calculation (Appendix 8 Reference 11) of
)
Samiscale Test S-LH-112 RELAP5YA had problems calculating ECC penetration into the downcomer.
As a result, most of the injected accumulator water wen? out the break and the core was not calculated to cuench.
The difficulties encountered in this assessment calculation led YAEC to reassess the modeling of interphase drag in vertical flow.13 After a considerable amount of work, YAEC developed a new flow regime map for vertical flow and new models for calculating interphase drag.14 This section briafly describes the new model and discusses the separate effects assessment calculations.
The effect of the new interphase drag models on the integral assessment calculations for Semiscale Test 5-LH-1 and LOFT i
Test LS-1 is d,iscussed in Sections 2.3.3 and 2.3.2, respectively.
The new flow regise map consists of two well defined regions, one for bubbly flow and one for annular-mist flow.
All other regio s, slug, churn, and froth, are lumped into one regime (called the churn regime) which
'[
providet a buffer between the two well defined regions.
These regions were not explicitly modeled because it was difficult to establish a mechanistic model for interphase drag in these regimes.
The churn regime is used to I
ensure continuity in the interphase drag over the entire range of void-fraction.
The bubbly-churn transitier. criteria were eased on the work of l
Taitel, et al.15 The churn-annular transition criteria were cased on the i
work of'Mishima and Ishii 16 and Wallis.17 A detailed description 5
of the'new models for calculating interphase drag in the flow regimes defined above is presented in Reference 14.
The main features of the new models include:
for vertical flow deriving the single-phase interphase drag limits as hetural extensions of the two-phase models, setting the single-phase interphase drag limit at 105 N-s/m4 for non-vertical components, mechanistically treating interphase drag in the bubbly and annular-mist regimes, calculating interphase drag in the churn regime by interpolating between the bubbly and acr.ular-mist regimes using a cubic spline fit, and removing the spatisi and temporal averaging of interphase drag used in earlier versions of RELAP!YA.
3 The new interphase drag models 4cr the vertical flow regime were assessed against data taken in FRIGG18.19 and GE lev 6) :well experiments.20 8ecause it is not possible to directly measure interphase drag, RELAP$YA's ability to calculate interphase drag was indirectly checked by comparing the measured and calculaten void fraction profiles, YAEC used RELAP5YA to analyze several groups of tests, at mass fluxes ranging from 472-1492 kg/s-m2, in the FRIGG test facility.
ret.APSYA did an acceptable job of calculating the void fraction'versus dynamic equilibrium quality profile in the uniform axial heat flux tests.
In tests where the inlet water was substantially subcooled (greater than 10*C),
the code underpredicted the measured void b action for void fractions less tnan 20%.
This was attributed to a lack of a subcooled boiling model.
The lack of a subcooled boiling model in RELAP5YA may affect the results of some steady state calculations but should have little effect on PWR 58LOCA f.
calculations where the system reaches saturatie.1 rapidly.
Measured void frsctions in the 20 to 70% range werc accurately calculated by the code, i
In some tests for void fractions greater thar. 70%, the code overpredicted tne void fraction.
From this YAEC concluded RELAP5YA overestimated
'h.
interphase drag at these void fractions.
However, this is conservative, c
Two additional FRIGG tests, with nonuniform axial heat flux profiles, were used to assess RELAP5YA.
Comparison of tho void fraction versus dynamic equilibrium quality for the calculations and the experiments showec 90c0 agreement below 70% void fraction and that RELAP5YA overpredicted tne test cata at void fractions above 70%.
This is consistent with the results fr:m
- tne uniform axial heat tiux testu and is conservative.
6
o
. GE levet swell Test 1004-3 addressed transient, two-phase level behavior in a cyttndrical vessel. Mass fluxes in the experiment ranged from 20-100 kg/s-m2 The code results match'ed the vessel pressure early t
in the-transient and slightly underpredicted the pressure net-
%. ena of l
t the transient.
The code did an excellent job of calculating t...
W.nsient void profile in the vessel.
i
)
i Based on these assessment calculations, the new interphase drag models in RELAP5YA are judged to be properly implemented and to provide an
[
adequate simulation of interphase drag related phanomena over the range of l
mass fluxes expected in a PWR $8LOCA.
l I
2.2.2 Moody Critical Flow Model,
)
l YAEC added the Moody Critical Flow mede121 to RELAP5YA as a user i
selected opi. ion to meet the Appendix K requirement that tne Moody model oe f
used at all break locations to calculate two-phase critical flow during a l,f
.U)CA.
The raodel was incorporated as a table of mass flux versus stagnation enthalpy and pressure. This critical flow table may also be entered with 2
l static pressure and enthalpy from either the donor cell or a user designated cell.
These options were included in RELAP5YA because of l
comoutational difficulties which arose when the Moody model was applied i
using stagnation conditions from the donor cell.
Difficulties also arose l
because of differences between the Moody calculat+on of the slip ratio, f
wnich is implicit in the Moody mass flux table, and the slip calculated I
from the basic equations in RELAPSYA.
To overcome this problem, Moody's i
l slip ratio was replaced by a salution of the difference momentum equation f
f using a relatively large interphase drag.
This resulted in slip ratios of i
i 1.0 to 1.3, which are more compatible with the slip ratios calculated by I
t the RELAPSYA equation set.
Code assessment work by YAEC discussed below y
-l a
showed this approach resulted in stable solutions when used with the static conditions from the donor cell.
i r
t l
When the Moody option is selected, the stindard RELAP5VA syecoolec l
critical flow model is used until the' void fraction is 0.05.
For void
~
j fractions greater than 0.05, the Moody n. cent is used.
YAEC notec in its f
1
-[
. ~ _. -
~
response to Q1.19 in Reference 11 that under herizontal stratified' choked s
flow conditior.s the Moody critical flow model is modified to account for vapor pull-through and liquid antreinment using the methods developed in
[
the RELAP5/M001 hase code.
Information was supplied in Reference 14 to j
show the stratifie.$ choked flow model was properly Nplemented with the
(
i Moody model.
Additicnal information on the isolecentation of the Moody l
model in RELAP5YA is found in Section 3.2 of the RELAP5YA ccde description
]
manual, Volume 1.
i y
As discussed above, the Moody critical flow table may be entered with the donor cell stagnation pressure and enthalpy, the donor cell static
(
j pressure and enthalpy, or the static pressure and enthalpy from a user I
specified cell.
YAEC performed a number of checkout runs, using each of I
th6.e cetions, in oreer to provide user guidelinet for selecting the appropriate uption when the Moody model is used in 4 (alculation.
lased on
(
i these runs, YAEC recomends the Moody model be usr.d '4ith the donor cell l
static preswee and enthalpy option (see Section 2.2 of the RELAP5'lA l
1
)
manual, Volume 3).
l
~I
{
The Moody criti n) flow model is impi nented in RELAP5YA without using j
Moody's theoretical slip modsl.
In its response to a question (Q.!X.17, l
Reference 6), YAEC noted RELAPbYA u buintes lower slip ratios than Moody's l
model. Thit,is conservative, however, becaust it results in more Itquid and less venor going out the break increasing the System mass loss.
It I
also decreases the systou depressurization rate relative to the f
r depressurintion rate,calcus.ted using the Moody slip model.
This results
{
l l
in reded ECC flow rates.
This change is judged to be properly j
l imelemented.
E I
t i
YAEC assessed the implementacion of the Moody model in RELAP5YA by
?
'l simulating Marvd.en Test 1022 using the recosamended Moody model option.
The results showed the calculated break flow exceeoed the measuced flow by l
f approximately 15% the vessel pressure decreased-more rapidly than tne test
]
data, and that the synten voided more rapidly than observec in the test.
These results indicata the Moody model un prope-ly implemented in RELAHYA j
S t
o ~
'and provides conservative break flow calculattens relative to the Marviken test data. These results also confirm the use of the recommended donor cell static pressure and enthalpy option with the Moody model.
2.2.3 Accumulator In its response to question Q.I.23 in Reference 7, VAEC stated the accumulator mocel in RELAP5YA is the same nodel found in RELAP5/M001,
~
cycle 18.
This replaced the model described in the RELAPSYA manual.
The details of the RELAPC/M001-accumulator model are found in Reference 2, Volurae 1, Section 2.3.T.
This is the accumulator model reviewed below.
YAEC presented the resultr of assessment calculations with the RELAP5/M001, cycle 18, accumulator 'nodel in response to questions Q.I.29 and Q.I.31 (References 7 and O.
The tests simulated in these calculations were the Maine Yankee Accumulator tasts and the accumulator injection in LOFT Test L3-1.23 The results snow d that the accumulator model in RELAP5YA accurately predicted the transient accumulator pressure and level response.
Therefore, the accumulator model in RELAPSYA is considerd adequate for lic.ensing analyses.
l 2.2.4 Forced Conysetive Boilino The Chen correlation 24 was originally used to calculate saturated nucleate boiling heat transfer coefficients and the modified Chen correlation was used to calculate subcooled nucleats boiling heat transfer coefficients.
YAEC's. assessment of th's Chen correlation indicated that the heat transfer coefficient is underpredicted at low quality and high mass fl u".
The magnitude of the nucleate boiling heat transfer coefficient is l
imprunt in steam generator heat transfer.
If the steam generator heat 3
transf*, cannot be accurately calculated, then all of the heat generated in the cc.cc canot be removed and accurate steady state initial conditions cacwt ;e calcula.:ed.
YAEC revwwed the data used to develop the Chen correlation and founa the tr,e data enest applicable to PWR applications was the data of Schrock and Grossman.25 This data was taken at low pressures (60 to 380 psia) 9 l
.A
and high void fractionsa Conditions in PWR steam. generators include subcooled and low quality saturated conditions up to 1000 psia.
To improve the nucleate boiling heat transfer calculations, YAEC added the Thom correlation 26 to RELAP5YA.
The Thom correlation was.selectec,because YAEC felt it was based on data more applicable to nucleate boiling conditions in PWR steam generators.
The H t tran*far package in RELAP5YA uses both the Thom and the Chen correlations to calculate nucleate boiling heat transfer.
The Thom correlation is used for void tractions less than 0.80.
The Chen correlation is used l'or void fractions above 0.90.
If the void fraction is between 0.80 and 0.90, then an interpolation t a, tween the Thom and Chen correlations is used.
YAEC feels this approach is more consistent with the range of data used in the development of each correlation.
This change was assessed by comparing the results from RELAP5YA to Bennett single tube tests.27 This comparison showed that the measured temperatures were more accurately predicted at void fractions below 0.8 when the Tnom and Chen correlations were used, as described above, than when the Chen correlation was used by itself.
%bove a void fraction of 0.8, the calculated temperatures were above the measured temperatures, 1
which is conservative.
This change to the heat transfer logic usulted in an improved simulation of the Bennett single tube data at low void t
fractions when compared to the original logic and, therefore, is recommended for acceptance for use in licensing calculations.
2.2.5 Critical Heat Flux
\\
As a result of its code assessment work, YAEC felt more accurate CHF l
j predictions were needed.
A new CHF option was added that uses two
[
correlations to cover the conditions expected in a PWR core during a LOCA.
28 is used.
l
~
At high mass fluxes a modified Srm of the Biasi correlation The modification accounts for the fict that the correlation was, originally developed from tube data as opposed to the bundle geometry in a reactor core.
At low mass fluxes, the Griffith-Zuber correlation 29 is used.
For l
intermediate mass fluxes, a linear interpolation between the two l
correlations is used to obtain the CHF.
For both correlations, a minimum 10 i
1 0
o 2
'CHF value of 1000 W/m2 (317 Btu /hr-ft ) is imposed.
In addition, a critical void fraction criterion is used so that wall dryout is assumed to occur when the void fraction exceeds the critical void.
More information on the new CHF option is found in Section 4.2 of the RELAP5YA manual, Volume 1 J
YAEC assessed this new critical heat flux option against steady state CHF tests performed at the ColumMa University Chemical Engineering Research Laboratcry,30,31 General Electric,32 and in the Oak Ridge National Laboratory Thermal-Hydraulics Test Facility (THTF).33 The results of steady state film boiling tests at THTF were also used to assess the new CHF option (Reference 33).
The tests chosen for assessment were selected to be ro resentative of PWR fuel red bundles, to cover a range of thermal-hydraulic conditions, and different axial and radial power distributions.
YAEC, in its response to question Q.VII.7 in Reference 9, stated the new correlations in RELAPSYA were best estimate CHF correlations and therefore were assessed for accuracy rather than conservatism.
?.
Five tests from the Columbia test program were used in the assessment.
The RELAPSYA results showed CHF was calculated within one computational cell of the measured location for all the tests (nine nodes were used to represent the 150 in. long rods).
For two of the tests, the i
calculated CHF location and quality were higher than u,e ~easured data.
YAEC attributed this difference to the fact that the high and low power rods in the test section were represented by ons average rod in the RELAP5YA model.
YAEC stated CHF would have been calculated at a lower quality and location if a high power rod had been modeled.
This is true l
and this ai: curacy is considered adequate.
