ML19332G124
| ML19332G124 | |
| Person / Time | |
|---|---|
| Site: | Yankee Rowe |
| Issue date: | 12/12/1989 |
| From: | Papanic G YANKEE ATOMIC ELECTRIC CO. |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| References | |
| BYR-89-175, NUDOCS 8912200189 | |
| Download: ML19332G124 (14) | |
Text
_
- t
"[ ~ NANKEEATOMICELECTRIC COMPANY.
"'C"*$g"g]((1l" n
~.
1m"~ -
580 Main Street, Bolton, Massachusette 01740-1g98
.u
- TAN KEE t.
December 12, 1989' BYR 89-175 United States Nuclear Regulatory Commission Document Control Desk.
Washington, DC 20555 Attention:
~Mr. Robert C. Jones, Acting Chief
' Reactor Systems Branch' Division of Engineering and Systems Technology Office of Nuclear Reactor Regulation
References:
(a) License No. DPR-3 (Docket No. 50-29)
(b) USNRC Letter to YAEC, " Request for Additional Information for Topical Report YAEC-1363," dated November 13, 1989
Subject:
Additional Information on YAEC-1363, "CASMO-3G Validation"
Dear'Mr. Jones:
Reference (b) regrested additional information in support of the subject topical ryort.- Enclosed are the responses to your questions. We trust that you will find this information satisfactory. However, should you desire
-further information or clarification, please contact Mr. Richard Cacciapouti of our staff.
Very truly yours, YANKEE ATOMIC ELECTRIC' COMPANY
'?
l w
I-Georg apanic, fr.
L.
Senior Project Eaginee l5 Licensing l
GP/gjt/0672v-
~
t.
Attachments cc: USNRC Region I USNRC Resident Inspector, YNPS go '
o t
t
~~
8912200189'891212 PDR ADOCK 05000029 ID _
Upcau)
1 2
RESPONSES TO REQUEST FOR ADDITIONAL INFORMATION CONCERNING YAEC-1363, CASMO-3G VALIDATION QUPSTIONS AND ANSWERS b
- 1) Q.
Nhat error is introduced in the gamma detector response / bundle power correlation by using the 10 group rather than the 18 gros:p library?
A.
The energy group structures for the 10 group and 18 group libraries are provided in Table 1.3 of YAEC-1363.
The 10 group library maintains essentially the same energy group definition as the 18 group library for the lower energy groups up to 1.0 MeV, with a coarser energy group structure in the high energy range.
The gamma energy distribution in a Westinghouse 3.1 w/o unshimmed assembly is provided in Figure 1.1 for both the 18 group and 10 group gamma cross section libraries.
Comparison of these energy spectra demonstrates that the loss in clarity at the high energy levels with the 10 group library results in a small effect on the overall gamma flux energy and opatial distribution.
The total detector responses were calculated for this problem using thc CASHO-3G default gamma detector sensitivity" functions for iron, and were 2.030 X 10" 1/cc-sec and 2.033 X 10 1/cc-see for the 18 group and 10 group libraries, respectively. This represents a detector response difference of 0.15 percent.
- 2) Q.
Nhat specific effect required the addition of a new energy group (group 32) to the U group neutron libraryt A.
Energy group 32 was added to the 09 group neutron library in order to provide a 1.855 eV boundary for editing purposes.
Addition of this group has an insignificant effect on the calculatic.al process.
- 3) Q.
Describe the adjustments to the U-238 and silver / indium resonances made to isprove agreement with measurement data.
A.
Most of the cross section data was obtained from ENDF/B-IV, with some fission spectra data from ENDF/B-V.
The U-238 resonances are adjusted to agree with Hellstrand's resonance integral measurements.
Ear: lier versions of CASMO-3G had minor adjustments in the strongly stielded silver and indium resonances for control rods. However, for the latast version of CASMO-3G, resonance self shielding in silver and indium is calculated with no adjustments to the silver and indium resonance library data.
