ML20114A977
| ML20114A977 | |
| Person / Time | |
|---|---|
| Site: | Mcguire, Catawba, McGuire  |
| Issue date: | 12/31/1991 |
| From: | Epperson K, Koontz D DUKE POWER CO. |
| To: | |
| Shared Package | |
| ML19303F029 | List: |
| References | |
| DPC-NE-2004-A, NUDOCS 9208240203 | |
| Download: ML20114A977 (199) | |
Text
-
. -~..
.. - ~....
r e
i 1
7
-*PWm4.4%Wh-%
i t
d e u-su-y Y
J
.[
4 rA,,,
%g i.
a
[-
I..
l w
i mm,'--
9208240203 920817 ADOCK 05000369 J' DR -
._ _ -. 7 DPC-NE-2004 A
-DUKE POWER COMPANY MCGUIRE AND' CATAWBA NUCLEAR STATIONS CORE THERMAL-HYDRAULIC METHODOLOGY USING VIPRE-01 l
Original Report: December 1988 i
Approved Report: December 1991 D. R Koontz K.' R. Epperson.
i i-
_ Duke Power Company
._-Nuclear Generadon Deparunent Nuclear Engineering Section Charlotte, North Carolina
i ABSTRACT
.This report presents Duke _ Power Company's methodology for performing core thermal-hydraulic analyses for the McGuire and Catawba Nuclear Stations tising the VIPRE-01 computer code.
The VIPRE-01 model that will be used for McGuire/ Catawba thermal-hydraulic analyses is discussed.along with the sensitivity study results used to select many of the input values.
This report also explains the methodology used to determine allowable operating limits in terms of power level, reactor coolant temperature and pressurt, and power i
distribution, McGuire/ Catawba thermal-hydraulic analyses will be based on a i'
.DNBR. limit that statistically accounts for the effects on DNB of the uncertainties of.important parameters such as power, temperature, and flow.
The statistical core design (SCD) methodology is explained in this report.
l i
I l
I
-ii i
l Table ~of Contents
.Section-Page
- 1. 0.E
'l#IE00UCIl0B-1
- 2. 0 CEEE_EESCRIPTION 1
3.0 STATION DESCRIEll0S 2
4.0 CORE MODELING 3
4.1.
MARK-BW VIPRF-01 INPUT 4.1.1 AXIAL N0 DING 5
4.1.2-ACTIVE FUEL LENGTH 5
4.1.3 CENTROID DISTANCE 6
4.1.4 EFFECTIVE CROSSFLOW GAPS 6
4.1.5 SPACER GRID FORM LOSS COEFFICIENTS 7
4.1.6 INLET FLOW DISTRIBUTION 7
4.1.7 VIPRE-01 CORRELATIONS 8
4.1.7.1 Friction Pressure Loss 8
4.1.7.2 Turbulent Mixing 10 4.1.7.3
'Two-Phase Flow' Correlations 12 4.1.7.4 BWCMV Critical Heat Flux Correlation 13
- 5. 0 ~
MCGUJRE/CATAVBA CORF THERH&L-HYORAULIC ANALYSES 14
, 5.1
- BACKGROUND 14 5.2-MCGUIRE/ CATAWBA THERMAL-HYDRAULIC DESIGN BASES 15 5.2.1-DNB DESIGN BA$15 16
' 5. 3 -
CORE PROTECTION 16
- 5. 4 -
CORE THERMAL LIMITS 17 5.5 CORE ONB LIMITS 18 5.5.1-REFERENCE POWER DISTRIBUTION 19 iii
,,.,. _..,... ~.. - _ _ _..
Table of Contents (Continued)
Page
-Section 21 5.5.2 RCS FLOW 21 5.5.3 PRESSURE 22
- 5. 5. 4 CORE INLET TEMPERATURE 5.6 MAXIMUM ALLOWABLE PEAKING LIMITS 22 25 6.0 SIAll5TICAL CORE DESIGN 26 6.1 SCD METHODOLOGY 26 6.2 SELECTION OF PARAMETERS 28 6.3 RESPONSE SURFACE MODEL 30 6.3.1 LEAST-SQUARES CURVE FIT 6.4 STATISTICALLY TREATED UNCERTAINTIES 32 6.5 PROPAGATION OF UNCERTAINTIES 38 40 7.0 M
40
8.0 REFERENCES
APPENDIX A Safety Evaluation Report A-1 APPENDIX B Responses to Request for Additional B-1 Information - September 14, 1990 APPENDIX C Responses to Request for Additional C-1 Information - November 29, 1990 APPENDIX D Responses to Request for Additional D-1 Information - August 29, 1991 APPENDIX E liandouts Presented in Oct. 7 & 8, 1991 E-1 Meeting iv
l l
List of Tables Table Page t,
1 Typical Mark-BW Fuel Assembly Data 43 2
Conditions for Comparison of Mark-BW Core Models 44 3
Mark-BW Model Comparisons 45 4
Axial Node length Sensitivity Study 46 5
Inlet Flow Sensitivity 47 6
Turbulent Momentum Factor (FTM) Sensitivity Study 48 7
Sensitivity to Void Models 49 8
SCD Statepoint Parameters 50 9
RSM Input and VI'RE-01 and RSM Results 51 e
10 VIPRE-01 Mar',-BW RSM 53 11 VIPRE-0F RSM DNBR Comparisons 54 12 Statistically-Treated Uncertainties 55 13 Sample of Random Operating Conditions 56 14 Monte Carlo Uncertainty Propagation Results 57 V
I l
List of Figures i
Fiqure Page 1
75 Channel Model - Subchannel Geometry and Power 58 Distribution 2
75 Channel Model - Assembly Geometry and Power 59 Distribution 3
12 Channel Model - Subchannel Geometry and Power 60 Distribution 4
12 Channel Model - Assembly Geometry ar.: Power 61 Distribution 5
8 Channel Model - Subchannel Geometry and Power 62 Distribution 6
8 Channel Model - Assembly Geometry and Power 63 Distribution 7
VIPRE-01 vs. LYNX 2 DNBR 64 8
VIPRE-01 vs. LYNX 2 Mass Flux at CHF 65 9
VIPRE-01 vs. LYNX 2 Quality at CHF 66 10 Typical Core DNB Limits 67 11 Typical Mark-BW OTaT MAP Limits 68 12 Basis For Mark-BW OTAT MAP Limits 69 13 MAP Limits vs. Predicted Peaking 70 14 Typical Mark-BW LOF MAP Limits 71 15 Mark-BW RSM vs. VIPRE-01 BWCMV DNBR 72 vi l
.. ~. _ _. _ - -
1.0l ItlIRODUCTION L An analysis of the thermal-hydraulic performance of a reactor core must be performed to' define core. thermal margin and allowable operating limits-.
The Lthermal-hydraulic analysis must accurately predict the local flow and enthalpy and departure' f rom nucleate boiling ratio (DNBR) in the high power channels of.
the core.
-This report presents the procedure used to apply the VIPRE-01 code, references 1 and 2, for thermal-hydraulic analyses of the McGuire and Catawba L:/C) reactor cores, The geometric representation of the core is illustrated and discussed along with the analytical models and experimental correlations used to. determine-friction pressure losses, coolant mixing, and subcooled voids.
This report'also explains the methodology used to determine the thermal-p hydraulic limits that are used:to define the regions of safe operation for the core in terms of.. power level, reactor coolant temperature and pressure, and power distribution.
l:
'McGuire/ Catawba thermal-hydraulic analyses will be based on a ONB limit that
! statisticilly accounts for the ef fects on DN8 of the uncertainties of important-L parameters such as power, pressure, and flow.
The statistical core design l.-
(SCD) meth'odology is also discussed in this report.
(
.0 CODE' DESCRIPTION' 2
VIPRE-01 is an open channel thermal-hydraulic code featuring diversion crossflow and turbulent mixing.
The VIPRE-01 code was designed to perform t
1 l
-steady-state.and transient _ thermal-hydraulic analyses of nuclear reactor cores
-for normal operating conditions and accident conditions.
The VIPRL-01 code has been reviewed by the NRC and was found to be acceptable for referencing in license applications with the requirement that "each or,ganization using VIPRE-01 for licensing calculations should submit separate documentation describing how they intend to use VIPRE-01 and providing justification for
'their specific modeling assumptions," ref. 2.
This report explains how the
.VIPRE-01 code.will be used for McGuire and Catawba thermal-hydraulic analyses-
~
and what code options and correlations will be used.
VIPRE-01 accepts input data which defines the geometry, hydraulic and thermal characteristics of the fuel, correlations, and method of solution.
The core is
. represented by parameters that define and describe the number of channels in the core and their individual characteristics, such as flow area, wetted and heated perimeter, and centroid distances between adjacent channels, Hydraulics of the core are defined by single and two phase flow correlations, spacer grid form loss. coefficients,-friction pressure losses, and the inlet flow distri-
~bution.
Thermal modeling of.the reactor core is described'by the core power, core radial'and' axial power distribution, and heat transfer correlations.
3.0f STATION BESfRIPTION The McGuire and Catawba Nuclear-Stations each have two Westinghouse ( W } units rated'at 3411 MWt.
Each reactor core consists of 193 fuel assemblies, each-sincluding-264 fuel rods, 24 control-rod thimble tubes, and an instrument-thimble tube arranged in a 17 x 17 array.
Three non-mixing vane spacer grids and five mixing vane grids provide lateral stiffness and position the fuel N
rods.
The M/C core models discussed in this report model Babcock & Wil !
(B&W) Mark-BW fuel that has been designed to be mechanicall cox y and hydraulically compatible with W standard and optimized 17x17 fuel, ref3.
Typical Mark-BW fuel assembly dimensions are given in Table 1 The transition cores
st will contain both B&W and W fuel will be analyzed using the methodolog y presented in this report except for appropriate modeling of the dif f erent fuel assembly designs.
Transition core results will be discussed in the Reload R eport for the first cycle of application of this methodology.
- 4. 0
_COLM k
The geometric representation of the M/C core is typical of a crossflow code model of a pressurized water reactor core.
A section of the reactor core is modeled as an array of adjoining channels realizing it i s important to include the effects of the surrounding core on the hot subcha nnel.
The hot subchannel and those adjacent to it are modeled individually with larger and larger channels modeled toward the periphery of t.he core, As discussed in references 1 and 2, VIPRE-01 has the cap bili a
ty to model the hot assembly in detail while simultaneously modeling th only slightly less detail.
e rest of the core with Selection of a final single pass model requires development of a number of different size models and at different conditions.
comparisons of the models Three Mark-BW models were developed.
1.
75 Channel Model 2.
12 rhannel Model 3.
8 Channel Model 3
l
~
-All three models were-developed assuming.1/8 core symmetry with the hot assembly located in'the center of'the-core.
The 75 channel model individually models all of the subchannels in the hot assembly (1/8 of a complete assembly)
(
and the 30. fuel-assemblies making up the rest of the 1/8 of the core.
The 12 channel model' individually models the hot subchannel and three rows of I
subchannels around the hot subchannel with the rest of the hot assembly lumped into one channel and the remaining 30 assemblies lumped into one large channel.
-The 12 channel model is simplified to form an 8 channel model that individually 5
models the hot subchannel and two rows of subchannels around th' hot l.
subchannel.
The 75, 12, and 8 channel models are illustrated in Figures 1-6.
i I
L To determine the-modeling detail required to accurately predict the hot channel coolant conditions and MONBR, the 75, 12, and 8 channel models were run using the-conditions given in Table 2.
The range of conditions were randomly.
selected-from the matrix of conditions analyzed as a part of the statistical core design analysis discussed in Section 6.0.
The MONBR-and mass flux and quality at the point of MDNBR are compared in Table 3 for the 75, 12, and 8 channel models.
The 8 channel model MONBRs are all within 1.0% of the 75 channel MDNBRs (a code /model uncertainty is included in the ca sculation of the ~ statistical DNBR limit as discussed in Section 6.4).
-The results given in_. Table 3 demonstrate that the 8 channel model can be used to~ accurately predict the MONBR for a wide range of operating. conditions.
The
.8 ; channel model will be used for all steady-state Mark-BW thermal-hydraulic analyses including the statistical core design analysis.
4 1
... ~...- - --
l L.1
' MARK-BW VIPRE-01 INPUT 4
.The Mark-BW fuel assembly data ~ used to calculate the geometric input (flow
- areas, wetted and heated perimeters, centroid distances, and gap widths).for the three different models is given in Table 1.
Other important VIPRE-01 input is discussed in detail in the subsections which follow.
4.1.1 AXIAL N0 DING Volume 4 of the VIPRE-01-manual, ref. 1, states as a. general rule that nodes on
~ he order of 2 or 3-inches long are recommended in the region where the MDNBR t
is likely to occur.
Results of_an axial node length sensitivity study per-ormed with the 8 channel model are given in Table 4.
A comparison was made for two different core conditions between the base case with approximately 3 inch
-axial nodes and a case with 2'irich axial nodes near the location of MDNBR (the remaining length is modeled with 3 inch nodes),
As shown in Table 4, the 3 and 2 inch node lengths yielded essentially identical MDNBRs; therefore, the 3 inch node length will_be used for all M/C core thermal-hydraulic analyses.
4.1.2 ACTIVE FUEL LENGTH 0
Uranium fuel both densifies and_ swells when irradiated, but densification t
effects are predominant'at-low burnup whereas swelling effects are predominant s t higher'burnup.
Fuel densification decreases the active fuel-length while-a l fuel swelling tends to-increase.the active length.
In the reactor the fuel o
5
i also thermally expands.
For B&W's low densification fuel, the fuel thermally expands more than it shrinks due to densification.
Thus, it is conservative to use the cold nominal active fuel length for thermal-hydraulic analyses.
4.1.3 CENTROID DISTANCE The centroid distance is the dimension that defines the lateral pressure gradient in the crossflow momentum equqtion.
By convention in VIPRE-01, the length is taken as the distance between the centroids of adjacent channels, which is the subchannel pitch for subchannels in a normal square array.
The centroid distance for a channel cut by a line of symmetry is assumed to be the same as the centroid distance for the complete channel.
For the lumped sub-channels the centroid distance is increased from the normal subchannel value in proportion to the number of rod rows between channel centroids.
4.1. ",
EFFECTIVE CROSSFLOW GAPS The product of the gap width and the axial node length defines the lateral flow area between channels.
The gap widths required to calculate the crossflow between it:dividually modeled subchannels are easily calculated given the rod pitch and fuel rod and thimble tube diameters.
The gap width between adjacent assemblies or it.mped channels is the sum of the subchannel gaps through which the assemblies or lumped thannels communicate.
6
4.1.5 SPACER GRID FORM LOSS COEFFICIENTS Form loss coefficients are used to account for the hydraulic loss caused by the variation in flow area and turbulence at a spacer grid.
The Mark-BW assembly includes five Zircaloy intermediate mixing vane grids, one Zircaloy intermediate non-mixing vane grid, and Inconel non-mixing vane lower and upper j
l end grics, reference 3.
Form loss coefficients specifically determined for the different types of subchannels (unit channel, thimble tube channel, peripheral
'~
channel, corner channel) are used to represent the hydraulic resistance of the different subchannels. Spacer grid form loss coefficients were developed by B&W based on full size Mark-BW fuel assembly flow tests.
The individual subchannel loss coef ficients were determined analytically based on the overall grio loss coefficients.
4.1. 6 INLET FLOW DISTRIBUTION c
To investigate the effects of local inlet flow maldistribution two cases were run using the Mark-BW 75 channel model.
1.
Uniform inlet flow distribution.
2.
Five percent less flow into the hot assembly (the remaining flow was distributed between the outermost assemblies).
Both flow distributions were used as an input boundary condition with the same core power distribution (Fig. 1 and 2).
The hot channel results for both cases are given in Table 5 for two different core conditions.
When the hot assembly inlet flow is reduced by 5 percent the MDNBR decreases by 1.1L 7
a'
'A..
McGuire/ Catawba thermal
- hydraulic analyses will be based on a 5% reduction in Lhe hot assembly, inlet flow, ref; 4.
The RCS flow anomaly observed at Catawba that results in a more severe core inlet flow'maldistribution will be evaluated prior to submittal of a Catawba reload report using -the methodology discussed in this report. -Margin will be included in the ONBR limit used for the generic M/C thermal-hydraulic analyses to more than account for any DNBR penalty resulting from the fl'ow anomaly.
r
~4.1.7 VIPRE-01 COPRELATIONS Empirical correlations are required to solve the continuity, energy, and momentum equations that' form the basic structure of the VIPRE-01 code.
The correlations selected for use in M/C thermal-hydraulic analyses are discussed in subsections 4;1.7.1 - 4.1.7.4 which follow.
4.1.7.1 Friction Pressure Loss Pressure losses due to frictional drag are calculated for flow in both the axial and lateral directions.
In the axial direction the friction pressure loss is calculated,by dP =~f G2v' di 2g 0 ch i
- where f = friction factor determined from an empirical correlation defined by user input G = mass flux, Ibm /sec-ft2
-v'= specific volume for momentum, fta/lbm l
8
.y m
s
,w--
-gs.
r-m.-
a
--sv * -
y r
-. ~.. _ - -. -, _..
g = force-to-mass units conversion factor, 32.2 lbm-ft/3bf-sec2 i
D = hydraulic diameter based on wetted perimeter, f t.
h Based on sensitivity study results that showed little change in hot channel coolant conditions and MDNBR using different friction factors, the default Blasius smooth' tune. friction factor expression 0.25 f = 0.32 Re
+ 0,0 will be used to calculate the friction pressure loss for turbulent flow.
Alse based on sensitivity study results, the friction pressure loss for two phase.
- flow will be calculated using the EPRI two-phase friction multiplier.
In the lateral direction the pressure loss is treated as a form drag loss that l
-is calculated by L
l SP=K w Ll v '
g 252gc 4
-~
4 l
where K = loss ccafficient in_the gap between adjacent channels g
t
=; crossflow through a gap, lbm/sec-ft w
1
- v' = specific _ volume for momentu-,
3 ft /lbm i
S = gap: width, ft.
l l
- g = force-to-mass units conversion factor, 32.2 lbm-ft C
lbf-sec2 1
When rod arrays are-modelea as lumped channels the effective crossflow l-
. resistance is_the sum of-the. resistance of the rod rows between the lumped i
channelicentroids.
The lateral loss coefficient becomes K.. = NK tj G
9 r
--ir
-smy 4
,.r"'+W
w--
f P
W' r-w
+-1' F
F=*
T W-
--'W W*-1P-"--P*T'
'"-WW"'W"9 l
--.,- ~_
i whereNisthenumberofrodrowsbetweenlumpedchannelsandK[isthenominal drag coefficient for afsingle gap.
Crossflow resistance coefficients are not precisely known, but sensitivity study results discussed in Volume 4 of ref.-1
-show that for applications where the axial flow is predominant relative to crossflow, crossflow resistance has an insignificant effect on mass flux and DNBR.
A subchcine' drag coefficient, K, of 0.5 will be used with the g
coefficients for lumped channels calculated internally by the code based on the input centroid distances between lumped channels and the standard subchannel fuel rod pitch.
'T
- 4. l. 7. 2 Turbulent Mixing The VIPRE-01 transverse momentum equation includes terms to calculate the exchange of energy and momentum between adjacent channels due to turbulent mixing.
Two parameters must be input to include turbulent mixing:
a turbulent momentum factor (FTM) and a turbulent mixing coefficient (s).
.i 1
. The turbulent momentum factor defines how efficiently +he turbulent crossflow
. mixes momentum.
FTM can be input on a scale from 0.0 to 1.0, where 0.0 l.
[
- indicates that-the crossflow mixes enthalpy only and 1.0 indicates-that it
' mixes enthalpy and momentum with the same strength.
Scnsitivity studies
- discussed in Vol. 4 of ref. 2 show that changes in the fraction of momentum l
-mi.xing have negligible impact on the-flow field.
As shown in Table 6, sensitivity studies performed using the M/C 8 channel model also showed that the MDNBR is'relativelyLinsensitive to changes'in the turbulent momentum' i
factor.
A turbulent momentum factor of 0.8 will be'used for M/C thermal-hydraulic analyses as-recommended in ref. 2.
10 L
w -
.~ -.
..,... ~
I Turbulent crossflow between adjacent channels.is. calculated by
-w' pSG
=
where w#is the turbulent flow per axial length, s is the turbulent mixing
~
coefficient, 5 is the gap width, and G is the average mass flux of the adjacent channels.
As discussed in Section 4.4.2.2.3 of the McGuire and Catawba FSARs, ref. 4, Westinghouse nas performed a series of tests to determine the mixing
' coefficient for 15x15 fuel with R mixing vane grids.
For 26 in, grid spacing the-tests yielded a mean mixing coefficient of 0.042 with 93% of tha data indicating a value of greater than 0.038.
Additional tests of 17x17 fuel with 26_in.-grid spacing yielded a mean mixing coefficient of 0.059.
The
-Zircaloy mixing. vane grid used with Westinghouse 17x17 optimized fuel was designed lto have. the same mixing characteristics as the 17x17 R grid design.
11his is verified by the fact that the ONB-performance of the Zircaloy grid fuel is similar to-that of the R grid fuel.
Also, since the actual grid spacing is approximately 20,5 in, and the mixing coefficient increases _as grid spacing decreases, the-Westinghouse design mixing coefficient of 0.038 includes added
- conservatism.
'A'mixingcoefficientof(
) will-be _ used for all M/C Mark-BW thermal-hyaraulic
- analyses.
The turbulent mixing.coef ficient for the ~ gaps between lumped channels lis' calculated based on the number of-rod rows between the lumped channel _centrcids 11 y
-e_.,
..,w
-c w
,w
-v---,--
--n-
+-, + - -
a----
--~- --- - - - - - - - - - - ~ -
.m _ _ _.. -.. _ _. _ _ _.. _ -.. _ _ --. - _ _ -. _ _. _ _ _ _ _ _
i
-B&W's mixing l vane grid design _is nearly identical to Westinghouse's' design which is verified by the fact that the Mark-BW confirmatory CHF tests, ref. 5,_are conservatively predicted.by the BWCMV CHF correlation which is l'
heavily based on Westinghouse mixing vane grid CHF data.
VIPRE-01 predictions of the BWCMV CHF data are nearly identical to B&W's LYNX 2 results as discussed in.Section 4.1.7.4.
The VIPRE-01 comparison cases used a mixing coefficient of[
_)
4.1.7.3 Two-Phase Flow Correlations L
(-
p
. A1subcooled void correlation is used to model the nonequilibrium transition l-from single phase to boiling flow with heat transfer from a wall.
A subcooled void correlation actually calculates a subcooled flowing quality for the fluid L
which is then used in a bulk void correlation to calculate the subcooled void.
I' Sensitivity studies were performed with three different combinations of t
subcoolea void and bulk void correlations to evaluate their effects on the hot l:
channel local coolant conditions and MONBR.
p Subcooled Void Bulk Void Levy Zuber-Findlay
. Levy Smith
-EPRI EPRI
-The hot-channel-local coolant-conditions and MDNBR are given in Table 7 for two
- ~
differentfcore conditions.
Table 7 shows that the Levy subcooled void correla-
-tion-yields conservative MDNBR results.
Section 3.3 of Vol. 4 of ref. 1 1
12 l
c
presents the results of VIPRE-01 predictions of the Martin void fraction tests at high pressure (1565 and 1991 psia).
Of the two phase correlations evaluated, the Levy /Zuber-Findlay combination compared most favorably with the test I
results.
In general VIPRE-01 results are insensitive to the selection of two phase flow correlations.
The Levy subcooled void and Zuber-Findley bulk void correlations will be used for M/C thermal-hydraulic analyses.
4.1.7.4 BWCMV. Critical Heat Flux Correlation The BWCMV critical heat flux (CHF) corrnlation, ref. 5, will be used for M/C thermal-hydraulic analyses.
The BWCMV correlation is based on Westinghouse and
. Nuclear Fuel Industries mixing vane grid CHF data.
Confirmatory tests have been performcd to verify that the correlation applies to Mark-BW fuel.
The BWCMV correlation was developed by B&W using the LYNX 2 computer code, ref.
6.
To justify use of the BWCMV correlation with the VIPRE-01 code a number of CHF: data points from each test section included in the BWCMV data base were predicted using VIPRE-01.
The data points from each CHF test section were selected using a-central composite design, ref. 7, for guidance.
The central composite design was based on three of-the-CHF test parameters; inlet enthalpy, inlet mass flux, and system pressure.. Based on the central composite design, 3124 data points wern selected to cover the ranges of test results and of anticipated operation:
Inlet enthalpy 345 - 656 Btu /lbm
. Inlet mass flux
- 1. 0 - 3. 5 Mlbm/hr-f t2 System pressure 1500 -=2455 psia 13
Figures '7-9 show the B&W LYNX 2 versus VIPRE-01 calculations for BWCMV DNBR, mass flux, and quality.at the CHF location, respectively.