YAEC also assessed the new CHF option in RELAP5YA against data from the THTF at Oak Ridge National Laboratory.
For the steady state film boiling tests (Tests 3.07.98, X, and X), comparison of the calculated and measured CHF locations and qualities showed the calculated location and quality were lower than the measured data in all three assessment runs.. In
'he steady state film boiling tests, YAEC noted, in its response to 11 S
S t
---..-.my
i c.uestion Q.VII.13 in Reference 7, that when compared to the CHF location j
inferred from the test data, the calculated CHF location was at the same elevation or lower.
These results are conservative relative to the data.
The GE CHF tests were performed in the Ni.7e Rod Test Section.
For the four tests analyzed, the calculated CHF location was higher than the measured location in three of them.
- 1. the fourth test the calculated location was lower than the measured location.
In terms of the calculattd and measured quality at CHF, the CHF quality was accurately predicted in two of the tests, was low by 5% in one test, and high by 16% in the last test.
The new CHF correlations in RELAP5YA were assessed agt 'st a wide variety of test conditions.
The results, summarized in F199re 2.2.5-1, showed that in general the correlations did a reascnable job of calculating CHF for the tests analyzed.
The new CHF option is, therefore, recommended for acceptance for licensing analyses.
2.2.6 Rowet and Quench Models As part of its code development effort, YAEC added a rewet and quench model to RELAP5YA as a user option.
The rewet and quench algorithm consists of four submodels:
rewet and quench initialization, quench front velocity, heat transfer enhancement, and quench front advancement.
Each of these submodelt is discussed in detail in the RELAP5YA manual, Volume 1 Section 4.3.
j i
The new reflood models in RELAP5YA were assessed against two separate effects tests for bottom quench and three integral tests which included both falling. film and bottom quench.
The separate effects tests will be h
discussed here.
The assessment of the the reflood models based on the integral test data will be discussed in Section 2.3.
Thi two separate effects tests used to assess the reflood' model were I
from THTF, Tests 3.09.100 and 3.09.10Q.34 The initial conditions far' these tests were pressures of approximately 565 psia, a refloo( rate of i
I 12
1.0 I
i i
l i
0.8 -
E O
D 0.6 o
'E 3
O I
0.4
- EW C
u O
..D..
l 0.2 A
Columbia o
O GE A
l i
l O
OLNI.
l i
I i
1 0.0 t
.I t
I O.2 0.4 0.6 0.8 1.0
?
l e
l Drperimental Critical Quality Figure 2.2.5-1.
Comparison of RELAP5YA calculated critical qualities l
' to measured critical qualities from assessment cases.
13 l
4.8 in/s and a linear heat rate of 0.62 kW/ft in Test 3.09.100 and a' reflood rate of 2.3 in/s and a linear heat rate of 0.31 kW/ft in Test 3.09.10Q.
The results of the assessment calculation for Test 3.09.100 showed the code accurately calculated the system vold Traci. ion profile as a function of time.
The calculated bundle mass inventory, the core collapsed liquid level, and the quench front in the core compare well to the test data.
When the calculated results differed from the test data for these parameters, they were in the conservative direction (low).
The calculated rod temperature profiles at four different points in time were compared to the test data.
This romparisvi1 showed that, except for the effect of the grid spacers, the code calculated temperature profiles compare well with the data.
A comparison of the calculated and measured quench times showad that the code calculated the rod quench at a time that was approximately the average of the quench times at a given core elevation.
The Test 3.09.10Q calculation gave results very similar to those calculated for Test 3.09.100.
The code calculated bundle mass inventory,
}
the core collapsed level, and the quench front did not compare quite as well to the measured data for Test 3.09.10Q as for Test 3.09.100 but the differences were still in the conservative direction.
Also, the calculated quench time was later than shown in the test data.
RELAp5YA did a reasonable job of calculating the bottom reficsd tests at T,HTF.
For the important parameters where there was a difference between the code calculated results and the test
- data, the code was conservative.
j Therefore, the new reflood models in RELAP5YA are recommended for use in l'
licensing calculations, i
u 2.2.7 Multiple Surface Radiation b
YAEC developed a multiple surface radiation heat transfer model for RELAPSYA.
Because radiation heat transfer is not usually imocrtant in PWR 1
SBLOCA analyses,, this model will not be discussed in detail.
The models are discussed in Section 4.4 of the RELAPSYA manual, Volume 1.
I 14 e
.1
~
YAEC performed a series of assessment calculations to ensure that the radiation heat tran:fer model was properly implemented in the code.
YAEC chose to assess the new model by running a number of sample problems.
The results of the assessment calculations indicate the model accurately calculates radiation heat fluxes based on a given view factor matrix and is properly implemented in the code.
Therefore, contingent on the justification of the view factor matrix used in the licensing analyses, or the licensing of the view factor matrix code (if one is used), the model is adequate for li, censing analyses.
2.2.8 Heat Transfer Looic Options Appendix K to 10CFR50 requiras a code lockout return to nucleate boiling heat transfer once CHF is predicted at an axial fuel rod location during blowdown.
Appendix K also requires a return to transition boiling lockout once the clad superheat exceeds 300*F during the blowdown phase of a LOCA.
These requirements force a degraded heat transfer calculation during the blowdown phase of a LOCA even though local conditions may allow a rowet and a return to nucleate boiling.
YAEC added the appropriate heat transfer logic to RELAPSYA to allow the user to meet these Appendix K requirements.
However in Appendix 8 of Reference 11, YAEC noted that, based on an agreement reached with the NRC during a June 22, 1984 meeting at Bethesda, MO, it does not plan to use the EM heat transfer lockouts for P' R SBLOCA analyses.
Therefore, this technical evaluation did not review chose options.
If tho' NRC should later require YAEC to use the lockouts durirg PWR SBLOCA analyses, these options will need to be reviewed.
4 l
2.2.9 Fuel Behavior Models Fuel behavior models were added to RtLAP5YA to enable the fuel rod behavior to be calculated during a LOCA.
These models include fuel rod internal pressure, fuel rod deformation, fuel rod gap heat transfer, and 2 rcaloy-water reaction.
The zircaloy-water reaction model js based on the 4
Baker-Just model as required by Appendix K.
All the fuel behavior models were adapted from the T000EE2-EM code 35 which is part of the Yankee Atomic approved water reactor evaluation model.36 Slight mo'difications
\\
15 4
L-.
4 were made to the clad deformation and rupture model and the internal pressure model to ease their implementation in' RELAPSYA or to reflect more recent modeling information.
These models are described in detail in Section 5.0 of the RELAP5YA manual, Volume 1.
Because the fuel behavior models in RELAP5YA were derived from the models in the T000EE2-EM code, YAEC assessed the RELAP5YA fuel behavior models by comparing RELAP5YA and T000EE2-EM calculated results.
For the assessment, YAEC chose a sample problem representing the adiabatic heat up of a single PWR type fuel rod.
Thi.s comparison showed the codes produced essentially the same results.
The differences between the calculations were not significant and could be explained by known differences between the two codes.
The major difference between RELAP5YA and T000EE2-EM was in the calculated metal-water reaction rate at high clad temperatures.
The RELAP5YA results showed a higher heating rate than those from T000EE2-EM which is conservativa.
To ensure that the algorithm for calculating the zirconium-water
[
reaction was implemented properly, YAEC compared RELAPSYA calculated results to hand calculations.
The two results compared exactly indicating the model was implemented properly in the code.
The red internal pressure model assumes the plenum gas temperature to be the temperature of the adjacent coolant plus a constant offset.
One of the modifications to the internal pressure model was to increase this offset.
Originally, the offset was set to 2*F (based on YAEC's response f
to Q.VII.3 in Reference 9).
Subsequently YAEC stated the offset was l
increased from 2 to 10*F (response to Q1.12 in Reference 10).
The 2*F offset was chosen to be consistent with the T000EE2 code input used y
for Maine Yanket and Yankee Rowe large break LOCA analys'es.
The 10*F 2
1 offset was chosen to be consistent aith FROSSTEY37 and FRAP-T1.38 Using a 10*F offsst, rather than 2*f, results in a more conservative
- alculation because the internal pressure predictions will be slightly i
higher.
~
16
~
-,--,.,,---,,,4
The assessment calculation discussed above was completed before YAEC's response to Q.VII.3 in Reference 9 was given.
Thus, it is as.tumed the i
calculation used the 2*F offset described in answering this question and, as a cesult, there was no direct assessment of the fuel behavior models with the 10*F of.fset.
Howev6r, as noted above, the increase t
from 2 to 10'F is conservative.
Therefore, it is acceptable for YAEC to use any temperature offset greater than 2*F in its licensing calculations.
Justification and assessment should be provided for any effset less than 2*F.
Because the RELAP5YA fuel behavior models were properly implemented and give essentially the same results as T000EE2-EM, which was accepted for licensing analyses, they are reconenended for acceptance for use in RELAPSYA licensing analyses provided the temperature offset in the rod internal j
pressure model is greater than 2*F.
2.3 Assessment Calculations RELAP5YA was assessed against a LOFT small break experiment, L3-1, and a combined i.0FT small break / severe core transient experiment, L3-6/LB-1.39 LOFT L5-1,40 an intermediate break size experiment, one Semiscale small break experiment, Test S-LH-1, and several THTF experiments were also used to assess RELAP5YA.
The LOFT small break assessment calculations will be discussed first followed by LOFT L5-1, the Semiscale calculation, and the THTF calculations.
2.3.1 LOFT Small Break Assessment Calculations j
t i
The main differences between the two small break tests used to assess RELAP5YA were:
(1) the break was located in the inactive loop cold leg l,
(simulating the pump and steam generator resistance only) for L3-1 and in the active loop cold leg (with the active pump and steam generator) for l
L3-6; (2) the pumos were trippect early in the experiment in L3-1 while they were left running in L3-6; and (3) ir. L3-1 the ECC was injected into the intact looo cold leg while in L3-6 it was injected directly into the vessel downcomer.
The results of the L3-1 calculation will be discussed first followed by the L3-6/L8-1 results.
17
~
1 The calculated system pressure was in good agreement with the L3-1 data for the first 350 s of the experiment (see Figure 2.3.1-1).
After 350 s, 1!he calculated pressure dec.' ease is less than in the experiment so that the calculated pressure is highir than in the experiment.
This is conservative because the higher systeo pressure will result in lower ECCS injection rat'es.
The secondary pressure calculated by RELAP5YA is lower than the test data after 725 s.
This was attributed to more secondary side condensation in the calculation than in the experiment.
The excess condensation was calculated even though the auxiliary feedwater temperature was raised to 200*F to limit the condensation rate.
However, YAEC showed that the steam generator secondary pressure has little impact on the primary system responsa.
The original L3-6 calculation also underpredicted the secondary A sensitivity calculation for L3-6, with the secondary pressure pressure.
forced to be the same as in the experiment, showed that the primary system results changed very little compared to the original calculation.
The calculated break flow agreed well with the measured break flow.
- i although between 350 to 1000 s it was slightly highar than the test data (Figure 2.3.1-2).
The data from LOFT L3-1 was also used to assess the horizontal stratified choked flow model in RELAPSYA.
The use of this data to assess the stratified choked flow model was discussed by YAEC in its response to question Q1.19 in Reference 11.
This response discussed why YAEC personnel believed the data indicated the break in this experiment was subjected to horizontal stratified choked flow.
Information was also presented to show the calculation used the horizontal stratified choked
{
flow model during a large part of the analysis to calculate the break i
flow.
A comparison of the calculated and measured break flow rate showed the two compared reasonably well.
This information indicates the f
'i horizontal stratified choked flow model was properly implemented with the standard two-phase choked flow model (i.e., not the Moody model).
i A rod heatup was not observed in the experiment as the core was weil cooled throughout the test:
The RELAP5YA simulation also did not calculate a heatup as the core was well cooled throughout the analysis.
~
18
.l l
)
_, _ _ _ __ _ _ L D
(
~
g o
O g
af a
E ma
- a. g
~
2-eflEUF5YA y
LE T L3-1
.s o f g e5
's es a 9 g_
g G
'd a
~.
gd
~
h"d
. u.
~;;
a d.
o.e sas.e aus.e aus.e tes.e ses.e aus.e Foo.e sou.s suo.e anos.e slee.e saos.e a3ou.e teos.e stes.e seca.e spen.e seus.e TIME (SEC)
Figure 2.3.1-1.
Comparison'of primary system pressure in LOFT Test L3-1 to RELAPSYA calculation.
i
. ~
j
..m i
o I
I I
i a
a g
g t
s t
a o, 8 ~
~
o u
N i
l ro m g_
a,..
v RELfPSYf)
Nct
=
c L0f'T L3-1 g-x
~ s roo
__1 @*
L
~
~
O' se a-uo x a.
m.-.
EWh# 9,^,-S~i-v,/v.-Eg I o
o_
I I
I I
I I
I I
0.0 20G.0 400.0 600.0 000.0 1000.0 1200.0 1400.0 1600.0 1800.0 TIME (SEC)
Figure 2.3.1-2.