- 4) Q.
Describe the "special" transport calculation for the fuel, clad, and 1
9
~
).
i*
. aoderator regions which is used to describe the unit-cell butter region.
A.
The aspecial" transport calculation in MICDURN-3, mentioned in page 10 of YAEC-1363, refers to thu homogenization process used to calculate buffer zone cross sections for use in the main transport calculation of the absorber pin, clad, moderator, and buffer zone.
The calculation is made using collision probabilities in the three regions (fuel, clad, and moderator) which defins the uniform lattice of the buffer zone.
The flux distribution obtained from this calculation is used to calculate cross sections for the homogenized buffer zone. The word "special" was used to differentiate the buffer zone calculaticn from the main transport calculation.
- 5) Q.
Describe the procedure used to treat the BNR control rod wing which preserve the rod blackness.
A.
Effective cross sections for a cruciform control rod which contains cylindrical absorbers are determined such that the region averaged reaction rates and fluxes are preserved in the 2D COXY calculation, where the control rod is represented as a homogeneous slab.
This process consists of the resonance calculation, microgroup calculation, macrogroup calculation, and cross section homogenization using control rod collision probabilities. These steps are described in detail below.
RESONANCE CALCULATION: The 40 or 70 group cross section library is adjusted for resonance absorbers via a collision probability calculation similar to that performed for a cluster control rod.
Dancoff factors are calculated using the same method as for fuelled cells with the fuel cell pitch replaced by the pitch between the absorber cylinders, and with the moderator replaced by the cruciform rod structural material.
An effective Dancoff factor is used which properly accounts for the cylindrical absorbers at the end of the cruciform blade.
MICROGROUP CAICULATION: The microgroup calculation is performed to obtain detailed neutron energy spectra to be used for energy condensation and spatial homogenization of the individual poison rods. This is performed with a homogenited buf fer region surrounding the poison rod / cladding / coolant cell.
The buffer region is defined by a microgroup calculation consisting of a lattice averaged fuel rod, cladding, and moderator regions. The buffer, region is 2.5 mean free paths (at.625 eV for the 70 group library or.500 eV for the 40 group library) in radial thickness.
Flux spectrum correction factors are calculated for the poison rod which properly account for the differences in the calculated spectra between a homogenized absorber cell and one explicitly modelled.
These factors are conder. sed to the 2D group structure for use in the 2
l L
(..
COXY calculation, along with the flux-volume homogenized cross sections, in order to preserve the reactivity and reaction rates in the lattice.
MACROGROUP CALCULATION: The macrogroup calculaticn is performed in cylindrical geometry with the cross sections for the regions deter 1nined by flux and volume weighting of the pin cella from the microgroup calculation.
The calculation is performed twice, once for the geometry with the wide water gap as the outermost annulus and once for the same geometry with the narrow water gap. The resulting fluxes are used to collapse the lattice cross sections to the 2D calculation energy group structure.
The flux energy spectra from this calculation are used for the homogenization of the control rod cross sections.
!!OMOGENIZED CROSS SECTIONS Cross-sections for the 2D COXY calculation are calculated using a Control Rod Collision Probability subroutine, CROCOP. The microgroup cross sections are collapsed into a
macrogroup cross sections by flux weighting using the moderator spectrum from the pin cell calculation. Condensation to the 2D group structure is accomplished via flux weighting with the control rod flux obtained from the macrogroup calculation.
The neutron current into the homogenized control blade is calculated in the macrogroup structure using escape probabilities. The neutron current is collapsed to the 2D group structure and then used to calculate the flux distribution within the blade. The geometrical model used to calculate the flux distribution is an infinite array of absorber cells, with each cell divided into two regions.
The inner region contains the absorber macerial and the surrounding rectangular region contains the structural material and moderator.
Effective cross sections are then determined such that the blackness is preserved for each energy group in the 2D calculation group structure.
- 6) Q.