All three figures show that the.VIPRE-01 DNBR and. local coolant conditions compare very well with
-B&W's LYNX 2 predictions.
Comparisonlof the LYNX 2 and VIPRE-01 BWCMV DNBRs yields the following Mean = 0.9962 o
= 0.0218 N
124 where Mean'
1 T LYNX 2 DNBR II L VIPRE-01 DNBR o
= standard deviation about the mean N
= sample size
- The VIPRE-01 and-LYNX 2 BWCMV DNBRs rgree to within less than 0.5 percent; thus, the VIPRE-02,6WCMV combination can be used for M/C thermal-hydraulic
. analyses.
The 0.5% VIPRE-01/ LYNX 2 difference is ii.cluded along with the LYNX 2/BWCMV correlation standard deviation of 10.2%, ref. 5, in the calculation
'of the statistical DNBR limit as discussed in Section 6.0.
- 5. 0_
.fiCEulRE/ CATAWBA C0_RE THERMAL-HYDRAULIC ANALYSES
~
5.1:
BACKGROUND A thermal-hydraulic analysis of the M/C core is required to define core thermal margin and_ allowable' operating limits.
The Reactor Protection System (RPS) is 14
l
'designea'to prevent excessive power, excessive coolant temperature and pressure,_or: combinations of these, The RPS philosophy is to define a region
- of allowable operation in_ terms of power, pressure, f. low,' coolant temperature, z_:_ _
and power distribution.
The RPS is set to automatically trip the reactor when the' limits of the safe operating region are approached.
Generic thermal-hydraulic analyses are performed to determine the thermal-hydraulic limits that-are_used to define the regions of safe operation.
A new analysis is performed for a reload core when there is a significant change in the fuel assembly design, a change in the input assumptions of the generic analysis, or a change in the regulatory criteria.
5.2 MCGUIRE/ CATAWBA THERMAL-HYDRAULIC DESIGN BASES
.The overall objective of the thermal and hydraulic design of the reactor core is.to-ensure that the reactor coolant provides_ sufficient heat removal capability to prevent fuel damage during Condition I and II events.
To ensure fuel integrity,_three' core protection design bases have been establ.shed:
1.
Departure from Nucleate Boiling (DNB) design basis-2.
Fuel temperature design basis 3.
Hot' leg boiling limit Only1the steady-state core thermal-hydraulic analyses that ensure that the DNB
' design basis is met are discussed in this report.
15 a.-
- 5. 2.1-DNB DESIGN BASIS
]
The McGuire/ Catawba DNB design basis is that there will be at least a 95%
probability at a 95% confidence level that DNB will not occur on the limiting feel rods during normal operation and operational transients and during any transients arising from fault:. of moderate f requency (ANS Condition I and 11 events), ref. 4.
Currently the DNB design basis is me' by performing core
(
I thermal-hydraulic analyses with all parameter (power, pressure, etc.)
uncertainties considered statistically using the Westinghouse Improved Thermal Design Procedure (ITDP).
The DNB design basis will now be met with an L
-analytical method known as statistical core design (SCD).
The SCD method statistically accounts for the combined effects of the parameter uncertainties as axplained in Section 6.0.
A statistical DNBR limit (SDL) will be calculated for Mark-BW fuel following
'the methodology given in Section 6.0.
Margin is added to the SDL to allow for
-mechanisms that adversely-impact DNB, such as the RCS flow anomaly and transi-c l
tion core effects. The larger DNBR limit is termed the design DNBR limit (DDL),
Generic core thermal-hydraulic analyses are performed using nominal input parameters (for the parameters whose uncertainties have been treated statistically) with the MDNBR compared with the design DNBR limit.
L l
5.3 CORE PROTECTION L
iThe-overtemperature AT (OTAT) and empower AT (0 PAT) trip functions, ref. 4, ensure complete core protection provided that; a) the pressure is between the' high and low pressure trips, and b) that the t asient is slow with respect L
16 l
~
.m.-
i l
to fluid transport delays from the core to the temperature sensors.
The OTAT trip ensures operation within the DNB design basis and the hot leg boiling limit (Tot <Tsat).
The OPAT trip and the low and high pressure trips provide a limit to the range over which DNB protection is required, but they do not specifically provide DNB protection.
The low reactor coolant flow trip provides protection when there is reduced DNB margin because of a decrease in the reactor coolant flow.
The OTAT trip provides a reactor trip before the MDNBR reaches the design DNBR limit based on the measured reactor coolant average temperature (Tave)'
the temperature difference across the vessel (AT), and the primary system pressure.
The OTAT trip setpoint is based on the core tharmal limits which are discussed in the following section.
A compensating term, F(AI), which is a function of the axial flux difference in the core, is factored into the trip setting to account for the effect on DNB of skewed axial tower distributions.
5.4 CORE THERMAL LIMITS The overtemperature AT and overpower AT trip functions define a region of allowable operation in terms of power, reactor coolant temperature and pressure, and axial power shape.
The three boundaries that define the region of allowable operation are:
1.
The core thermal limits which prevent DNB and hot leg boiling within the maximum and minimum pressure limits.
2.
The overpower limit-which protects against fuel melting.
)
17
t I.
3.
The locus of conditions where the steam generator safety valves open.
Ccre thermal limits are the loci of pcints of power and reactor ;oolant pressure and inlet temperature at which the MDNBR is equal to the design DNBR limit (core DNB limits) or the vessel exit enthalpy is equal to the enthalpy of saturated liquid (hot leg boiling 'imits).
The RPS uses the vessel average temperature dif ference (AT) as a measure..' core power.
To assure that aT is proportional to core power, hot leg bulk boiling must be prevented.
x The core thermal limits prevent overheating and possible rupture of the cladding.
Fuel clad overheating is prevented by restricting operation to within the nucleate boiling regime ensuring that the clad temperature is only slightly above the coolant temperature.
DNB is not a directly measurable pa'.ameter; therefore, power and reactor coolant temperature and pressure are related to DNB using the VloRE-01 code and the BWCMV correlation as explained in the following section.
5.5 COPE DNB LIMITS To ensure that the DNB design basis is met core DNB limits are determined for a i
range of operating conditions.
The core DNB limits are the combinations of power and coolant inlet temperature and pressure c' which the MDNBR equals the design DNBR limit.
The quality at the point of
'" is also limited to + 22%,
the upper range of applicability of the BWCMV c rcelation, ref. 5.
Other limits restrict the range of conditions over which the core DNB limits must provide protection.
The OPaT trip places an upper constraint on core 18
l power, the high and low pressure trips limit the range of pressure, and the hot leg boiling limits and the opening of the steam generator safety valves place an upper temperature limit.
Thi ore DNB limits are calculated using the 8 channel model discussed in section 4.0.
The VIPRE-01 input that is used to calculate the generic core DNB limits is discussed in the following section.
- 5. 5.1 REFERENCE POWER DISTRIBUTION The reference radial power distribution used to calculate the core DNO limits is shown in Figures 5 and 6.
The hot assembly power distribution is a relatively flat distribution that minimizes the benefits of flow redistribution.
The core DNB limits are based on the reference rad al power distribution shown in Figures 5 and 6, but the F(AI) portion of the OTAT trip function is determined to ensure DNB protection for any core power distribution.
The F(al) function is determined using maximum al',owable peaking (MAP) limits that are calculated as discussed in Section 5.o.
As shown in Figure 5, the reference peak pi. power (F H) 6 U * "" ""
the core DNB limits at power levels below 100% power, F is adjusted using H
the expression F
= 1. 50 (1 + 0. 3 (1-P))
where P is the fraction of pswer below 100% power.
19
The reference axial power profile used to generate the core DNB limits is a symmetri: chopped cosine with a peak to average value of 1.55.
A routine has been added to the VIPRE-01 code to generate axial power shapes with inlet, symmetric, or outlet peaks.
The routine is based on the following constraint 3 on an axial power shape:
F(x) is continuous from (B,E)
- F(x) is continuous from (B,[)
_1_ f F(x)ax = 1.0 B
E-B
/
where f(x) = axial power shape as a function of the axial location, x.
B,E
= beginning and ending normalized lucation of the active fuel length The reference 1.55 axial flux shape is generated using the new axial power shape routine.
The axia! power profile can changi as a result of rod motion, power change, or due to a xenon transient.
The axial power imbalance is measured using the excore nuclear detectors to protect the core from the effects of skewed axial power shapes.
The RPS provides automatic reduction of the OTAT trip setpoint on excessive axial power imbalance.
To determine the magnitude of the setpoint reduction, maximun allowable peaking (MAP) limits are calculated as discussed in Section 5.6.
20
5.5.2 RCS FLOW The core DNB limits will be based on a reactor coolant system flow of 385,000 gpm which is lower than the measured flow for any of tha four McGuire/ Catawba units.
This flow value could be increased to take credit for the measured flow for a particular unit.
The uncertainty in measuring the RCS flow is accounted for in the calculation of the statistical DNBR limit.
The difference between the flow into the vessel and the flow through the core is the bypass flow.
A nominal bypass flow of 7.5% will be assumed.
The uncertainty in the bypass flow is accounted for in the calculation of the statistical DNBR limit.
5.5.3 PRESSURE Core DNB limits are dependent on pressure and are calculated for four specific pressures to ensure that a bounding OTAT trip setpoint is calculated:
1.
Maximum allowable pressure 2.
Nominal pressure 3.
Intermed; ate pressure 4.
Minimum allowable pressure System pressure is cnntrolled using the pressure measured at the pressurizer.
The core outlet pressure, which is the lowest pressure in the core, is used in the VIPRE-01 code.
Due to the difference in elevation head, friction loss in the vessel and piping, and the velocity head aifference, the coolant pressure 21
is greater at the core outlet than at the pressurizer.
The M/C nominal operating pressure (pressurizer) is 2250 psia and the coolant pressure at the core outlet is 30 psi greater than at the pressurizer.
Thus, the nominal core outlet pressure is 2280 nsia.
The generic core DNB limits will be calculateo using core exit pressures based on the lowest high and low pressure trip setpoints for the McGuire/ Catawba stations.
The uncertainty associated with measurement and control of the system (pressurizer) pressure is accounted for in the calculation of the statistical DNBR limit.
5.5.4 CORE INLET TEMPERATURE For a given core power and pressure the vessel inlet temperature at which the MDNBR equals the design DNBR lirc.it defines a point along the core DNB limit lines.
The uncertainty in measuring and controlling the inlet temperature is accounted for in the calculation of the statistical DNBR limit.
VIPRE-01 is run at a number of power levels for each of the four pressures to determine the inlet temperatures tnat yield the design DNBR limit (or 22%
quality).
Typical McGuire/ Catawba core DNB limits are shown in Figure 10.
The nominal and minimum pressure core DNB limit lines are only shown for power levels greater than 100% since the hot leg boiling limits (Tout <
sat)
I' s
typically more restrictive than the core DNB limits below 100 percent power.
5.6 MAXIMUM ALLOWABLE PEAKING LIMITS The core DNB limits and OTaT trip function account for core power and coolant temperature and pressure effects, but the limits are based on the reference 22
~
i radial power districution and a symmetric axial power profile.
To ensure DNB protection for axially skewed power inapes an OTAT F(al) function is generated to mooify the OTAT trip setpoint.
To calculate the F(al) function, the change in core DNB limits with changes in c re power distribution (radial ano axial power distribution) must be determined.
A series of maximum allowable peaking (MAP) limits are calculated in the f orm of lines of constant MDNBR for a range of axial peaks with the location of *ne peak varied from the bottom to the top of the core.
For a given axial peak and location of the peak, the total peak (F H
- I ) that yields a MDNBR equal to the design DNBR limit (and ouality at z
the MDNBR location <22%) defines a single point along a MAP curve.
MAP limits are typically calculated for axial peaks of with the axici power shapes generated using the new axial flux shape routine discussed in section 5.5.1.
A typical set of OTAT MAP curves is snown in Figure 11.
MAP limits are compared in a mcneuvering analysis with peaks resulting from design power transients as discussea in ref. 8.
The OTAT MAP limits are used to determine a conservative envelope of allowable core power versus axial offset.
The allowable core power versus axial offset envelope is used to calculate the F(AI) portion of the OTaT trip function.
To determine the allowable axial offset as a function of core power the OTAT MAP limits will typically be determined at four power levels at the limiting inlet temperatures as shown in Figure 12.
Figure 12 shows how the not leg boiling limit line provides an coer 'imit for *9e inlet temperatures that must be considered.
23
l l
l l'
'/ r e.
v.
..;ine 9g:,; ins (predicted peaking greater than the appropriate li@ : m
.si -ined during a maneuvering analysis, ref. 8, the MDNBR will
_ 1.
M Q Qu' i.eu To. the limiting predicted power distribution (see Figure 13).
The predicted radial power distributio,i and axial power profile is input directly into VIPRE-01.
A second set of analyses is performed to determine the power distributions (F
x F ) that yield a MONBR equal to that calculated for the limiting H
7 Non-0 TAT ONB transient.
The limiting Non-0 TAT DNB transient (primary protection against DNB is not provided by the OTaT trip function) is a comolete loss of forced reactor coolant flow.
Thr complete loss of forced reactor coolant flos transient is analyzed using the RETRAN-02, ref. 9, and VIPRE-01 computer codes.
The transient bcundary conditions (power, pressure, temperature, and flow versus time) used in VIPRE-01 are calculated by the RETRAN-02 code.
m 9
.a 24
1 limits.
Typical loss of flow (LOF) MAP limits are shown in Fig. 14.
The LOF MAP limits are compared with predicted power distributions as discussed in ref.
8 to determine operating axial flux difference (AFD) limits.
- 6. 0 STATISTICAL CORE DESIGN Traditionally steady state and transient core thermal-hydraulic analyses have been performed with all core parameter (power, pressure, etc.) uncertainties applied conservatively.
Each parameter was assumed simultaneously to be at the worst end of its uncertainty range with respect to DNB.
The re,sulting MDNBR was comparea with a design DNBR limit that assumed that the CHF correlation was also performing at its worst.
The following text, Sections 6.1 - 6.5, describes a method for statistically combining the effects of the uncertainties of the parameters used to predict DNB.
Themethodology,knownasgtatisticalcoredesign(SCO), is used to determine an overall DNBR uncert'pinty which is used to determine a statistical DNBR limit.
Core thermal-hydraulic analyses are then performed using nominal input parameters with the calculated MDNBR compared with the statistical DNBR limit.
25
n 6.1 SCD METHODOLOGY The procedure for determining the statistical DNBR limit (SDL) includes five steps:
1.
Selection of parameters 2.
Response surface modeling 3.
Selection of uncertainties 4.
Monte Carlo propagation of uncertainties 5.
Calculation of the statistical DNBR limit j
-6.2 SELECTION OF PARAMETERS In a general sense the problem that must be solved deals with a function of i
many parameters that is only known point-by-point through a computer code, in this case DNBR, which is a function of many parameters such as power and flow, is only known using the VIPRE-01 computer code.
The problem is to determine the probability distribution of the DNBR given distributions for the-VIPRE-01 input parameters.
Since VIPRE-01 would have to be run thousands of time,s to determine the DNBR probability distribution -it is much more efficient to determine-an approximating function or response. surface model (RSM) to calculate DNBR.
The first step-in developing a response surface model is to select the parameters that should be taken into account.
The parameters _which significantly impact the calculation of DNBR are discussed below.
26
1 i
1.
Core Power, Q.
The core power represents the total thermal ostput of the plant expressed as a percentage of the rated thermal power (% of 3411 MWt for the McGuire/ Catawba units).
2.
Core ~ Flow, W.
The core flow is the coolant flow available for transferring heat from the core.
The core flow is the coolant system flow 1
- /
minus-the flow that bypasses the core through the thimble tubes, the gap between the reactor vessel and barrel, the core baffle-to-barrel region, the spray nozzles into the upper head, and the gap between the peripheral.
l fuel' assemblies and the core baffle.
3.
- Core Outlet Pressure, P.
The core outlet pressure is the lowest pressure in the core and for the purpose of determining coolant properties the-VIPRE-01 code uses the core outlet pressure uniformly
- throughout the core.
4.
Inlet-Temperature T.~
Core DNB limits define allowable core inlet temperatures as explained in section 5.5.4.
'The response surface model was developed.using inlet subcooling (Tsub) to better represent the DNB
- phenomenon.
-5.
- Nuclear Enthalpy Rise Hot Channel Factor, F F " is effectively the g.
3 maximum two-dimensional radial power factor.
6.
- Axial. Peak, F.
The axial power distribution expresses the power output 7
of each~ vertical linear unit of fuel to the average power output of all of
-' the. linear-units of fuel, The peak of the axial power distribution is 27
+-
--.,-......o,,.m,
.,,,~m.,..,e
,,,,---,---,...m-c
,,-,,..,-e.,m_-..,
..-em,
,-,....-,,..m.v.,--
. - - - - - _. _...,. -.- - - - - ~.~
l i
denoted by F,.
MAP limits are calculated as explained in Section 5.6 for a comDiete range of axial peaks.
t
-7.
Axial location of F, Z.
HAP limits are determined as a function of 7
the location of the axial peak.
[
All seven of the parameters are varied when performing reload thermal-hydraulic analyses since eath-of the parameters vary during reactor operation.
The parameter. ranges that were considered to develop the McGuire/ Catawba Mark-BW
-response surface model are given in Table 8.
The maximum and minimum values shown in Table-8 were selected to bound the values that will be analyzed in typical M/Cl thermal-hydraulic-analyses.
These ranges could change for I
-subsequent plant specific analyses.
6;3 RESPONSE SURFACE MODEL Considering the thcusands of cases that must be'run to determine the-statistical DNBR limit, an obviously more efficient anproach is to estimate the
_VIPRE-01/BWCMV DNBR predictions using a suitable function called a response i
surface.
A reasonable number of VIPRE-01 runs are made to develop the respunse L.
+
surface model which is then used to determine the ONBR probability distribntion.
F The RSM includes 4
E I
.28 a.-
. ~.
DNBR =
Instead of using the independent parameters directly when defining the RSM, each parameter is coded.
Coding is essentially a normalization process which results in an R$M where equal coefficients implies equal importa ce of the terms.
Coded values are signed and non-dimensional.
The center paint (coded values of zero) for the M/C RSM was placed at approximately nominal conditions as shown-in Table 8.
Coded Value = 2 [ Given Value - Center Value ]
Center Value Minimum Value Using the equation above the maximum value for each parameter was not always
-eauatedto[]asshowninTable8.
The oower and flow corresponding to a coded
~[wouldbeunreasonablyhighanditisnotimportantinthiscasefor value of the range of values to be symmetric.
The array of paraneter values given in Table 8 was developed referring to the central composite design idea discussed by Boxiano Wilson in ref. 10.
The factorial points (coded values of +/~1) for each of the parameters are also given in Table 8.
29'
For 7 parameters a total of 143 points (one center point, 14 axial points, and 128 factorial points) make up a full central composite design, ref.
7.
This is considerably grcater than the number required to determine the RSM coefficients sincetheRfMwillonlyinclude{~ coefficients.
To efficiently and accurately determine the RSM coefficients,]VIPRE-01caseswererun.
The M/C RSM
" design" matrix is given in Table 9.
The matrix of parameters given in Table 9 was developed only as a systematic way to generate a data case to determine the RSM.
The Mark-BW 8 channel model was used to determine the MDNBR for the combinations of parameters given in Table 9.
The reference radial power distribution used is shown in Figures 5 and 6.
For the axial and factorial point cases, the hot assembly (pins 1-8) pin peaks were adjusted maintaining the same distribution shown in Figure 5.
The radial power for the rest of the core (pin 9) was adjusted to maintain a normalized power distribution.
The VIPRE-01 MONBRs (BWCMV) are given in Table 9.
6.3.1 LEAST-SQUARES CURVE FIT A FORTRAN program was written to determine the RSM coefficients.
As for all least-squares curve fits, the coef ficients are determined to minimize the sum of the squares of the differences between the experimental values (in tilis case the VIPRE-01 PDNBRs) and the function being developed to fit the points.
30
4 Lin ref. 11 it is shown that a function Y = f(x) consisting of m functionally linear undetermined' coefficients b, be,...b f r a least-squares fit to n i
n data points, involves the simultaneous solution of m linear algebraic
-I equations.
The f unction Y = f(x) can be written in general form as Y = f(x)-= b fit (x) + b f (x) +... b, f,(x) 22 The-m linear equations can be written in matrix form and the matrix can be
-solved by any simulta'neous solution method.
The m simultaneous equations were solved using Cholesky's' method, ref. 12. The fitting program was verified by determining the coefficients for two response surfaces that are given as examples in ref. 7.
Using the VIPRE-01 MDNBRs given in Table 9.an RSM was calculated using the fitting program discussed above.
The coefficients are given in Table 10.
~
The standard error for degrees of freedom is DNBR points.
The VIPRE-01 and the RSM MDNBR$ for the
~ '
cases used to develop the RSM are given in Table 9 and are plotted in Figure 15.
For further verification that the RSM can accurately predict DNBR (compared
'with_the VIPRE-01 code), ten additional VIPRE-01 cases were run covering
. typical-core DNB limit and loss of flow accident conditions.
The VIPRE-01 and
'RSM DNBR results are given in Table 11. ' Comparing the VIPRE-01 and RSM DNBRs.
~
I he RMS error is which 'is less than the RSM standard error of
~
t For.th'e cases summarized in Table 9, the mean VIPRE-01 MDNBR/RSM MDNBR is and the standard deviation is l
31
-1
6.4 STATISTICALLY TREATED UNCERTAINTIES A statistical core design analysis is performed to statistically corr.bine the effects of the uncertainties of the parameters that significantly affect DNB.
A statistical DNBR limit is determined that accounts for the unctrtainties and then core thermal-hydraulic analyses are performed using nominal values of each parameter.
The uncertainties that are not included in the SCD analysis, such as the 5% reduction in hot assembly flow to account for inlet flow maldistribution, are included directly in the core thermal-hydraulic analyses.
The uncertainties included in the M/C SCD analysis are discussed below and are summarized in Table 12.
The normally distributed uncertainties are included at a 95%, one-sided probability level.
For example, there is a 95% probability that the RCS flow differs from the true flow by less than 2.2%, ref.
13, in the direction causing a decrease in DNB.
The RCS flow uncertainty of 2.2% is obtained by multiplying the uncertainty standard deviation of 1.337% by the 95%
probability factor (l.f>45) obtained from normal distribution tables.
When it is questionable whether an uncertainty probability is normally distributed the uncertainty was described using a uniform distribution.
The uncertainties discussed below were used to develop a statistical DNBR limit for Mark-BW fuel.
The uncertainties will be justified on a plant-specific basis in the Reload Report for the first application of this methodology.
Core Power The reactor power is monitored by performing a secondary side heat balance (power calorimetric measurement).
A bounding power uncertainty of
+/-2% is considered due to colorimetric errors in measuring feedwater flow, 32
temperature, and pressure, steam pressure, and moisture carryover.
The uncer-tainty--in core power is assumed to be normally distributed with a standard deviation of +/- 1.22%.
This uncertainty is a function of secondary side parameter: and is indepehdent of other primary side (i.e., 500) uncertainties.
RCS Flow --- The RCS flow is monitored by performing a precision flow calorimetric measurement at the beginning of each cycle.
The flow is calculated by measuring the steam-generator thermal output and the enthalpy rise of the_ primary coolant, correcting for the heat ' input of the RC pumps and
-the_ loop's portion of the primary systems heat losses.
The RCS loop elbow taps are normalized against the precision calorimetric-measurements and are used for surveillance of the flow.
The statistical combination of the
' instrumentation / measurement uncertainties for the loop elbow taps and the flow calorimetric errors result in a flow uncertainty. standard deviation of 1.337%.
Assuming that the flow uncertainty is normally distributed at a 95 percent probability, the uncertainty is 2.2% (1.645 x 1.337), ret. 13.
J Core Bypass Flow
- The coolant flow that is not effective in cooling the reactor core _is considered core bypass flow.
The uncertainty in bypass flow is assumed to be uniformly distributed about a mean value of 7.5% with a maximum 1
-value pf 9.0% (+/-1.5% range).
' Pressure Reactor Coolant System pressure is controlled at the pressurizer-by a system that compares the measured pressure.with a reference pressure.
The pressure _ uncertainty:is considered to be uniformly distributed over a +/- 30 psi range which_ accounts for measurement error and the allowance for steady-state I
fluctuations.
33
...a..
-.. =...
Inlet Temperature --- At the McGuire and Catawba Stations the core inlet temperature is indirectly controlled by controlling T,
, the average of th?
narrow range loop T and T s c n rolled by a system diat compares hot cold' avg the auctioneered high 1,
from the loops with a reference T based on the avg measured first stage turbine impulse pressure.
Statistically combining the uncertainties for the temperature measurement instrumentation and the controller deadband results in a total uncertainty of +/- 4 "F, which is considered to be uniformly distributed.
Three uncertainties that app y to the radial power (f g) are included in the M/C SCD analysis.
