ML20247C176
| ML20247C176 | |
| Person / Time | |
|---|---|
| Site: | Peach Bottom  |
| Issue date: | 07/13/1989 |
| From: | Helwig D PECO ENERGY CO., (FORMERLY PHILADELPHIA ELECTRIC |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| References | |
| TAC-67190, TAC-67191, NUDOCS 8907240288 | |
| Download: ML20247C176 (44) | |
Text
_ - _ _ -
j W
J PHILADELPHIA ELECTRIC ~ COMPANY.
{
NUCLEAR GROUP HEADQUARTERS 955 65 CHESTERBROOK BLVD.
l WAYNE, PA 19087 5691 (as s) e4o-sooo July 13, 1989 Docket Nos. 50-277 50-278 TAC Nos. 67190 67191 U.S. Nuclear Regulatory Commission Attn: Document Control Desk Washington, DC 20555
SUBJECT:
In-House Reload Licensing'for Peach Bottom Atomic Power Station Units 2 and 3
REFERENCE:
March 13, 1989 letter from R. E. Martin (NRC) to G. A.
Hunger, Jr. (PECo)
Dear Sir:
The attachment to this letter provides answers to the questions in the referenced NRC letter pertaining to PECo report PECo-FMS-0005, " Methods for Performing BWR Steady State Reactor Physics Analyses."
On March 30, 1989 we informed.you that the answers would be provided by May 12, 1989.
With the concurrence of the NRC Project Manager, the answers were forwarded, for preliminary review, to the NRC's contractor on May 12, 1989 and were forwarded to the NRC staff technical reviewer on May 17, 1989.
We have expanded our answers to address the questions NRC representatives raised during the preliminary review.
If you have any questions or require additional information, please do not hesitate to contact us.
Very truly yours, v
s D. R. Helwig[j Vi e President 8907240288 890713 PDR ADOCK 05000277 Nuclear Services Department PDC p
orl Jh M Attachment e
(
L__._--__
p r.
V
.Docum:nt Control Desk July 13, 1989 Page 2 1.-
l cc:
W. T. Russell, Administrator, Region I, USNRC l-T. P. Johnson, USNRC Senior Resident Inspector R. E.: Martin, USNRC PBAPS Project Manager T.
E. Magette, State.of Maryland J. Urban, Delmarva Power R. A.'Burricelli, Public Service Electric & Gas H. C..Schwemm, Atlantic Electric T.'M. Gerusky, Commonwealth of Pennsylvania i.
l I
~
[,
~
^]
D RESPONSE TO NRC REQUEST FOR' ADDITIONAL-INFORMATION ON PECo REPORT PECo-FMS-0005 l
l PEACH BOTTOM ATOMIC POWER STATION UNITS 2 & 3 DOCKET NOS.
50-277 50-278 9 -
I'
~ - - -
.4.
=,
P:
[
puESTIONfl:
- I I
llow is the bias in the SIMULATE-E K rt critical da'a relative to K rr =
t e
e 1.0 accounted for?. What is the standard deviation of the SIMULATE-E Kert predictions relative to K rr = 1.0?
e RESPONSE; The mean critical SIMULATE K-effective pre' dictions as observed during each of the five Peach Bottom operating cycles are summarized in Table 1 below.
TABLE 1 Cycle Average Critical K-Effective Predictions, Kgfr Unit / Cycle' flot Model Cold Model 2/5 0.9939 0.9889 2/6 0.9970 0.9934 3/4 0.9899 0.9868 3/5 0.9933 0.9921 3/6 0.9967 0.9947 All Cycles 0.9946 0.9916 Within a given cycle, the SIMULATE critical K-effective predictions exhibit an exposure dependent bias, f(E),~ represented as:
Kerr(E) = K ((
- f(E)
(1)
-f e
where different bias functions, f(E), are used in conjunction with the hot and cold models. The ftactions f(E) have been derived by PECo using a polynomial least squares fit in the cycle exposure parameter, E (See Figure 1).
By projecting past operating experience to future cycles, an estimate for Keff is used in equation 1 to forecast expected hot and cold critical K-effective values throughout the design cycle.
The forecasted critical K-effective (or more conservative) value is used as a target eigenvalues in searching for the SIMULATE-E predictions of cycle energy, hot operating control rod patterns, cold shutdown margins, etc.
In this manner, the critical K-effective bias of the model is directly accounted for.._--
__-- - _ - _ _ _ _ a
4 Response to NRC Question il Continued:
The overall standard deviation for the hot K-effective values cited in PECo-FMS-0005 (0.0033AK) was calculated about the overall mean K-effective value of 0.9946. Likewise, the overall standard deviation for the cold K-effective values cited in PECo-FMS-0005 (0.0035AK) was calculated about the overall mean K-effective value (K rr) of 0.9916. The use of a standard e
deviation derived around the mean value is consistent with PECo's intended use of a biased target critical K-effective directly in SIMULATE. Relative to a constant K rr = 1 value, the hot and cold RMS differences would increase to e
0.0063 and 0.0091, respectively.
In future reload design and licensing efforts, PECo will utiHze either the target critical K-effective eigenvalue as estimated by equai cn (1), or a value which is conservative and appropriate with respect to the intended application.
j PECo will continue to monitor SIMULATE predictions of plant operating data in i
order to update its critical database, and to ensure the reliability of the forecasted critical K-effective estimators.
)
j i
)
In actual practice Kegt s determined based on data from the most recent cycle for the unit being designed.
1 i
' l E________________-.----
)
7 Q
4
- s-FIGURE 1 PBAPS Hot Critical K-effective Bias Cycle Exposure Dependent Bios Function D
1.0060 C
1.0050 O
g a
a 1.0040 -
gc i
. a.0 0 ao U
d 1.0030 -
oa o
O C 00 a:
1.0020 -
0 C
U h
00CN CD g
7 1.0010 -
O c
a ca U
U O O
q a
-5 1.0000 -
O cc o
C C e-o cb 0.9990 -
U U
O 0
0.9980. -
c oo*
0 CE CD3 Co Ok2 U
0.9970 C
Co U
D.99 60 y
r 0
2 4
6 6
10 Cycle Exposure (uwo/ST) 0 NORMAUZED OATA CURVETIT PBAPS Cold Critical K-effective Bias Cycle Exposure Dependent B*cs Function 1.0060 1.0050 -
1.0040 -
1.0030 -
d I2 1.0020 -
O o
8 1.0010 -
u o
c o '.
O 1.0000 -
E U
C S.
U 0.9990 -
O C
Ct2 O
O 0.9960 -
O O
a g
0.9970 -
0.9960 0
2 4
6 8
Cycle Exposure (WWO/ST)
O NORMAUZED DATA CURvCFIT _ ___ _ -___ - - _-_ _ _- -
w l
QUESTION f2:
Does the PECo physics methodology require three SIMULATE / measurement normalization during a burnup cycle (B0C, MOC, E0C)? Since this is not typical industry practice, why are these additional normalization required? What par 6mters will be adjusted and will this information be used to update the precalculated cycle safety analyses such as those of Appendices A-D?