Comparison of break flow in LOFT Test L3-1 to REL/PSYA calculation.
9 M=8el>=4A
's e
n'
The comparison showed that RELAPSYA did a reasonable job of calculating the system response.in L3-1.
Where the code result's differed from the tes't data, the difference was in the conservative direction.
For Test L3-6, the calculated system pressure for the first 1000 s was in excellent agreement with the data (see Figure 2.3.1-3).
After 1000 s the calculated pressure dropped below the measured pressure but by no more than 50 psia. The underprediction of the primary pressure by RELAP5YA is not strictly a conservative difference because the lower primary pressure would result in a nigher ECC flow rate.
However, the lower calculated pressure only resulted in a maximum 0.13 lbm/s difference between the HPIS flow rate in the calculation and the total measured flow of 1.3 lbm/s.
(See response to question Q.V.6 in Reference 5.) Therefore, the total mass difference was less than 182 lbm for the time period from 1000 to 2400 s when the HPIS was terminated. Therefore, the effect on the overall calculation was not significant.
The comparisons of the measured and calculated break flow The (Figure 2.3.1-4) and intact loop densities showed good agreement.
comparison of the calculated and measured primary system mass inventory for L3-6 showed the code calculated the mass inventory withir the experimental uncertointy for most of the o.<periment (Figure 2.3.1-5).
Where the code differed it was in the conservative (low) direction.
After the L3-6 experiment was completed the rmactor coolant pumps and the HPIS were shut off resulting in a core heatup.
This began Test L8-1.
~
When the rods in the core reached a temperature of 650*F the break was f
isolated and the core reflooded with unscaled HPIS and accumulator flow.
l This provided data which YAEC used to assess the reflood models added to RELAP5YA.
~
The trends of the L8-1 system pressure were accurately calcuiated by When,the test was terminated by injecting the unscaled RELAPSYA.
accumulator flow, the primary pressure dropped due to condensation induced The code, because it overpredicted the condensation by the ECC water.
effect of,the accunulator water, overpredicted the system 21 e
.. ~
f WW W
TT WWWUWW WW WW W gy WW WW WV VTWWW WWVgW WWW WWWWgWWY WW W g*4W W
WWlWW WWWWWWgWWW W
WW gW l.
I.
- .a.
i
==
ii h
e n
a 6J
.J E
p :.
.N t
tort u-o s
e-Wo
+-3
.x t
=
anAPSYA g
i
........... _.i
..J..
.n....
-.i..
.%L as see see tse loos less tsee 6764 tece tabe; 24 4
TIME (Sec)
ComparisonofprimarysystempressureinLOFTTestL3-s[
Figure 2.3.1-3.
to RELAP5YA calculation.
s t
22 I
1 h
t
FWV yWWWW gy WWyyyyygyww yy w
yy y yy yy yyyy yyyyy yy y,,,,,
vvvv vyvP w
v v
vvy T V gy l
3l l
s, a
i ug; WwIl c
ia.1 6.
g i
W E. ;,
RELMSYA
/
Q.
.l
- u--
l......................:..................................
as age ses tee tone taas tese ties asse asse esse film (Sec)
Figure 2.3.1-4.
Comparison of break flow in LOFT Test L3-6 to RELAPSYA calculation.
?
4
'l 23 i
J
^
o
e 0
0 2
2 00 i
0 2
no 0
i t
0 a
8 l
1 uc l
ac 00 A
Y i
6 5
1 PA L
ER 00 o
4 t
1 6
h 3L 0
t 0 )C s
2 e
N E
T 1
S T
(
FO E
L 0
s H
0 n
4 l
0 i
T 1
ssap 0
m 0
e
]
3
)
t 8
6 s
y 3
s 3
fo N'
11 0
n 6
o s
ir
+
ap m
0 o
A 0
C Y
4 5P A
5 L
Ea 1
$?
00 3
O
?
2 er ug i
.7 f
i O
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 9
8 7
6 5
4 3
2 1
3 2
1 I
1 1
t cm ~ mME a_$ S 3s m*
j j
,l
'I i
,-I i
1 l
)
]
depressurization.
The original L8-1 calculation used the specified
~
accumulator water temperature of 88.5'F.
To demonstrate the effectiveness of the ECC modeling guidelines to be discussed in Section 2.4, YAEC presented the results of an L8-1 calculation in which the accumulator water temperature was ra'ised to 200*F.
These results were presented in Reference 9 in YAEC's answer to question Q.V.9.
In this calculation the pressure followed the test data better because the higher water temporai.ure reduced the condensation rate (Figure 2.3.1-6).
The rod temperature response comparison showed that the code calculated CHF to occur earlier than measured in the experiment resulting in higher temperatures.
The PCT in the calculation was approximately 50*F higher than the measured PCT.
The final quench was calculated to occur about the same time as observed in the experiment, however.
l These results showed that RELAP5YA did an acceptable job of l
calculatirg the LOFT system response during Tests L3-6/LS-1. Where differences were noted, the impact on the overall calculation was not i
l significant.
1 2.3.2 LOFT Intermediate Break Asses 3m,ent Calculation LOFT experiment LS-1 was used to assess RELAP5YA because it repr4sented an experiment where core uncovery/ recovery was observed in a large scale test facility.
These results were presented in Reference 14 The RELAP5YA model of tho LOFT facility was based on a model obtained from i
l FG40 Idaho, Inc., that was used to make the pretest, prediction for the experiment.
Various model changes were made before making the RELAPSYA -
l assessment runs.
The most imoortant of these' changes was modeling the annular downcomer as two ANNULUS components connected at the top by a single junction.
Other changes included adding the heat slabs needed to
?
~
represent the hot rod and changing the model to more accurately refi' t the actual system initial and boundary conditions during the test.
Two RELA 75YA calculations were made by YAEC.
The first one used 1.0 for the subcooled and saturated discharge coefficients.
The second calculation copresented a sensitivity study where the subecoled discharge coefficient I
25 l
l
4 4
a g
~
j
- p....
.m 1
~
=
1,
,\\
c
= -
=aean
-a mg
]-
SR i
i j
1 G
~
ga s '
LET LS-1
,A.
MQ-W RSYA REVISf3 (200*F)
M e-mRSVR ORIGINAL (88.S*F) i M-g g
a I
d I
o u
5 a
s 5
g a
u 5
5 a
e a
e assa. mas.s aus.s aass.e amia.s acco.s ano.s asas.a aess.e asas.s aess.e asso.e aus.s asas.a acca.e ases.e TIME (SEC) i figure 2.3.1-6.
Comparison of primary systein pressure in LOFT Test L8-1 to original (88.S*F ECC) and revised (200*F ECC) RELAPSYA calculations.
-.. ~. +
nc -
._s e
was 0.85 and the saturated discharge coefficient was 0.9.
The base case calculation will be discussed briefly followed by a more detailed discussion of the sensitivity calculation.
For the base case calculation, the calculated ar.d msasured system pressures compare well during the first 100 s, but then the calculated pressure dropped more r.pidly between 100 and 200 s.
As a result, the calculated pressure reached the accumulator injection pressure at 154.5 s in the analysis, which is approximately 30 s earlier than in the test.
Comparison of the calculated and measured system mass inventories shows the calculated inventory dropped below that esasured at 50 s and remained lower than the measured inventory for the remainder of the analysis.
YAEC attributed the more rapid depressurization and the underprediction of the system mass to the overprediction of the break flow during the first 50 s.
YAEC noted e break flow was not well known for this experiment because the quoted uncertainty was +16 kg/s.
It is also not clear how well the experimental system mass inventory was known because YAEC took the data used in the system mass comparison from the test quick look report.
The i
mass inventory was not included on the experimental data tapes.
Because core uncovery was calculated early and total core quench was calculated i
late relative to the data, the calculated PCT was 899 X versus a measured 1
PCT of 715 K.
l YAEC performed a discharge coifficient sensitivity study in which the subcooled coefficient was set to 0.85 and the saturated coefficient was set to 0.90.
Using these coefficients resulted in a higher calculated system i
pressure and better agreement with the data (Figure 2.3.2-1).
The accumulator was now calculated to inject on at 170.4 s as compared to 185 s in the test.
Y With the discharge coefficients, the break flow (Figure 2.3.2-2) agreed better with the data, and the comparison of the calculated and measured system. mass inventories (Figure 2.3.2-3) improved although the calculated system mass was still less than the dat,a.
The difference in the timing of accumulator. injection can also be seen in Figure 2.3.2-3.
The
)
27 i
..~w,m~-. _. _ -
-.._...-,.e-
I o
a l m y
t iv i
)
t i
s c
0 n
.S es 0D A
1
- A Y
i g
S Y 5 N P.
S L S 2
t P
T A e
O t
S L E
E E R
C T R o
M t
0E 1
.S S
a 0
L i 0 I
ts 2I e
s~
T TF E
O L
0.R n
i 0U e
r 5T 1
u s
1 P
s e
r U
p m
0 R e
ts y
0 R s
, 0 y
I r
E m
1 T
irp F
f 0R o n _
o n
i o
t s
a 0
i E
l I
r 5
u._
a c M
ml p
a o
c I
C r
oT
-a 1
n 2
.o 3
e 2
e O~o~
ne o.o, O.n oo r
u i.
a g
i f
iW) mD1Wn o( 3
}
r (r
=
1 i
o i
O ~
l I
I I
i l
O l
O El v> o _
\\$
cs 1
52 j h l
TEST L5-1
"' o M RELAPSYA o-I'c:
.I"'
C3 1
e o
k o.
hTh 1
'.mit "_ -0M 1 os m
i I
q 15 150[.0 i
gp vyg t i g-y 1 u i 1
0.0 50.0 100.0 200.0 250.0 300.0 i
j.
TIPE RPTER RUPTURE
(
SECONDS
)
I j
Figure 2.3.2-2 Comparison of break flow in LOFT Test LS-1 to RELAPSYA sensitivity calculation.
j
i
.u.
n s
I n
n m e. -g.(
i j.
]
m.e.
DATA (L5-1 Qt.a) a 3
- -MB RELAP5YA j
l mee. -
h 8
me.. -
l'.
noe..-
y i
i 1
1 I
I I
I
(
0.0 50.0 100.0 150.0 200.0 250.0 300.0
~
TIME flFTER RUPTURE I
SECONDS
)
Figure 2.3.2-3 Comparison of primary system mass in LOFT Test LS-1 to RELAPSYA sensitivity calculation.
s,'
4
g increase in the calculated system mass inventory was much more gradual than in the experiment once accumulator injection began.
This was due to differences in the calculated and measured accumulator injection rates (see Figure 2.3.2-4).
As shown in the figure, the accumulator in the analysis injected at a lower rate and for a longer period of time than the accumulator in the experiment.
YAEC suggested the longer period of accumulator injection in the analysis was due a lower calculated system pressure.
However, this is not supported by the comparison in Figure 2.3.2-1 which shows the calculated pressure in good agreement with the measured pressure after 200 s (if not a little higher than the data).
Also, in L5-1, the accumulator injection rate was not directly measured because of instrument failure.
The accumulator mass flow rate shown in Figure 2.3.2-4 was estimated by Sandia.41 1
Because more mass was left in the system, complete core uncovery was delayed to 95 s as compared t) 75 s in the base case calculation.
In the test the liquid level did not drop into the core until 108 s.
As a result, at all levels in the core, early rod heatup was calculated.
This is illustrated in Figure 2.3.2-5 which compares the calculated and measured clad temperatures at the middle of the core (0.76 m elevation).
Rod heatup, although later than in the base case calculation, was still calculated early relative to the data.
Once accumulator injection began, the core started to quench.
The whole core quenched at 275 s, 60 s later than in the experiment.
As a result, the calculated PCT in the sensitivity calculation was 847 K (52.K lower than the base case) versus a measured PCT of 715 K.
\\
i i
Finally, a significant amount of steam superheat was measured at the i
core outlet in the experiment.
The code was able to calculate the superheat observed in the experiment as shown in Figure 2.3.2-6.
k YAEC used the new vertical flow regime map and interphase drag models, discussed in Section 2.2.1, and a split downcomer model in the LOFT L5-1 assessment.
YAEC, in a code review meeting in February 1987, prinsented results of an eirlier RELAPSYA calculation that used a single downcomer ano
~
31
~
9
e O
I I
I I
f l
l O-n 4
f i
O
}
MO l
w f_
xn 4
i cs i.s = =
y
~mmu maio i
O
- d-1 1
\\
x N
=
a I
._.) O l
L d u_
O O
j m
I
@o L
(C rO I
I I
I d
l 1
l-0.0 50.0 100.0 150.0 200.0 250.0 300.0
)
TIME.RFTER RUP"URE I SECONDS
}
)
figure 2.3.2-4 Comparison of accumulator mass flow rate in LOFT Test LS-1 to RELAP5YA sensitivity. calculation.
o I
I l
o m
6a gn x
."2 m
=
U O
Cn 5
ooa
_ N Z s.
oe c.]