N F
This factor relates the heat generation rate in the peak power rod to 3g the heat generation rate of the average rod in the core.
Moveable incore instruments are used to measure the flux distribution within the core during operation.
Measurement uncertainties arise from instrumentation drift and reproducibility errors, integration and location errors, errors associated with the burnup history of the core, and errors associated with the conversion of instrument readings to rod power.
The measurement uncertainty is considered to be normally distributed with a standard deviation of 2.43%, ref. 13 ard 14.
E N
F This hot channel factor, which is applied as a direct multiplier on F 3g Ag, accounts for variations in the fabrication variables which affect the heat generation rate along the flow channel (pellet diameter, density, and Unas enrichmentt Uncertainties in these variables are determined from samples of E
manufacturing data.
An F design value of 1.03, ref. 4, is used considaring Ag 34
1 l
the combii, j uncertainty to be normally distributed about a mean value of L O.
This design value, determined at a 95% confidence level, applies to at least 95% of the pins in the core.
Thus, the standard deviation is 0.0182 (0.03/1.645).
F3g, Spacing In traditional thermal-hydraulic design analyses an uncertainty is applied to the hot assembly lateral flow area.
The hot assembly flow area is obtained assuming:
1.
Assemblies adjacent to the hot assembly shift in the upper and lower grid pads towards the hot assembly.
The grid pad openings are assumed to be manufactured to the largest allowable tolerance, while the end posts inserted into the grid pads are assumed t.
be mar.uf actured to the smallest allowable tolerance.
This creates the maximum allowable assembly shift and the sma!1est possible assembly-to-assembly pitch.
2.
Assemblies adjacent to the hot assembly then bow toward it, touching at the assembly midspan.
~
The hydraulic ef fect of the reduced hot assembly flow area is very small in terms of DNBR.
The hot assembly flow area uncertainty also results in a steeper power peaking gradient across the assembly.
As a conservative estimate of the overall effect of the hot assembly flow area variation, the uncertainty on peaking (F g) resulting from the variation in flow area is included in the M/C SCD analysis.
The uncertainty on F is assumed to be normally distributed H
35
. - -.. ~. - - - _
4 l
l with a-standard deviation of 0.02/1.645 = 0.0122.
The impact on F of'the 3
variation in hot assembly flow area must be calculated to verify the 2%
assumption.
F A chopped cosine axial flux shope with a peak-to-averaoa value of 1.55 z
i is usti to determine the core DNB-limits as explained in Section 5.5.1.-
Both the magnitude and the location of the axial peak can significantly vary due to i
control rod motion, xenon distribution, and fuel depletion.
To ensure that the core is protected, the core average axial offset is constantly monitored and a
. penalty isl applied to the OTAT setpoint to. : count for the effect of any adverse axial power distribution.
To determine the F(AI) portion of the OTAT trip function, which accounts for
-the effects of axially asymmetric power shapes, a series of maximum allowable peaking (MAP) limits are generated in the: form of lines of constant MONBR for 6
various axial peaks ativarious axial peak locations.
MAP limits are compared in a maneuvering analysis with peaks resulting from design power transients.
An axial peak calculational uncertainty must be accounted for in either the thermal-hydraulic analysis or the maneuvering analysis, ref. 8.
The.M/C SCO analysis assumed that the axial peak calculational uncertainty is normally distributed with a standard deviation of 2.92% (at a 95% probability the i
uncertainty'is.4.8%),'ref. 8.
'Z -
The M/C thermal-hydraulic analyses are performed using VIPPF.-01 with 3 inch axial nodes, but core-peaking-calculations.are based on 8 inch axial nodes, ref. 15. - In a. maneuvering analysis the allowable total peak fJr a given axial peak is determined by interpolating'between the appropriate MAP-linits a
36
..~_,e
..,...M..
-o
~.. _
,-a,_,,%,,..
_.._,m.w_..
,,,m.m..,.s
..,,,.,,.__E[m-
,ym.,m,,
u,y,-,.
using the location of the axial peak.
When pey' forming the interpolation, tne axia s ceak location can only be determined within +/-8 inches.
4his uncertainty is incirded in the M/C SCD analysis af suming that the uncertainty is unif ormly di str' outed.
Three uncertainties that apply directly to the calculation of DNBR are also treateo statistically.
l CHF Corr lation The BWCMV correlstion was developed by C&W using LYNX 2 with s
a mear-to predicted (M/P) CHF standard deviation of 0.1020, ref. 5.
As ais-cussed in Section 4.1. 7. 4, the BWCMV corre'ition can be used with the VIPRE-01
.:oce.
The VIPRE-01 versus LYNX 2 results justify using the correlation standard deviation of 0.1020.
The LYNX 2 and VIPRE-01 BWCMV DNBRs on the average agree to within less than 0.5% and a code /model uncertainty has been included as discussed below.
Code /Model Since neither VIPRE-01 uncertainties nor offsetting code E
conservatisms have bean quantified, a GNBR uncertainty was included in the M/C SCO an ny C s.
A total codennodel ur. certainty of was used to account E
for the 0.5% dif ference ir. the VIPRE-01 and LYNX 2 BWCMV results and the 1.0%
difference in the DNBR results for the 75 and 8 channel models.
The code uncertainty is considered to be normally distributed with a standard deviation of J
flSM The RSM is used to determine the effect cn DNBR of all of the s
uncertainties listed in Tab'.e 12.
The RSM uncertainty is assumed to be normally districuted with a standard deviation of For conservatism, 37
the stanaard deviation was increased to J The RSM uncartainty was i
developvd d ing the rat (os of the VIPRE-01 to RSM DNBRs given in Table 9.
6.5 PROPA^iATION OF 1(NCERT AINT Q A direct Monte Carlo and ysis was performeu tc determine the effect on DNBR of all of the uncertainties givm in Table 12.
To determine the most limiting DNBR coef',cient of variation (statida/d deviation /mean) ten core statepoints
~
covering a wide range of power, pressure, temperature, and flow were analyzed.
The ten core statepoints shown in Table 11 cover the range of conditions considered when determining cora DNB limits and it'e limiting Condition 11 DNB transient (complete loss of flow).
Wh'<ie these statepoint:, do not absolutely bound all possible core operating 4tes, the range of statepoints was analyzed to maximize the coef ficient of variation resulting from the propagation of uncertainties.
Sets of statepoints were randomly generated based on each " nominal" statepoint by independently varying each uncertainty according to the probability distributions given in Table 12.
The SAS random number generation functions RANNOR and RANUNI were used, ref 16.
RANNOR generates a normally distributed set of random numbers with mean zero and standard deviation one.
RANUNI generates a uniformly distributed set of random numbers on the interval (0,1).
Each uncertainty was independent'y varied to produce a set of random statepoints.
An exemple is shown in Table 13.
A total of 3000 randomly determined statepoints were generated for each of the "nomir.al" statepoints given in Table 11.
The mean, standard deviation, and 38
coefficient of variation of the 3000 DNBRs calculated for each " nominal" statepoint are given in Table 14.
All of the DNBR distributions were assessed for normality using the D prime test.
All of the distributions were found to
.l be normal at the 5% level.
As shown in Table 14, the coefficients of variation change very little considering the large variations in parameters that were l
i studied.
The two -core stet 6po'att wit h the largest coef ficients of variation (2' and 10) l were rerun to randomly determine a secoa.!J sef of 3000 combinations of parameters.
The DNBR results are given le Table 14 as cases 2R and 10R.
The coefficients of variation for the two repeat statepoints are not significantly different from the coefficients for the original runs.
For 3000 observations, the 95% Chi Square multiplier is 1.0218.
Thus, with 95%
confidence. the largest coefficient of variation (CV) is 0.1677 (1.0218 x 0.1641,-Case 2),
ToavoidLONSwith95%probabilityat95%'confidencelevel,thestatisticalDNBR limit (SDL) is calculated by SDL =-
1.0 1.0 -.(1.692 x.CV)
IThe M/C Mark-BW statistical DNBR limit is 1.396.
SDL =
- 1. 0
= 1.396 1,0 - (1,692 x 0.1677) 39
,.:._ u _~
a.
__._,._.._,..:.___..._2;_.
=.. -.
- M/C thermal-hydraulic analyses will be performed using a design DNBR limit which includes margin nuove the statistical DNBR limit.
The thermal-hydraulic analyses will be performed using nominal input parameters (power, pressure, flow, etc. ) with the MONBR compared with the design DNBR limit.
- 7. 0
SUMMARY
This report describes the VIPRE-01 models and methodology to be used for McGuire and Catawba (M/C) core thermal-hydraulic analyses.
Sensitivity studies have been performed to evaluate various modeling' options and the results-were used to select the final M/C model.
This report also describes the Statistical Core Design (SCD) methodology that will be used to account for the impact.on DNB of the uncertainties-of parameters such as power, pressure, and temperature.'
s
- 8. 0
-REFERENCES 1.
IVIPRE-01: A Thermal-Hydraulic 4nalysis Code for Reactor Cores, EPRI-NP-2511-CCM, Vul.1-4,- Battelle. Par.ific Northwest Laboratories, July 1985, 2.
Letter from L.E. Rossi (NRC) to J.A. Blaisdell (UGRA Chairman),
" Acceptance for Referencing of Licensing Topical Report VIPRE-01: A Thermal-Hydraulic; Analysis Coce for Reactor Cores, EPRI NP-2511-CCM, Vol.
1 1-4", May 1, 1986.
3.
Mark-BW Mechanical Design Report, BAW-10172, Babcock & Wilcox, Lynchburg, Virginia July 1988.
40 r,4.
,,,,, =
,e-
,*,,e---4 vw,,,,-
w--
w,tw sy,me-e.-----=
ev -m%-
-c e e w
.-w e + ws
,,,w,,--,.m.
- e ww-6-,=-mwa
~=a-,#.--e.-====,w-*
-e==-
i
4.
McGuire Final Safety Analysis Report, Docket Nos. 50-369/370, Catawba Final Safety Analysis Report, docket Nos. 50-413/414.
S.
BWCMV Correlation of Critical Heat Flux in Mixing Vane Grid Fuel Assemblies, BAW-10159, Babcock & Wilecx, Lynchburg, Virginia, May 1986.
6._
LYNX 2: Subchannel Thermal-Hydraulic. Analysis Program, BAW-10130-A, Babcock
& Wilcox, Lynchburg, Virginia, July 1985.
7.
R.H. Meyers, Response Surface Methodology, Allyn-Bacon, Boston, Massachusetts, 1971.
4 8.
Nuclear Design Methodology for Core Operating Limits of Westinghouse Reactors, OPC-NE-2011P, Duke Power Company, Charlotte,' North Carolina, April 1988.
9.
RETRAN-02.- A Program for Transient Thermal-Hydraulic > Analysis of Complex Fluid _ Flow Systems, EPRI NP-1850-CCM, Rev. 2, EPRI, November-1984.
10.
G.E.P, Box and K.B. Wilson, "On the Experimental Attainment of Optimum Conditions", Journal Roy. Stat. Soc.
B,- 13,.1-34, 1957.
11.
P.A. Stark, Introduction to Numerical Methods, MacMillan Publishing Co.
Inc., 1970.
12.
M.L.-James, et. al.. Applied Numerical Methods for Digital Computation with FORTRAN and CMSP, Harper and Row, 2nd Edition, 1977.
41
_.. _ _ _..~ _ __. _ _. _ _.,. _, - _. _ _ _ _ _ _.... -.. _,
13.
Tect.nical Specifications for Catawba Nuclear Station Units No. I and 2, Docket Nos. 50-413/414.
14.
Technical Specifications for McGuire Nuclear Station Units No. 1 and 2, Docket Nos. 50-369/370.
15.
McGuire and Catawba Nuclear Stations Nuciear Physics Methodology for Reload Design, DPC-NF-2010, Duke Power Company, Charlotte, North Carolina, April.1984.
16.
A. A. Ray, Ed., SAS User's Guide: Basics, 1982 Edition, SAS Institute Inc.,
Cary, North Carolina.
42
m E-Table 1. Typical Mark-BW Fuel Assembly Data i
i Dimensions at 70'F Fuel rod 00, in 0.374
-Thimble tube 00,-in.
0.482 i
Instrument tube 00, in.
0.482 ruel rod pitch. in.
0.496 fuel assembly pitch.in.
8.466
~ Fuel rods / assembly 264 Thimble tubes / assembly 24 Instrument tubes / assembly 1
t 1.
i k
L-43 M-g g-y y w+
---*c-mrw m
g r=-
w w
-ek y7 s
(wy py vy v Ty*W-'ev v wtce'T' a
f+"-+FT-1'h*T+r-'tB'--*Wt#
WS
- PN'D Ma*="
W *'"':*
Y Y*
m'8'*****
8'"'"
. ~ = _ _. _. _ _ _ _.. _ _. _ _ _.. _... _
Table 2.
Conditions f or Comparison of Mark-BW Core Models N
F F
Case Power, %
Flow, %
Prescure, psia
- Tin, F
AH z
Z*
.l 1
100-100 2200 569.5
- 1. 5 1.5
- 0. 5 2
115
'120 2400 562.1 1.7
- 1.7 0.7
[
3 125 100 2200 569.5
- 1. 5
- 1. 5 0,5 4
115.
120 2000 535.8 1.7 1.7
- 0. 7 5
115 120 2400 602.1
- 1. 7
- 1. 3
- 0. 7 6
115 120 2000 535.8 1.7.
1.3
- 0. 7
-7 100 100-2200 569.5
- 1. 5 1.9 0.5 0
100 100 2200 569.5 1.9 1.5 0.5 P
i f
f A
- Normalized location of axial peak. F 7
44
.x.
l Table 3.
Mark-BW Model Comparisons i
l i
f i
c c
MDNBR Mass Velocity Quality BWCMV M1bm/hr-ftr a
(
~
Case 1
l 2;
L 3
f 4
5 6
7-8 i
I i
I i
l I
a,
- see Table 2
-b.
75, 12, and 8 channel models l-c.
at MDNBR location t
t Y
45
=
y
.-+..e.a e
av
.=...,4r.
,--e...,-.
,we.
--+~-i.-%,w,-_.+.,-,%,
,ve..u~.w
--,m.-m--m-.w---,..-e.mce#,-,-=w.
.-*,,,,vrern+m--.-,-,-v
Table 4.
Axial Node Length Sensitivity Study 100% Power 100% Flow 2280 psia T
s 605.0or in MONBR Node Size Elevation Location (in.)
(in.)
MDNBR (in.)
3 1.95 - 145.95 1.487 109.3 a
2 105 - 125 1.491 110.1 120% Power 100% Flow 1990 psia Tin =564.0 F MDNBR Node Size Elevation Location (in.)
(in.)
MDNBR (in.)_
3 1.95 - 145.95 1.268 121.1 a
2 110 - 130 1.268 121,4 The rest of the active length was a.
modeled with 3 in, nodes.
46
l l
Table 5.
Inlet flow Sensitivity 100% Power 100% Flow 2280 psia Tin = 605. 0*F Mass Velocity" Exh inlet Flow MDNBR Mlbm/hr-fta Enthalpy*
Quality
~
Uniform Hot Assembly reduced 5%
120% Power 100% Flow 1990 psia Tin = 564.0 F Mass Velocity
- Exit Inlet Flow MDNBR M1bm/hr-ft2 Enthalpy*
Quality Uniform Hot Assembly reduced 5%
0
- at MDNBR location 47
Table 6.
Turbulent Momentum Factor (FTM) Sensitivity Study 100% Power 100% Flow 2280 psia T.
= 605.0 F in Ma. 3 Velocity Exit FTM MDNBR M1bm/hr-ft2 Quality
~
0.0
~
0.6 1.0 100% Power 100% Flow 1990 psia Tin = 564. 0 F
~
Mass Velocity Exit FTM MDNBR M1bm/hr-ft2 Quality
~
0.0
~
0.8 2.0 48
Table 7.
Senstitivity to Void Models 100% Power 100% Flow 2280 psia Tin = 605.0 F 2-Phase Subcooled -Bulk Friction Mass Velocity
- Exit Void Exit
. Void Void Multiplier MDNBR M1bm/hr-ft2
-Fraction Quality
~
-Levy-ZUBR EPRI LevyE Smith Homogeneous EPRI EPRI EPRI 100% Power 100% Flow 1990 psia-Tin = 564.0 F 2-Phase Subcooled t' ul k.
Friction Mass Velocity
- Exit Void Exit Void Vej d Multiolier MDNBR M1 bm/hr-f t2 Fraction Quality Levy ZUBR EPRI.
Levy-LSmith Homogeneous EPRI
-EPRI
-EPRI
- at MONBR axial location-49-
l l
l Table 8.
SCO Statepoint Parameters
- Core Flow = 0.94(382,000 gpm) = 359,080 gpm 6% Core bypass flow The RSM was developed using a core bypass-flow of 6% and a RCS flow of 382,000 gpm.
The core flow shown above was only used to develop the RSM which includes ilow as an independent variable.
The generic M/C thermal-hydraulic analyses will be based on a core bypass flow of 7.5% and a RCS flow of 385,000 gpm.
- The RSM was developed to cover up to a pressure of 2600 psia, but the BWCMV CHF correlation is only valid up to a pressure of 2455 psia.
50
1 -
l
}
Table 9.
RSM Input and VIPRE-01 and RSM Results h
W 9
51
Table 9.
RSM Input and VIPRE-01 and RSM Results (Continued)
We w
52
1 i::
~
p
)
I
. Table 10. - VIPRE-01 Mark-BW RSM 4
l '-
l Coefficients t
.)
1
-I f
I r'I I
l l
-- i l
1 I
1 I
l;
~
.1 The standard error I
t l
l 53 l
l
yz
.. ~ _
. ~. _
w
. Table 11, VIPRE-Ol'- RSM.DNBR Comparisons h
L M
e
- Case
- 1 2
- 3.
4
- 5 6-l7-8
- 3 110 -
r.
m a
,.e l;;
I.
l:
.54
~.
. Table 12.
Statis'-ically-Treated Uncertainties Standard Parameter Uncertainty
- Deviation Distribution
- Core Power
+/-2%
+/-1,22%
Normal
' Core Flow Measurement
+/-2.2%
+/-1.337%
Normal Bypass' Flow
/-1.5%
Uniform
+
Pressure
+/-30 psi Uniform Irlet-Temperature
+/-4 F Uniform h
Measurement
+/-'4%
+/-2.43%
Normal F
+/- 3%
+/-l.82%
Normal Spacing
+/-2.0%
+/-1.22%
Normal F-Calculational
+/-4.8%
+/-2.92%
Normal 7
~Z:
Calculational
+/- 8 in-Uniform DNBR.
. Correlation 10.2%
Normal Code Normal c
RSM Normal "The uncertainties and distributions wii' be justified on a plant-specific basis in the-Reloao Report for'the-first application of this methodology.
55
~
Table 13.
Sample of Random Operating Conditions Power Flow Pressure sub H
p F
psia F
AH z
Z Base Conditions 120 98.94 2430 70.8 1.5 1.5
- 0. 5 119.98 96.27 2417.6 67.73 1.640 1.501 0.463 121.84 98.49 2449.7 66.95 1.575 1.558 0.513 119.80 99.44 2409.2 68.35 1.600 1.506 0.573 7
119.94 96.38 2435.4 68.68 1.500 1.488 0.568 121.48 97.60 2403.7 70.73 1.591 1.544 0.490 121.58 98.30 2401.5 68.18 1.399 1.489 0.524 119.26 98.82 2452.6 72.76 1.452 1.502 0.460 118.51 97.30 2403.2 73.15 1.500 1.505 0.444 119.50 101.57 2454.1 67.25 1.443 1.540 0.475 56
i
-Table'14.
Monte Carlo-Uncertainty Propagation Results b-SAS ONBP, Results Stancard Coefficient
-a
-Case Mean Deviation
.of Variation
~
'1 3 4
5-
.6
'7I
-8 9
110-2R' C
'10R a,
Refer to Table 11 b.
3000 Observations c.
R2 peat case 57
FIGURE 1, 75 UIANNEL HODEL - SUBCilANNEL GEONETRY AND f uMER DISTRIBUTION
.6 1
58
n 4
FIGURE 2.
75 CHANNEL MODEL - ASSEMBLY GE0 METRY AND POWER DISTRIBUTION L
~
1' l
l l
l i
59 l
1 l
k'R 1 FIGURE 3.
12 CllANNEL 110 DEL - SUBCHANNEL GEOMETRY-AND POWER DISTRIBUTION 1.
l l
l l
60
FIGURE 4.
---1 12-CIMNNEL H0 DEL - ASSEMBLY GEOMETRY AND POWER DISTRIBUTION 61
y
+.-..a x---
FIGURE 5.
8 CllANNEL H0 DEL - SU8 CHANNEL GEOMETRY AND POWER DISTRIBUTION 62
FIGURE 6.
CHANNEL MODEL - ASSEMBLY GEOMETRY AND POWER DISTRIBUTION T
i t
d 63
FIGURE 7.
VIPRE-01 vs. LYNX 2 DNBR BWCMV CHF CORRELATION l
1.6
-- /
/
l
/,
1.4-
/
, t, i
1.2 -
z e
O e.
Z
- 2 1--
n M:.e 0.8 -
p
/
0.6 -
0.6 0.7 0.8 0.9 1
1.1 1.2 1.3 1.4 1.5 1.6 VIPRE-01 DNBR l
l
FIGURE 8.
VIPRE-01 vs. LYNX 2 MASS FLUX AT CHF BWCMV CHF CORRELATION 4
O J, -
- f 4A l$
b '
L d
a_
i NX l
g..
a 0
0.5 1
1.5 2
2.5 3
3.5 4
VIPRE-01 MASS FLUX AT CHF
FlGURE 9.
VIPRE-01 vs. LYNX 2 QUALITY AT CHF BWCMV CHF CORRELATION 0.3
/I
/
0.2 -
AZu
[
0.1 -
E-8 3
0.0 -
<t D
CY k
-0.1 x
>1
-0.2 -'
-0.3 -
-0.3
-02
-0.1 0.0 0.1 0.2 0.3 VIPRE-01 QUALITY AT CHF
4,,.
r.
+-.
+ ~. +
+.an,a+.--
s---
2
.~.-- - ms
_2.--
s 4
4-c
.~
~
18 3; g
a is 2
Q-O
- 4 1
-e
~
p a
_s a
CD cs 2;
e
-n a
_ o g*
La-c,2
~
-. Z
( g' w
WM a
~S 4
.O C
2
-o O
'* Q
_g-
- g:
S
- s g --
g g
o o
E
=
w o
5-o 8
T(d )~'3HD1VH3dW311TINI o
67 9
y 9,-
,,p
~
m l--
-a
i 5 -a
. Eg m
=
5 A
3 ea a.
68
?
I I
H A 0
l I
,2 L S 1
4 P
l iN0 U3 l
f E2 R
6 O
,1 1
C T
3 I
H 1
0 l
1 L
,1 M
1 J
G l
I l
l P
l A
l 0
6 M
,0 )7 1
1 1
T G
(
A E
T L
R O
E T
0 0W 2
W 1
1 B
S 0O 1
1 E
P I
I RU K
s M
E G
R I
L R
I F
A N
E O
M P R
,6 A U 9C 4
t S
R i
S D
~
ARE I
F I
P S
O I
0 I
RN
,9 S
OA A
FT B
S SN
_s O
I s-.
SC AH
,68 A'
so '-
0
- 8 0
0 0
0 0
3 2
0 1
8 6
a 6
8 6
g%h4M a
l l
Figure 13.
MAP Limits vs. Predicted Peaking l Maximm Allowable Peaking Nuclear Design l
(MAP) Limits 3-D Power Distributions Ref. 8
/
Maneuvering Analysis Allowable '%ak vs. Predicted Peak Ref. 8 Any Core Design &
Negative No-assumed Rod Insertion Margins limits are acceptable Yes Analyze specific Power Distributions in VIPRE-01 Calculated DNBR <
No Core Design &
Design DNBR Limit assumed Rod Insertion limits are acceptable Yes Revise Rod Insertion Limits or AFD Limits 70 1
ST I
M IL P
AM 4
F 1
O L
ER W
U B
G I
F KM L
AC IPYT
FIGURE 15'.
MARK-BW.RSM vs.VIPRE-01 DNBR' r
-BWCMV CHF CORRELATION w
N i
-4 4
l f
k i
t pm r
i I
4 I
P R
t i
k 5
l-rp-APPENDIX-A Safety Evaluation Report 1
~
't n
s d
a f
A-1
l l
,pm *%
e o
l UNITED STATES l
8 NUCLEAR REGULATORY COMMISSION E
WASHINGTON, D C. 20555
/
Novemoer 15. 1991 Docket Nos.