RESPONSE
PECo statistical methods have classified the bias and uncertainties in the SIMULATE hot power distribution prediction at beginning, middle, and end of cycle (80C, MOC, EOC) conditions. This was performed in order to account for an increased axial bias in the SIMULATE power distribution predictions observed at the middle of cycle (M0C) condition.
In future reload design applications, PECo will either utilize statistical adjustment factors classified in this manner, or will otherwise ensure that the observed MOC SIMULATE biases are accounted for by the use of adequate design margins. Statistical adjustment factors will be derived from measured versus predicted TIP comparisons from previous cycle (s).
This statistical technique is restricted to the adjustment of the power distribution only in the axial direction, and only at MOC conditions. Thus, it should impact neither the calculation of ACPR, nor the calculation of reactivity kinetics parameters, both of which are performed in support of reload licensing. As further demonstrated in PECo-FMS-0005, the licensing events analyzed in Appendices A-D (e.g., RWE, LFWH) utilize inherently conservative procedures, with conservative margins.to adequately account for the modeling uncertainties discussed here. Consequently, PECo plans to directly use the predictions from the SIMULATE model, together with the conservative procedures as qualified in Appendit.as A-D, in the analysis of reload licensing events.
a QUESTION f3:
What isL he effect on the PECo calculation uncertainty estimates when only the t
.- PECo. calculations'are included in the benchmark comparisons?
l
RESPONSE
When the YAEC derived results are removed from.the statistical database, the 1following changes' occur in the statistics reported in PECo-FMS-0005.
(A) On page 4-121 (Table 4.2.1.2):
Overall RMS% Difference = 2.2%
(B)
On page 4-218 (Table 4.3.2.2):
VOID COEFFICIENT STATISTICS:
KENO-IV CASMO-1 Hean Void Coefficient:
-1.47x10-3
-1.43x10-3 Mean. Difference (CASM0-KENO):
4.27 x 10-5
^
-STD Difference:
8.26 x 10-5 Mean Difference X 100 2.95%
Mean STD Difference X 100 5.70%
Mean (C) On page 4-226 (Overall Void Reactivity Statistics):
For Uncontrolled,d Nodes:
CASM0-1 Code NORGE-B Code TOTAL p(BIAS) 3.0%
0%
3.0%
i o(UNCERTAINTY) 5.7%
2.1%
6.1%
1 l
For Controlled Nodes:
CASMO-1 Code NORGE-B Code TOTAL I
p(BIAS) 3.0%
0%
3.0%
1 o(UNCERTAINTY) 5.7%
4.4%
7.2% _ _ _ _ _ _ _ - - _ _
m - -
- (
_. g.
e
' Response to NRC' Question -#3 (Continued):
'(D)' On page 4-229 (Table 4.3.3.1)
CONTROL ROD WORTH STATISTICS:
KENO-IV-CASM0-1 Mean:
0.2793 0.2776 Mean Difference: (CASMO-KEN 0)
-1.76 x 10-3 STD Difference:
1.88 x 10-3 Mean Difference X 100
-0.63%
Mean STD Difference
.X 100 0.67%
Mean (E) On page 4-233 (Overall Control Rod Worth Statistics):
CASM0-1 Code NORGE-B Code TOTAL r
p(BIAS)
-0.7%
0%
-0.7%
o(UNCERTAINTY) 0.7%
0.9%
1.1%
Thus, deletion of YEAC-derived results has little effect on the overall statistics or conclusions as reported in PEco-FMS-0005.
Differences between the above values and those reported in PECo-FMS-0005 are within the statistical uncertainties associated with the individual KENO-IV calculations.
) )
L;-
r QUESTION #4:
Are the KENO-IV and CASM0-1 calculations independent? For example, are the nuclear cross sections used in the two < calculations different?
RESPONSE
i l
The KENO-IV and CASM0-1 nuclear calculations as reported in PECo-FMS-0005 were j
performed in an essentially independent manner. The CASMO-1 calculations utilized the ENDFB-III based 25 group neutron cross section library as supplied
)
i by Studsvik for production use.
Included in this library are corrections made by Studvik to the resonance parameters and cross sections for U-238 and Pu-240 to normalize code results to those from integral experiments. The KENO-IV i
calculations utilized the ENDFB-IV 27 group neutron cross section library as transmitted by ORNL in the SCALE-3 computer code package. The CASH 0-1 and KENO-IV codes themselves utilize different theoretical and computational formulations. CASM0-1 performs essentially a two dimensional Sg transport theory calculation in 10 energy groups, which have been collapsed from the original 25 energy group structure using collisional probability calculations in each of the fuel pin cells, absorber rods, and other non-fuel regions of the lattice. The PECo KENO-IV model performed a Monte Carlo calculation in 27 energy groups, tracking 360,000 neutron particle histories through their simW ated flight paths in the lattice geometry. Each of the fuel and absorber reO ons were discretely represented in the KENO-IV geometry to allow an accurate evaluation of the flux depression effects in these important regions.
l It was necessary to tailor the input streams to the CASMO and KEN 0 models in order to provide a consistent, and executable problem description in both.
In this regard, minor adaptations were made to the problem input descriptions in the following areas.
o Since KENO-IV lacks an internal thermal expansion model, the CASM0-1 thermally expanded dimensions and isotopic number densities were input into KENO-IV.
Because of KENO-IV code limitations, the rounded corners of the channel box o
wall were input as square (90 ) corners in the geometry input to both KENO-IV and CASMO-1.
o
- Again, because of KENO-IV code limitations, the internal absorber cylindrical rods in KENO-IV were modeled with a square cross sectional area equivalent to their actual circular shape. The spacing of these internal absorber rods was modified slight' to allow better alignment of subvo%me boundaries in the control rod wi
. hose in the fuel lattice.
The intent of each of these adaptations was to allow a consistent evaluation of the results derived by the independent nuclear calculations in the two codes.
No attempt was made to normalize results from the two codes, and the results from both codes were compared directly.
l i
QUESTION f5:
Ilow do the uncertainties derived f rom the EPRI and AB Atomenergi Doppler coefficient comparisons account for actual core conditions such as fuel burnup, fuel rod geometrical changes, fuel temperature uncertainty, spectral effects due to the presence of voids, etc.?
RESPONSE
CASM0-1 explicitly accounts for primary Doppler coefficient sensitivities related to fuel burnup and the spectral effects due to the presence of voids.
The CASMO-1 fuel burnup isotopics calculations have been ink pendently qualified by PECo in Section 4.4 of PECo-FMS-0005. Likewise, the CASH 0-1 calculation of void-related spectral effects has been independently qualified by PECo in Sections 4.2.1 and 4.3.2 of PECo-FMS-0005. Although the Swedish D 0 2
reactor measurements were performed at low irradiation, zero void conditions, the same CASMO-1 Doppler calculation techniques are applied in the BWR flux spectrum and isotopics environment for which separate CASM0-1 qualification has been provided. Thus, the comparisons cited for the test conditions reported in Section 4.3.1 of PECo-FMS-0005 are not expected to deviate significantly from those representative of other burnup and void conditions.