Discuss in detail the applicability of the COXY neutron transport collision probability code to the calculation of gamma t:ansport.
For exangpie, what is the effect of the " nearest neighbor" coupling assun\\ption in the gannan transport calculations?
A.
CASMO-3G uses the transmission probability routine COXY for the gamma transport calculation.
This model is equally applicable to g.tmma transport problems as it is to neutron transport.
Although the numeric solution to the transport problem is solved with coupling only between neighboring meshes, the gamma (or neutron) transport from any individual mesh to another mesh anywhere in the bundle is accurately modeled by the transmission probabilities and interface currents on mesh surfaces.
An older CASMO version, CASMO-1G, used the collision probability 3
[
~
j
)
l
'1 module CPM for the gamma transport calculation. The CPM approach is accurate but very slow. The more efficient COXY model was therefore chosen for CASMO-3G.
To illustrate the validity of the COXY module, gamma detector responses were calculated using CASMO-3G and CASMO-1G at 0%, 40%, and 704 void for exposures from 0 to 20 GWd/T for Bundle Type 5 in the Hatch BWR, as reported in Section 2.6 of YAEC-1363.
The two models agreed closely, with an RMS difference of 1.16 percent and a maximum difference of 2.73 percent.
- 7) Q.
Is the gamma transport theory option applicable to PNR assemblies?
A.
Yes.
Resulta using this optie.
,re presented as part of the answer to Question 1.
The option 1.
.ed to calculate Seabrook Station's fixed gamma detector power to signal response functions.
- 8) Q.
How large are the G-factor material dependent multipliers for typical BNR and PNR assemblies?
A.
G-factors are used as multipliers on the absorption cross section in areas of strong gradients to match fine mesh diffusion solution (DIXY) reaction rates to the more accurate transport solution (COXY).
This improves the accuracy of fine mesh diffusion theory codes such as PDQ. However, the new methods described in YAEC-1363 and in YAEC-1659 do not involve the use of a fine mesh diffusion theory code such as PDO, so G-factors are not used.
Nevertheless, G-factors for unshimmed and 8 shim 14 x 14 CE PWR lattices were provided in Table 2.4 of YAEC-1363 and are presented in Table 8.1 with the relative absorption rate differences presented for the individual regions of the assembly (Table 2.4 provided the maximum relative difference).
G-facters for a Vermont Yankoe 8 x 8 CE BWR high energy bundle at 40%
void are presented in Table 8.2.
This bundle contains seven fuel rods of three different U-235 enrichments which contain 5 w/o gadolinia burnable absorber. There are also four water rods, and the wide and narrow water gap regions. G-factors are calculated for each of these regions and are presented in Table 8.2 along with the relative absorption rate differences between the DIXY and COXY calculations, with and without use of the G-factors.
Tables 8.1 and 8.2 demonstrate the difficulty tha't diffusion theory has in modeling strongly heterogeneous lattices without using G-factors.
The CASMO-3G / SIMULATE-3P methodology utilizes the COXY calculated intra-cssembly detail when calculating the pin-by-pin power distribution in an LWR core.
Therefore the application of G-factors is eliminated.
4
- 9) Q.
Now does the CASNO-3G definition of the flux discontinuity factors ensure the bundle surface (flux) gradients are preserved in the nodal calculationT A.
Discontinuity factors are defined in CASMO-3G as the heterogeneous assembly surface flux (collapsed from the multigroup transport calculation) divided by the homogenized surface flux.
The homogonized surface flux for each direction is computed by solving the one dimensional two group diffusion model of the assembly in which 1) cross sections and diffusion coefficients are defined by flux-volume weighting with the COXY calculated fluxes 2) the COXY net surface currents are taken as boundary conditions, and 3) the flux is assumed to have a fourth order polynomial flux shape that is the same as that used in the SIMULATE-3 code.
Since this flux shape is determined with the surface currents as boundary conditions, this assures that a SIMULATE-3 calculation corresponding to the CASMO-3G geometry will reproduce the CASHO-3G net currents, or flux gradients, on each surface.