50-369, 50-370 50 413 and 50-414 Mr. H. B. Tucker, Senior Vice President Nuclear Generation Duke Power Comoany P. O. Box 1007 Charlotte, North Carolina 28201-1007
Dear Mr. Tucker:
SUBJECT:
SAFETY EVALUATION ON TOPICAL REPORT OPr NF-7001, " CORE THERMAL-HYORAULIC METHODOLOGY !! SING VIPRE-01." f TAC NOS. 72032/73765/737%
73767/73768)
The'NRC staff with the support of its contractor has reviewed Duke Power Company Topical Report DPC-NE-2004, " Core Thermal-Hydraul'; !
hodology using VIPRE-01," submitted January.9,1989, as revised and supplementea by latters dated February 22, 1990, September 14, 1990, November 29, 1990, August 29, 1991, October 16, 1991, and Hovember 5, 1991. The staff has found the topical report to be acceptable for referencing in the core thermal-hydraulic analyses for the McGuire and Catawba Nuclear Stations subject to the conditions in section 4.0 of the attachec Technical Evaluation Report.
This concludes our review activities in response to your submittals regarcina Topical Report DPC-NE-2004 accressee by TAC numoers 72032, 73765, 73766, 73767, and 73768.
Sincerely,
)M$
T4 mothy A. Reed, Project Manager Project Directorate 11-3 Division of Reactor Projects - I/I' Office of Huclear Peactor 9egulation
Enclosures:
As stated cc:
See next page A-2
~
e McGuire Nuclear Station Duke Power Company Catawba Nuclear Station cc:
Mr. R. C. Futrell Mr. Alan R. Herdt, Chief Pegulatory Compliance Manager Project Branch e3 Duke Power Company U.S. Nuclear Regulatory Comission
- Catawba Nuclear Site 101 Harietta Street. NW, Suite 2900 Clover, South Carolina 29710 Atlanta, Georgia 30323 fir. _ A.V. Carr, Eso.
North Carolina Electric "embership Duke Power Company-Corp.
422 South Ch.'ch Street P.O. Box 27306 Charlotte. -Horth Carolina - 28242-0001
-Raleigh, North Ca.olin.
7611 J. Michael McGarry, III, _ Esq.
Saluda River Electric Cooperative.
Winston and Strawn Inc.
1400 L Street,1N.W.
P.O. Box _929 Washington, DC 20005 Laurens, South Carolina 29360 North Carolina MPA-1 Senior Resident inspector Suite 600 ttoute 2, Box 179N P.O. Box 29513_
York, South Carolina 29745 Raleigh, North Carolina 27626-513 Regional Administrator, Region II Mr. Frank Modrak:
U.S. Nuclear Regulatory Comission
, Project Manager, Nid-South Area ~
101 Marietta Street, NW, Suite 2900 ESSD Projects.
Atlanta, Georgia 3032.,
-W::stinghouse Electric Corporation l MMC West Tower _ Bay 241.
Mr. Heyward G. Shealy7 '91Mo sief-P.O.; Box 355 Bureau of Radiologicas Pittsburgh,. Pennsylvania 15230 South Carolina Dept. of hesiti ano Environmental Contr,1
.: County Manager of-York County 2600 Bull Street' York County Courthouse Columbia, South Carolina
9201 York, South Carolina 29745 Ms. Karen E. Long Richard P. WilsorJ Esq.-
Assistant Attorney General Assistant Attorney General horth Carolina Dept. of. Justica S.C._ Attorney General's Office P.O. Box 629 P.O.: Box 11549 Raleigh, North Carolina 27602 Columbia,. South Carolina 29211-Mr. R. L. Gill,-Jr.
Piedmont Municipal Power, Agency Licensing
- 121 Village Drive' Duke Power Company Greer, South Carolina' 29651 P.O.-Box.1007 Charlotte, Norti. Carolina 28201-1007 1
A 3
c Catawba Nuclear Station Duke Power Company McGuire Nuclear Station County Manager of Mecklenburg County Dr. John M. Barry 720 East fourth Street Department of Cnvironmental Hu lth Charlotte, North Carolina 28202 Mecklenburg County 1200 Blythe Boulevard Charlotte, North Carolina 28203 Mr. R. O. Sharpe Mr. Dayne H. Brown, Director Compliance Department of Environmental Health Duke Power Company ano Natural Resources McGuire Nuclear Site Division of Radiation Protection 12700 Hagers Ferry Road P. O. Box 27687 Huntersv111e, No
.h Carolina 28078-8985 Raleigh, North Carolina 27611 7687 Senior Resident Inspector Mr. M. S. Tuckman c/o U.S. Nuclear Regulatory Ccmtssien Vice President, Catawba Site 12700 llagers ferry Poao Duke Power Company Huntersville, North Carolina 28078 P. O. Box 256 Clover, South Carolina 29710 Mr. T. C. McHeekin Vice President, McGuire Site Duke Power Company 12700 Hagers Ferry Road Huntersv111e, North Ca olina 28078-8985 A-4
f%
v UNITED STATES
[ *'
~ I, NUCLEAR REGULATORY COMMISSION
.,, r q
y WAsw. atom o. c. :esss y,,
'% %o.
- l 4
SAFETY EVALUATION BY THE OFi1CE OF NUCLEAR REACTOR REGU' ATION RELATING TO TOPICAL REPORT DPC-NE-700(
CORE THERMAL hydraulic HETHODOLOGY USING VIPRE 01 DURE 00VER COMPANY MCGUTFE AND CATAUBA NUCLEAR STATIONS DOCKET N05. 50-369, 50-370, 50 413, 50 414 1.0 INTFOCUCT'CH Duke Power Company (DPC) submitted Topical Report DPC-NE-2004, "McGuire anc Catawba Huclear Stations Core Thermal-Hydraulic Methodology Using VIPRE-01" in a letter cateu January 9,1989 (Pef.1), as revised by a letter dated February 22, 1990 (Ref. 2), and amended by letters of September 14,1990 (Ref. 3), Novemoer 29, 1990 (Ref. 4), August 29,1991 (Ref. 5)s October 16,1991 (Ref. 6), and November 5, 1991 (Ref. 7).
This topical report and supplemental information d:cument the use of the VIPRE-01 computtr code and the statistical core cesign (SCD) methocology for the McGuire and Catawba core thermal-hydraulic analyses.
TM VIPRE-01 comeuter code has been approved for PWR licensing calculations for he-c transfer regimes up to critical heat flux (CHF) (Ref. 8).
The SER for VIPRE-01 recuires each user to document and submit for staff review and approval a separate report describing how they intend to use VIPRE and pr;viding justification for its specific modeling assumptions, choices of particular nocels and correlations, and input values of plant specific cata such as the turbulent mixing coefficient and grid loss coefficient.
Th2 statistic 31 core design methodology is a technique that provides a more realistic assessment of nre DNB protection by statistically comoining uncertainties associated with the core statepoint carameters, code /model, ano CHF correlation.
The traditional rethort cf treating uncertainties is to assume all the core statepoint parameters are at extreme levels of uncertainty with r::spect to DNB simultaneour;ly.
The SCD methodology is proposed by DPC to replace the Westinghouse improvec Thermal Design Procedure (ITDP) methodology which also ccnvolutes the core statecoint parameters uncertainties statistically (R;f. 9).
DPC intends to use the BWCMV CHF correlation with the VIPRE-01 code for B&W Fuel Company's (BWFC) Mark-BW 17 x 17 fuel design as well as the Westinghouse 17 x 17 Optimized Fuel Assembly (OFA) design.
The BWCMV correlation has been previously approved for use with LYNXT and LYMXE comouter codes for BWFC ano Uestinghouse fuel designs with a DNRR limit of 1.21 (Ref s.10 to 12).
A-5
\\
1 2.0 STAFF EVALUATION The staff review ano evaluation of the Topical Report OPC-NE-2004 covert- (1) the acceptability of the VIPRE mocel, as described in the topical report and supplemental information, with respect to the VIPRE-01 SER requirements, (2)
OPC's statistical core design methodology which statistically combines uncertainties assoc 1ateo with the core statepoint parameters, coce/model, and CHF correlation, ano (3) use of EWCMV CHF correlation with VIPRE-01 for BWFC's Mark-BW and Westig 'use OFA fuel oesigns.
The review was perfonned with technical assist ua om International Technical Services, Incorporated (ITS).
The findings from tu !TS review 6re containea in the Technical Evaluation Roport (TER) which is attached.
The staff has reviewed the TER and concurs with its findings.
Tho NRC staff'!: prircical fincines, based on the TER ana our review, are the following:
(a)
Selections for input parameters, model assumptions and correlations for VIPRE-01 were ;ustified by showing them to be suitably conservative or showing that calculational results were insensitive to changes in the selection; as such, the VIPRE-01 model for McGuire and Catawba is acceptable; (b)
Use of the BWCPV CHF correlation with VIFRE-01 and a DNBR limit of 1.21 is acceptable because VIPRE-01/BWCMV calcuhtions were found to compare well with those of the LYNX 2/BWCMV code which has been previously approved by the NRC with a ONBP limit of 1.21; (c)
The OPC developed statistical core design methodology described in the topical report, with a statistical design limit on DNBR of 1.40, is acceptable for analysis of McGuire and Catawba reactors, and only for those reactors oue to the licensee's use of specific uncertainties ano distributions based upon plant data and its selection of statepoints used for generating the statistical design limit. The SCD is approved because, (1) it is based on analytical concepts and statistical techniques acceptable to the NRC staff, (2) sound bases were used for selection of parameters, their ranges and distribution functions for their uncertain-tics, (3) the stancaro error between DNBR values computed with the response surface ecuation and VIPRE-01 is acceptably small, (4) uncertain-ties for the coce/mocels, CHF correlation and fitting the response surface equation to the VIPRE-01 comeutations are taken into account.
3.0 CONCLUSION
The staff has reviewea the Topical Report DPC-NE-2004 and finds it acceptable for referencing in the McGuire and Catawba core thermal-hydraulic analysis, subject to the conditions celineated in Section 4.0 of the TER.
A-6
l l
l l
4.0 REFERENCES
1.
Letter from H.B. Tucker (DPC) to USHRC, "McGuire and Catawba Nuclear Stationt Core Thermal-nycraulic Methooology Using VIPRE-01," January 9, 1989.
2.
Letter from H.B. Tucker (DPC) to USilRC, "McGuire and Cata.<ba Nuclear Stations Core ThermalJydraulic Methodology Using VIPPE 01," February 20, 1990.
3.
Letter from H.B. Tucker (DPC) to USNRC, " Topical Report OPC-Ni 04,"
September 14, 1990.
4 Letter from M.S. Tuckr.an (DPC) to USNRC, " Topical Report OPC-NE-F004,"
Novemoer 29, 1990.
5.
Letter from M.S. Tuckman (CPC) to USNRC, " Supplemental Information to Assist in Review of Topical Reports DPC-NE-3000 and DPC NE-2004," August 29, 1991.
6.
Letter from H. B. Tucker (DPC) to USNRC, " Handouts Presented in the October 7 and 8,1991 Meeting with NRC Staff and Centract Reviewers,"
October 16, 1991.
7.
Letter from H. B. Tucker (DPC) to USNRC, " Final Response to Questions Regarding the Topical Reports Associated with the MICB Reload Package,"
November 5, 1991.
g 8.
Letter from C.E. Rossi (NRC) to J. A. Blaisdell (UGRA), " Acceptance for Referencing of Licensing Topical Report VIPRE-01: A Thermal-Hydraulic Code for P.eactor Cores, EPRI NP-EC11-CCM, Vols.1-4," May 1,1986.
9.
WCAP-8567-P-A, "!mproveo Thermal Design Procedure," February 1989.
10.
Letter from A. Thacani (USHRC) to J.H. Taylor (B&W), " Acceptance for Referencing of Augmenteo Topical Report BAW-10159P "SWCMV Correlation of Critical Heat Flux in Mixing Vane Grid Fuel Assemolies" May 1986," May ?2, 1989.
11.
Letter from A. Thadani IUSNRC) to J.H. Taylor (B&W), " Acceptance for Referencing of Topical Report EAW-1C159P "BWCMV Correlation of Critical Heat Flux in Mixing Vane Grid Fuel Assemblies" May 1986," February 17, 1989.
12.
" Safety Evaluation by the Office of Nuclear Reactor Regulation Relating to Topical Report EAW-10173P, Revision 2, Mark-BW Reloao Safety Analysis for Catawba and McGuire," February 20, 1991.
- ate:
,ovemoer 15, ;391 g
~
l ITS/NRC/91 1 October 1991 TECHNICAL EVALUATION:
Core Thermal-Hydraulic hethodology using VIPRE-01 Tootcal Eeport DPC NE-2004 for Duke Power Company McGuire and Catawba Nuclear Stations Prepared for Reacter Systems Branch Division of Systems Technology Office of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Wasnington, D.C.
20555 m.
l International Technical Services Inc.
420 texington Avenue New York, NY 10170 A-8
l ITS/NRC/91 1 TECHNICAt EVALUATION 0F NE CORE THERMAL-HYORAULIC METHODOLOGY USING VIPRE-01 TOPICAL _?EPORT OPC NE 2004 E.QB THE DU.(E DOWER COMPANY MCGUIRE AND CATAWBA NUCLEARl TATIONS 1.0 INTD0 DUCTION DPC NE 2004
- ated December 1988 (Ref. 1) and as revised (Ref. 2), was submitted by Duke Power Company (DPC) for NRC revie.v and approval.
' Additional information was submitted on September 14,1990 (Ref. 3), Novemoer 29, 1990 - (Ref. 4), August 29, 1991 (Ref. 5) and October 25, 1991 (Ref. 5) togather with the 10/7&B/91 NRC/ Duke meeting handouts (Ref. 7).
This topical report and the aforesaid supplemental submittals document the development of core thermal hydraulic analysis based upon the statistical core design (SC01 methodology using the VIPRE-01 computer code for the McGuire and Catawba (M/C) Nuclear Stations.
The SCD method is a thermal hydraulic analysts technique whien computes ONB margin by statistically combining core and fuel bundle uncertainties.
The submittal provides a description and justification for applying uncertainties to the DPC DNBR limits calculations using a statistical rasher than a deterministic method.
DPC intends to replace the Mestinghouse I: proved Thermal Design Procedure by the SCD methodology as part of the DNB design basis approach.
The objective of the subject topical report. therefore, is twofold: (i) to ful fill VIPRE 01 SER requirements (Ref.
8) by providing geometric representations of the core, and DPC's selection of thermal-hydraulic models and correlations for use in analysis of M/C cores in support of the SCD methodology, and- (ii) to describe OPC developed SCD methodology.
The core thermal-hydraulic methodology of this report is based on the BWCMV CNB 1
a-9
correlation wnich was previously approved for use with LYNXT and LYNX 2 for
-B&W and Wettingneuse fuel designs (Refs. 9 ano 10).
The purpose of this review, based upon a review of the submitted materials (Refs. 1-7), is to determine acceptability of (i) conformity of the DPC topical report and suoplemental information (Refs.1-7) to the VIPRE 01 SER rQquirements (Ref. 8),
(ii) use of the EWCMV critical heat fl'Jx (CHF) correlation with VIPRE 01, and (iii) DPC's SCD methodology.
Therefore, this review of :PC's VIPRE-01 core odel was concucted in the context of its use in support of the SCD methodology.
The SCD methodology was reviewed for acceptability of the generic methodology and its application to the M/C reload analyses.
McGuire ano Catawba Nuclear Stations, each having two Westinghouse units, are assumed by OPC to be identical for the purpose of core thermal-hydraulic calculations. The analyses presented in the submittals were performed for a core containing all B&W Mark-BW fuel assemelles, except as noted for transition cores.
2.0
SUMMARY
OF TOPICAL DEPORT and SUPPLEMENTS The topical report DPC NE-2004 and its associated submittals (refered to as DPC NE-2004) document descriptions of DPC's VIPRE 01 steady-state models for roload type analysis with all BAW Mark-BW fuel assemblies for McGuire ano Catawba Nuclear Stations' (all Westinghouse plants).
DPC's objective in submitting the topical report was twofold: to fulfill conditions required by the VIPRE-01 SER, and to document DPC deve_looed statistical core design mQthodology as described below.
2.1 VIPRE 01 Comcuter Code VIPRE-01 has been previously -eviewed and approved for application to pressurized water reactor (PWR) plants in steaoy-state and transient analy:;es with heat transfer regimes up to critical heat flux (CHF).
The NRC safety evaluation report (SER) on VIPRE-01 (Ref. 8) includes conditions requiring 2
A-10
racn user to document ano submit to the NRC for approval its procedure for using VIPRE 01 and provide justification for its specific modeling assumptions, choice of particular two-pnase flow mooels and correlations, heat transfer correlations. CHF correlation and DNBR limit and input values of plant specific data such as turbulent mixing coefficient and grid loss coefficient including defaults.
The conservative core model, use of certain thermal hydraulic correlations, and other key input selections were justified through a series of sensitivity studies.
Since DPC's intended use of VIPRE-01 is to perform ec-
- hermal-hydraulic calculations in support of the SCD methodology, wherever
-opplicable, the core conditions used in the VIPRE 01 sensitivity analyses were selected from the set of conditions used to covelope the response surface for the SCD analysis.
Because the methodology is based on use of the BWCMV CHF correlation with VIPRE-01, DPC qualified use of EWCMV correlation with VIPRE 01 by predicting a set of data points from the original set of BWCMY CHF data base that cover the ranges of anticipated operation, obtaining a DNBR limit of 1.21.
BWCMV is approved for use with LYNXT and LYNX 2 computer codes with the DNBR limit of 1.21 (Refs. 9 and 10).
2.2 Statistical tore Desian Methodoloav The traditional method for accounting for the design and modeling uncertainties that enter into the determination of a ONBR assumes that key input parameters to the core thermal-hydraulic code are simultaneously at their worst level of uncertainty.
The methodology described in the DPC Topical Report OPC NE-2004 assumes that, while the input parameters are cccasionally at their worst case values, the input uncertainties are indepenoent and it is highly unlikely that all the input parameters will take on_their worst case values simultaneously.
Therefore, the application of the SCD method differs from previous techniques in that the thermal-hydraulic
-limit analyses are performed by statistical analysis of a series of computations " perturbed" from a "Zero" point computed at nominal plant 3
A-ll
(
l l
l conoitions.
DPC has appiled the SCD method to simulate the direct comoutation of DNBR with V!f..E 01.
The SCD methodology statist 1: ally :mbines uncertainties associateo with key parameters used in determination of the DNBR.
In oroer to ;erform the recuired statistical combinat1:n of the various input uncertainties, the DPC SCD method employed a non linear response surface model (RSM).
'he response surf ace equation, variables, and the s6lection of the test cases which were useo :3 determine the coeff':1ents that appear in the response surface equation were described. The topical report describes the proc 9ss by which the CNBR is determined with the SCD methodology and ultimately used in combination with other operating limits.
The statistical DNBR lirait (SDL) to replace the traditional CHF correlation limit was determined for the cases inalyzed in the submittal:
the calculated SDL was found to be 1.40 for tne M/C Mark-BW core.
The range of applicability of the SCD method (therefore the RSM) is defined by the range of values from which the composite design points, used to determine the RSM equation, are selected.
For statepoints which fall outside of the SCD range but which must nevertheless be analyzed for certain transients, DPC developed a simplified method which used VIPRE 01 cirectly and avoided use of the RSM.
- or these cases, values of the variables were nenerated by use of a Monte Carlo, metW according to the uncertainty distribution for each state variable.
The DNB was computed for each such set using VIPRE-01.
Statistical analysis was performed of the set of DNBRs so computed and the computec DNBR is com:ared against the SDL for acceptability.
3.0 EVALUATION 3.1 VIPRE Model Descriction The M/C core models discusseo in the topical report were assumeo to contain all B&W Mark-BW fuel.
.t A-12 l,
s- ~ - " - ' - - - ^ ^ ^ ' -'-
I l
3.1.1 Core Nodalization In develeping the core models. DPC assumed a 1/8 core symmetry with the hot assembly located in the center of the core.
The set of thermal hydraulic models anc correlations useo by OPC in the nodalization sensitivity studies were those which OPC intenos to use in future licensing analysis.
These models and correlations were found to yield acceptably conservative results.
3.1.1.1 Dadial Nodino Ser,sitivity i
A parametric study was performed to determine the sensitivity of predicted DNBR to the subchannel model :1:e.
The thermal-hydraulic calculations were performeo for three different core succhannel models using steady-state conditions.
The coarse channel model was found to yield acceptably conservative MDNBRs.
Therefore. DPC intends to used the coarse channel medel for steady-state Mark-BW thermal hydraulic analyses for McGuire and Catawba Nuclear Stations.
3.1.1.2 axial Nodino Sensitivity A sensitivity analysis for axial node length was performed with the coarse core channel model using two different axial node lengths under two different core conditions.
Both of these node lengtns correspond to the range of code developer's recommended values.
The results indicated that the agreement was adequate enough that the large size noding is acceptably conservative.
3.1.2 VIPRE-01 Geometric al 'nout Data OPC's approach to generation of incut to the VIPRE-01 code was reviewed for acceptability.
No review was concutted of the input data in comparison to the actual ;nysical geometry.
5 A-13
3.1.2.1 Active ruel teneth For B&W's low densification fuel, the amount of fuel densification is off set by the fuel thermal expansion.
OPC chose to use the cold nominal active fuel length for calculation since this is more conservative.
+
3.1.2.2 Soacer Grid Form Coefficients The vendor cetermined coefficients based on test data were used to acco for the~ hydraulic loss caused by the variation in flow area and turbulence at each spacer grid.
3.1 ?. 3 sore Bvoass Flow Since the _ bypass flow depends on the number of control rod and burnacle poison rod assemblies in the core, this is a cycle dependent parameter.
Therefore, the core bypass flow data used in the analysis should be based on a bounding value or on cycle specific data.
For the purpose of generating the response surface model using the SCD methodology presented in the topical report, a 6% core bypass flow was kssumed and thus the resulting core inlet flow was 94% of design -flow of
'382.000 gpm.
3.1.2.5 Inlet now Distribution CHF is decreased and the probability of CNC is enhanced if flowrate is recucco cue-to a
flow maldistribution.
The use of 5%
inlet flow maldistribution to the-hot assembly, while the remaining flow was distributed between the ~ outermost assemolies, yielded slightly more conservative ONBR
. prediction than did a uniform inlet flow distribution.
Pri'or to submittal of a Catawba reload report using the SCO methodology, the licensee will evaluate the impact of RCS flow anomaly observed at Catawoa 6
A-14
}
=
that resultea in a more severe core inlet flow maldistribution.
3.1.2.7 Radial Dower Distribution The reference radial power distribution was used to calculated tha core ONB limit, but the F(AI) portion of the OTDT trip function was determined to ensure DNB protection for any core power distribution.
The hot assembly power distribution was assumed to be relatively r~lat to minimize flow redistribution.
The reference peak pin power factor, for the power levels below 100?. power, was adjusted to determine the core DNB limits.
3.1.2.8 Arial-Dower Distribution A symmetric choppea cosine was used for the ax1t. oower distribution with a peaking factor of 1.55.
A routine has been added to the V! PRE 01 code to generate axial power shapes with inlet, symmetric, or outlet peaks.
3.1.2.9 Hot Channel Factor The hot channel factor F[g used for the McGuire/ Catawba analysis was 1.03 and accounted for the allowance on enthalpy rise to account for manufacturing tolerances.
3.1.2.10 Numerical Solution Technioue r
The RECIRC solution method was used for the McGuire/ Catawba analyses presented in the submittal and will be used for all of the steady-state core thermal-hydraulic analyses.
3.1.3 VIPPE-01 Corralations
'VIPRE-01 requires emoirical correlations for the following models:
a.
turbulent mixing b.
two-phase flow correlations (subcooled and saturated voic, and 7
A-15
void-quality relation) c.
critical heat flux 3.1.3.1 Turbulent Hirina i
t The lateral momentum equation requires two parameters: a turbulent momentum factor (FTM) and a turbulent mixing coefficient.
The turoulent momer.tum factor (F1M) cescribes the efficiency of the momentum mixing: 0.0 indicating that crossflow mixes enthalpy only; 1.0 indicating that crossflow mixes enthalpy and momentum at the same strength.
OPC selected a more realistic value for FTH: however, the MONBR was found through a sensitivity study to be insensitive to this parameter.
Since the turbulent mixing coefficient determines the flow mixing rate, it is an important' parameter.
A mixing coefficient was determined based upon 'osts performed by Westinghouse.
Although the grid spacing in the test was different from that of the MARK-BW fuels, since the mixing coefficient increases as grid spacing decreased, the value obtained by Westinghouse snd used by OPC is conservative.
-3.1.3.2 Two-Phase Flow Correlations For subcooled and bulk void correlations, a sensitivity study using three different combinations of subcooled void and bulk void correlations was performed for two sets of steaoy< state core conditions.
The results
-indicated that the use of OPC -selected combination of correlations in conjunction with Columbia /EPRI two-phase friction multiolier predicted conservatively computed DNBR relative to other combinations of correlations.