In order to further demonstrate conservatism in its estimate of the Doppler coef ficient uncertainty, PECo has performed an additional parametric study using the CASM0-1 lattice physics code. This evauluation quantified the relationship between the uncertainties in the actual core conditions (i.e., fuel burnup, fuel rod geometrics, fuel temperature, and in-channel void fraction) and the
' implied uncertainty in the predicted Doppler coefficient. Results from this parametric study are sumarized in Table 2.
In this table, the variations cited for each of the parameters were conservatively estimated based on information documented in the technical literature (1) (2), together with results from the PECo FROSSTEY (fuel performance) and SIMULATE-E computer models. As noted in Table 2, a conservative estimate of 4.9% results for the overall percent uncertainty in the predicted Doppler coefficient, as attributed to differences between assumed and actual core conditions. This value is bounded by the 10%
experimental uncertainty in the Swedish measurement data, which has been additionally applied to the overall Doppler reactivity statistics, as cited in Section 4.3.1.3 of PECo-FMS-0005.
r l
l l
l l t
.;; y )
6: i-
~i-
.e,
c m
- c. ;
Risp'.nst ty NRC'Questinn #5'Csntinusd:
n'
..j
- In its statistical ? analysis of the limiting'BWR delta-CPR pressurization events (i.e., generatorf oad rejection',~ feedwater. controller failure). PECo has; l
accounted for the Doppler uncertainty by the use of a 0.85 multiplier, as:
applied to the nominal, axially' dependent-core average Doppler coefficient y
predicted by the PECo: SIMULATE-E and SIMTRAN-E computer models.- ' This:is further
- described in PECo-FMS-0006, " Methods for Performing BWR Reload Safety..
Evaluations", which has been submitted to NRC for separate. review.
I f-I I
4 i
REFERENCES:
(1) ' NEDE-24011-P-A-US "Gornwal Electric Standard Application for Reactor Fuel *, September,1988.
(2) - EPRI-NP-2246-SR, ' Mechanistic Model for Prodicting Two Phase Void Fraction for Water in Vortical Tubes,
, Ctannels, and Rod Bundles", Fotxuary,1982.
I
(: I
- y, 1
TABLE 2 Doppler Coefficient Parametric Evaluation PERCENT UNCERTAINTY VARIATION IN PARAMETER IN DOPPLER C0 EFFICIENT PARAME.ER FROM NOMINAL VALUE PREDICTION 1.
Fuel Burnup i 10%
i 1.5%
2.
Fuel Rod Geometrics A) Pellet Radius
+ 2%
+ 2.0%
i B) Rod Pitch
[+ 2%
i 2.0%
3.
Fuel. Temperature i 14%*
1 3.0%
'+ 2.0%
4.
In-Channel
-+ 0.05 VF**
Void Fraction Overall Doppler Coefficient Uncertainty..
1 4.9%
- Based On Predicted AT As Calculated Between The Pellet and Coolant
- Based On Core Average Void Fraction Conditions (VF=0.40) i {E__ _
~ t :.
- s' f.'
f : QUESTION 16:-
-t:
How is the expression for (AK/K)DOP-(Page 4-209) derived?-
-What is the relationship between-Knop _and the CASM0-1 cross sections?
RESPONSE
~The two group,'Km equation.u' sed by SIMULATE is:
,,a v[fi v[fa~
-[r-K(T)
- +
=
[r la2 la1(T)+[,
'In.this expression, only [ai(T) is represented as a function of fuel temperature, T:
-[as(T)-
a'T18 + b
=
Where a-and b are constants.
Define:
V1ft V1f2
. [,
o
=
__ +
Eri las as a parameter independent of T.
Then:
0 K(T)
=
[ai(T)+[,
1 1
Ko - K(T)
'o.
=
[ai(To)+[r
[at(T)+1r Ko - K(T) lai(T) - lat(To)
____s_.
a K(T)
[, + [at(To) a (T18 - To18)
=
[r + lai(To) _ - _ _ ___ ---
R~spons,ti NRC Questirn #6 (Continued).
l or:
KDOP (T 8 - To18)
Ko - K(T)
=
K(T)
.Where:
a KD0P 1r+1ai(To) 2 9 _ _ _ _ _ _.
e QUESTION #7:
t flow are the rodded and unrodded void and Doppler reactivity and coefficient uncertainties combined to determine the core reactivity coefficient uncertainties?
RESPONSE
The void and Doppler reactivity statistics cited in Section 4.3 of PECo-FMS-0005 are representative of the bias and uncertainty in the 3-D model's local prediction of these parameters at a single node (or axial segment of a single assembly).
Because of the greater complexity associated with fitting the rodded cross section data, the statistical uncertainties in the prediction of reactivity components for the rodded nodes are slightly greater than for the unrodc'ed nodes. The way these nodal uncertainties are accounted for depends upon the specific licensing transient analyzed (e.g., rapid pressurization events versus slow cold water insertion events), and the method of kinetics solution employed (e.g., 1-D versus point kinetics). For slow events in which point kinetics solutions are employed, conservative multipliers are applied to each of the core average reactivity parameters as calculated by the 3-D model.
For rapid pressurization events in which 1-D kinetics solutions are employed, conservative multipliers are applied to the appropriate 1-D neutron cross sections on a nodal basis. For end-of-cycle licensing analyses unrodded uncertainties are applied to reactivity coefficients and nodal cross sections, consistent with the all rods out condition typical of end-of-cycle. For the limiting 1-D events, generic statistical adjustment factors are derived to conservatively account for model uncertainties associated with the nominal cycle-specific transient analysis calculations.
The techniques for deriving cross section input for the 1-D kinetics model are ccmplex and require the use of the cross section collapsing program, SIMTRAN-E.
The description and qualification of conservative methods employed by PECo in the analysis of each of the reload licensing transients will be presented in more detail in PECo-FMS-0006, " Methods for Performing BWR Reload Safety Evaluations",
to be submitted to NRC at a later date.
\\
. i 0
. g~
)
s QUESTION #8:
What is the sensitivity of the Doppler and void coefficients to changes in the core flux distribution during a transient? How is this uncertainty accounted for?
RESPONSE
The PECo point kinetics procedures described in section 5 of PECo-FMS-0005 calculate instantaneous core average values for void, Doppler and scram reactivities. These are determined with the 3-D simulator using a fixed power shape assumption but allowing the neutron flux distribution to converge to the perturbed void or fuel temperature state. The application of the point kinetics solutions is limited to slow trantients in which the core power shape varies slowly during the course of the eient.
It-is also important to note that the fixed power shape point kinetics assumption will actually predict conservative results for some transients, as in the Loss Of Feedwater Heating event discussed in Appendix C (see response to NRC question #22). For events in which this is not the case, multipliers are applied to the point kinetics reactivity parameters to ensure conservative predictions for core power response in the point kinetics model.
Events involving rapid changes in neutron flux and power distributions (e.g.,
pressurization events) are modeled by PECo using one-dimensional space time kinetics. This method allows a detailed evaluation of the neutron flux and core power distributions at each instant of time during the transient.
In this respect, one inherent limitation of the point kinetics approach (i.e.,
uncertainties attributable to the fixed power shape assumption) is eliminated.