OW" i
cn
~
o i
F 9
na e
a o <v :
W s
.a o*
]
e w
'e
~n'E 1
o 8 w ~9
=
-M5 e
r, c
82 o LL,
.:: 4 ts t
-o g gii u
6@
to i
rv
- 2 J
e.
,.4 9
o'g 3
D e
d i
"I l
=
0'000I 0'009 0'009 0'00F0'00E f
(
)
6.d W b
bJ e
G 33
d O
O g
i j
g g
o O
C3 g-v g.
"O TE-SP-001 g
._.uire 7mnram Mb j
m
.o_
[-*
O k
0- O k
TO_
[2_] O
. a_ 8 [
m n
L _1 8 -
[H tg
=
_ _ j 3l1_ _
1 y
O P
I i -
I I
I 1.
0.0 50.0 100.0 150.0 200.0 250.0 300.0
~
TIME FTER RUPTL'RE
{
SEC0hDS
)
~
Figure 2.3.2-6 Corparison of Core outlet temperatures in LOFT Test LS-1 to RELAPSYA sensitivity Calculation.
w -
a different vertical flow regime map and interphase drag models.
In the earlier calculation, the core did not quenco because ECC injected in the cold lag could not penetrate the downcomer.
When using the new models, the core quenched by 275 s.
While this is still later than in the experiment where the core was quenched by 213 s, it is a significant improvement over the earlier calculation.
Thus, use of the new models resulted in an improved simulation of the test.
The L5-1 assessment indicated a conservative PCT can be calculated even if the system depressuriza*. ion is werpredicted ar.d accumulator injection begins early.
The difference between the calculated and measured accumulator injection rates could 5e ona possible reason a conservative PCT was calculated.
To determine whetur this was the case, the tin.e the PCT occurred in the calculation was compared to the time accumulator injection began.
This comparison, as seen in Fi,1ures 2.3.2-4 and 2.3.2-5, shows that the calculated PCT occurred at the time accumulator injection began.
- Thus, the difference in accumulator injection rates did not influenc: the calculated PCT.
The conservative PCT is, therefore, due.to the underprediction of the core inventory as a result of overpredicting the break flow rate.
This resulted in early rod heatup in the calculation and, because the calculated heatup rate was comparable to that in the experiment, a higher ?CT was calculated.
Overall, the calculated results indicate an adequate simulation of the test data.
A conservative PCT was calculated with RELAP5YA.
The results of the sensitivity calculation showed an improvement in the calculated depressurization rate and the timing of accumulator injection.
The system thass comparison also improved.
2.3.3 Semiscale Assessment calculation l
?
A RELAPSYA assessment calculation using data from Semiscale Test S-LH-1 was presented in Reference 14 This test was chosen because it represented an integral test designed for PWR $8LOCAs where core i
uncovery/ recovery was encountered.
Two assessment calculations were completed for Test S-LH-1, a base case calculation using, single-enase and 35
~
two-phase discharge coefficients of 1.0 and a sensitivity calculation where the two-phase"discharge coefficient was changed to 0.8.
The base case calculation will be discussed briefly followed by the sensitivity calculation.
For the bose case calculw' M1, RELAP5YA did reasonably well simulating the U-tube level response in the intact loop, steam generator and on the upflow side of the broken loop steam gc eratur.
On the downflow side of the broken loop generator, the U-tubes were calcu'ated to drain earlier than observed in the experiment.
No explanation for this difference was given by YAEC.
Ou.ing the loop seal elearing period, a deeper core level depression was calculated and this resulted in a more extensive core heatup than was measured.
The early drain of the downflow side of the broken loop steam generator U-tubes in the calculation resulted in the broken loop pump suction clataing first followed by the intact loop pump suction.
This order of 'oep seal clearing is the reverse of that observed in the oxperis;ns.
But, as noted by YAEC, the loop seal clearing process was completed in both the test and i'1 the calculation by 300 s.
Once the loop C
seals cleared, the fluid in the core began to boil off, and the core began to uncover in the exparitant.
Less core uncovery was calculated because, due to a more rapid depressurization after 200 s, the accumulator setpoint was reached at 400 s as opposed to 500 s in the test.
With the early accumulator injection, the core did not uncover as much as in the test, and l
a core heatup was not calculated during the core holloff phase of the experiment.
YAEC tried to address the effect af the faster depressurization in the base case calculation by changing the two-phase i
discharge coefficient in the sensitivity calculation discussed below.
l However, overall, the break flow rate comparison showed the break flow in i
the base case compared well to the measured flov.
e i
YAEC completed a sensitivity calculation of Semiscale Test S-LH-1 l
usiAg the same model as before 9.4 changed the two-phase discharge coefficient to 0.8.
Results fe the sensitivity calculatien for pnenomena such as steam generator U-tube drain, loop seal clearing, core level l
36 l
a
depression (Figure 2.3.3-1), core heatups (Figure 2.3.3-2), and break flow (Figure 2.3.3-3) were similar to the base calculation.
With the 0.8 two-phase discharse coefficient, the calculated depressurization rate matched the data better after 200 s and the accumulator setpoint was reached at 470 s in the calculation as opposed to 500 s in the experiment (see Figure 2.3.3-4).
However, lowering the discharge coefficient lso left more mass in the system, as evidenced by the integrated break flow
(' Figure 2.3.3-5), and core (Figure 2.3.3-1) at the beginning of the coro boiloff period.
As a result, the calculation still did not show a core neatup during the boiloff period (see Figure 2.3.3-2).
YAEC concluded that in order to batter calculate the core thermal-hydraulic response during the boiloff phase, the energy lost out the break needed to be lowered in the calculation (to decrease the depressurization rate) without lowering the system mass loss (1n order to calculate the core uncovery during the coiloff phase).
YAEC did not present any suggestions on how tnis could be done.
YAEC used the new vertical flow regime map and interphase drag models, discussed in Section 2.2.1, and a split downcomer model (to represent the small annular part at the top of Semiscala's external downcomer) in the Semiscale S-LH-1 assessment.
YAiC, in Reference 11, presented results of an earlier RELAPSYA calculation that used a single downcomer and a different vertical tiew regime map and interphase drag models.
In the earlier calculation, long term cars cooiing could not be maintained because ECC injected in the cold leg could not penetrate the downcomer.
When using
(
the new models, the calculation showed the core did not uncover during the boiloff phase because of early accumulator injection (base case) or l
insufficient system mass depletion (sensitivity calculation).
ECC penetration of the downcomer, however, was not a problem as in the earlier
)
calculation.
Thus, use of t,he new models resulted in an improved l
simulation of downcomer phen aena.
[
l Because RELAP5YA was'not able to calculate the core heatuo during tne boiloff phase of Test S-LH-1. YAEC was ask'ed to provide its oosition witn resoect to RELAPSYA's ability to calculate Test $-LH-1.
The response given by YAEC is discussed below followed by INE' 's assessment.
37 l
o
=
r j
I i
l l
u
\\
5 g
5 B
e.
e g
5 3,
a a
54 o.
a*
-g g3
)'t NW o
tr%
l 3 "m
s\\.'
ox x
s x.
SC
=
g N.
x 3
,.t '.,)
O g.
-h x
./
U 3
H m
'.e C. L t.
-g C 8
/
8
,U
.r
/.
- ~
83
-g H
.3
[
-r) L - k-3 ST i
c.
2 m
g.
/
2 i.
m "8
<.....<7 c.
-5 A
J t
9 N
e*
v a
3 6
i i
6 0'001-0* DOE-0'00E-0'00F-0'005'- 0'009-f (WD) 13A31 OI-n0Il 13SS3A I
38
=
ik f
f i
t 5
llE C
o 5*
i,
- E.
u 5
~l-A AE o
E3 FM le i
[
- N.n u
Nr XN 3
C,
-8 s d
x'.
- cc
.v O
jag
-g x es N
.h?
- o. C li c
l
-8 u
z3 cr
-O*x i
aH 8:
_jil r.2
't k$
I
.g u.
i t
t I
?
e m
t, "O
N h
i U
~
e.
o.
8.
o w
i i-i i
0*006 0'000 0'004 0'009 0'009 0'00>
(N) 38DIW83dW31 0W10 l
l l.
Il- ~ - - -- __.,._'_._,_....____ _, _ _ _ _ ____.__,..., _ 3 9_ _, __,. _
Q I
t t
t I
t
_ 3 T
2 o
T
-8 s
p 5
}a o
Pi 8,.
s 4
e a
g 1- -8 n
s 9
c o u,
,4
_g-5 2.b m u c
V N
E
.h a
-5 0
C%
M 3
-g'u i
t s
m
,4b g
- o. c r
y
_g u 3_
_r W
.3d G-e i--
.2 '
=B
.g
~,. _
st c$
N C.
o-
/
-g U-
~e
.b.
vv o.
s 1
e e
c o,
4 4
m n S.
4 i
S4'I OS*I-SE*I 00*I S4*O OS*O
.SE*0 00'0 h
e (S/SM) 2198 M01J SSWW NW389 c
0 4
o o.
D i
e i
i i
i o
a 8
f 5
_g 55 a
o 5
'5<
m=
4y sg a
S Uw e.
r=
_g 5
n s
m
~
a a
.M a
/-
~
~
x
,/
- o. C 5
LJ
/
_8 ca ce 5
m oe tg
-8 H h
- ta_
a
./
o, C Ea
/
o ta e
_8 r 5
i g
oH e
seg QJ 83 o
i
% 'E
-8 Is.
v s.
F og 4
o 4
n A
8 4
t i
g l
S*4I 0*Si S*ET 0*01 S*4 0*S S*E 0*O
.e l
waW) 38nss38d i
e
.. y r.
O.
an 1
i a
e i
i i
i i
I h
.-e o.
Q 4
G an ~
~
j M"
O e
4' W ' tD _
1 W N i
(n
]
E o.
18-i Tm
.l Ed E
LD o.
sn -
O s.
N Ed TEST S-LM-1 g_.*
O.
- -4 RELAF5YA (I g -
x G
nd E-+ o j
z g_
1
~n 1
o J
o m.
"I I
I I
I I
I I
I I
i
-50.0 0.0 50.0 100.0 150.0 200.0 250.0 300.0 250.0 100.0 450.0 500.0 1
1 TIME AFTER BREAK (S)
~
Figure 2.3.3-5 Comparison of integrated break flow in Semiscale Test S-LH-1 to RELAP5YA l
sensitivity calculation.
]
n r.<
YAEC stated that, based on the good results obtained with RELAP5YA during the separate effects assessment and against the LOFT data, the physics of the code were correct.
The separate effects assessment included comparisons against FRI;3 and GE Level Swell data, Marviken critical flow data, a variety of critical heat flux data, and Bennett single tube data for forced convective boiling.
The LOFT tests used to assess RELAP5YA included LOFT L3-1. L3-6/L8-1, and L5-1.
LOFT L5-1 included a core uncovery/ recovery similar to that observed in Semiscale Test S-LH-1.
With respect to not calculating the heatup during the boiloff phase of Test S-LH-1. YAEC stated that they concluded the problem was with the input model representing the Semiscale facility or that possibly some experimental data was missing or not measured that was key to understanding the system response.
YAEC came to the conclusion regarding the input model because:
(1) a number of different codes, such as RELAP5YA, RELAP5/M002, or TRAC have been used to calculate Test S-LH-1, (2) each code has a different formulation of the basic conservation equations, (3) similar input models were used for each of the simulations, and (4) all the codes had prcP :ms calculating portions of the test.
This led YAEC to conclude the problem was in the RELAP5YA code input, not the code.
YAEC also noted that some preliminary. plant calculations completed with RELAP5YA show core uncovery and heatup due to core boiloff (see recconse to question Q.X.6, Reference 9).
YAEC stated this further reinforced its conclusion that M e j
code has modeled the physics cor" tly.
In general, the INEL agrees witn YAEC's conclusions.
The good comparisons between the calculated and measured results for the separate effects and LOFT integral assessments provide assurance that the code is
~
working properly and the physics are correct.
The INEL reviewed the RELAP5/M002 result's for S-LH-1 (see Reference 12) and the RELAPSYA
]
calculated results were very similar to those calculated with RELAPS/M002.
In the RELAP5/M002 calculation, the calculated depressurization rate was l
also greater than in the experiMt resulting in early accumulator injection and no core uncovery bei o calculated during the core boiloff onase of the experiment.
Thus, Setiscale Test S-LH-l'accears to be a very l
difficult test to simulate.' The ersuits presented by YAEC indicate i
43
RELAPSYA did better than RELAP5/M002 in simulating the experiment because RELAPSYA calculated a core heatup during the loop seal clearing period, as was observed in the experiment, while RELAP5/M002 did not.
The results of the two-phase break flow sensitivity calculation showed an improvement in the calculated system depressurization rate and timing of accumulatc*
injection, although the boiloff heatup was still not calculated because of a larger amount of mass left in the system.
Therefore, it is recommended that the RELAP5YA simulation of Semiscale Test S-LH-1 be treated as a single data point in the assessment matrix and that approval of RELAP5YA for use in licensing calculations not be withheld because the core boiloff heatup of Test S-LH-1 was not calculated.