DPC intends to use this combination in McGuire and Catawba analysis.
l This is consistent with the VIPRE 01 SER findings.
3 A-16 w-
~
-gy-ry,-'*f--w-w e-u r-e,,-
1r g..
1 l
l l
3.1.3.3 Friction Pressure loss Selection of (i) the axial friction factor in the Blasius friction pressure factor sna (ii) the EPRI two onase friction multiplier was based on l
sensitivity studies which resulted in conservative prediction of MDNBRs.
I 3.1.3.3 BWCMV Critical Heat nux Correlation Use of BWCMV CHF correlation with the LYNX 2 and LYNXT codes has been approved by the NRC with a DNBR limit of 1.21.
DPC provided qualification of its use with the VIPRE-01 code based upon prediction of CHF data points from each test section included in the BWCMV data base to cover the range of anticipateo operation.
OPC cotained comprable results from these calculations when compared with LYNX 2 results and determined a DNBR of 1.21.
Therefore use of BWCMV CHF correlation is acceptable for use with VIPRE 01 with a DNBR limit of 1.21.
3.2 McGuire/ Catawba Core Thermal-Hydraulic Analyses Core thermal margin and allowable operating limits define the safe operating region (which DPC defines as preventing excessive power, coolant temperature and pressure, or any combination thereof).
The primary objective of DPC's use of this approach is to assure the ability to maintain coolability of the core to prevent fuel damage during Condition I and II events.
Three core protection design bases were used:
1.
Departure from Nucleate Boiling (DNB) design basis:
2.
Fuel temperature design casis; 3.
Hot leg boiling limit.
The DNBR limit determined by use et SCD method is a statistical DNBR limit (SDL).
The cesign DNBR limit (DDL) Inu;rpoiates additional margin to the SDL to allow for adverse impacts on DNB from other parameters, such as the RCS flow anomaly and transition core effects.
The core DNB limits are the 9
A-17
comoinati:ns of power, coolant inlet temperature and pressure at which the MDNBR equals the design DNBR limit.
Due to the range of applicability of.he BWCMV correlation, the cuality at the point of MDNBR is also limiteo to R..
{
The core CNB limits were calculated using the coarse channel model referred to in a previous section of this report.
The use of the statistical core design (SCD) methodology as part of the DNB design basis approach was proposed by DPC to replace the Westingneuse Improved Thermal Design ProcPdure Which considers Core I/H analysis parameter uncertainties statistically.
The description of the SCD method is given in the next section.
3.3 Etat 1stical Core Des 1cn Methooolcov The traditional method for accounting for the design and modeling uncertainties that enter into the determination of a DNBR assumes that key input parameters to the core thermal hydraulic code are simultaneously at their worst level of uncertainty.
The proposed DPC methodology assumes that, while the input parameters are occasionally at their worst casa values, the input uncertainties are independent and it is highly unlikely that all the input parameters will take on their worst case values simultaneously.
Therefore, t.ie application of the SCD method differs from traditional techniques in that the thermal-hydraulic limit analyses are performed by perturbation analyses from a
"Zero" point computed at nominal plant conditions.
DPC has applied the SCD method to simulated direct e.omputation of DNBR with VIPRE-01.
The SCD metnodolocy statistically comoines uncertainties associateo with Key parameters used in determination of the DNBR.
In order to perform the required _ statistical combination of the various input uncertainties the DPC
.SCD method employed a non-linear response surface model (RSM).
The-response surface equation is an equation for MDNBR as a function of seven state variables.
A probability distribution of MDNBR as a function of 10 A-18
those variables can be obtained if one knows the distribution of probabilities of the seven variables.
A probability distribution is established for each of the seven variables with the nominal state conoitions as the center and with normal distributions for core power, core flow, radial power peaking factor and axial pewer peak and bounded uniform distributions for core outlet pressure, axial location of the axial peak and core inlet temperature.
The core flow uncertainty is comprised of two parts:
measurement and bypass flow uncertainties.
Similarly three components contribute to determination of the radial power peaking factor.
In addition three other variables are assumed to impact computation of DN8R Je/model uncertainty, CHF correlation uncertainty, and error associated w.
the fit of the response surface equation to the VIPRE-01 computations used to develop the RSM equation).
The uncertainty due to code /model includes a difference
- etween VIPRE-01 and LYNX 2 EWCMV results and the difference in the DNBR results due to the difference between sitas of the core channel models.
A Monte-Carlo computation is used to select sets of values of each of the seven state variables at random (weighted by the distribution functions) and a
. resultant MDNBR is computed from the response surface equation.
The SCD method is used to determire an overall DNBR uncertainty.
The core thermal-hydraulic analyses are performed using nominal values for the parameters that are treated statistically; all other input parameters are assumed at their conservative values.
3.3.1
. Selection of Parameters Seven parameters were identified as those which would significantly impact the - calculation of DNBR: (1) core power, (2) core flow, (3) core outlet
!) core inlet temperature, (5) peak racial power factor, (6) axial
- pressure, peak, and (7) axial location of the axial peak.
These variables, assumed independent, define a core statepoint on the response surface.
Since these parameters vary during reactor operation, parameter ranges were developed which would bound the values that would be expected to be encountered ta typical M/C T/H analyses.
These ranges are plant and transient set dependent.
11 A-19
i 3.3.2 Resoonse Surface Model Assuming independence of the seven selected variables. the equation of the response surface is defined as a quadratic function of the statepoint variables including cross. terms.
The coefficients in the RSM equation were determined by performing a least-squares fit of the response surface equation to a set of values of MDNBR actually calculated with VIPRE 01 using a OPC eight channel eighth-core symmetric VipRE-01 model.
The basic statepoints at which MDNBR was calculated for the purpose of computation of coefficients were determined by a Central Composite Design technique.
Nearly one half of the 143 possible-points in a full central composite design were selected to i
include all extreme points but one plus a series of other points to enable the response surface to cover a wider range of parameter variations including those expected to produce the worst MDNBR.
The standard error between the RSM predicted and VIPRE-01 calculated MDNBR values for the set of statepoints considered was found to be acceptable.
3.3.3 Statistically Treated Uncertainties t
in order to statistically combine the effect of the uncertainties of the parameters DPC determined the uncertainties, uncertainty distributions and the uncertainty standard deviation.
A probability distribution was establishec ar each of the seven variables with the nominal state conditions as the center and with normal distributions for core pcwer, core flow, radial power peaking factor and axial power peak, and with bounded uniform distributions for core outlet pressure, axial location of the axial peak and core inlet temperature.
The core flow uncertainty is comprised of two parts:
easurement anc bypass flow.ncertainties.
Similarly three components contribute to determination of the radial power peaking factor.
-OPC's rational for assignment of uncertainty distribution was that a normal distribution was assumed wnen the uncertainty was due either to measurement uncertainty or a known statistical uncertainty distribution.
Whenever such assumption could not be reasonaoly made. DPC chose the conservative approach 12 A-20
of assuming a uniform distribution with estimated reasonable upper and lower bounds.
The licensee stated in the topical report that the uncertainties and distributions will be justified on a plant specific basis in the reload report for the first application of this methodology.
3.3.4 Procacation of Uncertainties In order to combine the uncertainties to compute an overall DNBR uncertainty, a Monte Carlo method analysis is performed using the distribution of uncertainties defined with each variables.
A probability distribution is established for each of the seven variables with the nominal state (loss of flow) conditions as the center.
A Honte-Carlo computation is used to select sets of values at random (weighted by the distribution functions) and a resultant MDNBR is computed for each _ such set from the response surface equation.
Statistical analysis is then performed on tht set of MONBRs so generated.
This process is repeated for ten core statepuints that cover the range of conditions considered when determining core DNB limits Ed the limiting Condition 11 DNB transient in order to maximize the coefficient of variation resulting from the propagation of uncertainties.
In addition three other factors are assumed to have the dominant impact on computation of DNBR (code /model uncertainty, CHF correlation uncertainty, and orror associated with the fit of the response surface equation to the VIPRE-
.01 computations used to develop the RSM equation).
Finally, the statistical DNBR limit (SDL) to replace the traditional CHF correlation limit is determinea.
The statistical design limit is determineo from the largest coeffic':ent of variation based on the DNBRs computed by the Monte Carlo computations referred to above which avoid DNB at a 95Y.
probability /95Y. confidence level.
For the cases analyzed in the submittal the calculated SDL is 1.396 for the M/C Hark-BW core.
The resulting 50L is 1.40.
'3 L
A-21
i i
3.3.5 Sirolified SCD Methodoleov The range of applicability of the SCD methoo (therefore the RSM) is defined by the range of values from whic h the composite design points are selected.
Therefore, for this methodology to be generally applicable, the coolposite design points must be selected so that values expected during any expected transient would fall within the envelope defined by these variables.
For the statepoints which fall outside of the SCD range but which must nevertheless be analyzed for certain transients, DPC developed a simplified method which
)
used VIPRE-01 directly and avoided use of the RSM.
For example, the lowest value of RCS flow used in the composite design point is 80% of nominal core inlet flow which is 359,080 gpm.
However, during at least three licensing type transients (loss of forced flow, feeowater line break and uncontrolled control rod assembly withdrawal from a subcritical or low-power condition), the flow is expected to be less than 80% when MDNBR Therefore, for these cases, the RSM as developed is inapplicable.
occurs.
The OPC developed simplified method is generally usable and will be applied to conditions such as:
(a) a significant change is made in the fuel assembly design; (b) a new or revised CHF correlation is developed: and (c) operating conditions falling outside of the range of conditions listed in revised Table 8 (Ref-. 5).
The simplified method bypasses the RSM by directly computing DNBR with the VIPRE 01 code based on the values for the seven variables generateo by the propagation of uncertainties througn the use of the Monte Carlo method.
An SCD limit is determined for each case as before and compared against the SDL.
DPC stated (Ref. 5) that a submittal will be made to the NRC for review if a new SDL is required to be calculated unoer the following conditions:
i I
14 i
l t
A-22
.. ~..
a 11 A completely new fuel design would require developreent of a new SDL.
However, development of a new design feature, such as a new spacer grid design, may only require propagation of the most limiting statepoint to validate the existing limit.
The NRC would only be notified if the design changes result in the calculation of an SQL > 1.40.
2)
Development of a new version of the BWCMV correlation or use of a different CHF correlation would require calculation of ntw SDL and a submittal to the NRC.
3) emoutation of the SDL for any statepoint outside of the range given in revisec Table 8 would be propagated using VIPRE 01 directly as was done for FSAR transient 15.4.1.
A submittal would be made to the NRC only if this propagation results in an SDL >
1.40.
3.3.6 DNBR pena, ties The SDL for McGuire/ Catawba was computed to be 1.40.
The design DNBR limit (DDL) was selected to be 1.55, allowing 10.77. DNBR margin.
Against this margin, several penalties'were applied:
a transition core penalty against OFA fuel of 3.87.;
a.
b.
a total instrumentation DNBR penalty of 1.9?.; and c.
a rod bowing penalty of 3.57..
Therefore, the total DNBR penalty against OFA - fuel is 9.27. leaving a 1.5Y.
margin to DDL.
3.3.6.1 Transition Core Analvsis The transition core effect was analyzed using VIPRE 1 by modeling both the 15 A-23
. so
Optimized fuel assemolies (OFA) and Mark-BW fuel.
The BWCMV CHF correlation t:as used for this analysis.
Although the data base on which the development of the correlation is -based cover the range of parameters of 0FA fuel, it does not specifically include OFA data.
The application of the BWCMV orrelation for analysis of 0FA fuel has been approved by the NRC (Ref.11) and was further validated by comparison of VIPRE/BWCMV to LYNX 2/BWCMV results.
The power peaking associated with each fuel type is calculated and compared
-against MAP limits.
Using the simplified SCD method, the SDL was computed for OFA fuel and Mark-BW fuel and was bounded by the SDL of 1.4.
4.0 Conclusions We find that the subject topical report, together with OPC responses, contains s sufficient information to satisfy-the VIPRE 01 SER requirement that each. VIPRE-01 user, submit a document describing proposed use, sources of input variables, - and selection and justification of correlations as it relates to use by OPC for development of statistical core design methodology.
Acceptability.of the OPC'. VIPRE 01 model for steady-state application to analysis of McGuire and catawba Nuclear Stations is based upon selection of m:dels/ correlations supported by the sensitivity study results submitted.
Should DPC change any of these items, DPC should submit justification;for the change to the NRC for approval.
Use; of the BWLF' M 1rrelation with VIPRF 01 is found acceptable with a DNBR 1imit of 1; M We further-' find that-the' manner in which the code is to be used for such analyses,' selection of nodalization, models, and correlations provides, except as listed below, adequate assurances of conservative results and is therefore acceptable.
16 A-24 2
i The following limitations and restrictions are recomended regarding the use of DPC's VIPRE-01 model and its associated statistical core design methodology presented in P -NE 2004 and its supplemental materials for analysis of McGuire and Catawoa Nuclear Stations:
1.
The OPC developed statistical core design methodology, as described in the submittal, is a generic cathodology and is conceptually acceptable and generally applinble to other PWR plants; however, the approval we recomend at this time is for only McGuire and Catawba Nuclear Stations due to OPC's use of specific uncertainties and distributions based upon plant data and its selection of statepoints used for generating the statistical design limit.
2.
Whenever conditions provided in response to NRC question 4 in Reference 5 are present, either the response surface must be re-evaluated or the
" simplified method" must be used.
The licensee is further required to make a submittal to the NRC for review if a new SDL is calculated as a result of conditions outside the range of parametars set forth in revised Table 8 of Reference 5.
3.
Core bypass flow is cycle dependent.
DPC will veri fy, in future applications, that its use of a particular core flowrate resulting from a bypass flowrate for that cycle is bounded by the range of values used in the subject topical report.
Otherwise, OPC will reassess the need for regeneration of a new response surface.
5.0 :EFEoENCES 1.
" Duke Power Company McGuire and Catawba Nuclear Stations Core Thermal-Hydraulic Methodology Using VIPRE-01," DPC-NE-2004, December 1988.
2.
Letter from H.B.
Tucker (DPC) to USNRC, "McGuire and Catana Nuclear Stations Core Thermal-Hydraulic Methocology Vising VIPRE-01,' February 22, 1990.
3.
Letter from H.B. Tucker (DPC) to USNRC, " Topical Report JPC-NE-2004,"
17 A-25
,.w_
4
1 September 14, 1990.
4.
Letter from M.S. luckman (DPC) to USNRC, " Topical deport DPC NE-2004 "
Novemoer 29, 1990.
5.
Letter from M.S.
Tuckman (DPC) to USNRC, " Supplemental Information to Assist in Review of Topical Reports OPC-NE-3000 and OPC-NE 2004," August 29, 1991.
6.
Letter from H.B.
Tucker (DPC) to USNRC, " Handouts Presented in the October 7 & 8, 1991 Meetin9 with NRC Staff and Contract Reviewers,"
October 16, 1991.
7.
Letter from H.B.
Tucker (OPC) to USNRC, " Final Response to Questions Regarding the Topical Reports Associated with the MIC8 Reload Package,"
Novemoer 5, 1991.
8.
Letter from C.E. Rossi (NRC) to J.A. Blaisdell (UGRA), " Acceptance for Referencing of Licensing Topical Report VIPRE 01: A Thermal Hydraulic Code for Reactor Cores. EPRI NP-2511-CCM, Vols.1-4 " May 1,1986.
9.
Letter from A.
Thadani (USNRC) to J.H. Taylor (B&W), " Acceptance for Referencing of Augmented Topical Report BAW-10159P "BWCMV Correlation of.
Critical Heat Flux in Mixing Vane Grid Fuel Assemblies" May 1986 " May 22, 1989.
10.
Letter _ from A.
Thadani (USNRC) to J.H. Taylor (B&W), " Acceptance for Referencing of Topical Report BAW-10159P "BWCMV Correlation of Critical Heat Flux in Mixing Vane Grid Fuel Assemblies" May 1986," February 17, 1989.
11.
Letter from USNRC to N.B. Tucker (DPC), " Safety Evaluation by the Office of Nuc', tar Reactor Regulation Relating to Topical Report BAW-10173P, Revision 2 Mark-BW Reload Safety Analysis for Catawba and McGuire,"
February 20, 1991.
18 A-26
AFPDiDlX B Responses to Request for Additional information September 14, 1990 D
A B-1
II I
O.ht 1%91 ComtD v W 0 T&"
f f) fa.) f })q
- V'* stat'nt Chanotte. A C,%212
%rau! P" suction
- tot *Tif31 Ch udKE POWER September 14, 1990 U. S. Nuclear Regulator / Cc::: mission ATT!h Document Control Desk Washington, D.C.
20555
Subject:
McGuire Nuclear Station Docket Numbers 50-369 and -370 Catawba Nuclear Station Docket Numbers 50-413 and -414 Topical Report DPC-NE-2004 By letter dated January 9, 1989, Duke submitted the subject Topical Report for review.
By letter dated August 2, 1990, the NRC staff requested additional information.
Attached are responses to the 18 questions transmitted by that letter.
Please note that this submittal contains proprietary information, pursuant to 10 CTR 2.790, which should be withheld from public disclosure.
An affidavit which supports the proprietary designation is included in the January 9. 1989 submittal.
If thete are any questions, please call Scott Gewehr at (704) 373-7581.
Very truly yours, af
- k. /] h' 335 Hal B. Tucker SAG /232/lcs B-2
1 i
U. S. Nuclear Reguissory Commission September 14, 1990 Page 2 cc t.
Mr. Stewart D. Ebneter, Regional Administrator U. S. Nuclear Regulator *, Commission - Region II-101 Marietta Street, Suite 2900 Atlanta Georgia 30323
)
Mr. Tim Reed,-Project Manager Office of Nuclear Reactor Regulation U. S. Nuclear Reactor Regulation Washington, D.C.
20555 Mr. W. T.' Orders NRC Resident Inspector Catawba Nuclear Station Mr. P. K. VanDoorn
.NRC Resident Inspector McGuire Nuclear Station Dr. Kahtan Jabbour, Project Manager Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission Washington, D.C.
20555 B-3 m
-a
1.
Tor-HCGuire and Catawoa application, specify the user-deterr. red input used in the VIPPo-01 models for: (a) heat transfer
. correlations, (b) UPPLOW versts RECIRC cptions, and (c) damping facter (D.).
Also provice bases or Oustifications for these selections.
i al. As stated in Section 5.0, Only the steady-state core thermal-l hydraulic analyses that ensure that the DNB design basis is met are
-discussed in this report.
Heat transfer correlations are used in VIPRE-01 to obtain the heat transfor solution only when the conduction model is used and the conduction mocel is not used for steady-state analyses.
The followiny heat transfer correlations are input since rome of the flow correlations maxe use of the neat transf er c0rrelations:
EPRI single-phase f crced convection cor'flation Thom subecoled nucleate boiling correlation Thom saturated nucleate boiling correlation BWCMV CHT correlation defining the peak of the boiling curve j
[
b)
The RECIRC solution optien will be used for all of the steady-state core thermal-hydraulic analyses discussed in DPC-NE-2004.
The VIPRE-01 l
SER, ref.
1, concluded that "the UPFLOW and RECIRC options are properly implemented and these solution techniques are acceptable for licensing 1
calculations".
l l
L
-c).The default.value (0.9) is used for the damping factor applied to L
'the tentative. axial flow and crossflow.
As stated in the VIPRE-01 SER,
?ref.
1, if a convergence problem occurs, the calculation would stop and
'sufficent information would.be printed to allow the user to determine the state of convergence.
Therefore, the use of damping factors and-
'their~effect on numerical stability is not a concern.
.Roterence-1.
Letter from C.
I.
Res :
MRC)- to J. A.
31sindell (UGRA),
" Acceptance for Referencing'of Licensing Topical Report, VIPRE-01:
L
'A Thermal-Hydraulic Analysis Code for Reacter Cores", EPRI-NP-2511-CCM, Vol, 1-5,-May 1, 1996.
o l
{
r l-B-4.
.m Justify that the gener:.: 1/8 core VIPRE model with the smallest number of channels and the assumed core center hot assembly
~.
location is conservative and representative of the future McGuire/ Catawba reloac core designs, including the mixed fuel designs.
The-genc.ric 9 channel 1/8 core.VIPRE-01 mocel described.in DPC-NE-2004
-is used to determine the regions of safe operation in terms of power level, reactor-coolant temperature and pressure, and power uistributien.
The allowable-space is first determined in terms of power level and reactor coolant pressure anc temperature based on a reference power distributien which is discussed in Section 5.5.1 and shown in Figures 5 and 6.
The reference hot assembly pin power distribution, Fig.
5, is relatively flat : ) con **rvatively minimize the benefits of crossflow (refer to the _ response for Ouestien 6).
The lumped channel power shown in Fig. 6 is based en the relatively flat assembly power distributien shown in. Fig. 2 to eliminate any DriBR impact due to assembl' power.
Af ter ' calculating the allowed operating ' space using the reference peaking, tne cc:mnations Of racial anc axial peaking are ceterminea wnich provide equivalent DNB protection.
These-limits are known as Maximum Allowable-Peaking (MAP) limits and they are compared with cycle-specific' predict ed. peaking 1.. a maneuvering _ analysis, ref.
1.
The peak pin powels calculated in a maneuvering analysis may be predicted in any core _ location (1/4 core symmetry _ is assumed).
The peak pin powers are compared with the MAP limits' conservatively based on the power distributions discussed above.
If any negative peaking margins are determined (predicted peaking greater than the MAP limit) during a maneuvering analysis, the MDNBR will be calculated using the limiting predicted power _distribut4on, The predicted radial power distribution and axial power profile is input directly into VIPRE-01.
Mixed core analyses are addressed in the response to question 5.
i l
L l
Reference 1
Nuclear Design Methodology..
C.
6 Opprating I.imits of l
Westinghouse Reactors, :Pr E-20ilP, April 1988.
l l'
-5 Y
T r~-
m-
3.
Provide eith 11 comparison to experimental-data on pressure
.-drop, or (5 asults of sensitivity studies to demonstrate that the-use of 11us fricti:n pressure f actor expression and the EPRI two-phase.._ction multiplier yield conservative results f or both single and two-phase flow.
A sensitivity study was performed to select the axial friction f actor.
The default smooth tube friction factor was compared with the following correlations:
f = 0.32Re*"
default f = 0.184Re**'
f = 0.092Re
f'= 0.368Re**'
The -sensitivity study results are given in Table 3-1.
The MDNBR and local hot channel conditions for the first two correlations are nearly identical.
Halving or doubling the
- ading coefficient of the correlatien. yields the expectec decrease and increase in pressure drop and a fairly significant change in the local conditions and MDNBR.
Doubling the lending coefficient, although yielding a conservative MDNBR, unrealistically increases the pressure drop across an assembly.
. Based on the sensitivity study results given in Table 1 and the
- recommenLation in Vol. 4 of the VIPPI-01 manual, the default Blasius friction factor will be used, e
The EPRI two-r::u se friction multiplier was selected based on the sensit 'rity s : a:"
,asults given in Table 3-2.
The EPRI two-phase d
friction mult4.;.
yielded conservative MDNBRs.
1 B-6
- Table-3-1.
Sens10 vity to Axial Friction Fact 0:
Hot Channel Results 1
Fr10:10n Mass Velocity Exit Case-Facter MDNBR 3 MDNBR Ouality AP 1
0.32Re-3'S 1.467 1.6621 19.68 18.78 3.184Re'88 1.485 l' 6674 19.72 19.36 0.092Re-43 1.526 1.6644 18.96
- 6.37 3.3 6BRe*4 #
'l.450 1.6554 20.38 25.37 2
- 0. 3 2Re-o. rs 1.487-1.6657
.7.48 19.52 0.18 4 Re-* 3 1.482 1.6677 17.58 20.11 0.092Re-43 1.544 1.8813 16.42 17.08-
- 0. 3 6 8 Re.s.2 1.422 1.6269 18.74 26.32 Case Pevar Pressure i
'l 2
.5.7
Table 3-2.
Sensitivity to Two. Phase Friction Multiplier Two-Phase Mass '/elocity Exit Case
- >.u l t i c l i e r MDNBR
@ MONER ogi a l i t */
1 IPRI 1.187 1.6621 13.68 Homogeneous 1.506 1.7020 19.32 2
IFRI 1.487 1.6657 17.48 Mc=ogeneous 1.506 1.7085 17,04 The Levy subcooled void model and Zuber-Findlay bulk void model were usec (see Secti:n 4.1.7.3 in DPC-NE-2004).
h._
g 4
9 6
l l'
l l
)
4 Previce comparisons
- Mark-BW confirmatory CHF tests that demonstrate that these results are conservatively predictec by the BWC:f/ C3F C0rrelation.
Given.nn fact that the DNBRs calculatec hy BWCMV/V! PRE-01 and BWC:f// LYNX 2 disagree by as muen as 10 % for sc=e data points (see Figure 7), explain how the ONBR will always ce conservative.