- 10) Q. Since the CLOSEUP library only includes P, scattering, how is the gamma scattering saisotropy treated in the CASNO-3G gamma transpore calculation?
A.
The CLOSEUP library contains P, scattering cross sections, however cross sections are used by CASMO-3G to calculate the only the P3 transport corrected cross sections by the formula ot, -
E o'...
o.....
g'2g where c'
and o' are the P,
and P scattering cross sections, i
respectively.
The gamma fluxes within an assembly are typically used te calculate the detector response function for the assembly.
The flux distribution is flat and only sources which are close to the detector contribute significantly to the response.
Therefore the gamma detector response is insensitive to the choice of the transport correction.
- 11) Q. Provide a description of the method used in CASNO-4G to calculate the baffle / reflector cross sections.
Indicate the assumptions made in deriving the haffle cross sections and discuss their Lapact on predicted power distributions. How do these cross sections vary with fuel loading and core state variables? Are any ad-hoc adjustments or normalizations made to these cross sections?
How is the baffle / reflector region treated in cold shutdown margin or cold critical calculations?
5
l?..~,*
A. CASMO-3G computes baffle /ref16ctor cross sections by performing a detailed multigroup transport flux calculation in a one or two dimensional model containing one fuel assembly, the baffle, and the reflector.
Pin cross sections for this calculation are generated using the same microgroup and macrogroup calculations as in a standard assembly calculation.
The fuel assembly is used primarily to generate a flux spectrum in the baffle and reflector, and the few group reflector data is insensitive to the choice of fuel assembly.
The baffle and reflector can consist of from one to three material regions.
.The baffle / reflector data needed for few group core calculations is defined by the flux-volume weighting and collapsing of cross sections in the baffle and reflector regions.
These cross sections alone, when used in a few group diffusion calculation, will not properly predict the core leakage.
Therefore, CASMO-3G uses homogenization theory to define appropriate values of discontinuity factors for the baffle and reflector regions such that a few group diffusion solution will produce the same net current at the fuel / baffle interface as the multigroup transport solution. Subsequent use of these discontinuity factors and cross sections assures that few group diffusion solutions will properAy predict leakage in a core calculation.
No ad-hoc adjustments of the baffle / reflector cross sections are used in defining the CASMO-3G reflector data.
The SIMULATE-3P model uses the baffle / reflector cross section and Assembly Discontinuity Factor (ADF) data with a one node thick reflector, and an analytic boundary condition which assumes an infinite reflector.
This provides essentially the same solution as when an infinite reflector is modelled.
The reflector cross sections for the power operation range are typically functionalized by TABLES-3 versue moderator density and boron concentration for a PWR and versus void fraction for a BWR.
Therefore, cross sections for the reflector nodes in a SIMULATE-3P model are based on the actual reactor conditions.
For cold conditions, an isothermal cross section library is gesierated along with appropriate reflector cross sections.
Upper and lower axial reflectors are treated in the same manner as the radial reflectors.
The one dimensional axial model is represented by radially homogenizing the structural components, fuel rod plenums, and moderator above or below the fuel.
Cross sections and ADFs are generated as a function of boron concentration and moderator density.
Within a typical range of fuel assembly desig'n parameters, the spectral differences do not significantly affect the homogtnization.
A study was performed to demonstrate this.
It analyzed the effect of using reflector cross section data developed with a fresh assembly for a typical low leakage reload core, where highly burnt fuel is located in the core periphery. Figure 11.1 presents a comparison of the power distributions calculated with a ' fresh' reflector and a 6
r
' burnt' reflector.
The only difference in these models is the fuel spectrum used to homogenise the reflector cross sections and calculate the reflector ADFs. The ' fresh' reflector was calculated using a fresh 2.7 w/o U-235 assembly and the ' burnt' reflector was calculated using the same assembly depleted to 40 GWd/Mt.