2.3.4 THTF Assessment Calculations YAEC used data from TNTF to assess various models in the code.
In Section 2.2.5, data from THTF steady state film hoiling Tests 3.07.9B, K, and X was used to assess the CHF models in RELAP5YA.
This data, along with data from transient film boiling Test 3.08.6C, was also used to assess RELAP5YA post-CHF heat transfer.
The comparison of the RELAPSYA results to the steady state film bailias data showed that RELAP5YA overpredicted the rod temperature above the CHF point.
For Test 3.08.6C, RELAPSYA either calculated rod temperatures that were higher than the data or matched the test data.
l These results indicate RELAP5YA tends to provide conservative calculations of post-CHF heat transfer under conditions representative of those expected f
in a PWR.
l 2.3.5 Assessment Summary
- (
b The assessment calculations presented by YAEC indicate that ths various models and components in RELAPSYA worked together to provide a reasonable simulation of the LOFT small break experiments analyzed.
The
~
diffe-ences noted between the code results and the experimental data were either in the conservative direction or were determined not to have a significant effect en the' overa11' analysis.
For LOFT LS-1, the calcula se 44
~
i
results were adequate.
A conservative PCT was calculated in the RELAP5YA analysis.
The results presented for Semicale Test S-LH-1 showed that while many of the phenomena observed,in the experiment were calculated reasonably well, the core heatup due to the core boiloff was not calculated.
These results were very similar to those obtained at the INEL with RELAP5/M002.
This would indicate that Test S-LH ~1 is very difficult to simulate and, thus, it is recomended this test be treated as a single data point in the assessment matrix.
The results of the TNTF assessment calculations for post-CHF heat transfer, as well as the LOFT integral calculations, indicate the code was able to calculate the PCTs or was conservative.
Therefore, based on the overall assessment results, it is recommended RELAP5YA be accepted for use in integral systems 58LOCA licensing analyses including the calculation of peak clad temperatures.
For LOFT LS-1.no :imiscale S-LH-1, tne results of the base case eticulations shoved the system depressurized toc quickly and, as a result, accumulator injection begr.n early.
In addition, the system pressure was t.nderpredicted in the LOFT L3-6 assessment.
The RELAP5YA assessment calculaticns used the RELAP5YA critical flow model to calculate two-phase critical flow.
Use of the Appendix X required Moody model to calculate two-phase critical flow in Itcensing calculations will have tha same effect. The calculated system pressure will decrease too rapidly for a given break size and accumulator injection will begin early.
When asked to discuss the effect of the more rapid depressurization on plant licensing calculations YAEC noted, in Reference 14, that sensitivity calculations for L5-l'and S-LH-1 that varied the subcooled and two-phase discharge i
coefficients matched the measured pressure better.
YAEC stated that the effect of discharge coefficients on the calculated results is the same as varying the' break area.
Thus, the effect of underpredicting the system l
pressure would be to shift the' limiting break towards a smaller break l
size.
YAEC, in its response to question Q.!X 1 (Reference 9), stated it will perfors a break size study.
This study is sufficient to account for l
any uncertainties in the break flow model and its effect on system l
depressurization rates and accumulator injection titues.
However, the breat i
size study should be reviewed to ensure it is detailed enougn to account for these effects.
l 45 l
2.4 Phenomena Important to PWR SBLOCA In this section, the code's ability to simulate the important phenomena during a PWR SBLOCA will be discussed.
This includes its ability to represent a non:ondensible gas; condensation heat transfer; and single-phase, two-phase, and reflux natural circulation.
2.4.1 Noncondensible Gas The RELAPSYA hydrodynamic model has the capability to model the presence of a norcondensible gas along with the vapor and liquid phases of water.
The nonco.idensible model in RELAP5YA is the same one used in RELAPS/M001 and discussed in Reference 2.
This model was reviewed and found to be adequate to model the presence of a noncondensible gas.
However, YAEC did not assess the noncondusible gas model in the code for the raasons discussed below.
YAEC stated, in its response to Q.I.3 in Reference 7, that it would not include noncondensibles in its calculations because of the expected negligible effect on SBLOCAs.
The information given in response to Q.I.3 was supplemented by additional analyses and discussion in Reference 14 YAEC discussed the potential sources of noncondensible gases at Maine Yankee and Yankee Rowe as well as the effect of noncondensibles on the system response.
The noncondensible sources considered by YAEC included:
(1) hydroger, dissolved in the primary coolant, (2) air dissolved in the ref0eling water, (3) hydrogen contained in the pressurizer vaper space, (4) nitrogen dissolved in the accumulator water, (5) hydrogen released from j.
zirconium-water reaction, and (6) fission and fill gas in fuel rods, i
Sources excluded by VAEC include nitrogen cover gau in the accumulators, hydrogen in the makeup tank, and hydrogen released from radiolytic
'h decomposition of injected water.
Hydro. gen in the makeup tank.was excluded because this water is not injected into the primary during a SBLOCA at either Maine Yankee or Yankee Rowe.
YAEC divided SBLOCAs into three categories: those where the primary depressurizes to the accumulator 4&
..-L.
.N
g----------------------------------------
m setpoint, those that establish a quasi-steady pressure plateau, and those that may repressurize.
The. applicable sources of noncondensibles and their potential effects on the three SBLOCA categories are discussed below.
Foe the breaks that depressurize to the accumulator setpoint (approximately 0.1 to 0.5 ft2 at Ma N Yankee and larger than 0.09 ft2 at Yankee Rowe), YAEC showed that condensation heat transfer would only occur early in the transient while the secondary was still a heat sink for the primary.
During this period the only significant sources of noncondensible would be :ources 1, 2, 3, and 5 listed above.
YAEC presented an analysis on the effect of these sources of noncondensible on the primary system response its answer to question Q.I.3 and Reference 14.
The analysis included the hydrogen released by the zirconium-water reaction assuming 1% core wide oxidation.
When the primary to secondary pressure difference is greater than 200 psi, YAEC showed the haac transfer rate with all the noncondensibles in the steam generators would differ from the heat transfer rate without noncondensibles by 33 to 35% for Maine Yankee and
' Yankee Rowe. As the system depressurized and the primary pressure approached the secondary pressure, the effect would be greater.
- However, because the primary continues to depressurize to the accumulator setpoint and eventually the secondary becomes a heat sink for these break sizes, the I
time period where this would effect heat transfer from the primary to secondary would be short.
4 l
Later in the transient for these larger break sizes, the core could uncover and superheated steam could flow to the steam generators and the i
l secondary would alternate between being a heat sink or a heat source to the l
primary.
YAEC, in Reference 14, listed the potential heat transfer modes l
on the primary side of the steam generator z.s laminar natural convection, turbulent natural convection, or forced convection.
In the same reference, e
YAEC show d that even if the steam generators contained 100% noncondend ble heat transfer degradation would range from 10 to 30% for these modes of heat transfer.
YSEC stated this was within the uncertainty associatid witn the correlations used for these heat transfer modes.
47
)
YAEC also assessed the effect of noncondensibles on the system depressurization for these larger break sizes in Reference 14.
At the time of accumulator injection, not including noncondensibles in the analysis could result in the time of accumulator injection being off by approximately 5 s.
Noncondensibles considered included sources 1 - 3, 5, and 6.
At the end of the transient, the pressure could be about 21 psia higher if noncondensibles were modeled.
Because the core is quenched and all the decay heat is being removed at the break, this difference is not important.
All six sources listed above were considered in this analysis.
For medium small breaks (from approximately 0.02 to 0.1 ft2 at Maine Yankee and 0.001 to 0.09 ft2 at Yankee Rowe), noncondensibles could have l
a more significant effect because the system pressure is above the accumulator setpoint for a long time.
Therefore, for these break sizes the steam generators must be maintained as a heat sink until the decay heat I
decreases to the point where all the energy can be removed through the break.
YAEC's response to Q.I.3 and additional information in Reference 14 were presented to assess the effects of noncondensibles on this size break.
In its assessment, YAEC included the amount of noncondensible gas added to the primary system from sources 1, 2, 3 and 5.
The analysis
[
included the hydrogen released by the zirconium-water reaction assuming 1%
core wide oxidation.
It concluded that even if all the gas were to migrate j
to the steam generator tubes and stay there, the gas would occupy about f
4.3% of the U-tube volume at stable conditions.
Thus, natural circulation would be unaffected by the noncondensible. With respect to heat transfer
{
f for medium size breaks. YAEC concluded the effects of noncondensible gases i
l are (1) reduction in the available primary to secondary temperature l
difference at a given primary pressure and (2) reduction in condensation l
heat transfer due to the resistance of the noncondensible gas boundary
{
'l layer.
YAEC showed the primary to secondary temperature difference would i
be reduced by at most 10*F if noncondensibles were modeled. Thersfere, neglect of noncondensibles would not significantly change the 58LOCA calculations.
The potential impact of noncondensible gas 'n c ndensation o
heat trans'for was greater but still small.
As noted above,' when the l'
orimary to secondary pressure difference is greater than 200 osi, the heat transfer rate with all the noncondensibles in the steam generators woule I
48 1
differ from the heat transfer rate without noncondensibles by 33 to 35Y,for Maine Yankee and Yankee Rowe.
YAEC determined the effect of noicondensible becomes more pronounced as the primary pressure approaches the secondary preseure, however, the not result would be to stabilize at a slightly higher primary pressure (50 to'73 psi) than for the case without noncondensible.
For breaks small enough to result in a repressurization of the primary system, YAEC stated that the cenclusions drawn for the medium size breaks in response to Q.I.3 and in Reference 14 were also valid for this class of breaks.
The INEL reviewed YAEC's work and agrees with its conclusion that the effect of noncondensible gas from sources 1 to 6 will be small on the class of breaks where they apply.
Therefore, not including noncondensibles from these six sources in RELAP5YA analyses is considered acceptable.
- YAEC, however, did not include two sources of noncondensible in its analyses or in its discussion.
These were nitrogen cover gas in the accumulators and q
radiolytic decomposition of injected water.
Only if YAEC provides i
additional justification for neglecting noncondensible gas from these sources is it recowmended that RELAP5YA be used for cases where the accumulators empty and nitrogen cover gas enters the system or when the hydrogen concentration falls below 5 cc H /kg of water and radiolytic 2
decomposition of injected water becomes a source of noncondensible.
The additional justification would include discussing the effect of the nitrogen cover gas and hydrogen from the radiolytic decomposition of g
l injected water on the system pressure, core heat transfer, core level, i
bror.k flow, and ECC injection and showing that neglecting these sources of noncondensible did not have a significant effect on the calculated results.
j q
e
?
Alternately, the code may be used in the situations discussed above if i
YAEC accounts for the noncondensible gas from these sources in its analyses.
As noted at the beginning of this section, RELAP5YA includes a f
noncondensible model but the modal was not assessed by YAEC.
Therefore, if l
YAEC chooses to account for noncondensible gases from these sources in its l
. analyses, then the noncondensible endel in RELAP5YA,sheuld be assessee anc l
the results providui for review by the NRC, f
l 49
1, 2.4.2 Conde.nsation Heat Transfer One of the areas of concern in rhodeling SBLOCAs that was identified in NUREG-0737, Item !!.X.3.30, was the ability to accurately calculate i
condensation heat tra'nsfer rates.
The need to confirm this feature of the small break model against applicable experimental data was recognized.
The condensation / vaporization models in RELAP5YA are identical to the models used in R'ELAP5/M001.
These models are discussed in Section 2.1.3.1 of Reference 2.
Both the condensation and vaporization models arv empirical and are based on the work of Jones and Saha.
I The empirical constants in the vaporization model were determined mainly from depressurization experiments.
The large depressurization rates in these experiments resulted in relatively high vaporization rates that are reflected in the empirical constants.
Because the empiri 41 constants were based on depressurization data, it raised a question concerning the applicability of this model to processes dominated by wall heat transfer.
YAEC addressed this issue during its code assessment work.
As noted in YAEC's response to question Q1.13 in Reference 11 because there is a lack l
of detailed data to assess the interphase mass transfer models, assessment of the vaporization model was completed by reviewing its effect on the l
analysis of integral system tests (LOFT Tests L36/L8-1, LS-1 and Semiscale Test S-LH-1) as part of the overall PW SBLOCA analysis method.
In most I
cases it was concluded by YAEC, and the INEL agrees, that the model worked l
reasonably well when considered as part of the overall PWR SBLOCA analysis j
l method and the overall method should provide conservative results.
l i
e l
50
The high vaporization rates also affect RELAP5YA's ability to calculate vapor superheat.
YAEC noted in its response to questions Q.VII.25 and 26 (Reference 9) that, due to the high mass transfer coefficient, the code is not able to calculate vapor superheat in a control volume unless the volume void fraction is 1.0, For this to happen, the nodalization must be detailed enough to show a two-phase level and the heat transfer above the mixture level must be sufficient to vaporize all entrained droplets.
Thus, in calculations where it appears that vapor superheat cotld play a significant role in the overall system response or nave a significant offect on the peak clad temperature YAEC must show that the nodalization used was de ailed enough to account for superheating.