PECo point kinetics and one dimensional kinetics modeling procedures and uncertainty analyses will be qualified in PECo-FMS-0006, " Methods for Performing BWR Reload Safety Evaluations", to be submitted to NRC at a later date. These will include techniques employed by PECo to ensure adequate consistency between the reactivity characteristics displayed by the steady state and transient models. -
i
~
QUESTION #9:
How does the Peach Bottom 3 Cycle 7 (for example) Doppler and void coefficients and control rod worth, calculated by SIMu uTE-E, compare with the fuel vendor values?
I
RESPONSE
The values of void and Doppler coefficients routinely reported in the vendor nuclear design reports and reload licensing submittals involve the use of computer codes and procedures which are specific to the vendor's own licensing analysis process.
In this regard, the vendor-derived reactivity parameters are processed for input to ODYN/REDY, while the corresponding PECo-derived parameters are processed for input to RETRAN. This makes it difficult to determine raw reactivity parameters derived from the vendor's 3-D simulator which are directly comparable to those calculated by PECo using SIMU MTE-E. For example, the 3D void coefficient reported in the vendors nuclear design report is not the raw value calculated by the PANACEA 3-D simulator, but instead is a value derived using a void model transformation as applied to the PANACEA result, for consistency with ODYN and REDY'. Likewise, the values for the void coefficient reported in the vendor's reload licensing submittal are based upon results from a point reactivity model, not the PANACEA 3-D simulator.
The vendor-calculated Doppler coefficient, on the other hand, is predicted using a perturbation theory technique in conjunction with the PANACEA 3-D simulator.
Finally, the vendot no longer calculates a cycle-specific scram reactivity curve, but instead relies upon the use of a generic exposure independent scram function. These reactivity parameters, as reported, are not directly comparable to the values derived by PECo using the SIMUMTE 3-D simulator. Consequently, any comparisons between the two would be speculative and could lead to misleading conclusions. PECo has instead relied upon its own independent qualification of SIMU uTE reactivity parameter predictions for input to RETRAN, as presented in sections 4 and 5 of PECo-FMS-0005.
1 The void model transformation is necessary to correct for void model differt.cos which oxist betwoon the vendors steady stato and transient computer codos. Noither of thoso vendor-developed void models are used in either the PECo SIMULATE or RETRAN computer codos.
4 3
QUESTION #10:
How is the dependence of the void coefficient on changes in control rod insertion and void fraction during a transient accounted for? What uncertainty is introduced by the treatment of these effects?
RESPONSE
As in the response to NRC question no. 8, the treatment of void coefficient dependencies on control rod insertion and void fraction depend on the transient analysis technique chosen to analyze the event.
For rapid pressurization events, the one dimensional space time kinetics model is employed.
In this model, core average axially dependent cross sections are determined at various intermediate control rod states between the initial rod configuration (typically all rods out) and the final rod configuration (all rods in). Thus, the dependencies of the neutron cross sections on steam void fraction and fuel temperature are determined at each axial node, and with consideration for the different control rod states during the transient.
Therefore, the one dimensional kinetics model explicitly accounts for the void coefficient dependencies on changes in control rod insertion and void fraction at each axial node.
For the slower (i.e., non-pressurization) events, the point kinetics model is employed.
In these cases, the reactivity (void and Doppler) inputs to the f
transient model are derived on a core average basis using the initial control rod configuration and void distribution at the start of the transient. As stated earlier in response to NRC question no. 8, conservative multipliers are applied to the reactivity parameters to account for model uncertainties and possible changes in flux and void fraction distributions. This approach is adequate since these slower events are less severe, and the worst power condition generally occurs prior to any scram occurrence. Thus, changes in the reactivity parameters caused by changes in control rod position and void fraction during the transient can be conservatively ignored.
. t 1
7
.c i.
,q if i
f..:
-QUESTION #11:
How is the reduced leakage probability for delayed neutrons calculated?-
RESPONSE
The reduced leakage probability for delayed neutrons'is determined in the l
calculation.of Serr using the CASM0-1 program. CASM0-1 values for Serr are appropriately averaged for use in both the point kinetics and
.one-dimensional kinetics transient methodologies. The CASM0-1 formulation for.
Determination of Serr is as follows :
- The average onor0y of the dotayed fission spectrum is much lower than that of the prompt fission spectrum. Thorofore, delayed neutrons have sinaller chanco to leak out of (or into) the system and do not causo last fissions.. An ((fpguyn,Dalaypd Neutron Yigjd may bo defined as:
Sotg. t =
D.
Sg (10.5.5) k Prod q.
Prod = E E Prodm.g '
(10.b.6) rn g k
= Tito mulbplication factor kD = delayed noulton multiplication factor All quantities above are the elloclive, leaka00-depondent quantitics.
Pr"sd = 1 k = k rt = 1 o
Sirco the slowing down of tho delayed neutsons relatos to a specific spectrum kg should be calculated by a separato spoetrum calculation with a dolayed noutron source at about 0A5 MoV.
This would be complicated and inskiad we utilito the spectrum calculated with the ordinary fission source. kg is then given by lig Prud-E Prodh h*I kg =
(l0.5.7) hD hD 4 leak - E 1.cakh Abs E Absh h=1 h=1 1 Excerpts taken from the CASMO-1 code retorence manual.,
_________.__.__._________.______n_____
(.
Response to NRC Question 111 (Continued):
Prod. Abs and Leak are given by the cell noutron balance in the macro groups h. The Group ho is tho highest macro group havin0 the ion onorgy boundary below 0.45 MoV. Usat is:
6 Eno, 3 0.45 10 eV > E o h
With the energy boundaries of Die present library, the most correct estimation of kg is obtained when the lower enorgy boundary EhD of macro Oroup ha is 0.5 McV. This corresponds to micro group 6 in the 69 group library.
PECo directly utilizes the CASM0-1 edits of B-effective for transient safety analysis applications. These values are appropriately averaged for point kinetics ano one-dimensional space time kinetics applications.
d
{
i,
.-.-__----__---._..-_._..--_d
e QUESTION #12:
-How is the highest worth rod determined in the calculation of core shutdown margin?
RESPONSE
At each cycle exposure condition evaluated, the highest worth control rod is determined based on results from the SIMULATE 3-D simulator. Assuming octant symmetry in the design loading, each non-edge control rod in one octant is withdrawn, in turn, from the initially all rods in configuration. The 3-D core simulator is executed with this single control rod withdrawn and the predicted core K-effective is noted. After exhausting each of the twenty-four non-edge control rods in the octant, the single rod out K-effectives are compared, and the Lighest value (i.e., with greatest control rod worth) is selected. As a
~
further verification, control rods from each of the other octants which are in locations symmetric to the highest worth control rod, are each withdrawn with the 3-D simulator.
In this way, the control rod with the maximum rod worth in the entire core is determined. Using this procedure, the shutdown margin is evaluated at each cycle exposure point. All SIMULATE rod worth calculations are
~
performed using an explicit, full core geometry (764 bundles) model.