The maximum assembly relative power difference was.005, with an RMS difference of.002.
Initial fuel enrichment could have been used in the same problem with similar results. This example demonstrates the insensitivity of the fuel model used to generate reflector cross section data on the global calculation.
- 12. Q. When the fundamental node solution is used to nodify the infinite lattice results to account for leakage effects, the calculation is carried out in either diffusion theory or the B, approximation. For typical Not and PNR isttices, what is the difference in the calculated lattice parameters obtained by these two methods and which nothod is recommended?
A. The fundamental mode calcu1Ltion provides a buckling (B') that solves the equations (E - E',) $ + D B' $ = 1/k T $
wheret E = the total cross section matrix T', - the zeroth moment of the scattering matrix
$ = the flux vector D = the diffusion coefficient B' = the buckling k = the system eigenvalue T = the fission spectrum The fundamental mode calculation can be performed in three modes:
- search for k. with D' set equal to zero (normal eigenvalue calculation)
- calculate k.,s..u,, with a user input value for B' (used to for pin cell critical calculations), or
- search for B' so that k.,,..o,, equals 1.
D' is then the equal to the material buckling B'..
This is used for assembly depletion and provides the spectrum for cross section collapsing.
In the diffusion theory approximation (or P approximation), the 4
i diffusion coefficient is defined as D," = 1 / 3E.,,,
(
For the B approximation, the diffusion coefficient is approximately:
j i
l l
D, = 1 / 3E.,,, { 1 + 4 /15 (B/E.,,,) ')
l 1
7
l',,-.
The B approximation resu2'.4 in a leakage term which is sensitive to 3
the buckling.
For the situation where k. is unity, the two methods result in the same diffusion coefficients.
For the situation where
- k. is not unity, for example a fresh unshimmed assembly, the B approximation provides incorrect values for the diffusion coefficient.
This is illustrated in Table 22.1, which compares the fast diffusion coefficients calculated using the B
and P 3
approximations for a 3.1 w/o PWR assembly with 0 and 24 integral fuel burnable absorber B.C shims. The shims are a thermal absorber, hence would not be expected to significantly af fect D.
However, the B 3
approximation results in a *l.3 percent change in D, compared to a 0.8 percent change in'Di for the P: approximation.
The change in the B diffusion coefficients with geometric buckling 3
creates a bias in cores which contain segregated regions of fuel of differing reactivities. Use of the B, fundamental mode approximation l'
has shown an out-in power tilt compared to measured data for fresh cores which have the higher enriched fuel on the periphery, and an in-out power tilt for low leakage reload cores with the fresh fuel located inboard.
The P method is therefore recommended over the B approximation because it results in better power distribution comparisons to measured data.
a 8
s a,
4 J.27, _..
Table 8.1
-DIXY G-f actors for CE Assemblies Percent Percent Initial Final Relative G-factor G-factor Relative Realon Difference 1 Group 1 Group 2 Differencel CE-0 Guide Tube ~14.29
.966228 1.18189 0.01 CE-0 Water Gap
-2.29
.996330 1.02988 0.03 CE-8 Guide Tube -17.06
.969943 1.19873 0.00 CE-8' Shim 16.06
.980833
.75787 0.01 CE-8 Water Gap
-3.70
.997647 1.03122 0.04 1 100 X (Absorption Pate,3,,- Absorption Ratecos,)/ Absorption Rate,,
n __
I.
9
L
! ' y,f.' 0. _.
E -. _.
Table 8.2 DIXY G-factors for a GE Hioh Enerav Bundle Percent Percent Initial Final Relative G-factor G-factor Relative
[.
Realon Difference 1 Group 1 Grouc 2 Difference 1 i
Gad Pin 1 18.00 1.00152 0.645233
-0.01
_ Gad Pin 2-19.15
_0.993092 0.661948
-0.08 Gad Pin 3 17.40 1.01724 0.619952
-0.02 i
-20.57-0.991679 1.16298
-0.03
- Box & Wide
-12.46 0.951264 1.13987
-0.06
?