This would include any SBLOCA analysis where the two-phase mixture level dropped below the top of the core.
The core will generally uncover during the loop seal clearing process due the balance of hydrostatic heads around the loops (unless the upper head and upper plenu~a to downcomer bypass flow is sufficient to relieve the pressure buildup in the upper plenum), but this is a relatively rapid uncovery/ recovery of the core.
The core uncovery due to the long term bolloff of the core fluid'is of the m'est concern.
In this type of core uncovery, YAEC must show that the core nodalization is detailed enough to account for vapor superheat.
The empirical constants for the condensacion model were also determined mainly by comparison to depressurization experiments.
From its code assessment work, YAEC concluded that the condensation model tends to overpredict the condensation rate, especially when subcooled ECC is injected into a steam environment.
This results in lower system pressures j
than would otherwise be expected.
When questioned on how the overprediction of the condensation rate
~
would affect the calculated response during a SBLOCA, YAEC stated (question Q.I.9, Reference 9) that the conditions of pressurizer refill and condensation of vapor in the system high points are not,important because these occur beyond the scope of the analysis.
They occur after the core is completely recovered so'that there is no longer a question of plant safety.
This is a acceptable resolution of the question but it is recomended that use of RELAP5YA be limited to the period before condensation of vacor in system hign points and pressurizer refill begins.
51 e
Also, in response to Q.I.9, YAEC stated the conuensation model in
)
RELAP5YA may not be conservative during ECC injection.
However, YAEC developed ECC modeling guidelines which it feels will result in conservative ECC injection flow rates (questions Q.II.3 and 5 - 8 in Reference 9).
The approach taken at YAEC was to inject the ECC at the actual injection location and to increase the EC temperature from its normal value of approximately 100'F to 200'F.
Use of ECC with a temperature of 200*F was found to mitigate the ar,tificial, rapid decrease in the system pressure that was calculated when the ECC was injected with the actual temperature of 100*F.
YAEC chose a temperature of 200*F because it is closer to the saturation temperature corresponding to % e containment pressure yet low enough to avoid the injection of superheated liquid.
This approach was assessed against the LOFT L8-1 data as discussed in Section 2.3.1.
similar techniques' are used to mitigate unrealistic pressure decreases calculated with other codes and applications.
Therefore, this approach is adequate to limit ECC induced condensation.
The effect of nodalization on condensation calculate'd during ECC injection was discussed in YAEC's response to question Q.II.5 (Reference 9).
YAEC presented the results of a nodalization sensitivity calculation for the accumulator injection in LOFT Test L3-1.
The study was performed by repeating the LOFT calculation with the nodes upstream and downstream of the injection point divided in half relative to the base calculation.
The study used an ECC water temeerature of 200*F based on YAEC's recovenendation discussed above.
he results of the sensitivity j
calculation were essentially the same as the base calculation indicating i
i that, when the ECC water is assumed to be 200'F the nodalization used i
in the base calculation was sufficient to adequately represent the condensation associated with ECC injection.
However, if YAEC uses an ECC 3
f temperature colder than 200*F in a licensing analysis, the effect of c
cold leg nodalization on condensation must be reassessed and the results presented for review.
Question'Q !V.9 of Reference 3 asked YAEC to discuss how the condensation that' could occur on the primary side of the steam generator U-tubes was assessed.
In its response (Reference 9), YAEC noted the 52
E'.
condensation heat transfer in vertical flow is based on 1!wo comonly used correlations documented in J. G. Collier's book, Convective Boilino and Condensation.
For laminar flow the Nussel theory of film condensation is used and for turbulent flow the Carpenter and Colburn correlation is used.
The assessment of the condensation heat transfer model was limited to the two LOFT SBLOCAs discussed in Section 2.3.1, LOFT L3-1 and L3-6.
YAEC noted,in its response to question Q.!V.9 that the calculation of the system response in these tests depended in part on the calculation of the condensation heat transfer in the U-tutes.
As discussed in Section 2.3.1, RELAP5YA did a reasonable job of simulating the system response in these tests.
This indicates the code simulation of the condensation heat transfer in the steam generator U-tubes was adequate.
Based on the information provided by YAEC on its modeling techniques and on the model performance, the condensation model in RELAP5YA is adequate for modeling Maine Yankee and Yankee Rowe SBLOCAs.
~
2.4.3 N,atural Circulation NUREG-0737, Item II.K.3.30, also identified the need to experimentally and analytically verify the various modes of single-phase and two-chase natural circulation predicted to occur in each vendor's reactor during small break LOCAs.
The Semiscale Natural circulation test series provided a data base ta verify the proposed small break model's ability to calculate natural cir:ulation.
YAEC assessed RELAD5YA against data from Semiscale Test S-NC-242 because this test examined the various modes of natural i
circulation.
Three power levels were examined in Test S-NC-2, 30, 60, and i
i.
100 kW, corresponding to decay heat levels of 1.5, 3, and SiL, respectively.
At each power level, the system started at 100% primary mass
)
inventory with the pressurizer maintaining subcoolei conditions.
The pressuri:er was then valved out of the system and mass was drained frem the primary system in discrete amounts.
Data was recorded when steady state conditions were establishe'd at each mass inventory.
YAEC used RELAPSYA to simulate two of tho' three power levels examinec in the test, the 30 and 100 kW core sower cases, at.various primary system inventories.
YAEC provided these assessment results in its resoonse to 53 e
'.l 4
questien Q.V!.9 in Reference 9.
With a core power of 30 kW, RELAP5YA did an adequate job of calculating the primary system hot leg temperature, the' i
primary syst,em pressure, and the natural circulation flow rate as a function of primary system mass inventory (Figure 2.4.3-1).
Although the l
peak flow rate was overpredicted by about 25%, the rest of tne calculated
]
flows agreed very well with the measured data.
The arrows in l
Figures 2.3.4-1 and 2.3.4-2 indicate the magnitude of the oscillations in the natural circulation flow rate calculated at certain mass inventories.
l i
These oscillations are discussed in more detail below.
In the 10C kW core f
l power case, the calculated hot leg temperature and system pressure compare well with the experimental data but the calculated natural circulation flow l
i rates do not.
The flow rate comparisons provided by YAEC for the 100 kW case (response to Q.VI.9, Reference 9, and Q1.15. Reference 11) showed the j
measured flows for mass inventories from 90 to 60% were overpredicted by 25 i
to 100%.
l A review of the Semiscale Test S-NC-2 100 kW data used to assess l
RELAP5YA and discussions with Semiscale personnel involved with the Natural
- (
Circulation test series indicated that the 100 kW data used by YAEC to
'l 4
1 i
assess RELAP5YA was not the best data available.
The 100 kW data from Test 4
I S-NC-2, for system mass inventories above 72.5%, was superceded by 100 kW data taken during Test 5-NC-10.*3 The data from Test S-NC-2 was still i
j the only data available for system mass inventories below 72.5%.
i l
A comparison of the RELAP5YA results, the S-NC-2 data, and the S-NC-l')
data is shown in Figure 2.4.3-2.. The S-NC-10 and S-NC-2 data show the sam, i
f j
trends but the S-NC-10 peak flow rate is shifted to a mass inventory i
i i
approximately 5% lower than chat observed in S-NC-2.
This difference was
)
attributed to system leaks in Test S-NC-2 that were not accounted foe in l
the system mass inventory (Reference 43).
Figure 2.4.3-2 shows that the
)
RELAPsYA calculated flow rates compare much better to those measured in i
Test S-NC-10 than the T'est S-NC-2 data.
The RELAP5YA flow rates generally
{
l lie within, or just outside, the uncertainty bands of the data for system mass inventories greater than 75%.
However, the calculated flows below 75*.
l
)
I
.l 54 k
~
1
[
e 0.9 i
4 a
i a
DATA A
RELAPSYA UNCERTAINTY FLOW =
- 0.033 KG/S
~
0.6 INVENTORY =
- SK i
7 c.
N en O
r g
kl C
0.3 C
Q d
-l m
}
Q I
s
~
O y
9 i
-0.3 50 60 70 80 90 10 0 inventory (7.)
t Fi gu r e 2.4.3-1.
Comparison of natural circulation flow rates from f
Semiscale Test S-NC-2 to RELAP5YA co culation -
30 kW cose.
e 9'
l 5
i t
l ss
I 1.2 -
i i
i i
i c
SNC2 DATA x
SNC10 DATA A
RELAP5YA CALCULATIONS 0.9 XX X Q
g 3
x E
N I
.0.G a
g_OO d ]l B
e
.2u.
0.3 a
na UNCERTAINTY W
O O
Ft.0w =
- 0.033 KG/S a
INVENTORY =
- SM FOR > 72%
O
= + 5%. -05 FOR s: 723 i
-0.3 50 60 70 80 90 10 0 i-Inventory (7.)
FI gur e 2.4.3-2.
Comparison of noturoI eircuiatlon flow rates Semiscale Tests S-NC-2 ond S-NC-10 to RELAP5YA calculation - 100 kW case.
e 56
system mass inventory were outside the data uncertainty bounds for the S-NC-2 data.
The reason for this was adequately discussed in YAEC's response to Q1.15 Reference 11.
t lt should also be noted that the 100 kW case represents decay heat powers of approximately 5%, a relatively high level of oecay heat.
The 30 kW case, representing 1.5% decay heat, is more typical of conditions to be found during a long ters' natural circulation cool down.
The natural circulation flow rates for the 30 kW case were more accurately calculated by RELAP5YA as noted above.
At some of the inventories examined in the RELAP5YA natural circulation study, the flow did not achieve a steady state but oscillated l
until more mass was drained from the system.
In its response to Q1.15 in Reference 11, YAEC stated it reviewed the model used in the original natural circulation assessments and found and corrected several errors in the model.
YAEC reran the 100 kW case and presented the results in its answer to Q1.15.
The new results showed that oscillations were now calculated at only one inventory, 60%.
The other results, the natural circulation flow rates, system pressures, etc., were not significantly different from the previous calculation.
Based on YAEC's response to Q1.15,. Reference 11, discussions with YAEC at a meeting in Bethesda, MO on February 5, 1987, and a review of some of the literature on flow modes in inverted U-tube systems.44 the oscillations calculated by RELAP5YA. while not observed in Test S-NC-2, are considered plausible.
i The following factors were also considered in concluding the i
calculated oscillations were acceptable.
First, the oscillations occurred j
at an inventory, 60%, where the core was still covered and wel' cooled.
Semiscale small break tests have shown the core remains covered and cooled j
~
with system inventories as low as 35% (Reference 42).
Thus, the oscillations, if they occur in a small break licensing analysis, will not have an effect on the calculated PCT.
Also, similar oscilfations were calculated in a RELAP5/M001.5 study'of the Semiscale natural circulation j
test series.45 They were calculated for the 100 kW case at approximately the.same inventory, 61.3%.
In this recort it was stated that the 1
57 e
'O e
+
calculated flow transitions from two-phase nr.tural circulation to reflux natural circulation at mass inventories ranging from 60 to 70%.
Because the RELAP5YA oscillations were calculated for an inventory at the lower end of this range, they could be due to an instability in the single tube system as it makes the transition from two-phase to reflux natural circulation.
Finally, during a SBLOCA analysis the system mass inventory is constantly changing.
At a result, the amount of time the plant model would spend at an inventory where the oscillations would occur is limited.
Because RELAP5YA was able to accurately :alculate the natural circulation flow rates for the more typical condi'. ions in the Semiscale system and did and an adequate job on the other case, the models in RELAP5YA are considered adequate to simulate natural circulation during Maine Yankee and Yankee Rowe SBLOCAs.
2.5 Calculation of S-VT-8 phenomena Question Q.X.6 of Reference 3 requested YAEC to validate the ability 0
of RELAP5YA to calculate phenomena similar to that observed during Semiscale Test S-UT-8.46 This included liquid holdup in the upside of the steam generator U-tubes and consequent core level depression.
To demonstrate RELAPSYA's ability to calculate this phenomena, YAEC presented the results of a Maine Yankee SBLOCA calculation (Reference 9).
The calculated results clearly show liquid holdup and a core level depression thereoy demonstrating the code's ability to calculate the type of phonemena observed in the experiment.
Because the code's ability to calculate S-UT-8 i
type phenomena was demonstrated, this calculation is sufficient to meet the l
requirements of auestion Q.X.6.
The Semiscale Test S-LH-1 assessment presented in Reference 14 also demonstrated RELAP5YA has the capability to calculate liquid holdup phenomena accurately, p
m 53 m
- 3. COMPLIANCE WITH NRC REQUIREMENTS Appendix K to 10CFR50 specifies the required and acceptable features of any model to be used for licensing analyses..
YAEC discussed each Appendix K PWR-related requirement individually in the RELAP5YA manual, Volume 1. Section 6.
Only YAEC's responses having a bearing on this licensing assessment or areas of potential concern are addressed in this section.
Any Appendix K requirements not addressed in this section were found to have an acceptable response or were not applicable to this PWR licensing review.
All requirements related to simulating the metal-water reaction were m.t.
YAEC incorporated the taker-Just model into RELAP5YA.