Explain now the value of the percent difference between the DNBRs calculated by VIPRE-01/BWCMV and LYNX 2/BWC:f/
indicated in Section 4.1.7.4 is obtained.
Tho _VIPRE-01/SWCMV measured-to-predicted (M/P) CHF values for the 84 "igh ! bubbly) flow confirmater*/ test points are given in Table 4-1.
T '. e measurec CHT values are plottec in Figure 4-1 versus tne VIPRE-01 predicted CHT values (BWCMV).
The VIPRE-01 results are nearly identical to B&W's LYNX 2 results, ref.
1, as shown below:
Mean M/P Std. Devi ation VIPRE-01/BWCMV LYNX 2/BWC:f/
~
The V! PRE-01 resulte snow, as cc the LYNX 2 results, that the BWC:f/
correlation can ce conservatively used for Zircaloy mixing vane grid fuel with a CNBR limit of 1.01.
Secti:n 4.1.7.4 should state that the mean percent difference between the VIPRE-01 and LYNX 2 BWCMV CNBRs is less than 0.5 %.
The VIPRE-01 DNBRs f0r the 124 data points avaluated are given in Table 4-2 which also lists the percent difference between the VIPRE-01 and B&W's LYNX 2 results for eact test run.
The maximum percent difference between tne VIPRE-01 and LYNX 2 results is[f [
As discussed in Section 6.4, a
total code /model uncertainty o was included in the calculation of the statist: cal DNBR limi';. along with the BWCMV correlation stancard deviati n cf 10.2 s
?,eference BAW-10159-A, BWCMV - Correlati:n of Oritical Heat Flux in Mixing Vane Grid Fuel Assemblies, Babecck & Wilecx, July 1990.
B-9
5 i
1
- Table 4-1.
VIPPI-01/BWCW COnfir=atory CHT Test Results VIPP2-01 CHF Test !5
-Measuree Precie en M/p 9
10 1
11 13 14-16 17 18 20 21 22-24 25 26 27
'28 29 30 31' 32 33 34 35 36 37 38 39-40 42~
43 44
- 45-46 47.
48 49 50 51 54 55 56.
5 y.
5S 59 60 61-62 63 64 B-10
Table 4-1.
7!PPI-01/3WC:n C:nfir:.atery CHF Test Results (COntinuco)
VIPPI-01 CHT Test ::
Measuram Precieted M/P 65 66 42 69 69 70 71 72 73 74 c.
M 73-79 80 31 82 83 84 85 38 89 90 91 92 33 94 95 96 97 99 99 100 101 B-11
u.
Table:4-2.,
SWCEN ONBR Results Test' Run-
.VIPRE-01 isse-Feetion No.
?NBR t.
4
' W108 9
2
'18
- 31 61-4' W114:
-217 5
-228 6
244
-7 247 8-W121' 402 9
416 10.
421
'11 425 12;
'W122 442
'13 455 14 459
-.15 W124' 482-16
~491
.17 496
- 18 501 L19 W125 518
' 2 0 -~
532
- 21 -
537 22~
540
-23 W127 558!
24 572-25 580 26
.W131 621 27 628-L28
'648' 29 651 30-
-W132 657 21 674 32-679 33-684 34 W133 694-35-
-703 36 704 37 718 A =. LYNX 2 ON3R t/TPRE-01 DNBR x
100 LYNX 2 ;NBR B-12
,n-..
, k.
i, A ;, ><
_=_
1 Table _4-2.-
BWCMV CNBR Results-
-(Continued)
. Test Run-VIPRE-01 CSSe" 3 e ct i "U NO.
DNBR A. (
38-W134 39
.40-41-42-43 44 W138 46.
~47-4 8_.
N133
_49
_50
'51
.52 W153 53-54
- 55J 56 W15 7.'
57.:
58
- 59 160
-61 62-
.63 -
'64
-W158' 65-66->
6 7 -'
68-69 70 71 72; W160.7 4 -
- 75~
76'
'77.
-W161 78:
79.
'90 31 b-14
I Table 4-2.
SW2MV :::3R Results (Continuec)
Test Run VIPRE-01
' 2; SMBR A
4 d
Case Section 82 W162 1891 83 1902 84 1911 85 1534 36 W163 1951 87 1953 88 1955 89 1961 90 W164 1984 91 2005 92 2030 93 2037 94 2041 95 W166 2019 96 2032 97 2033 98 2042 99 NO3 13 100 25 101 32 102 44 103 N04 13 104 33 105 46 106 62 107 75 108 92
-109 107 110 116 111 149 112 NOB 43 113 44 114 53 115 70 116 71 117 95 118 N09 23 119 30 120 59 121 "1
122 102 123 228 124 229 B-14
.j.
Figure a1 BWCMV Measured vs. Predicted CHF B-15
~
5.
-Since the Westinghouse Standata and Optimized 17 x 17 fuel designs and-S&W Mark-BW fuel designs are not identical, provide a description of the thermal-hydraulic and mechanical differences and
-their impact en the VIPRE methodolcgy presented.
.As-stated in Section 3.0 of the report, Mark-BW fuel has been designed
-to be hydraulically and mechanically compatible with Westinghouse (W) standard and optimized 17 x 17 fuel.
The current McGuire and Catawoa cores contain all 'd optimized fuel assemblies (OFA).
There are two main differences in the W and B&W fuel designs.
The flow areas are different (OFA and Mk-BW rod diameters are different) and the Mark-BW first intermeciate spacer grid is a non-mixing vane grid.
To verify the hydraulic compatibility between the OFA and Mark-BW designs, B&W performed a series of flow tests for both designs using their-transportable flew test rig (TFTR)
The flew test results demonstrated tnat the t0tal pressure crop of the W OFA design is 2.4 %
higher than that of the Mark-BW design.
A transiticn or mixed core analysis' was perfornec to determine the DNBR impact of this difference.
-For each mixed core 1 configuration during the~ transition to the new fuel design,' performance of each design is evaluated relative to a reference e
analysis.
For the McGuire and-Catawba transition cycles in which the resident OFA fuel is being displaced by Mark-BW fuel, the reference thermal-hydraulic. analyses are performed for a full Mark-BW core.
A transition core DNBR penalty was determined for the CFA fuel which is applied against the DNBR margin included in the design DNBR limit (DDL).
Since Mark-BW fuel has;a lower overall pressure drop-than the OFA design, a Mark-BW assembly in a mixed core will tend to have more flow through it and consequently more margin to the DNBR limit than the same assembly in an all Mark-BW, core.
Similarly, flow will tend to be forced out_of the OFA fuel in a mixec coret thus, a' DNBR penalty was _determinea for the OFA fuel.
An OFA transition core DNBR penalty was determined generically moceling a bounding core configuration with one optimized assembly (the hot assembly) and the rest of the core modeled as Mark-BW-assemblies.
A number of statepoints and peaking conditions were
. analyzed, yielding a maximum DNBR penalty of 3.2 %.
This penalty is
-applied against the 10.7 % margin included in the DDL (DDL= SCD limit
+
margin =.1.55).
For a specific reload where it is desirable to reduce this transition core penalty, a core-specific analysis will be performed moceling.tne-actual transit;cn core configuration.
B-!6
6.
Explain why (see - Section 5. 5.1) flow redistribution in the hot.
assembly due to radially peakea power prefile would not cause lower r
For'the same maximum pin peak (Fa n), a steep power gradient around the hot pin will intuitively result in a higher MDNBR compared to the reference, relatively flat, distribution.
Three different pin power distributions,-all with a peak pin power of 1.50, were analyzed and the results are given in-Table 6-1.
The change in MDNBR for cases 2 and 3 is relatively small,- hut the reference power' distribution, which is tne
-flattest, does yield'the most conservative MDNBR as expected.
Additional operating statepoints have been analyzed with nearly -
identical results to those shown in Table 6-1.
The hot channel flow for the three different power distributions is shown in Fig._6-1 which demonstrates that the flow in the hot channel changes very little if the peaking gradient around the hot pin is changed (maintaining a constant hot pin peak),
e I
c l'p l
l B-17
.~
Table 6-1.
Pin-Power 01stribution Sensitivity.Stucy Power Distribution MDNBR 6.
4 1-1.612 2
1.627 0.9 3
1.649 2.3 Core Exit Pressure Inlet Temperature Ocre Power
-Core Flc*.1 Pin Penk' Pin No.
1 2
3 1
1.500 1.500 1.500 2
.1.488 1.476 1.464 3
. 482 1.464 1.446 4
1.491 1.461 1.4415 5
1.450 1.401 1.3515 6
. 480 1.461 1.4415 7
1.482 1.464 1.446 8
1.4158 1.4184 1.4211 1
r
- ~
All three 'dist:1but;:ns maintain the same peak pin and assemmly average power.
- Reference-cistricut;:n, Elg. 5-in 0;C-hE-2004
-l B-18 1
l l
l o
-e e
o Thi
-m llD a
A
- ri b
=
o
)
Z zz
- -r O O C 6
EEE o
Z 8EE
-M g$h dl D
Ap O
oa
-N r_
g6 cr: e W
p-
~
=
o C ~#
299 6
a e A r,,.
u o
=== m
_o Z
- i Jz l
=
M ra Z
- o Z
~
~
i-e Z
_o 4
Z o
C-6 x
s m
6C
_o u
e R* g m
w Pv
_o
~
mm m
0"*
- w
_o m%
v Ab
_o rA y m
- d a
7 m
f 8
o mv r
.a.
m o
s s
s s
s s
s s
s m
o m
o m
o m
o m
o m
N N
n o
o cn ca e
e s
N N
N N
N cd a
4 a
J (E13 Hil/Will10 Xfl13 SSVW as-le
7.-
Explain the differences between the MAP limit methodology used in DPC-NE-2004-and t hat use1 in DPC-NE-2003.
i The-MAP limit methodology discussed in both reports is identical except for the cperating statepoints analyzed.
]
8.
Provide'jestification that the assumption of 7.5 % for the core bypass flow (p. 21) is conservative.
Explain further why a bypass flow of 6 % was used in the statistical core design (SCD) stucy.
e The McGuire reactor vessel internals are being modified to mitigate fuel rod failure due to baffle jetting.
This modification effectively changt s the direction of flow in the barrel / baffle region of the vessel
-from down-to up.
The total nominal-core bypass flow has been calculated by B&W for the current. (downflow) - configuration
(< 6.0 %) and the modified (upflow)1 configuration
(<- 7.5 %).
The McGuire core bypass flow is greater _than that calculated for-Catawba.
The response surface-model (RSM) described in DPC-NE-2004 was determined
-prior to the decision to-procead with the--upflow modification at
,McGuire.
The RSM was therefere-: based on a nominal core bypass flow of 6.0 %.
A change in the_ nominal. core bypass flow only changes the core
. flow analy:ed; thus, there is no need to develop a new RSM to consider the-increase in bypass flow to 7.5 %.
The statepoints analyzed to.
determine the SDL were analyzedrafter calculating the increase in bypass flow.for the upflow modification.
As stated below Table 8 in DPC-NE-
-2004, the generic M/C thermal-hydraulic analyses, including the calculatien of the statistical DNBR - limit - (SDL), are based on a nem:,nal core-bypass-flow of.7.5_%'.
s B-20
n l
1 3.
Explain how-tne-application'of'the response surface model (RSM)-
methodology to licensing analysis will account ftr the fact that analysis of,00nciti:ns not expressly included in the determination of the RSM crof11: lent: yielded results deviatit; frcm the VIPRE-01 calculaten results cy nearly 8 % (see Table 11).
, The RSM results given in-Table 11'for cases 4 and 5, wnich yield the
- largest differences between the VIPPI and RSM DNBRs, are incorrect.
The 00rrect RSM results are given below:
Power riow Pressure T,.
Th F
DNBR Case 5
6 ma
'F 2
VTPPE PSM A.1 4
5
= VTPRE ONBR - :SM DNBP x
100 VIFRE tNBR
'The.RSM uncertainty ha b en shown tc fit a normal' distribution with a standard deviation of The SDL was calculated using an RSM
' uncertainty conservat:.ve y increased to[two, and three standard] deviations will contain 68.3 For a standard normal 1 distribution: One,
' %,. 95. 5 -. %, and 99.7 %,
respectively, cf the values cc:.:prising the
- dist ribution.
The VIPRI and RSM DNBRs for the ten statepoints evaluated to determine the SOL are ccmpared in Table 1.
The RSM DNBR is within two-standard deviaticns of the VIPRE-DNBR f:r 9-of the 10 cases and the
- ONBRs are within' :nrae stancard ceviaticas f:: all ten cases.
- Thus, ne RSM and VIPRE results agree within the e:mectaticns cf a normal distribution..Also, as discussed in the response
- question 12, the
'RSM = is used to propagate various uncertaintiec to cetermine an overall DNBR uncertainty, not := calculate acsolute DNBR values.
i 6
B-21
.q aJ p-Tele 9-1.
- /IPRI-01 vs. RSM Results
,sp DNBR Casea viPPE-01 RSM A.
t r
2 3
4.
=
I 6
.y.
.8 9'
10 Refer :: ~'stle l',... in DP C-NI-2 0 0 4 '
(.
l-F j-'-
l:
I B-22 i.
l:
1
10.
Regarcing the statistically treated uncertaincies:
a.
Present _the data supporting the bounding uncertainties and the shape of distributien for uncertainties fer eac: 0-. the twelve parameters referred to in Secta n 6.4.
.b.
Section-6.4 indicated that ' the uncertainties will be
- ustified on a plant-scecific basis in the Reload Report for the'first application ' ? the SCD methocology.
Do you intend 10-re-evaluate the stm aatical DNBR limit if the plant-specific uncertainty and distribution of any of the 12 parameters are not bounded by the current values?
a)-
Q, W,
P, T - Plant specific core power, flow, pressure, and tcmperature uncertainties have been calculated by statistically cc=ciping the uncertainties of the associated process indication channels anc centrol channels.
Each enannel uncertainty is calculated fr a normally cistributec ranac= errer terms:
e.g.,
sensor calatrat:On
. accuraci', sensor drif t, rack drift, etc., using square-root of the sum of tne squares -(SRSS) methodology.
Since each channel uncertainty is calculated from normally distributed values, the channel uncertainties and parameter uncertainties are also normally distr buted.
The largest McGuire Lor Catawon uncertainties are listed below along with the uncertainties used to calculate the - statistical DNBR limit (SDL) which are conservative,; historical values that bound the plant specific values:
McGuire/ Catawba Uncertainty Parameter Unce rt ainty Included in SCD Core Power (% RTP)
'RCS Tlow (%)
Pressure (psi)
Temperature (*T)
J Duke may choose to recalculate the SDL using the McGuire/ Catawba
. uncertainties f:r a specific reload cycle where DNBR margin as neeced.
' Core.sypass riow - The total core bypass flow includes guide chimale coeling f1:w, head-cooling flow, cuter fuel assencly/ baffle gap leaxage, vessel :utlet no::le gap leaxage, anc carrel / baffle annulus fi:w.
S&W has calculated a core bypass flow uncertainty for McGuire and Catawoa by
.first identifying the key f eatures which control the flow rate in the five flow paths listed above.
Tocusing on tnese key features,
- parameter :atiens were perf rmec on key dimensions, loss coeffi: lent correlati:ns, and the effect of the uncertainty in the driving pressure drop en the flow rate in each flow path.
Since they are indepencent, ithe calculated changes in: flow rate due to dimensional tolerances were comminac _using the square-root-of-the-sum-of-the-squares (SRSS) approach to c=tain a ccmcined uncertainty for possible dimensional variati:ns for eac core. bypass _ flow path.
For conservatism, core bypass flow enanges due to uncertainties in the loss coefficients and uncertainties in the driving pressure drops were added together to obtain a combined unce rt ainty.
The total uncertainty was then calculated using the SRSS B-23
approacn to c =mine tne resu.t: Of all three effects.
The calculated total c re bypass flew-uncertainty for McGuire and Catawoa is less than the 1.5 % usec in the SCD analysis.
GL As discussed in Secti:n 6.4, the T2g uncertainty is a measurement uncertainty for-the moveacle anc:re instruments.
-A measurement uncertainty can arise f rom 1..strtmentati:n drif t or reproducibility error, integration and locati:n error, error associated with the burnup
-history of the core, and err : associated with the conversion of instrument-readings to rod-power.
The 4 % measurement error (2.43 %
. standard deviation) includec in Ine SCD analysis was calculated by Wostinghouse and is given in the McGuire and Catawba Technical Specifications.
GN This uncertainty, typically applied directly as a hot channel
-factor, accounts for manufacturing variations in the variables affecting
-the heat generation rate along the flow enannel.
Westinghouse and B&W fuel is manuf actured to tole:ances which ensure that a 3 % uncertainty I Ff, = 1-. 0 3 ) conservatively s:00unts f or the possible variations in the pellet ::ameter, aensity, an 1*n s snri:nment.
F, u, This uncertainty :n F, ia applied to account for the effect 6
A on peaxing of a recuced hot assemmly flow area and spacing between assemblies.
The reduced fltu area and spacing is calculated as explained in Section 6.4.
S&W has detem=ined that the power peaking gradient across the het assemcly becomes steeper due to the reduced flow
-area and spacing.
Althougn the assembly average power does not change, the maximum rac ancreases by 1.5 i, ref. 3.
This uncertainty was conservatively increased t0 1. 0 % in Duke's SCD analysis.
F. _ The.4.8 % axial peak calculational uncertainty included in the SCD analysis was calculated ano appr ved by tne NRC ir ref.
1.
An axial peak Observec Nuclear Reliacility Facter (CNRFi was calculated using McGuire-and Sequoyan measurec peaxing cata, 3
An uncertainty on axial peak location is included in the calculat1on of the SDL since :n a' maneuvering analysis the predictec peaks are cc= pared with the c.aximum allowaole peaking (MAP) limits basec on the predicted axial peak ( T,) and its location (2), ref.
1.
An uncertainty ofJ+/- 8 inches was used which is +/-one axial node for tne
'EPRI-NODE-P physics model usec-by Nuclear Design, ref, 2.
This uncertainty was : nservatively assumec :: ce unif:rmly castr:Outec.
'Three uncertainties that apply directly :: the calculation of DNBR are treatec stat:stically.
All :..ree uncertainties are incepencently var:ec and then applie as multipliers :: the DNBRs calculated by the RSM.
The CHF1correlati:n uncertainty and c:de/mocel uncertainty ar{e.di cussed in
- Section 6. 4 of :ne tcpical report.
An RSM uncertainty of
% was included in the SCD analysis.
The ratic of the VIPRE-01 udB to the RSM DNDR is-given in Table 9 of the report f:: the 78 statepoints used t:
develop the RSM.
The meanL V! PRE-31 DNBR/RSM DNBR is 1.000' and the standsrd deviation is
] The stancard deviation was increased to
{
, to -- add acditiona conservatism t: the calculation of the SDL.
B-24 w-'*
+ --
- ~.
b)
'All of th'e uncertainties listed in Table 12 bound the current McGuire and Catawba uncertainties:-thus,-the statistical-DNBR limit (SDL) of-1.396 is applicable until a significant fuel design change, instrumentation change, or nuclear design code. change is made.
When any change is-made that raises any of the uncertainties above the values
.given in Table 12, a new SDL will be determined.
.Peterence 1.
Nuclear Design Mathodology for Core Operating Limits of Westinghouse Peacters, DPC-NE-20llP-A, March 1990.
2.
Nuclear Physics Methodology for Reload Design, DPC-NT-2010-A, May 1985.
-3.
-BAW-10170-A, Statistical Core Design for Mixing vane Cores, Babecck
& Wilcox, December 1988.
r
-B-25
t o
~
- 11 l.Section 6.4 briefly-describes the propagation of uncertainties.
Provide a more detailed-description of how the propagation of the.
uncertalnties=with normal distribution and uniform distribution are treated.
Section 6.5 describes how multiple sets (3000) of random statepoints are 3
-generated around-each nominal statepoint by indepencently varying each statistically-treated parameter according to the probability
- distribution-of its uncertainty.
As an example, randomly varied core
~
f power levels are generated about a nominal core power level of 100 %
using;the4 expression Core' Power = nominal valus + (std. deviation)(Random No.)
= 100.0 + (1.22) (Random No. )
where 1.22 = the standard deviation cf the power uncertainty Random No. = a normally distributed random number with mean zero and standard. deviation +/- 1.0 calculated using the SAS function RANNOR.
Repeated calculations.with'this expression result in a normal distribution of core power levels with mean of 100 %.
All of the
_ uncertainties 1given in Table'12 are independently varied-in a similar manner to produce a set of 3000 statepoints for each nominal statepoint.
12.
Justify using nominal-plant conditions,.instead of the limiting DNB
-conditions, as ":ero point" for the RSM coefficient determination.
The-primary purpose of the response surface is propagation of uncertaintiesL to determine an overall DNBR uncertainty, not absolute 1 calculation of DNBR.
The response-surface model (RSM).is used to 1 determine the chany-in DNBR.for relatively small changes of each of the seven parameters (Q, W,
?,
etc.).
The RSM coefficients are determined to bestifit the VIPRE-07 DNBRs over~a wide range of operating conditions.
The : larges'. DNBR uncertainty may not be at a statepoint 4 yielding a limiting DNBT of 1.4.
The statistical DNBR limit is based en the -largest Eccef fician of variation' (std. deviation /mean) cf DNBR, snot a' minimum;mean'value c; DNBR.
Using a "zero point"Hof approximately nominal conditions also allows for a distribution of DNBRs '(over the range of conditions the RSM is ideveloped for).with values > 1.0.
If a limiting set of conditions-is used as-the ":ero point", the RSM will have to fit a narrow range of conditions or a significant number of the data points will yield unrealistic DNBRs1<< l.0.
B-26 v
v g
my w
a v
'v
L xplain why at "tero point" the VIPRE-01-and RSM calculated DNBRs 13.-
E do not match, The!RSM 1s a least
~
squares curve fit which minimizes the sum of the sqaures of the differences between the VIPRE-01 MDNBRs and the expression calculated to-fit the data.
The RSM is the best fit to all 78 VIPRE-01 MDNBRs, of the other points. it does not-exactly predict the "zero point" or any 14.
-Since in the-range of ONBR below 2.0 where the limiting MDNDR the RSM predicted MDNBRs are less concervative than the
- occurs,
.VIPRE-01 predicted values in about 50 and 11),--discuss how necessary conserva% oof the cases (see Table 10 tism is introduced into the DNBR. prediction in the-licensing analyses.
As discussed in the responses to questions 12 and 13, the RSM is a-1-ast squarcs curve fit to the VIPRE-01 data and thus-the RSM can be expected to'under and.over-predict the VIPRE-01 results over the entire range of conditions analyzed.-_
licensing analyses, The RSM is not used to calculate MDNBRs for determine an overall DNBR uncertainty.it is used for propagation of uncertainties to in the calculation of the statistical DNBR limitAn RSM uncertainty is included to account for the ability of the RSM to predict the VIPRE-01 results, i
i B-27
f 15.
Justif9'the assumption that each of the 79 points, some of which are first order-and some of which are second order, computed with VIPRI-01.and computed by variation of the coef ficients of the RSM equation should be treated with equal weight in determinatien of the 36 coefficients.
To determine the 36 coefficients of the RSM, greater than 36 values of the dependent variable (DNBR) must be found along with an equal number of ccmbinations of the independent variables (Q, W, P,AT, Fa n, Fe, 2).
An efficient-way to define-the matrix of conditions needed to determine the. coef ficients is _using a Central Composite Design (CCD),
ref.
1.
For
.7 independent-variables, a total _of 143 points (one center point, 14 axial points, and 128 factorial points) define a full CCD.
Although a full CCD could be used to determine the RSM coefficients, this is considerably greater than the number of points required.
To accurately and efficiently determine the coefficients, 78 statepoints were selected.
The=78 statepoints were selected generally following the CCD apprcach i:ne centar point, 13 axial points, 50 factorial points, anc 14 additional points).
The CCD approach was only followed as a systematic way to select the data base for determining the RSM coefficients.
As discussed in Section 6.3, the independent parameters are coded instead of'being used directly in the RSM.
Coding each of the variables using the expression given on pg. 29 results in an RSM-in which equal coefficients means equal importance of the terms.
Using coded variables, the coefficients of the RSM are direct indicators of the level of significance of each term and a term can be dropped if its coefficient:is very very small.
The coefficients are determined using a least-squares curve. fitting program equally weighting all of the terms, p
Reference 1.
G..E. P. Box and K.
B. Wilson, "On the Experimental Attainment of Optimum Conditions", Journal Royal Stat Soc B, 13, 1-34, 1957 B-28
1 l
16.
Discuss the impact en MONBR and on safety of applying the statistical core design (SCO) methodology to the DNBR computation instead of the current computational method.
Westinghouse (W) currently performs all of the thermal-hydraulic analyses for McGuire and Catawba using the Improved Thermal Design Procedure (ITDP), ref.
1, which is a different, but similar, approacn to statistically account for uncertaintiet.