PECo is also actively engaged in developing and qualifying a two-dimensional rod worth esti&4 tor program which will be employed in the approximate determination of BWR cor...ol rod worths. When qualified, the code will allow for a significant reduction in the totai number of explicit, cold rod worth calculations to be performed with the 3-D simulator, while assuring with a high level of confidence, that the highest worth rod in the core is used as the basis for shutdown margin evaluations. The proposed 2-D estimator will be extensively benchmarked against no less than 400 full core 3-D simulator cases, with the following restrictions placed on the database:
o Only the top five worth rods, as determined by the 3-D simulator, will be evaluated for a given core condition.
Only blades in one octant of the core will be analyzed; symmetric rods will o
not be included in the database.
o The database will be representative of both the Peach Bottom (D-lattice) and Limerick (C-lattice) units.
o Core conditions will be representative of the entire cycle burnup range.
Benramark calculations will typically be executed at 1000 MW9/ST intervals.
o Approximately 50% of the database will be composed of cases which include current (axially zoned) fuel designs.
No plant / cycle combination shall comprise more than 20% of the database.
o _
P f'
Responsr2t~t Question #12 (Continued):
- The program will be implemented by PECo for application to design / licensing cold shutdown margin-evaluations when it ir demonstrated that the mean difference (absolute) between the 2-D estimator and the 3-D simulator rod worths is less than 0.0008AK with.a difference ~ distribution variance (about the mean) of less than 0.0000003. Once the program is shown to perform within these acceptance criteria, PECo will execute 3-D rod worth calculations for an appropriately reduced number of blades, as determined by the 2-D estimator, to determine the strongest rod in the core.
4 0 - _.. - _ _
y i
- i. -
i-QUESTION #13:
_How are the " projected cold critical eigenvalues" determined from the " data base of cold critical projections" (page 5-32)?_ How are the projected eigenvalues represented by a polynomial?
E RESPONSE:.
The determination of the projected cold critical eigenvalue is based upon the predictions of the cold critical measurements recorded at Peach-Bottom during recent operating cycles. Referring to the PECo response to NRC question no. 1, the SIMULATE cold critical K-effective predictions exhibit an exposure dependent bias which can be represented as:
CH tical Cfold (E)
(2)
. (i) o K rf
- 4 e
In this equation:
Keff = The average SIMULATE K-effective prediction within a single operating cycle. For core design purposes data from the r ceding cycle a-e used to determine this parameter.
COLD f(E) = A cycie o posure dependent bias function obtained from a polynomial least squares fit of SIMULATE cold K-effective predictions versus cycle exposure, E1 As an illustration, the cold critical predictions for Peach Bottom 3 Cycle 6 have'been plotted as a function of cycle exposure in Figure 2.
In this figure, the actual cold critical K-effective predictions are displayed, together with
'i the polynomial curvefit results (equation 2). The latter curve was used to project the cold critical eigenvalues for the Peach Bottom 3 Cycle 7 design shutdown margin calculations described in PECo-FMS-0005.
j f
For future cycle design purposes PECo will determine a cold shutdown margin based on a critical eigenvalue estimator from equation (2), or a more conservative approximation.
1 Refer to Figuro 1 used in responso to NRC question no.1..___________-_____L
9-4 4.
e-FIGURE 2 PBAPS Cold Critical K-effeqtive Data Pecch Bottom 3 Cycle 6 0.9990 -
0.9980 -
0 0.9970
.5 E
0.9960 h.
d 0.9950 -
U U
O 5
0.9940 -
15 0
0.9930 -
o 0
0.9920 -
0.9910 -
0.9900 0
2 4
6 8
Cycle Exposure (MWD /ST)
O K-effective DATA CURVEFIT _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ - _ _ - _ -
~
_ UESTION #14:
Q How is the increased uncertainty in the shutdown margin calculation of Keff for the state with the highest worth and rod withdrawn accounted for?
RESPONSE
As stated in PEco-FMS-0005, the PECo design criteria for evaluation of the-minimum predicted Shutdown Margin (SDM) during the cycle is 1% AK. The uncertainty in the prediction of cold critical K-effective has been conservatively estimated as 0.35% AKN. The 1%AK SDM design criterion thus provides a conservative margin of 1.770 above the minimum value required by the Peach Bottom Technical Specifications (0.38% AK).
This provides a large conservative margin to ensure that adequate shutdown
~
margin is provided 1,6 the reload design pricess.
PECo's use of cold in-sequence criticals as a basis for evaluation of the uncertainty in the strongest rod out K-effective prediction is discussed in response to NRC question no. 29.
This is based on the quoted standard deviation in the prodiction ol 31 critical measurements over five operating 1
cyclos. This starxiard deviation is calculated around a constant mean K-effectivo value of 0.9916. This uncertainly estimato does not take credit for the observed variation in the expected K-effective prediction as a function of coro exposuro. -
.v's QUESTION #15:
What are the calculational uncertainties when all measured TIP signals are included in the calculation / measurement comparisons (e.g., MCPR, MAPLHGR, PPLHGR, etc.)?
RESPONSE
- When all strings are included in the TIP calculation / measurement statistica, evaluation, excluding the top and bottom 18 inches of the core, pointwise and integral RMS statistics are calculated to be 7.5% and 5.2%, respectively. Based on the methods set forth in section five of PECo-FMS-0005, these power uncertainties propagate into the following thermal limit uncertainties:
THERMAL LIMIT RMS UNCERTAINTY MCPR 6.6%
MAPLHGR 7.5%
PPLHGR 8.3%
These values, as updated, will be employed in the development of thermal limit design margins.
l l
\\ -
g 4
l j
QUESTION #16:
In the analysis.of the rod withdrawal event, how is the error-rod yielding the-minimum RBM setpoint determined? How is the uncertainty introduced by this j
- procedure accounted for?
i 1
RESPONSE
I The control rod pattern and error rod used in the analyses of the Rod Withdrawal 1
Error (RWE) event is conservatively determined when compared to normal BWR operating practice. Conservatism is achieved in the selection of the control rod j
pattern in each of the following respects.
o The analysis is performed at a condition of maximum core reactivity. This tends to cause the worth of the error rod to be larger than at other core exposure r.onditions.
o At this core exposure condition, the error rod location is selected to be one of high reactivity worth. The search for the highest worth control blade is based on the highest cell average relative power in the full power all rods out condition for the statepoint. The worth of the error rod is further accentuated by assuming no credit for xenon in the event analysis.
o The error rod selected is initially fully inserted. This assumption, again, tends to maximize the worth of the error rod in the analysis when compared to initially partially inserted control rods.
o The remainder of the rod pattern is configured such that at least one fuel assembly close to the error rod location is placed at or near the MCPR and LHGR operating limits with the error rod fully inserted and the core in a critical state.
This assumption minimizes the initial margins to the transient safety limits at the beginning of the transient.
These criteria are conservative for the RWE analysis and result in highly abnormal control rod patterns when compared to those used in ordinary BWR operations. The results of the RWE analysis when performed in the manner described in PECo-FHS-0005 are thus conservative in a deterministic sense. As such, PECo believes that consideration of additional uncertainties attributable to selection of the error rod to be unnecessary. This approach is consistent with current industry practice as utilized by the fuel vendor and other BWR utilities. _-
QUESTION #17:
In the analysis of the rod withdrawal event, how are rods with less than four adjacent LPRM strings treated considering worst-case LPRM failures?