YAEC also stated that renodalization will be performed, if needed, to prevent the ruptured node from being more than 3 in. in length.
The fuel rod behavior requirements were met by using the fuel rod j
behavior models from the licensing aporoved T000EE2-EM code and the cladding rupture and flow blockage tables from NUREG-0630.47 I
l The Moody Critical Flow model was incorporated into RELAP5YA and will l
be used in all licensing calculations when the break flow is two-phase and
]
has a void fraction greater than 0.05 (see Section 2.2.2).
Appendix K requires the break size and discharge coefficients be varied to determine i
the limiting break size, i.e. the break size yielding the highest PCT.
For PWR $8LOCAs, YAEC will perform the break
_e study and RELAP5YA has the
\\
I capability to vary the discharge coefficient.
i To meet the CHF requirements of Appendix K. YAEC added a new CHF algorithm to RELAP5YA.' The algorithm uses the modified liasi correlation
]
at high. mass fluxes, tne Griffith-Zuber correlation at low mass fluxes, and j
an interpolation between the two at intermediate mass fluxes.
The.CHF l
algorithm was assessed against a wide variety of data and was found to j
provide reasonable results.
!l-59 i
i j,.
l The post-CHF heat transfer requirements of Appendix K specify return j
~
to nucle.;e boiling be locked out once CHF has occurred during blowdown and a return to transition boiling be locked out if clad superheat exceeds 300*F during blowdewn.
YAEC added the appropriate logic to RELAP5YA to meet these requirements but does not plan to use them for PWR $8LOCA i
i analyses based on an agreement reached with the NRC (see Section 2.2.8).
The centrifugal pump model in RELAP5YA is the same model used in i
RELAP5/M001.
TiieRELAP5/M001modelwasastraightforwardconversionofthe RELAP4 centrifugal pump model.
The model was completely reorogransned but l
no changes were made to the physical model.
The RELAP4 pump model is used by YAEC (Reference 36) to perform licensing calculations and was approved l
I by the NRC staff.
The pumo homologous curves used in the Maine Yankee and i
1 Yankee Rowe plant models were discussed in YAEC's response to question Q.V.13 in Reference 7.
The information provided showed the homologous curves used are based on acceptable data sources.
I Appendix K requires the effect of the compressed gas in the accumulator on the reflood rate be considered in licensing analyses.
This 3
requirement is part of the post-blewdown requirements for a large break l
LOCA and, therefore, is not a concern for $8LOCAs.
In Section 2.4.1, it i
i was recomended that use of RE! AP5YA be limited to the period before the accumulator empties and the nitrogen cover gas enters the system.
This
}
reconenendation was made, not because of this Appendix K requirement, but j
because of concerns over the affect of the nitrogen on SSLOCA condensation.
I i
The Appendix K requirement relating to the thermal-hydraulic i
i 4
l interaction between the steam and the (CC water is met in two ways, RELAPSYA is.a two-fluid thermal-hydraulic code with the. capability to treat l
i
)
~
the steam /ECC water interaction included in the basic fluid models.
In
, addition. YAEC developed ECC modeling guidelines, discussed in l
{
Section 2.4.2 to orevent excessive condensation when ECC is injected into f
a steam filled pipe.
These guidelines were reviewed and found adecuate.
l
{
Refill and reflood heat transfer requiremen'ts were met by inclusion of l
reflood heat transfer models in RELAP5YA as discussed in Section 2.2.6.
l i
l 60 j
t.
- 4. RELAP5/M001 CODE UPDATES RELAP5YA was developed from the publicly released RELAP5/M001, cycle 18, code.
During the development of RELAP5YA, RELAP5/M001 was updated to correct errors or to add model improvements.
This resulted in the creation of cycles 18 to 29, where cycle 29 was the final released version of the code.
Approval of RELAP5YA included a review of the status of these updates.
YAEC was asked (Q1.20 of Reference 4) to state for each update whether it was (1) incorporated as written by RELAP5 code developers, (2) incorporated but in modified form, or (3) not included in RELAP5YA.
In response (Reference 11), YAEC showed the updates were generally included in RELAP5YA as written by the code developers.
Those updates which were not included and have an impact on licensing calculations are discussed below.
The specific modeling areas not recommended for acceptance for use in RELAPSYA licensing calculations because the update developed by the RELAP5/M001 code developers was not included in RELAP5YA are:
1.
Control variables as power input to a heat structure.
2.
Reactor kinetics when no power is generated from gamma decay heating.
3.
VALVE component with form loss coefficients, t
i l
4 Control variables with reactor kinetics feedback.
With the exception of these scocific correction updates, which YAEC l
chose not to include in RELAP5YA and, therefore, wi11' restrict certain code applications RELAP5YA contains the error corrections from RELAP5/M001, cycle 19 to 29, which ensures that known errors were corrected.
9 61 O
g e *
- 5. SYSTEM MODELING TECHNIQUES f
A number of the questions to which YAEC responded in References 5 to 9 dealt with the nodalization to be used in its plant analyses.
YAEC's j
responses to these questions are discussed.
Also, the nodalizations l
proposed for use in the Maine Yankee and Yankee Rowe licensing calculations and those used in the integral system assessment calcul4tions were reviewed l
to determine if they were consistent.
I i
The steam generator nodalizations proposed for YAEC's plant models was l
discussed in its response to question Q.!V.2 (Reference 7).
YAEC provided l
plant nodalization diagrams for Maine Yankee and Yankee Rowe as
{
Figures IV.2-1 and IV.2-2, respectively, in its response to this question.
These node diagrams are shown in Figures 5-1 and 5-2.
YAEC stated the eight nodes would be used in the U-tubes in the Maine Yankee model and six f
nodes in the Yankee Rowe model.
When asked why the Maine Yankee and Yankee f
Rowe steam generator nodalizations were different. YAEC stated in its
,[
response to question Q2.3 in Reference 10 that it was YAEC's understanding
(
that the steam generator nodalizations need not be identical for the two
[
plants.
However, YAEC will perform nc.dalization sensitivity studies to I
1 determine the effect of steam generator nodalization on the PCT.
i Convergence of the PCT would determine the plant specific nodalizations for f
the two plants. The results of the sensitivity studies would be submitted with the plant specific submittals.
In the same response. YAEC noted that uncertainties in the initial conditions for the steam generator secondary l
i inventory would be applied in the direction of a conservative PCT.
j i
i Baced on YAEC's response to question Q.!V.3 (Reference 7) it was l
d inferred the steam generator nodalizations determined by the sensitivity s
studies discussed above would also be used to calculate low flow and reflux cooling conditions in SBLOCAs.
It is not known how these nodalizations will compare to that used to assess RELAP5YA against Semiscale natural l
circulation Test S-NC-2 (discussed in Section 2.4.3).,
The nodalization l
YAEC used in its.5-NC-2 assessment had 16 volumes to represent the primary i
side of the U-tubes.
The study in Reference 45, bewever, used eight f
volumes to represent the U-tubes and showed RELAP5/M001.5 could p
62
_ = -.
.--7 rp e
I..
l l
C I
i
. =.
~[,hl [_.
, ei i
~~
< n.
il I se I
nWs t
' x:..- A uu **
~
m M w
~
~
~
g e
=
r l
. '.[.5.1.I l
- TWu 1
=.
a af e
J f7, ~1 lY I
i r
I I
l i
Figure 5-1.
Preposed Maine Yankee noda11 ration for SKOCA EM calculations.
i I
j a r ar +
l w---n-,
.n,,---n- - -..
.,--,-----,n----,
---n-
- - -,,, - - - - -. - - <,. ~ - - -, -. - -,, - ~ ~ ~ - -,. - - - - ~ + - -, -, ~.,,, -, -. - -, - - - - -
v --
i
!=1
- ==*====
1
- *"' N
- esame,
{
s.s ec, 34 em se i.
s:9.e
(,, 3
[ (h g
a T
l I._ "_ '_
r
?
s m
L 1
sa 1
=D r
.s 1
I 5
Q
>-....p.
l
...g...
Q,,
T ll 11 I
I i
n l
l 1
- M.
L as j i
pam.. p.m nas amas sesrs
~
-.s s.s.,s s
s u m
- es s. s.e - a s e s e.
s e u.w..s,.
.. s I
{
l Figure 5-2.
Proposed Yankee Rowe nodalization for SBLOCA EM calculations.
I I
)
g-g 8
adequately calculate.he Semiscale natural circulation response.
Therefore, eight nodes,should be sufficient to calculate low flow and reflux cooling conditions.
YAEC discusssd the propcsed pressurizer nodalization in its response to questions Q.II.12, Reference 6, and Q2.2, Reference 10.
YAEC stated the pressurizer would be modeled by two regions with each region having three to four nodes.
Thus, six to eight nodes would be used to represent the pressurizer in plant analyses.
However, the final nodalization used will te presented with the SBLOCA submittal for each plant.
Plant nodalization diagrams show that the core in the Maine Yankee plant will be modeled with five nodes and the core in Yankee Rowe will be modeled with three nodes (see Figures 5-1 and 5-2).
While the RELAP5YA model used in the LOFT SBLOCA assessment calculations had three nodes in the core, a submodel with six core nodes was used to calculate the core thermal-hydraulics when the core uncovered in Test L8-1.
In addition, six nodes wire used in the LOFT LS-1 assessment and twelve nodes were used in the Semiscale Test S-LH-1 assessment.
Therefore, while three nodes may be adequate in cases where there is no core uncovery, at least six nodes should be used in analyses with core uncovery and more nodes may be needed if vapor superheat is important to the analyris.
4 In question Q.IX.5 of Reference 3 YAEC was asked to discuss how the treatment of ECC injection in its plant models would affect the calculated break flow.
In its response in Reference 7, YAEC noted that for the cold i
leg break accidents analyzed, the ECC would inject into the cold legs very i
close to the actual injection location.
Also, the break would be modeled in the injection node.
This nodalization would maximize the ECC going out the break and at the same time limit the deprer :trization rate because of
[
the presence of two-phase fluid at the break.
is modeling is adequate.
Based on Reference 14, YAEC will use a sp.it c?wncomer to represent the annular downcomer in plant analyses.
Use of this nadalization in LOFT LS-1 and Semiscale S-LH-1 assessm,ents, as discussed earlier, gave adeouate i
results for phenomena where downcomer modeliag would be important such as ECC penetration into the downcomer.
65
~
- 6. CONCLUSIONS The RELAPSYA submittal by YAEC was reviewed to determine the code's acceptability for usa in PWR SBLOCA licensing analyses.
Based on this review, it is recomended the code be approved for use in licensing analyses with the following coments and restrictions:
1.
It is recomended that RELAP5YA be approved for use in integral systems licensing analyses including the calculation of peak clad temperature.
The largest break size YAEC used to assess RELAP5YA was equivalent to a 0.7 ft2 br dak in a PWR.
RELAP5YA may be used to analyze breaks larger than this if the important phenomena for the larger break sizes are similar to that assessed for the 0.7 ft2 and smaller breaks.
2.
YAEC will not model noncondensible gases in its system analyces.
YAEC demonstrated that the effect on the overall system response of aoncondensible from the sources analyzed will be small on the class of SBLOCAs where they apply.
Because of the assumptions in YAEC's analyses, it is recc mended the code be restricted to analyzing SBLOCAs where the accumulators do not empty'and the hydrogen concentration remains above 5 cc H /kg of water so 2
that radiolytic decomposition of water is not a source of noncondensible gas unless additional justification is given (see Sect':.,2.4.1).
i 3.
RELAPSYA overpredicts the condensation reto during a pressurizer insurge.
YAEC stated condensation of vapor in system high points i
and the pressurizer refill portion of a SBLOCA are beyond the point of interest because the core in covered and there is no longer a question nf plant safety.
1herefore, RELAPSYA cannot be used to analyze SBLOCAs during periods of condensation of vapor in system high points and pressurizer insurge unless further justification is given.
9
~
d 4.
Due to the large interphase mass cransfer coefficients used in RELkPSYA, 'no code is unable to calculate the presence of superhested vspor unless the volume void fraction is 1.0.
Therefore, for any analysis where it appears the presence of vapor superheat would have a significant effect on the system response or the peak clad ter..perature, YAEC should justify that the model nodalization was detailed enough to accoun't for vapor superheat.
This would include any SBLOCA analysis where the two-phase mixture level dropped below the top of the core due to long term core boiloff.
5, RELAPSYA also overpredicts the condensation rate when subcooled ECC is injected.
To mitigate the effects of injecting subcooled ECC, YAEC developed ECC modeling guidelines.
These guidelines specify that the ECC water be modeled with a temperature of 200'F instead of the nominal plant tamperature of approximately 100*F.
This guideline is acceptabl.e and any deviation from it must be justified.
In addition, the ar.alysis used to justify that the cold leg nodalization at the injection point is sufficient to represent condensation induced by the ECC used 200*F water.
If colder ECC is used in a licensing analysis, the effect of cold leg nojalization on condensation must be reassessed and the results provided for NRC review.
l l
6.