The W ITOP DNBR limit using the WRB-1 correlation is 1.32 for thimble channels and 1.34 for unit channels, ref.
2.
Since the Duke SCD analysis and the W ITDP analysis both determine DNBR limits that statistically account for uncertainties relateo to CNB, there will he ess9ntially no impact On MDNBR or safety.
i References 1.
- Chelemer, H.,
- Soman, L.
H.,
and Sharp, D.
R.,
- mproved Thermal Design Procedure, WCAP-8568, July 1975.
2.
McGuire and Catawba Final Safety Analysis Reports.
e 17.
Does DPC intend to perform all DNBR calculations using the SCD methocology?
If not, discuss where DPC intends to use SCD methodology and where DPC intends to retain the current DNBR computational method.
All steady-state McGuire/ Catawba thermal-hydraulic analyses will be performed using the SCD methodology.
All of the DNBR limited FSAR Chapter 15 analyses will use the SCD methodology except the rod ejection and steam line break analyses.
These transients will directly acccunt for all of the uncertainties incluced in the SCD metnodology.
B-29
18.
Explain the process by which DPC-plans to deal with less than four
-pump operation in terms of (1) sa:e of channel models to use, (2) validity of sensitivity stadies used to select a ecc se model, (3)
-applicability ~of 1/S. core symmetry bouncary condition, (4) location 1
-of hot assembly, (5) inlet flow distribution, -(6) justification and applicability of the.SCD methodology based on 1/8 core symmetry,
.and (7) applicability of the ":ero point" chosen to evaluate the coefficients in the response surface equati -
1The McGuire and Catawba units are not licensed for less than four pump operation and there are no plans to perform any analyses discussed in DPC-NE-2004 assuming less than four operating pumps.
e 1
I :
l' B-30 l
l-o I
\\'!
~1 I
I
)
m t
T APPENDIX C
.-Responses to Request for Additional Information November-29, 1990
-i a
4 5
1
.C-1.
ll l
nnn em.re.-~.on
, o.
i.
1 11 <t,, '
t herr e n.- \\ t
,.llr
- DUKEPOWER rst November 29, 1990 U. S. Nuclear Regulatory Commission ATTNt Document Contr:1 Desk Washington, D.C.
20555
Subject:
McGuire Nuclear Station Docket Numbers 50-369 and -370 Catawoa Nuclear Station Docket Numbers 50-413 and -414 Topical Report DPC-NE-2004 (TACS 73765-68)
Sy letter dated January 9, 1989, Duke submitted the subject Topical Report for review. On November 1 and November 15, 1990, telecons were held to discuss the Topical.
Participants in the telecon included representatives of the NRC staff, the contract reviewer, and Duke Power.
Attached are formal responses to the questions which were asked during the telecon.
Please note that this submittal contains proprietary information, pursuant to 10 CFR 2.790, which should be withheld from public disclosure.
An affidavit which supports the proprietary designation is included in the January 9 1989 submittal.
If there are any questions, please call Scott Gewehr at (704) 373-75S1.
Very truly yours,
- n. s. 3 A M. S. Tuckman Vice President Nuclear Operations 1
SAG /232/lcs l
I C-2
- - _ - _ -.~.
i U. S. Nuclear Regulatory Conmission November 29, 1990 Page 2 cc Mr. Stewart D. Ebneter, Regier.a] Administrator U. S. Nucioar Regulatory Cornission - Regior 'I 101 Marietta Street, Suite 2900 Atlanta, Georgia 30323 Hr. Tim Reed. Project Manager office of Nuclear Reactor Regulation U. S. Nuclear Reactor Regulation One White Flint North Mail Stop 9H3 Washington, D.C.
20555 Mr. W. T. Orders NRC Resident Inspectoe Catawba Nuclear Statta..
Mr. P. K. VanDoorn NRC Resident Inspcct.r McGuire Nuclear Station Mr. R. E. Martin-Project Manager Office of Nuclear Reactor Regulation U. S. Nuclear Regulatory Commission One White Flint North, Mail Stop 9H3 Washington, D.C.
20555 t
C-3
1 1.
Justify use of the SCD limit for Ch. 15 transient ant.yses.
The range of conditions for wnich the McGuire/ Catawba statistical CNBR limit (SDL) is applicable are given in Table B.
The statepoint at whici; the MDNBR occurs for eacn Ch. 15 transient can be compared with the ranges of conditions given in this table.
If the statepoint is within the ranges of conditions, then the SDL calculated in DPC-NE-2004 is applicable for that transient.
If the statepoint is outside of-the ranges of conditions, a new S;L can be calculated.for nat statepoint as explained in the response to question :7 or the transient can-_be analyzed with all of the uncertainties applied directly with the MDNBR compared to the CHT torrelation limit.
This apptcach to determining the acceptability of the statistical DNBR limit is the same as that used by both-B&W and Westinghouse.
The most limiting DUB transient-for McGuire/ Catawba is a loss of flow (LOT) transient which is analyzed using the RETRAN and VIPRE-01 codes.
The tr=~51ent boun"ry conditi ns (power, flow, pressure, and temperature) are c.uculated by RETRAN and input into VIPRE-01 to calculate the MDNBR.
As explained-in Section 5.6 of DPC-NE-2004, one set of maximum allowable peaking (MAP) limits are calculated at the LOT statepoint (the conditions during_the transient at which the MDNBR occurs).
The MDNBR for the LCT statepoint is slightly more conservative than the MDNBR during the transient.
The LCF statepoint is one of the statepoints about which the uncertainties are propagated thus the statistical DNBR limit (SDL) is an appropriate DNBR limit for the LOF transient.
The LOT transient MDNBR (and the MDNBR for all other transients based on the SCD methodology) must actually te greater than the design DNBR limit -(DDL) which incluces margin above the SDL.
If another transient is found to be more limiting than the LOF-transient, an SDL can be calculated for this new statepoint as explained in the response to question 7.
1 2.
Discuss how the VIPRI-01 input is handled that is not treated statistically in the SCD methodology?
<The uncertainties treated statistically are listed on Table 12.
.1 other. uncertainties such as the inlet flow maldistribution factor
( Se cti on_ 4. l'. 6 ) are considerea directly in VIPRE-01 in all of the core thermal-hydraulic analyses.
C-4
~
1 i
s 3.
Is the assumption made that the statistically treated parameters are independent?
Explain how the ur.ce tainty districuticns were selected.
The standard methed for accounting for the uncertainties that enter into the calculation of DNBR assumes that every uncertainty is simultaneously at its worst level.
The SCD method described in Secticn f.? acetunts for the f act that the input uncertainties are independent anc it is highly unlikely that 111 of the uncertainties will be at their worst values simultaneously.
The SCD method acc:unts for the andependance of the input uncertainties by treating the uncertainties statistically.
A non-linear response Jurface model (RSM) is used to calculate MONBR so that thousanes of conditions can ce analy:ec to statistically comcine the uncertainties.
The variables used to define a reactor core statepoint on the response surface are assumed to be uncorrelated and independent.
Inter-variarle affects are accounted for by the cross-product terms in the RS9 (Q x W, P x T,,).
The independent variables and the RSM are discussed in Sections 6.2 and 6.3.
The levels of the uncertainties are assumed to be independent anc they are the maximum that could be encountered at any limiting statepoint.
The core power uncertainty represents the error associated with a nermalization of the reactor primary side power to a secondary side heat balance.
The core power uncertainty is a function of secondary side parameters (feedwater flow, tcmperature, and pressure, steam pressure, and moisture carryover) and is independent of other primary side (i.e., SCD) uncertainties.
Tha individual uncertainties that are statistically combined to calculate the core power uncertainty are normally cistributed and by the Central Limit theorem, ref.
1, even if the individual uncertainties were uniformly distributed, the statistically combined uncertainty would be normally distributed.
The reactor coolant system (RCS) flow uncertainty is determined ficm elbow meter AP measurement uncertainty, precision calorimetric measurement uncertainty, and the RCS pressure and temperature measurement uncertainties.
Thus, the flow uncertainty is not truly incepencent of tne RCS pressure and temperature uncertainties, c' u t the dependence '
very weak it can be shown that it does not need to be accounted for when propagating the uncertainties.
The flow uncertainty, like the core power uncertainty, is calculated by statistically combining a number of different uncertainties; thus, the flow uncertainty was assumed to be normally distributed.
The core bypass flow uncertainty is a calculated value that is insensitive to core power, RCS flow, pressure, temperature, and power distribution.
Since this uncertainty is a calculated value primarily based on manufacturing tolerances of the reactor internals, the uncertainty is assumed to be uniformly distributed.
The RCS pressure and temperature uncertainties reflect measurement allcwances for two indepencent. Instrument strings.
The measurements, l
C-5
1 and the uncertainties, are independent of other primary side uncertainties.
Both the temperature and pressure uncertainties are conservatively assumed to be uniformly distributed.
The radial peaking factor (F.) measurement uncertainty is independent 6
of other primary side uncertainties.
The F, measurement uncertainty A
also results frem a statistical combinaticn of several uncertainties:
thus, a normal distribution was assumed.
1
.The hot channel factors, Tj,and Fan,u.m., acccunt for fuel assembly i
manufacturing variations and are thus clearly independent of the uncertainties on power, pressure, flow, and temperature.
These uncertainties are determined frcm samples of manufacturing cata wnich can be shown to fit a normal distribution.
The axial peaking calculational uncertainty, rg, is not dependent on any.of_the other DNBR parameters (power, pressure, etc.).
The r
icalculational uncertainty was shown in ref. 2 to be normally distributed.
The uncertainty on axial peak location, 2, accounts for the difference in node sizes between the Nuclear cesign codes and VIPRE-01.
This uncertainty is obviously not dependent on any of the other DNBR uncertainties.- The axial peak location uncertainty was conservatively assumed to be uniformly distributed.
The three uncertainties on DNBR were assumed to be independent and wore individually propagated.
The BWCMV correlation uncertainty was shown in ref. 3 to be norma 11,, distributed.
The RSM uncertainty was shown-to be normally distributed using the D' test.
The code /model DNBR uncertainty was assumed to be normally distributed.
Reference 1.
Statistical Methods, G. W. Snedecor and W.
G. Cochran, Iowa State University Press, Sixth Edition, 1973.
2.
Nuclear Design Methodology for Core Operating Limits of
= Westinghouse-Reactors, DPC-NE-20llP, Duke Power Company, April 1988.
-3.
.BWCMV Correlation of Critical Heat Flux in Mixing Vane Grid Fuel Assemblies, BAW-10159-A, Babcock & Wilcox, July 1990.
C-6
4.
' Clarify that CPC-NE-2004 and the SCD methodology and SDL apply to McGuire and Catawba.
The statistical methodology discussed in DPC-NE-2004 is a generic methodology for statistically ecmbining uncertainties to calculate a DNBR limit.
The ranges of parameters, response surface model, uncertainties, and the statistical DNBR limit (SDL) given in the report are spectfically f tr the McGuire and Catawba Nuclear Stations.
i 5.
Discuss the rationale for selecting the range of variables given in Table B and for selecting the 78 points used to develop the RSM.
'The. range of the independent parameters used to develop the McGuire/ Catawba (M/C) RSM are given in Table 8.
The maximum and minimum values were selected to bound _the values that will typically be analyzed in M/C core thermal-hydraulic analyses (including the loss of flow transient).
The center point was-placed at approximately nominal conditions resulting.in a MDNBR much greater than 1.0 (VIPRE-01 2.321)..
As shown in Table 9, there are other points that
-broad range of DNBRs (f rom.5. 4 69 to 1.2 69).
cover a The matrix of conditions for determining the RSM was defined in a systematic way.using a Central Composite Design (CCD).
For n independent variables: one center point, 2n axial points, and 2" factorial points define a full CCD.
Thus, a full CCD for 7 variables includes 143 points.
This is considerably greater than the minimum number of points required to determine the 36 coef ficients of the response surface.
To accurately, as well as efficiently, deterndne the coefficients, 78 points were selected as shewn on Table 9.
These points were selected to adequately cover the entire range of c:nditions.given in Table B and to ootain a "small" RSM standard deviation.
Any errcr in the RSM feeds directly into the uncertainty
_ propagation; thus, a large error (poor fit) results in a large SDL.
An RSM was first determined based on cases 1-64 in Table'9.
A second L
LRSM was_ determined' including the-next-14 cases (65-78).
The RSM based on_78 points "as a slightly better fit (smaller RSM error), but the reduction ~in,ne RSM error was so small there was no reason to expect
-that the inclusion _of any additional points would result in a better fit.
C-7
.... ~ -
~.
6.
How were the ten statepoints used for prcpagation of the uncertainties selectea?
Describe the ten statepoints given in Table 11.
The ten statepoints analyzed to determine the statistical DNDR limit (SDL) were selected to cover the limiting conditions typically analyzed for McGuire/ Catawba thermal-hydraulic analyses.
Statepoints that-define the core ZNB li.?its (see Section 5.5) were analyzec along with the Loss of Flow (lor) statepoint (the conditions during the lor transient at which the MDNBR occurs).
The nominal operating statepoint was also analyzed.
While-these statepoints do not absolutely bound all core operating stater
' 'y do cover a wide range of power, pressure, flow and temperature iu; c -
wyta.cm coefficient of variation is used to determine the SDL th.st t,
.w
%d for all operating cenditions within the ranges gis in
-y
- i Case Definition of Statecoint i
1 OTAT MAP statepoint (see Fig. 12) 2 OTAT MAP statepoint (see Tig. 12) 3 MAP limit for 1.7 axial peak G Z = 0.5 4
MAP limit for 1.7 axial peak 0 Z = 0.3 l
5 MAP limit for 1.7 axial peak 0 Z = 0.7 6
Core DNB limit statepoint, 2280 psia (see rig. 10) 7 Core DNB limit statepoint, 1990 psia (see rig. 10)-
8 Nominal conditions 9
Loss of Flow statepoint, 1,5 axial peak G Z = 0.5 l
10 Loss of Flow statepoint, 1.5 axial peak G Z = 0.7 i
deb
_ _ _ ~ -.. -,. _. _ - _ _.. _ ____ _..
. ~... _.,...
-7 Explain when a new RSM will be generated.
The SCD methodology discussed in PC-NE-2004 is used te generically determine a statistical DNBR limit.
The RSM used to generically calculate the SDL must be evaluated to determine if a new RSM must be developed when the following occur 1) a significant change is made in the fuel assemoly design 2) a new or revised CHT correlation is developed 3) operating conditions outside of the range of conditions listed in Table 8 are of interest Each item is further discussed below.
1)
A completely new fuel assembly design would obviously require a now RSM.
The development of a new feature, such as a new spacer grid dh,'gn, may not require a new RSM.
An evaluation may show that the ncw grid design does not sianificantly change the MDNBR and the current RSM may remain valid.
2)
Development of a new version of the BWCMV correlation or use of a different CHT correlation would require a new RSM which would be developed in the same. manner as discussed in Section 6.3.
Use of a now version of VIPRE may not require a new RSM, but would require an ovaluation of the current RSM.
A subset of the RSM design matrix conditions given in Table 9 would be analyzed using the new version of VIPRE and the' results compared with the-RSM predictions.
3)
Any operating conditions outside of the ranges given in Table 8 that become of concern would first be evaluated in VIPRE-01 to d3termine the MDNBR.
The MDNBR would then be compared with the RSM 1 prediction and if the difference is within the error included in the SCD uncertainty propagation, the uncertainties (including the original RSM error)-would be propagated about the new statepoint to calculate an SDL.
If a larger RSM uncertainty is required for the extended conditions, a new, larger SDL can be' calculated by propagating the larger RSM error.
C-9
1 8.
Provide a complete, corrected Table 11.
i Table 11.
'/IPRE RSM O!!BR Comparisons i
l l
Power. Flow Pressure DNBR Can*
5 5
sta A_
r.
?
vippr.01 a s;M 1
1 3
J d
5.
1 6
y.
B i
9 10.
J h
'?
'C-10
APPENDIX D Responses to Request for Additional Information August 29, 1991 D-1 1
l
I A
~
DUKC POWER 9
August 29, 1991 U.
S.
Nuclear Regulatory Commission ATTN: Document Control Desh Washington, D.
C.
20555
Subject:
McGuire Nuclear Station Docket Numbers 50-369 and -370 Catawoa Nuclear 5tation Docket Numbers 50-413 and -414 Oconce Nuclear Station Docket Numbers 50-269, -270, and -287 Supplemental Information to Assist in Review of Topical Reports DPC-NE-3000 and DPC-NE-2004
References:
- 1) DPC-NE-3000, " Thermal-Hydraulic Transient Analysis Methodology," July, 1987
- 2) DPC-NE-2004, " Core Thermal-Hydraulic Methodology Using VIPRE-01," December, 1988 References 1 and 2 provide, in part, the basis for the Technical Specification changes required to support the startup of McGuire Unit 1 Cycle 8.
These topical reports are currently under NRC staff review.
During the review, items have been identified which require clarification to assure that appropriate limitations and restrictions on the methodologies are observed.
Accordingly, contains a list of transients which use Statistical Core Design, and Attachment 2 is a compilation of commitments which apply to References 1 and 2.
Please note that Attacnment 2
of this supplement contains information vhich is proprietary to Duke Power Company. Therefore,
.;e request tnat this report be withheld f rom public disclosure, in accordance with 10 CFR2.790.
Affidavits documenting the proprietary nature of the information in References 1 and 2 were provided with the original submittals, dated January 9, 1989 (DPC-NE-2004) and September 27, 1987 (DPC-NE-3000).
Also please note that approval of these Topical Reports is needed for startup of McGuire Unit 1 Cycle 8 following its upcoming refueling outage.
The outage is scheduled to begin in late september, 1991.
Cycle E is expected to start up in late November
- r early Decencer.
l D-2
U.
S. Nuclear Regulatcry Commission August 29, 1991 Page 2 If there are any questiens, please call Scott Gowehr at (704) 373-7581.
Very truly yours, c $'
M.
S.
Tuckman cvr/ sag cca Mr. T.
A.
Reed, Project Manager Office of fauclear Reactor Regulation U.
S.
Nuclear Regulatory Commission Mail Stop 9H3, OWFN Washington, D.
C.
20555 Mr.
R.E. Martin, Project Manager office of Nuclear Reactor Regulation U.
S.
Nuclear Regulatory Commission Mail Stop 9H3, CWFN Washington, D.
C.
20555 Mr.
S..D.
Ebneter, Regional Administrator U.S. Nuclear Regulatory Commission - Region II
'101 Marietta Street, NW - Suite 2900 Atlar.ta, Georgia 30323 Mr. P.
K. Van Doorn Senior Resident Inspector McGuire Nuclear Station Mr. R.
C.
Jones Reactor Systems Branch Office of Nuclear Reactor Regulation U.
S. Nuclear--Regulatory Commission Washington, D.
C.
20555 Mr. W. T.
Orders NRC Resident Inspector Catawba Nuclear Station D-3
I 1
Act::r. ent 1 McGuire Unit : - Cycic 8 Reload List oi Transients That L'se Statistical Carc Design (SCD)
Table 15.0.3-3 (p.8-107) lists the transients that use the SCD approach.
15.2.3 and 15.6.3 were inadvertently left eut.
15.2.3 Feedvater Line Break 15.3.1 Partial Loss ei Tiow P
- 15.3.2 Complete Lass c: Flow 15.4.1. Uncontrolled 3ank Withdrawal At Zero Power (W-3S is used below the mixing vanes without SCD) 15.4.2 Uncontrolled Bank Withdrawal At Power
- 15.4.3 Dropped Rod 13.4.3 Uncontrolled Rod Withdrawal At Power
- 15.4.3-Statically Misaligned Rod 15.6.3 Steam Generator Tube Rupture NOTE:
"*" indicates that the FSAR markup did not explicitly mention that SCD was used.
This will be corrected.
"**" indicates that the dropped rod analysis
~
was suomitted'with DPC-SE-3001. which ia separate from the MICS reload report D-4 l
DPc-?ir-? * ? 4 "Pc 2-etrie* ten? And 'i-itstters t'/TPPE 61) 1.
Determination of acceptability is based upon review of selection of models/ correlations for symmetrical, steady-state conditions only.
The review of OPC's use of VIPRE-01 in transient analyses for McGuire and Catawba Nuclear Stations, documentec in DPC-NE-3000, resultec in several :;ncatitns anc Open ;tems to ce Justifico ;n future applications.
Fearense OPC will use the V1 PRE-01 : ore: thermal-nycraulic models developed in OPC-NE-2004 for McGuire/Catawoa steacy-state analyses only.
- ransient analyses will be performec with the mocels cocumentec in DPC-NE-3000 or OPC-NE-3001.- The restrictions imposed en tne CPC-NE-3000 VIPPI-01 mocels are addressed in a separate section of tnis submittal.
5 b
4 F
L D-5
l 2.
Studies presented for McGuire/ Catawba plants in this report are performed using design data for Mark-BW fuel assemblies.
During the transition phase, CPC will add a DNBR penalty (to be reviewed and approved) due to Mark-BW fuel.
Perrenta The Mark-BW fuel assembly was designed to be hydraulically compatible with Westinghouse optimized fuel (OFA).
Babcock & Wilcox has performed a seriesofflowteststoverifythecompatibilityofthetwofueldesGgns.]
The tests showed that the total pressure drop across the CFA fuel is higher than the pressure drop across the Mark-BW fuel, ref.
1.
A generic transition core analysis was per!0rmed to determine the DNBR impact of this cifference.
Since the Mark-BW fuel has a lower overall pressure drcp than the CFA design, a Mark-BW assembly in a mixed core will tend to have more flow througn it and consequently more DNB margin than the same assamoly in an all Mark-BW core.
Conversely, flow will be forced out of the CFA fuci in a mixeo coret thus, it is necessarry to calculate a DNBR penalty for the OTA fuel.
A generic transition core DNBR penalty was determined by modeling a conservative core configuration with one OPA assembly as the hot assembly.
The rest of the core was modeled as Mark-BW fuel.
A number of statepoints and peaking conditions were analyzed, yielding a maximum DNBR penalty of 3.B % for the CFA fuel.
The transition core penalty was increased f rom 3.2 to 3.8 % af ter analyzing the limiting SCD transient statepoints (see the response to question 3).
DPC will verify that this penalty is appiacable for each transition core.
When it is destranle to reduce this penalty, a cycle-specific transition core analysis will be perf ormed modeling the actual mixed core configuration.
To provide design flexibility, margin is added to the SDL to determine a design DNBR limit (DDL).
For the generic Mark-BW and McGuire 1 Cyclo 8 analyses, the DDL is 1.55'(10.7 % margin above the SDL).
The DNBR penalties that must be assessed against the margin included in the DDL are given in Table 2-A.
Justification for use of the 1.40 statistical DNBR limit for CFA fuel is discussed in tne response to question 7 Feference 1.
BAW-1017:P-A, Mark-SW Mechanical Design Report, 3abcock & Wilecx, Lynchburg, '/i rginia, Decemcer 19, 1989.
D-6
Table 2-A.
DNBR Penalties Statistical ONBR Limit 1.40 Design DNBR Limit 1.55 DNDR Margir.
- ".7
"?!P R - Da n s ! * -
virk
=W
^rA Transition Core 0
3.8 %
Instrumentation / Hardware 4.7 %
1.9 %
Rod Bow-0 3.5 %
Total DNBR Penalty 4.7 %
9.2 %
Available CNBR Margin 6.0 %
1.5 %
i l
D ~
i i
i I
3.
The ;PC developed statist::a1 coro dest;n methodology, in the submittal, as ces:::ceo I
acceptable and generally applicaole to etner PRR plantstis a generic method the approval we reccmmen:
however.
at this time is f or only McGuire an:
Catawna Nuclear Stations due to OPC's use of specafic uncertainties i
and distributions baseo upon plant data and its selection of statepoints used for generating the statistical design limit.
Furt?.ermore, should tne 2 00 mets:cology ce extended for use ;n Chapter 15 type transient analyses, CPC should assure that tne 1
uncerlying use of the V PRE-01 c ce f or that particular appl::sta:n has been qualified and approved by the NRC or is covered by ::ner SERs and should develcp a new response surf ace appropriate f::
ccmputations.
nose Pasconsa' Duke interprets this questten as icil:ws:
(1) Confirm that the Res; nse Surface Model (RSM) is valid f:r all of the FSAR Chapter 15 transient analyses whien use the SCD rethocology ttrans ent numbers 15.2.S,
'I.2.;,
15.3.2,
'5.4.1, 15.4.2, 15.4.3, and 15.6.311 (2) demonstrate that :ne SCE developed frem the Duke DPC-NE-2004 report is valid for the core TH m:cel developed in the Duke DPC-NE-3000 report; and (3) limit the use of OPC-NE-
- 2004 to the McGuire and Catawoa units enly.
Duke accepts the limitation Of restricting the current use of DFC-NE-2 004 to the McGuire and Catawba react 0rs only and agrees to notify NRC pr:cr Oc use of this technology on ether reactors.