Water Gap a
Box.& Narrow
-10.00 1.00949 1.07267
-0.04 Water Gap f
1 100 X (Absorption Rateen,- Absorption Rate.,) / Absorption Rate a, co I
l t
r 4
I l
10
- t,,
'.6 s,
.o
~ '.
Tnble 12.1' Dependence of D, Versus Preeenee of a ThernAl Absorber Westinahouse PWR Assembly i:
l' f
U-235 Diffusion Coefficient
'W/0 Number o.f BP's B.
P, 3.1 0
1.321-1.451
~ 3.1 8
1.352 1.448 3.1 20 1.401-1.442 I.'
l' 3.1 -
24 1.418 1.440 i:
i,;I o
1 e
11
i I
e..
\\
~
i
- 9 i
t r'IGURE 1.1 GAMMA ENERGY SPECTRUM AT DETECTOR WESTINGHOUSE 17X17 3.1 W/0 0 SHIM ASSEMBLY E
18 GROUP GAMMA LIBRARY E
6-l i
I b
a' 5 -
O P
)
N 4-5 i
h
)
3' N
i N-b
~
2-x 1-0' i
b 0
1 2
3 4
5 6
7 8
9 10
?
GAMMA ENERGY (MEV) 10 GROUP GAMMA LIBRARY i?i n6 Y
- a.,;
i O
?
g 4-o l
)
3 N7 i
e l
2-I I
i$
1-l L
s0 -
'k 0
1 2
3 4
5 6
7 8
9 10 GAMMA ENERGY (MEV) 12 e
(.
L..
e
-o e, '
<.g*
e o-Piaure 11.1
)
Core Power Distribution Comoarison Low Leakace PWR Reload Cvele
' Fresh' and ' Burnt' Reflector Cross Sections 9
10 11 12 13 14 15 16 17 9
0.912 1.189 1.223 1.237 1.125 1.014 1.056 0.883 0.914 1.192 1.226 1.239 1.127 1.015 1.056 0.882
[
.002
.003
.003
.002
.002
.001
.000
.001 10 1.189 1.325 1.101 1.350- 0.997 1.331 1.112 1.085 0.349 1.192 1.327 1.103 1.353 0.998 1.332 1.112 1.084 0.348
.003
.002
.002
.003
.001
.001
.000
.001
.001 r
l-11 1.223 1.108 1.005 1.220 1.309 1.151 1.099 1.071 0.306 i
1.226 1.110 1.007 1.222 1.311 1.152 1.099 1.067 0.304
.003
.002
.002
.002
.002
.001
.000
.004
.002 i
[
12 1.237 1.353 1.222 1.333 0.949 1.247 1.070 0.861 1.239 1.356 1.224 1.335.0.949 1.247 1.070 0.856
. 002
.003
.002
.002
.000
.000
.000
.005 13 1.125 0.996 1.308 0.946 0.854 1.093 1.002 0.345 1.127 0.998 1.310 0.947 0.854 1.093 1.001 0.343
.002
.002
.002
.001
.000
.000
.001
.002 14 1.014 1.329 1.148 1.241 1.006 0.989 0.397 l
1.015 1.330 1.149 1.241 1.086 0.988 0.394
.001
.001
.001
.000
.000
.001'
.003 15 1'.056 1.111 1.095 1.064 0.992 0.383 1.056 1.111 1.095 1.063 0.991 0.381
.000
.000
.000
.001
.001
.002 16 0.883 1.082 1.067 0.855 0.342 0.882 1.081 1.064 0.851 0.340
.001
.001
.003
.004 002 s
17 0.335 0.305 0.334 0.302 001
.003 0.335.
' fresh' reflector assembly relative,pewer 0.334
' burnt' reflector assmebly relative power
.001 difference = ' burnt'
' fresh' i
1 13 1
w