The nine modifications reviewed in Section 2.2 that apoly to pWR l
SBLOCA analyses are recorrsnended for acceptance for use in i
l licensing analyses with the following coments or restrictions:
a.
The Moody critical flow model is recomended for acceptance.
as implemented.
However, plani: licensing analyses should use the model options recomended by YAEC, i.e. the donor cell static pressure and enthalpy.as input to the Mcody critical flow table (see Sec. tion 2.2.2).
l l
l 67 1
1
~
b.
The accumulator model is recomended for acceptance as implemented.
However, it is also recomended that use of RELAP5YA be limi:ed to the period before the accumulator emptias and nitrogen cover gas enters the primary system.
See Item 1 of this section.
c.
The multiple surface radiation model is recommended for l
acceptance for licensing applications provided justification of the view factor matrix used is given and, if a code is i
used to generate the view factor matrix, the code receives licensing approval (see Section 2.2.7),
d.
The fuel behavior models are recomended for acceptance as implemented provided the rod internal pressure temperature offset is greater than 2*F (see Section 2.2.9).
7.
There are a large number of options available to the user in
~
RELAP5YA.
Therefore, to ensure that the code is used within its
~
capability and with croper input, it is recomended that each user suomit, with future licensing submittals, information i
justifying all selected options and input data, including defaJ1ts.
8.
YAEC noud the followir.g items will be performed on a plant specific basis and documented in the plant submittal:
time steo i
sensitivity studies, break size sensitivity studies, a list of
,f metal heat slabs used at various nodes ir.cluding justification of any metal structures not represented in the analysis, clarification of how the multiple surface radiation model is used in EM analyses, justification that clad geometry changes during a SBLOCA do not change the radiation model view factors, 4
justification of any nonzero value for the direct moderator heating fraction, and justification of the pressurizer nodalitation used in SBLOCA licensing analyses.
i 68 L -
9.
YAEC will also provide justification of the steam g'enerator nodalization'in SBLOCA licensing analyses on a plant specific basis with the plant submittal.
The steam generator nodalization will be determined based on sensitivity studies to show convergence of the PCT.
In order to calculate low flow and reflux cooling conditions any nodalization proposed by YAEC should have at least eight nodes to represer.t. the U-tubes (see Section 5).
10.
The analyses that assessed RELAPSYA's ability to calculate the core thermal-hydraulic response during core uncovery and recovery had nodalizations that used at least six volumes in the core.
Therefore, it is recomended that YAEC use at least six volumec in the core for its plant licensing analyses.
More detail may be needed ii calculating vapor superheat is important in the analysis.
The complete nodalization used in a plant licensing analysis should be reviewed to ensure it is consistant with i
current modeling guidel.ines or the nodalizations used in the assessment calculations.
Any differences should be justified.
11.
Certain models, listed in Section 4, are not recomended for acceptance for use in licsnsing applications.
These models l
include known errors and YAEC chose not to incorporate available update corrections.
l 12.
In several of the integral assessment calculations, the system
['
pressure was underpredicted and, as a result, accumulator injection began earlier than in the experiment.
The system deprescurization rate was calculated better when a discharge j
coefficient less than 1.0 was used.
The effect of discharge coefficients on the calculated results is the same as varying the break area and YAEC will perform a break size study.
The break size study should be reviewed to ensure it is detailed enough to account for break flow uncertainties arid their effect on the calculated depressurization rate and accumulator injection time.
I l
69 l
I
13.
YAEC's RELAP5YA submittal indicated the code was to be applied to
~
Yankee,Rowe, a two-loop Westinghouse plant, and Maine Yankee, a three-loop Combustion Engineering plant.
The review found notning in the code that was plant specific in nature or that would preclude the ap;:lication of RELAPSYA to either plant.
Therefore, RELAP5YA can be applied to both Maine Yankee and Yankte Rowe.
l i
9 Is
-l e
h e
6 70
0 0
- 7. REFERENCES 1.
R. T. Fernandez, et al., RELAp5YA:
A Computer pecoram For Licht Water Reactor System Thermal-Hydraulic Analysis, YAEC-1300P, Volumes 1,2,3, October 1982.
2.
V. H. Ransom, et al., RELAP5/M001' Code Manual. Volumes 1-and 2, NUREG/CR-1826, March 1982.
3.
Letter J. R. Miller, NRC, to J. A. Xay and J. 8. Randazza YAEC, "Review of YAEC II.K.S.30 58LOCA Model," May 11, 1984.
4.
Letter P. D. Wheatley, INEL', to C. Graves and S. Sun, NRC, "Questions Resulting from the Review of RELAP5YA," POW-8-86, September 4, 1986.
5.
Letter J. A. Kay, YAEC, to J. A. Zwolinski, NRC, "Respons tc NRC Questions on RELAP5YA," March 1, 1985.
6.
Letter J. A. Xay, YAEC, to J. A. Zwolinski, NRC, "Response to NRC
[
Questions on RELAP5YA," April 30, 1985.
7.
Letter G. Papanic. YAEC, to J. A. Zwolinski, NRC, "Response to NRC.
Questions or RELAP5YA," July 1, 1985.
8.
Letter G. Papanic, YAEC, to J. A. Zwolinski, NRC, "Response to NRC Questions on RELAP5YA," August 15, 1985.
i b
9.
Letter G. Papanic, YAEC, to J. A. Zwolinski, NRC, "Response to NRC Questions on RELAp5YA," November 1, 1985.
10.
Litter R. W. Capstick, YAEC, to V. L. Rooney, NRC, "Response to Additional NRC Questions on the RELAP5YA Computer Code,"
October 16, 1986.
11.
Letter R. W. Capstick, YAEC, to V. L. Rooney, NRC, "Response to Additional NRC Questions on the RELAP5YA Camputer Code,"
November 4, 1986.
1 j
12.
G. G. Loomis and J. E. Streit, Results of Semiscale Hod-2C Small
.i, Break (50 Loss-of-Coolant Accident Experiments S-t.H-\\ ana S-l.H-2, i
~
NUREG/CR-6438, EG4-2424, Novener 1985.
13.
Letter G. Papanic, Jr., Y4EC to E. M. McKenna, NRC, "RELAP5YA,"
'(
l December 5, 1986.
k 14.
YAEC transmittal of additional work on interphase drag, assessment of RELAP5YA against LOFT Test L5-1 and Semiscale Test S-LH-1, and noncondensible gas, to be published.
15.
Y. A. Taitel, O. Bornea, and A. E. Dukler, "Modelling Flow Pattern I
Transitions for Steady Upward Gas-Liquid Flow in Vertical Tubes,"
AtchE Journal..Vol. 26, No. 3, May 1980, pp. 345-354 i
t
\\
71 l
?
16.
X. Mishima and M. Ishii, "Flow Regime Transition Criteria Consistent with Two-Fluid Model for Vertical Two-Phase Flow," NUREG/CR-3338.
./.
C. 8. Wallis, One-0imensional Two-Phase Flow, New York:
McGraw-Hill, 1972.
18.
O. Nyland, et al., H_yfrodynamic and Heat Transfer Measurements on a Full-Scale Simulated 36 Rod Marviken Element with Uniform Heat Flux Distribution, FRIGG-2, R4-447/RTL-10007, 1968 (Sweden).
19.
O. Nyland, et al., Hydrodynamic and Heat Transfer Measurements on a Full-Scale Simulated 36-Rdd BHWR Fuel Element with Nonuniform Axial and Radial Flux Distribution, Report FRIGG-4, R4-502/RL-1253, 1969 (Sweden).
20.
B. C. Slifer and J. E. Hench, Loss-of-Coolant Accident and Emeraency Core Coolina Models for General Electric Boiling Water Reactors, ATDO-10329, April 1971.
21.
F. J. Moody, "Maximum Flow Rate of a Single Component Two-Phase Mixture," Journal of Heat Tr-sfer, Trans, of ASME, 87, 1, s
February 1965.
22.
L. Ericson, et al., The Marviken Full-Scale Critical Flew Tests Interim Report: Results from Test 10. MX 3-63. Marviken Power Station, November 1978 (Sweden).
23.
P. D. Bayless, J. B. Marlow, and R. H. Averill, Exoeriment Data Reoort for LOFT Nuclear Small Break Experiment L3-1, NUREG/CR-1145, January 1980.
24 J. C. Chen, "A Correlation for Boiling Heat Transfer to Saturated Fluids in Convective Flow," Process Oesign Develooment, Vol. S, pp. 322-327, 1966.
25.
V. E. Schrock and L. M. Grossman, Forced Convection Boiling Studies, Final Reoort on Forced Convection Vaeorization Project, TID-14632, University of California at Berkeley. 1959.
26.
J. R. S. Thom, et al., "Boiling in Subcooled Water During Flow Up Heated Tubes or Annuli," Proceedinos of Institute of, Mechanical.
i gaineers,3C180,1966.
27.
A. W. Bennett, et al., Heat Transfer to Steam-Water Mixtures Flowing a
in Uniformly Heated Tubes in Whien tne Cri,tical Heat Flux Has Been e
Exceeded, AERE-R5375, Atomic Energy Researen Estaolishment, 1967 (Great Britain).
28.
R. E. Phillips, R. W. Shumway, K. H. Chu, "Improvements to the Prediction of Boiling Transition During Boiling Water Reactor Transients," 20th ASME/AIChE National Heat Transfer Conference, Milwaukee, Wisconsin, August' 1981.
29 P. Griffith, et al., "Critical Heat Flux Ouring a Loss-of-Coolant Accident;" Nuclear Safety, 18, May-June 1977.
72
~_
,r ;,
~
30.
Critical Heat Flux Correlation for CE Fuel Assemblies with Standard Grids. Parts 1.2. Non-Uniform Axial power Distributions, Combustion Engineering Topical. Report CENPD-207, June 1976.
31.
Electric Power Research Institute Report, EPRI-RP-813-1, to be published.
32.
E. Janssen, Two-Phaso Flow and Heat Transfer in Multirod Geometries.
Final Report, Genera' Electric Company Report GEAP-13347, March 1971.
33.
G. L. Yoder, et al., Dispersed Flow Film Boiling in Rod Bundle Geometry - Steady-State Heat Transfer Data and Correlation Comparisons, ORNL/5822, to be published.
34.
C R. Hyman, et al., ORNL Small Break LOCA Heat Transfer Test Series II:
High Pressure Bundle Boil-off and Reflood Test Analysis, Fina)
Report for THTF doil-Off and Reflood Tests 3.09.100-X-ORAFT, Oak Ridge National Laboratory, September 1981.
35.
G. N. Lauben, T000EE2-EM. A Two-Dimensional Time-DependenU Fuel Element Thermal Analysis Program, NUREG-75/057, May 1975.
36.
Yankee Atomic Electric Company, WREM-Based PWR ECCS Evaluation Model (Version YAEC-058), YAEC-1160, July 1978.
37.
Kevin St. John, Methods for the Arialysis of Oxide Fuel Element Thermal Analysis Program (FROSSTEY). Code /Model, YAEC-1249P, April 1981.
38.
R. N. Cehlberg, W. V. Johnston, and J. A.
Dearien,
"FRAP Fuel Behavior Computer Codes," Nuclear Safety, Vol. 19, No. 5, September-October 1978.
39.
P. O. Bayless and J. M. Carpenter, Experiment Data Repott for LOFT Nuclear Small Break Experiment L3-6 and Severe Core Transient Experiment L8-1, NUREG/CR-1868, January 1981.
40.
D. B. Jai roll and J. M. Divine, Experimental Data Report for LOFT k
Intermediate Break Experiment L5-1 and severe Core Transien*
i Experiment L8-2, NUREG/CR-2398, EGG-2136, November 1981.
i; 1
41.
J. L. Orman and L. N. Kmetyk, RELAP5 Assessment:
LOFT Intermediate Briaks L5-1 and L8-2, NUREG/CR-3406, SAN 083-1575, August 1983.
42.
G. G. Loomis and X. Soda, Results of the M00-2.8 Natural Circulation
.d Experiments, NUREG/CR-2335, EGG-2200, Septemeer 1982..
43.
P. North, INEL, ltr, to R. E. Tiller, 00E-ID, PN-130-81, "Transmittal of Selected Results from Semiscale Mod-2A Test S-NC-10," '
October 9, 1981.
j 44 C. Calia and P. Griffith, Modes of Circulation in an Inverted U-tuce Array with Condensation, NUREG/CR-1699, Octeoer 1980.
e 73 w
n e
~--
45.
R. Dimenna, RELAPS Analysis of Sem+ scale M002A Single-loco Single-Component Steady-State Natural Circulation Tests, EGG-SEMI-6315, June 1983.
46.
M. T. Leonard, J. L. Perryman, and G. W. Johnsen, "The Influence of Liquid Holdup in Steam Generator U-tubes on Small Break Severity,"
ASME reprint 83-WA-NE-2.
47.
D. A. Powers and R. O. Meyers, Claddino Swellino and Ruoture Modes for MCA Analysis, NUREG-0630, April 1980.
b i
i e
b i
O e
74 e
o e
.n,
,---r
- ---