Table 8 of OPC-NE-2004. provides ne tsnges tr.at were considered fer ne seven indepencent variables used in development of the RSM.
The finai
'(coded) ranges of these seven incepencent variables that 78 VIPRE cases frem which the RSM was sucsequently developed were usec :. the DNBRs-were > 1.0)
(ie, wnese are listed -ir Table 9 (atta:ned and unchanged).
Table i has been-rev: sea to clarify snat the 1:w flow limit for the RSM is i? \\
t-1)c of the zero point (nominal) flow.
A comparison of the RSM indepencent the SCD transient limiting MDNBR statepoints is providedvariable ranges - (revised Table it and in Table A attached.
The RSM independent.*ariable ranges do not envelope three
-transients: -; 5. 2. 8. 15.3.2..and 15.4.1. - Ouke has perf ormee an ace::::nal Monte Carlo prepagation of 3. 000 'pe,nts f or transient 15. 4.1 using :ne core 1TH.(14' channel) model frem OPC-NE-3000.
A 3,000 point propaga:::n cf transient 15.4.1 was also repeate us;ng the ::re TH (5 enannel) m: e1 from DPC-NE-2004.
+
?
Tor'each prepagatien, a simplifiec, but more accurate SCD approach was used. :Tigure A is_a flow enart
-The: shaded blocks of Figure A were deleted for the simplified SCD.f.-the standarc SC 3-presents a flow chart of tne samplified SCD approach which deletes :ne Tigure
- RSH and performs the Monte Carlo. propagation directly using VIPRE-01.
This is a more straight forwarc, more accurate,
.inl which to cetermine the SDL f:r a single statepoint.and more convenient manner F
D-8
i 1.s can be seen. n Table 3, tne two core TH ::eis CPC-bT-2 00 4 and :P C-bT3000) yield nearly identi:a1 SDLs.
Table 5 a.so snews the p.epagata:n results for transient 15.4.1 for 500 cbservat; ens.
The SDLs are h que:
than tha limits for 3000 ebservations as expe:ted wnen using a larger F.
factor for the smaller numcer of points.
Nevertneless, the SDLs are sta.:.
below the RSM based limit of
- 1. 40 and tne EO:.s f:r tne two VIPPI-01 ecels are nearly identical.
Monte Carlo propagatt:ns f or 5 00 points were also perf ormed f or transients 15. 2. 8 and 15.3.2 using the CPC-!fE-3000 VIPPI-01 model.
The SDLs for tnese cases are also tel:w tne ESP based limit f
1.40.
Two ecnclue cns resuit:
1.--
The existing ESM SDL of 1.4 s t:uncing f r all Chapter ;5 E 0 transient ases.
2.
Prepagation of either core model CPC-hT-2 004 o r DPC-bT-3 0 00) yields a nearly identical SCD linit (1 ;38 vs. 1.291); thus, tne 500 ilmat :f 1. 40 can te uses f:: steady-state and
-_ t ransient anal"4ses.
l 5
D-9 c
-~w
,,.r
..,w.,---,,...m.,,
....-wcy..r%m-m,3y.,.,.-.-..-,..-m--
r.
.... - ~ -..=- -.-
3 lable 9.
SCO Stateooint Parameters l
l G
I i
em
" Core Flow = 0.94(382,000 gpm) = 359,080 gpm 6% Core bypass flow The RSH was cevelopea using a core bypass flow of 6% ana a RCS flow of 382.000 gpm.
The core flow snown acove was only used to develop the RSM wnich includes flow as an incepencent variable.
The generic M/C thermal-hyoraulic analyses will be cased on a core bypass flow of 7,5% ano o RCS flow of 351.000
- gpm, The F5M was cevelopea to cover up to a pressure of 2600 psia, but the n*C",' C'd?
- rrela
- ::n u eniv valid u: to a cressure of 455 psia.
D-10
Table 9.
RSM Ir.out ano V! PRE-01 ano RSM Results n.w D-Il
i l
)
I J
F Table 9.
RSM (nout-anc v! PRE-01 ana RSM Results l
(Continueo) l i
i I
t l
-h I
i s
I e
(
F F
E l.
f
)
D-12
_ _... _,. _ _. _. _. _.. ~.,.. _... _ _ _. _ _ _ _...... _ _.
Y 1
f 740.e 4 20;
- a n t. e ra i
- 1.. ; i t & & eD C a r.t 3 00re-
- re n.et
-i'
- re /: net
- .et f
't cerat.:e
?:ess.te negr:r.:
+65
.:s
_reng es est
>eri r,
. at m
'R$" ranceri 15.2.8
- c
- a
+.
3..e
- es
- a. a. 3. 6 4
.. 4.
l-
.c.4
.a 4,
- 00.%
ftt ies b,
$0 % -
Yes
.0.
'.D Yes Yes 15.4.3 4
A. Oroppect fes
' Roa j
D.~-dingle RJS tet W i t r.3 r a w a.
. Stati:AA.y fet Misaligneo):c L
- 5.6.3
- unceo cy ;5.0.9.
1
- -.ra. ::re.*.e: :.:.
.smina. ::re tr.et f.:w = ;-: 06 6 (152,0 00 ;;:m
= 31).030 qcm 1
two =t!'erent reac::v;;y.nsert;:n rates D-13
...:.w-.-._.-.-,.-.,...,-
I t
Table B.
Monte Carlo Uncertainty P::pagatien Results (VIP PI-01 )
Stancard Coefficient case
'1 Wan Oeviatten of */ariatten
- DL' 15.2.8
- PC-NE-3000 VIPRE-01 mooel 100 15.3.2.
.OPC-NE-3000 VIPRE-01-moos 1 100 f
.15.4.1 DPC-NE-3000 VIPRE-01 mocal 500 CPC-NE-200 4 VIPRE mooel 500 OPC-NE-3000 VIPPI-01 memel-2000
'OPC-NE-2004 VIPPI-01 mooel -3000 3DL :.imit 1.40 (OPC-NE-2004)
=
D-14
FIGURE A RSM B ASED S C D Develop Matrix of
(~ 78 Cases)
VIPRE: Cases-Run VIPRE: Casest Develop RSM
( ~ 36 Coels. )
Core Power, Tin, RC Flow, Bypass Flow, Pressure.
Determine Uncertainties E
Fo H measurement,F nH, F6 H Spacing, FZ, Z, RSM, Code /Model, CHF Correlation I
10 Oper.
Propagate Uncertainties Statepoi1ts
( Monte Carlo - RSM )
~
F(
Determine SCD Limit and Design DNBR Limit (DDL) 1, To Applications l
was
FIGURE B
" SIMPLIFIED" S C D Core Power, Tin, RC Flow, Bypass Flow Pressure, Determine Uncertainties E
FA H-measurement, F aH, Fo H Spacing, FZ, Z, Code /Model, CHF Correlation o
10 Oper.
F,&,; agate Uncertainties Statepoints
( Monte Carlo - VIPRE )
1F Determine-SCD Limit and Design-DNBR Limit (DDL)
.; n
-- To-Applications -
vis
..._______.m.
4.
The response surface must be re-evaluatec wnenever conditi:ns provided in response to the ::RC :r.:estien 7 in Reference 7 are present.
1
)
Pescense-As explained in the response o Questi:n 3 above, a simplified method for determining an SDL using VIPRi-01 directly will be used whenever one of the following occur:
1) a significant change is made in the fuel asserraly design 2) a new or revised CHF correlation is developed 3) operating conditions outside of the range. of conditions listed
.in revised Table S are of interest A submittal will be made to tne imC for review if a new SDL is calculated due to the one of the following:
1)
A completely new fuel design would typically. require development of a new SDL.
However; development Of a new design feature, suen as a new spacer grid design, may only require propagation of the most limiting statepoint to validate the existing limit.
The imC would only be notified if the design changes result in the calculation of an SDL > 1.40.
2)
Development of a new version of tne BWC!W correlation or use of a different CHF correlation would require calculation of a new SDL and a submittal to the tmC.
3)
Any statepoints outside of the ranges given in revised Table B would be propagatec using VIPPI-01 as was done for FSAR transient :,5. 4.1.
A submittal would be made to the :~AC.cnly if tnis propagat:.on results in an SDL > 1.49.
t I;
^
l D-17
5.
Whenever CPC intenes to use other. CHF correlations, power distributions, fuel pin conduction models, or any cener.nput parameters and def tuit ptiens wnien were not part of the original review of the VIPP.E-01 code or part of this review, DPC must suomit its justif t:atien for !!RC review and approval.
OPC further must justify that either til their use would not impact the response
= surface already generated and the new C:!BR is bounded by that determined cy the current CCD metnco er (11) a new surface must ce generated.
Pasvonsa Justification will be submitted for ?!RC review whenever the approvec selection of VIPP.E-01 mocels and correlati ns is cnanged.
f D-18
1 6.
Core bypass flow is cycle cependent.
The RSM presentec :,n :ne tcpical report was developec using 6 % and a RCS ficw of 382,000 gpm.
DPC :tates that :ne genera: M/C nermai-nyorculi: anal /ses were performec based en a 0:re b'/ pass fi w of 7.5 % anc RCS flow of 385,000 ypm.
OPC will st:fy, in future applications, :nat its use of a part:.:ular core nypass flowrate is conservative or bounced by that value used in the sub e:t tcpical report.
In such event, CPC sill reassess :ne neec f:
regenerat;:n of a new response surface.
Remonse The generic M/C thermal-nydrauli: analyses casec cn the 500.etnoce.
.y assume a nominal core cypass flew of 7.5 4.
PC will perform an evaluation eacn cycle to verify nat this cypass ficw is greater inan or equal to the cycle-specific bypass flow.
- PC will also verify eacn cycle that the cycle-specific core flow (system flow minus the cypass flow) is bounced by the RSM tiow range given in the revisec Table 8.
If not, an evaluation will be performea as curlined in the response to quest:On 4 D-19
~~
Justify use of the 500.ami
.f
. 4; for Westingnouse Opti 1:en fuel (CFA), initially determanec f:r M3rx-BW fuel.
Pe monme An SCD analysis nas ceen perf:rme to /er fy tnat ne SDL ceterminea f:r Mark-BW fuel (1.40) also appites t: Westinghouse (!!.) optimized fuel.
A Monte Carlo propagation of 3:00 points was performed using an CFA 6 channel VIPRE-01 model.
The.imiting starepoint for Mark-BW fuel (the conditions that yielded the n:gnest fDL) was used for this propagati:n (case 2 in Table 11 of DPC-!!E-2 004 ) :
n w
e s-e r
.w
~
100 %
100 %
2430 45.4 518.5 1.50 1.5 0.5 The same set of uncertainties and associated distributions used for Marx-BW fuel (Table 12 in DPC-NE-2004) were used in the OFA analysis.
Only three of the uncertainties:
Q,,,
T and F,,
are fuel dependent and u,,,,, y,,
the values given in DPC-UE-2004 conservatively bound the uncertainties for OFA fuel.
The mean, standard deviation, :ceffielent of variation, and SDL for Marx-BW and CFA fuel are given in Table C.
The results in Table C demonstrate that the SDL of 1.40 is conservative for CFA and Mark-BW fuel.
1 D-20 l
i l
i 1
1 i
1
-l Table C.
SDL Oc:::parison
' Mark-EW and CJ A Tuel Fuel.
Coefficien*.
Desien "I
of variatien SDL*
Marx-EW 2;;0 G4 A -
2000 SDL = 1.40 (DPC-NE-2004) e e,
Z n
W p
n
.s-,
- r m
100 U -100-i 2430 45.4 618.5 1.50 1.5 0.5
+
D-21
-+
' APPENDIX E' l
Handouts Presented in the October 7 & 8, 1991 Meeting with NRC-Staff and Contract Reviewers i
f d
a J
fi j'
i 6'
P I
5 6
ll,
i' 5e i
k' L
.E-1 I
i'.
. -. -.. -. =. -
Use of BWCMV CHF Correlation for OFA Fuel The BWCMV correlation is based on an extensive data.
base with parameters that span the OFA parameters, Z
although specific OFA data is not included.
Two important fuel design parameters are hydraulic diameter and rod-to-rod gap The OFA hydraulic. diameter (0.510 in.) and rod-to-rod gap (0.136 in.) are within the range of the BWCMV data base.
lj m
7 BWCMV results given in.BAW-10159P-A.show.
negligible trendLor bias to hydraulic diameter and
- rod-to-rodigap.
This led theLNRC to conclude in the SER that the BWCMV correlation applies to OFA fuel.
I.
f
.=
+
VIPRE-01/BWCYIV VIPRE-01/BWCMV and LYNX 2/BWCMV results I
have been compared to show that both codes give essentially identical results.
/
f Mean M/P Std. Dev.
VIPRE-01 124 0.9962 0.1045 LYtlX2 124 1.0006 0.1082 4
k A
Measured / Predicted CHF Vs.
Hydraulic Diameter 14 7
n= 12 a
i
(>
- LYNX 2
[
,L--8--
D-t g
a o
o VIPRE-01 ii0g 08 i
06 e
1 a
f 0 35 04 0 45 05 0 55 Hydraulic Diameter -Inches 00
T
\\
\\
l 9
\\
m
.e q
m C.
4 ZC a~
nM g )-
W CQ gM Z
g%
Q 14
~O
,.a O a
O I
M til r.
g g
4
=s G a:
Ex:1 >-
_g i
%-d
<a 6:is:
C y cc a
-o 9
9a o
9
.-o c
o
~
MEKG EXNXI E-6
10 Mixed Core Modelling
. Mixed core analyses specifically model both the OFA and MKBW fuel. The peaking associated with each fuel type is explicitly calculated and compared against MAP limits.
. MAP limits are the total allowable peak that would reproduce the DNBR limit functionalized against the magnitude and location of the axial peak.
. Mixe ' core MAP limits apply to both OFA and MKBW fuel.
. AFD limits are set based on the allowable peaking as dictated by the MAP limits.
C-7
T r
Mk-BW MAP limits are applicable to OFA fuel in a transition core MAP limits are calculated for a full Mk-BW core based on a DDL of 1.55 A conservative transition core DNBR penalty for OFA fuel of 3.8 % is applied against the magin included in the DDL e
The SDL for OFA fuel is virtually identical to the SDL for Mk-BW fuel i
N SDL l
Mk-BW 3000 1.337 OFA 3000 1.322 n'/
I
r a
McGuire DNBR Penalties 1
f Statistical DNBR Limit (SDL) = 1.40 a
Design DNBR Limit (DDL) = 1.55 i
DNBR Margin = 10.7 %
t
-e
~pc ',
Transition Core DNBR Penalty
~
Optimized Fuel Assembly (OFA) total AP is
~
Mark-BW fuel assembly AP E
Flow will be forced out of the OFA fuel into the Mk-BW. fuel A generic transition core analysis was performed to
. determine a DNBR penalty for the OFA fuel w
1 3 R01R ETARY ftARK.RW vs nra porttitor hono e
E-Il
For the transition cycles containing Mk-BW & OFA fuel, the core s;afety &' operating limits are. based on analyses 1of a full Mk-BW core as described in.
DPC-NE-2004 7
- A conservative transition. core VIPRE-01 model.was developed consisting of one OFA assembly (the hot-assembly) and the rest of the core'is modeled as Mkl-BW assemblies h.
14 75 CHAfillEL ftARK-BW/0FA VIPRE-01 f:0 DEL 0FA i
45 l
\\
t N
i i
l fik-Bil 1:k-BW aN i
I
\\x fik-BW f k-Bit 11k-Bil 50\\
!4s de l
N i
N f!k-Bli fik-BW fik-BIl itk-BW 51 52 53 54 \\s N
N Ik-BW fik-BW f1k-Bil fik-Bt1 fik-Bil F55
!6 57 58 i
- 9 \\
N I
\\
N l
i
{ fik-B'i l
FA Design
( c fik-Bil
- lik-Bl!
ilk-Bil f!k-Bil itk-Bil r
t so 1
s2 s3 l
$4
- 5 N._
'2 i
Channel flo, I
l i
, Mk-BW Mk-BW '
itk-BW f k-Bil f1k-Bl!
ilk-Bil ion 37 sa 39-l o
t l-
.! fik-Bil.
i I
itk-Bil f:k-Bil I, lik-BW I
E-13
f.
4 i
1 The following conditions were analyzed:
- core safety limit:statepoints Ch.15 SCD transient limiting statepoints v
inlet & outlet skewed axial power shapes &
corresponding max. allowable radial peaks I
s
The transition core MDNBR is compared with the corresponding MDNBR for a full Mk-BW core to-determine the tiansition core penalty T
C Transition Core DNBR Penalty = Mk-BW DNBR - Transition Core DNBR.
Mk-BW DNBR 1
'h y
+u
- - -----w
b.
OFA Transition Core DNBR Penalty = 3.8 %
The generic transition core penalty 'is applied against-i the margin (10.7 %) included in the Design DNBR Limit (DDL) of 1.55 W
)
Instrumentation DNBR Penalty The instrumentation penalty accounts for measurement biases that are constant in direction and magnitude.
McGuire instrumentation biases:
E Power.(feedwater venturi fouling) l Pressure (Barton transmitters)
RCS flow (Barton trans. & FW venturi fouling)
The instru. mentation penalty is based on individual parameter sensitivity studies.
l fotal instrumentation DNBR penalty = 1.9 %
(McGuire)
N O
}
Flow Anomaly Several1 Westinghouse 4-loop plants have observed an anomalous condition concluded to be.an aperiodically i;
occurring vortex flow disturbance in the lower plenum.
Data was taken at 19 plants to determine if the anomaly was.present or not-(WCAP-11528, April 1988)
The flow anomaly does occur at both Catawba units, but is not observed at either of the McGuire units.
A DNBR penalty was calculated to account for the flow anomaly at Catawba.
n
f r.
Catawba Flow Anomaly I
The flow. anomaly core inlet flow. distribution was determined based on Catawba core exit thermocouple
- r and power distribution data.
The DNBR penalty was calculated in the.same manner
. as the' transition core penalty, comparing the flow-anomaly inlet flow distribution DNBR results with the generic DNBR results
,}
L N
- r*.
r-e 1
McGuire DNBR Penalties f
/
DNBR Penalty-Mk-BW
_OFA Transition Core
.0 3.8 %'
i Inst rumentat i on/Ila rdwa re 4.7 %
1.9 %
7 5
Rod Bow 0
3.5 %
' Flow Anomaly 0
0 Total DNBR Penalty 4.7 %
9.2 %
Available DNBR Margin 6.0 %
1.5 %
/
J v4 i
7 VIPRE-01 SCD Methods and Application to Transients The Statistical Design Limit (SDL) of 1.40 calculated in DPC-NE-2004 is conservative for all conditions where the SCD methodology 1s.used.
t
+
4 1
i t
VIPRE-01 SCD Methods 1
The " Simplified" or Brute Force method runs the VIPRE-01 code instead of using the Response Surface Model (RSM) to calculate DNBR.
FIGURE A depicts the analysis flow path using the RSM. FIGURE B shows the same process with the Simplified method.
i l
c
LT FIGURE A RSM B ASED S C D D~evelop::! Matrix:Lofr
(~ 78 Cases)
VIRRt= Casesi..
1P
,. 1k
.f.,
... +
' s,
~ ' ' 'Y
+
I
'RUrirVIPRECasesk 1
ir p
yyyelop-RSMi
~
~
( ~ 36 Coefs. )
n C+w l
l
[
Core Power, Tin, RC Flow, L
Determine Uncertaintias Byp ss Flow, Pressure.
l FA H-measurement, F H,
E Fe H Spacing, FZ. I RSM,
[
Code /Model, CHF Correlation y
I 10 Oper.
Propagate _ Uncenaikties
,Statepoints
-( Monte Carlo - RSM )
Jk.
Determine SCD Limit and L
j
. Design DNBR Limit (DDL)
To Applications g3 l
zs FIGURE B
" SIMPLIFIED" S C D Core Power, Tin, RC Flow, Determine Uncertainties Bypass Flow, Pressure.
FA H-measurement F".1H, FA H Spacing, t-l., Z.
Code /Model. CHF Correlation Oper.
Propagate Uncertainties
~
'Statepoints -
-( Monte Carlo - VIPRE )
e t
1P Determine SCD Limit and Design DNBR Umit(DDL)
To Applications E-24
TABLE 1 The following values are from the simplified method: analysis:of theistatepoint;for the LFSAR Section=15 312Ttransient,1 Complete Loss of RCS Flow..
~
VIPREH01:
3 Case.
Power-
- RCS Flow Teold Pressure-FdelH-Fz; Z
h1DNBRE 0
,95.7%
- 71.79 %
1561.9: F-2320.0 #
1.500 1.550 0.70
- 1.601 2:
~ 95.9%
170.18 %
- 564.8 F 2325.5.#
1;515-L528 0.67 1.519-10 '
96.9 %
L71.38%
561.8 F 2328.2-#
1.489
. l.473..
0.69 1.731' a
100L
- 95.6%
171 A0%
564.0 F-2330.0 #.
1.460 1.561:
0.66 1.787 250 94.5%.
74.65 %
- 563.0 F 2329.0 #l 1.539 1.575 0.72 1.467.
7 2
500 95.2 %
171.67 %
563.9 F 2301.8 #
1.463 1.594 0.68 2.000-The S.DL is calculated as described in DPC-NE-2004..
1 l
SDL= 1/ l 1- (Chi Square
- Kfactor.
- CV) l i
where MeanL= 1.619 Kfactor = 1.763 Stand. Dev = 0.2119 Chi Square = 1.055 Coeff. of Variation.= 0.1309 SDL = 1.322 I
There are no differences in the method between the 500 and 3000 case runs.
_.s.
=.
I
Application To Transients l
I The application of the SCD methodology to a transient l
l is determined by the conditions at the point of MDNBR.
7 l
l 1
l m
31 Table A.
SCD Transient Lin...ing Statepoints Core Core Inlet Core Power Inlet Flow Temperature Pressure Basis TSAR f% FP) f%)*
f'F) fosia) 1 F.
?
for SDL (RSM range)
- L.2.8 Simplify
- S.3.'
RSM
,1.3.2 31mplify
'. 5. 4.1 Simplify
- 5.4.2 a.
100 %
RSM RSM b.
50 %
RSM c.
10 %
RSM RSM
'. 5. 4. 3
- a. Dropped RSM Roc
- c. Single Rod RSM Withdrawal I
- c. Statically RSM Misa11gned Rod L
5.6.3 bounded by 15.2.8
% of. nominal core inlet flow Nominal core inlet flow (1-0.06) (3 82,000 gpm)
=
359,080 gpm
=
two different reactivity insertion rates E-17
32 cj. -
U..
able B.
Monte Carlo Uncertainty Propagation Rewits (VIF PI-01 )
standard Coefficient tasa
'J Mean hviat1on of Varsat1on
__?9t*
- !.i 8
OPC-NE-3000 */:P.5-01
..oo e l
!00 1
- !.3.2 pPC-NE-3000 */IPRE-01 model 500 i
13.4.1 DPC-NE-3060 V: PRE-01 model 500 DPC-NE-2004 V: PRE-01 mod '
500 DPC-NE-3000 */IPRE-01 mot 3000 DPC-NE-2004 */ PRE-01 modei 3000 SDL Limit = 1.40 (DPC-11E-2004) i -
E-28 9P4
7 Application To Transients 1
The Statistical Design Limit of 1.40 determined in DPC-NE-2004 conservatively bounds all SCD i
y transients.
i l
I i
L i
1
5Y UCBW From Zero Power Peaking Analysis:
. A three dimensional peaking analysis is performed at the MDNBR statepoint.
. Power distributions are evaluated for the two highest worth (differential and integral) sequential control banks moving in 100% overlap. Fully inserted to fully withdrawn bank positions are analyzed.
. The resulting power distributions are compared to the allowable peaking provided by the MAP limits.
. Acceptable results are no DNB.
. Typical maximum radial and axial peaks are less than 2.25 and 3.0, respectively.
E-30
55 UCBW From Zero Power Peaking Analysis cont'o:
In the event that neuative DNB mareins are found, the following analyses would be performed.
- a. Analyze the suspect shape (s) in VIPRE, or
- b. Reanalyze the UCBW fro.n zero power with cycle-specific values, or
- c. As a last resort, redesign the core.
E-31
UCBW AT Zero Power SCD StatepointS e
The SDL for the UCBW transient from Zero Power is calculated for two peaking conditions. The first condition is the transient statepoint. The second condition is the largest radial and axial peaking pennitted by the MAP limits.
Condition Power RCS Flow Tcold '
Pressure FdelH Fz
-Z 4
l 2
- t..
e:
A 500 case simplified method analysis was performed for'each condition.
The results are as follows:
Standard Coefficient i
Condition Mean Deviation of Variation SDI.
I 1.914 0.2381 0.1244 1.301 2
1.596 0.1798 0.I127 1.267 h
?
~
~~'~
q q<
'. ~ ' " ' ' ' ~ '
-