RESPONSE
Figure 3 depicts the assignment of LPRM strings to control rods used in the Rod Block Monitor (RBM) system. Of the total of 185 possible control blades which may be selected by the operator, 40 are in core edge locations (unmonitored by the RBM) 14 are in core locations monitored by 2 LPRM strings,11 are in core locations monitored by 3 LPRM strings, while the remaining majority (120) are in core locations monitored by 4 LPRM strings.
l The 40 core edge control rods are in locations of high neutron flux gradients, and are consequently of low control rod worth. These edge rods are usually fully withdrawn or in shallow core insertion positions during full power operations. For these reasons, the edge rods pose no threat to fuel integrity during the postulated Rod Withdrawal Error (RWE) event, and are consequently unmonitored by the RBM. They are not considered as candidates for the error rod in PECo RWE analyses.
The 25 control rods monitored by the RBM using less than 4 LPRM strings are likewise near the core edge (one control rod location removed). Consequently, these control rods are usually in lower power (i.e., lower rod worth) regions of the core, although there may be instances, especially near the middle of cycle, where these rods may be of low to intermediate worth.
Control rod locations where less than 4 adjacent LPRMs are usca in the RBM system have not been selected for the error rod in the Peach Bottom RWE licensing analysis performed by either PECo or the fuel vendor. This is attributable to the following conditions.
(1) As stated earlier, these control rod locations are not associated with the highest rod worths. Upon additional inspection of the RWE analysis performed for Peach Bottom 2 Cycles 6 and 7 and Peach Bottom 3 Cycles 6 and 7. PECo has determined the control rod worths at these locations to be at least 29% lower 1
than those associated with the worst control rod location used in the analysis.
The larger error rod worth used by PECo to analyze the RWE will tend to result in a more severe (i.e., conservative) delta-CPR evaluation for this event.
(2) Deeply inserted control rods are uncommon at these locations in full power l.
operating control rod patterns.
In order to flatten the core radial power shape, deeply inserted control rods are generally used only in the core central locations. This fact significantly reduces the probability of the full withdrawal of an error rod in locations where less than 4 LPRM strings are used l
by the RBM. l l
Response to NRC Question #17 (Continued):
l Because of these conditions, PECo does not anticipate determining the RWE error rod to be located where less than 4 LPRM strings are used by the RBM. However, in order to ensure a comprehensive evaluation of the RWE event, PECo will monitor these locations as possible candidates in the PECo RWE error rod selection process. As described in PECo-FMS-0005, the selection of the error rod location is governed by the largest rod worth criterion.
Based on the failure modes allowable, the worst case LPRM/RBM failure condition analyzed is associated with the failure of half of the LPRM strings used by the RPM to monitor the control rod. Consequently, if the error rod is determined to be located where less than 4 LPRM strings are used by the RBM, PECo will consider the following LPRM failure conditions in the RWE analysis:
(1) for rods monitored using two LPRM strings, at most one string will be assumed to be failed (3 possible failure modes).
(2) For rods monitored using three LPRM strings, at mosti two strings will be assumed to be failed (7 possible failure modes).
As stated in PECo-FMS-0005, for each channel taken separately, the lowest RBM response as a function of error rod position is used in the prediction of the RBM setpoint, as required to block the rod.
l RBM System responso is ovaluated for all possibio LPRM string failure combinations within the limitations 1
spx:thod heroin. For conservatism, the most hrniting string failuro combination is assumed in the Hod Withdrawal Error ovent evaluation. _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
n,_-
p:
d.
, s.
-7.-
Core Top View.
0*
Control Rod Group -
/b b b@
E * **
'dVJ
~
~
V49
 Mm mV4 VAm YA y
s c
Ms.
,x 74 g
2cggg VA,,.4
~
M JillNL fJ m
2e se ys s
V4 i
,. x wgy Vp 2
"'"4 K2 W/Z-VAk
/
to p Colm!
1 2
3 4
5 6
7 e
LPRM Detector String
@ LPRM. Assembly Assigned to RBM for Rod Group selected h Rod Selection Resulting in two-assembly assignment l
h Rod Selection resulting in three-assembly assignment k Rod Selection resulting in four-assembly assignment
(([JREM automaticaHy bypassed (Reading Zerced)
Figure 3 LPRM String Assignment Geometry Used In RBM System i
- l. _. _ _ _ _ _ _ _ _ _ _ _
Q' a:
o.
JQUESTION #18:
How is-the disoriented fuel bundle treated?
RESPONSFj The disoriented.(i.e., rotate.d) fuel bundle methodology will require execution of the PECo transient analysis codes-(RETRAN - TCPPECO). 'As'such, the qualification'of these methods has not been. included in the PECo steady-state physics methods report, PECo-FMS-0005. The rotated fuel bundle methodology will instead be qualified in the PECo reload safety evaluation methods report (PECo-FMS-0006), to be submitted to NRC at a later date.
I I
l l- _ _ _ _ _ _ _ _ _ _ - _ _ _ _ - - _ _ _ _ _ - _ _
j,-
4
- 1. ;
- js
.. QUESTION 319:.
j
~I
'What~ is the sensitivity of the ACPR regression fit of Figure B-1, for the mislocated bundle loading error, to the core operating conditions (power, ficw, rod pattern, xencn, etc.)? How is the resulting uncertainty accounted for?
RESPONSE
The Mislocated Bundle Loading Error (MLBE) event has been evaluated by PEco in a statistical manner, using SIMULATE results associated with a number of cycle exposure conditions for Peach Bottom 3 Cycle 7.
In this evaluation, rated core operating conditions (e.g., power, flow, rod' pattern, xenon) were chosen since these generally more yield more limiting core average conditions in the evaluation of licensing events. Throughout the cycle exposure range, the local fuel assembly power, flow, xenon, and rod pattern conditions were allowed to vary in the manner predicted by SIMULATE. The variations in delta-MCPR data reported for the MBLE event resulted from these local variations in the power, flow, xenon, and rod pattern near the error assembly location in the core.
Thus, the scatter _in Figure B-1 data about the regression fit line is a measure of the sensitivity of the delta-CPR results to these local variations in the subject parameters. Consequently, the sensitivities attributable to these parameters have been included in the regression model uncertainty (20) of 0.064 ACPR cited for the MBLE event.
PECo has also evaluated the effects of a dislocation on assemblies innediately adjacent to the dislocation. The procedures set forth in Appendix B of PECo-FMS-0005 were repeated, expanding the scope of the analysis to include all assemblies within one assembly pitch'of the mislocated bundles. Results from this expanded analysis verify that for Peach Bottom 3 Cycle 7, a worst case M8LE event would not drive either the mislocated assembly or any adjacent assembly to the CPR safety limit. Further, a second MBLE generic evaluation like that presented in Appendix B of the subject report was performed based on a MBLE adjacent assembly assessment of ICPR and ACPR data. The MBLE adjacent assembly generic licensing basis delta CPR, including a 2a uncertainty and 0.05'ACPR adder, was calculated to be 0.10 ACPR. This value is bounded by the 0.13 ACPR value reported originally in PECo-FMS-0005 for the actual, mislocated bundles.
) _
y
[.i f.
- g. %
+
j;.
.$3 QUESTION'd20:-
L.
L RJ In the Loss of Feedwater Heating event, is the feedwa'ter flow increased as a result of the increased power level?- If not, how isfthe resulting increase in-g r
power and axial peaking accounted for in the analysis?
y
- RESPONSE:
L 4
l The Loss of Feedwater Heating Event results reported in PEco-FMS-0005 (Appendix
. C) include the effects due to increases in the feedwater flow resulting f rom the increased power level. This is accomplished by matching the inlet. feedwater flow with the core exit steam flow during the power search iteration.. The power-
~ level at the end of the transient is considered converged when a thermal power is determined such that:
(1) the ~ critical-K-effective prediction at the end of the transient.
. matches the critical K-effective prediction at the beginning of the transient i
(2) the feedwater flow used to determine the SIMULATE input conditions (e.g..' inlet subcooling) at the end of the transient matches the resulting core exit steam flow. Of course, the effect of the assumed j
100 F decrease'in feedwater temperature is also included -in the determination of SIMULATE input conditions at the end of the transient.
4 l -
1
, )
)
gs R..
l
.w v...
Vl,
E.-
QUESTION #21:
~ 1 What'is the' sensitivity'of the Loss of Feedwater Heating Event to the core-conditions-(power, flow, xenon, rod pattern,' exposure, pressure, inlet
'a subcooling).and how is the resulting uncertainty accounted for?
-RESPONSE:
The sensitivity of the Loss of Feedwater Heating event to the core conditions-
.(power, flow, xenon, rod pattern, exposure) will be reported-in the PECo reload
. safety evaluation methods report, PECo-FMS-0006.. This report will be submitted.
to NRC at a-later date.
6 0.
M
'f p
s cp
-y
+
f"
$ QUESTION #22:l
<w
. ThePECoIossoffeedwaterHeatinganalysisassumes;a.fixedcorepowershape 1during the transient.
In fact, the: axial power distribution becomes more.
. bottom-peaked'during the transient, resulting in an additional reduction in CPR.
margin in.the bottom of the' core. How is this'effect' accounted for in the PEco
' methodology?!
p
RESPONSE
y
[.'.
i; h
' SIMULATE-E calculations for the: Loss of Feedwater Heating event show that'a more
- bottom-peaked distribution will result at the end of
- the transient if the power-
' redistribution is: allowed to occur. The axial power redistribution. Din turn, would result in the generation of more in-channel voids which would tend to
. limit the core average thermal power response. By use of the bundle critical
-power correlation in SIMULATE-E, the combined effects of'both the core average thermal' power and the axial power. distribution.are incorporated directly into.
t the evaluation'of delta-CPR for this event. Tables C-3 and C-4 in PECo-FMS-0005
~
suianarize the overall conclusions'in this regard. From-these tables it is
! observed that the core average thermal power increase in this event would be reduced by approximately 3.6% of full power.as a result of the variable power.
shapeL(i.e., power redistribution) assumption. Likewise. the combined effects-of-the reduced core thermal power response, together with the more bottom peaked power distribution, would result in an overall reduction of approximately 0.06 ACPR in the variable power redistribution case. Consequently, the fixed power. shape assumption as used by PECo in the Loss of-Feedwater Heating event
- .results in an^overall conservative over-prediction'in the severity of the
' delta-CPR for this event.
{
l
% l 1
n'
- Q'.
...o
)
- QUESTION 123: -
)
l The'4.1% RMS-error in the SIMULATE assembly integral power calculation is based'
~
on the elimination of the top and bottom 18 inches of the core from the
-statistics. What is the effect of this deletion on the calculation uncertainty?
-Are these regions ever limiting?
RESPONSE:-
Inclusion of TIP data from the top and bottom 18 inches of the core in the power
' distribution statistical evaluation results in a small decrease in the integral-power, RMS error from 4.1% to 3.8%. This result is indicative of a small conservatism in PECo's original evaluation of SIMULATE integral power.
calculational uncertainties.
MCPR directly reflects. integrated assembly power deposited-in the fuel rods..
Due to the BWR flow path,.the uppermost nodes in the core therefore typically.
experience minimum margin to the CPR operating limit.. However, due to the relatively low power densities characteristic of the axial core boundaries, nodal powers in the regions in question have a small relative impact on channel MCPR..
9 I
ip -
QUESTION #24:
Describe the fuel loading' t,. Peach Bottom 2, Cycles 5 and 6 and Peach Bottom 3, Cycles 4, 5 and 6 which were included in the benchmarking of SIMULATE. Does PECo intend to use fuel designs and loadings which are not represented in the.
benchmarking? If so. what are they and how will they affect the qualification of the PECo methods?
RESPONSE
The five Peach Bottom cycles used to benchmark SIMULATE were Control Cell Core (CCC) design loading patterns, consisting of mixed cores containing both GE'7x7 and 8x8 fuel assemblies. The GE 7x7 fuel was generally depleted beyond the gadolinium exposure lifetime, and was essentially gadolinium-free. The GE 8x8 fuel ranged in exposure between 0 GWD/T (fresh) to approximately 25 GWD/T (thrice burned) and contained, for the most part, seven gadolinium-bearing fuel 1
rods with initial gadolinium loadings between 2 and 4 weight percent. The initial average U-235 enrichments in these fuel designs ranged between 2.50 to 2.99 weight percent.
These. fuel and core loading designs are characteristic of 18 month reload fuel.
cycles which are currently implemented at both Peach Bottom and Limerick.
Future BWR fuel design improvements are expected to allow 24 month cycle operation in PECo units. This will probably result in higher fuel and gadolinium loadings, a greater use of axial zoning in the fuel and poison rods, as well as a larger number of fuel rods per assembly (e.g., 9x9 or 10x10 lattices). Also probable is the use of more water carrying rods to increase thermal moderation and to improve fuel utilization.
Many of these advanced designs have already been tested and used on a demonstration and/or production application basis in US, European, and Japanese reactors. With the exception of the ASEA Atom Water Cross Design, all of these advanced designs can be analyzed using the PECO CASMO-1/ SIMULATE-E code sequence. The ASEA Atom Water Cross Design would require a modification of PECo lattice physics methods. Utilities which are currently using CASM0/ SIMULATE (or the closely related CPM / SIMULATE) methods for the analysis of these advanced i
designs have not observed'a significantly adverse reduction in calculational accuracy.
Thus, PECo intends to utilize its current physics methodology in the analysis of advanced fuel designs wherever practicable.
If fuel design changes occur which are outside the capability of the current methods, any new methods to be employed will be re-qualified in a manner similar to that described in PECo FMS-0005.
In the interim, PECo will continue to monitor the performance of the CASM0/ SIMULATE models with the introduction of any advanced designs into Peach Bottom or Limerick.
In this way, the SIMULATE model bias and uncertainties will be properly tracked and accounted for in the design and licensing of future cycles. _ _ _ _ _ _ _ - _ _ _ _ _ _
(
[
t s
QUESTION #25:L
,a.
i
.The hot critical eigenvalue results for Peach Bottom 2 and Peach Bottom 3 shown d
-in Figures 3.1.1 through 3.1.9 show a pronounced upward trend with exposure.
This trend is clearly seen in the multicycle plot of Figures 3.1.6 and 3.1.7.
I What is causing this exposure dependent bias, and how is it accounted for in the-SIMULATE' predictions?
1i L
RESPONSE
The PECo SIMULATE hot critical K-effective predictions have been statistically 4
analyzed over each of the operating cycles as follows:
l P2C5
'P2C6' P3C4 P3C5 P3C6 Keff 0.9939 0.9970 0.9899 0.9933 0.9967 o
0.0015 0.0024 0.0019 0.0027 0.0018
.i 1
The cycle mean K-effective (Keff) thus displays an increasing trend of
,j approximately 0.3%AK per cycle in each of the two Peach Bottom. units. This is attributable partly to the transition in fuel designs from the 7x7, 144" active length fuel to the 8x8, 150" active length fuel.
As described in response to NRC question no. 1 the changing SIMULATE critical K-effective prediction is accounted for by using a target critical K-effective
- in the SIMULATE calculations which is given by:
Keff(E) =Keff
- f(E),
where Keff is the projected cycle average K-effect.ive prediction, and f(E) is the cycle exposure dependent bias function derived by a polynomial least squares fit of previous K-effective results.
l In the application of this equation to core design activities, Keff for the design cycle is estimated to be equal to the average critical K-effective prediction obtained from the analysis of the preceding cycle.
In this way, predictions for the design cycle are based on the results for the most current operating' data for the same unit.
l 1 t
3,.
2:
.3 7.,
'LQUESTION #26:
L.
Which specific SIMULATE-E normalization parameters are adjusted from one cycle
.to the next?
i
RESPONSE
No specific SIMULATE normalization parameters are adjusted in the PECo model-from one cycle to next. Normalization parameters (e.g., horizontal and vertical albedo factors) were set.to' cycle-independent values in the initial Peach Bottom core ' analyzed, and have remained set to these same values in the SIMULATE analyses for each of the subsequent cycles. These cycle-independent values have been demonstrated to provide acceptable accuracy in the SIMULATE predictions of core critical K-effective, power distribution, and other associated reload design and licensing parameters.
I 1
l l
l l
l 1
l
)
I 4
l. _ _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ _ _ _ _ - _ _ _
1
' t ~.
10
~
QUESTION #27:
L^
Explain the systematic underprediction of the core average axial-power
~ distribution near the bottom of the core for Peach Bottom 3, Cycle 6.
. RESPONSE:
PECo has observed a systematic underprediction in the SIMULATE core average power distribution near the bottom of the core during~many of the middle of
. cycle (MOC) statepoints in the Peach Bottom cycles.
It is believed that this is most~probably due to an underprediction of the gadolinium burn out rate in the
-lower core axial locations. This effect has been tracked in a statistical manner, and the average bias' exhibited by the model is used to correct the SIMULATE power distribution predictions at MOC statepoints.
It is noted, however, that no such systematic axial power distribution bias has been observed at Beginning of Cycle (B0C) or at End of Cycle (EOC) statepoints. Thus,
' SIMULATE power distribution predictions at or near BOC and EOC conditions are used directly, without correction.
1
7; '
v
-1
-o,f
-QUESTION i28:-
l-Describe the procedures used in correcting for temperature and reactor period.in
.the calculation of-the critical tests.
RESPONSE
.The PECo cold SIMULATE model contains explicit cross sectional dependencies on moderator temperature, THOD. - Consequently, the moderator temperature-
- associated with each plant critical test is. input directly into-the SIMULATE model, and no external TMOD correction of the K-effective results is required. A small external correction to the SIMULATE K-effective results is made to account for the observed cold critical test period. To do this, the core average B-effective is calculated for the core design. Next, the in-hour equation is used to relate the reactivity correction (p) to the
' measured critical test period (T). The period correction is typically small (less than 0.1% AK).
4 9 _ _ _ _ _ - - _ _ _ _ _
3
.t -
.o
- 1,.
QUESTION #29:
Have any few-rod criticals been evaluated with SIMULATE-E and, if so, how do these calculations compare with the measurements?~
RtSPONSE Results from few-rod local critical predictions have been evaluated by another 1
utility in its qualification of the SIMULATE-E computer program. This study essentially compared the accuracy of SIMULATE-E predictions of few rod criticals with those associated with in-sequence criticals. Based on 25 few-rod local critical predictions, it was demonstrated that there is essentially no difference in the calculational accuracy in the SIMULATE predictions of few-rod criticals as compared to that associated with in-sequence criticals.
It is therefore inferred that the PECo cold critical K-effective bias and uncertainty, as derived from the prediction of in-sequence criticals, apply to the prediction of few rod criticals as well.
9 1
Oyszel, A., K. C. Knoll, " Qualification of Steady-Stato Coro Physics Methods for BWR Design and Analysis",
PL-NF-87-001, Pennsylvania Power and Ught Company, March 1987.,
i
pm, p
g' y
l s.: y,y j
L
[ QUESTION 130:.
l JAre,theisystem variablei; such as pressure, feedwater flow, steam flow, etc.,
. assumed c'.,nstant-during the Loss o.f-Feedwater Heating. transient? If so, provide-
.the br,isifor this. assumption. Can changes in these variables result.in a-4
' limiting ACPR during the transient, making the final-state ACPR' l
calculation not-bounding?-
a RESPONSE:-
As stated in.the' response to NRC question no. 20, changes in the feedwater flow and steam flow are. accounted for in the analysis of the Loss of feedwater Heating (LFWH) transient. Changes'.in system pressure are expected to have a-
.very small effect and, as such, are ignored in the LFWH evaluation. Because of-the~ slow nature of this. transient, the core thermal power. increases gradually -
from the initial-' state. Changes in the other system parameters (e.g...feedwater-and steam flows) likewise change in a slow, gradual manner. As long.as a.high flux. scram does not occur, the reactor will continue to proceed gradually to its:
final equilibrium state at higher core power. Thus.the SIMULATE-E 1
model,can be used to accurately determine the limiting ACPR which occurs between the initial and final reactor equilibrium states.
LThis' analysis' approach was verified by evaluating'the LFWH event using the RETRAN. transient analysis computer code.2 The evaluation demonstrated:
that the transient CPR responds -in a gradually decreasing manner, with the
- minimum CPR. occurring at the final equilibrium state.
In addition, the use of
. the" 3-D simulator for the analysis of the LFWH event is consistent with the current NRC approved licensing basis.3 3-
'1-In the LFWH ovaluation, no credit is taken for a possible high flux induced reactor scram. This is a conservative assumption since the reactor power in this event is ordinarily limited by the APRM system sotpoint (l20% of iriitial APRM signal at rated conditions)
NHC approval for PECo's use of the RETRAN computer code is outlined in the SER for topical report
'F 2
' PECo-FMS@04, 'Mothod For Performing BWH Systems Transient Analysis' 3 General Electric 't iconsing Topical Report', NEDE-240ll-P-A-9, September,1988 -
--_.--__um;--_._____m.______.
m.__
___