ML20059K999
| ML20059K999 | |
| Person / Time | |
|---|---|
| Site: | Maine Yankee |
| Issue date: | 09/20/1990 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML20059K996 | List: |
| References | |
| NUDOCS 9009260018 | |
| Download: ML20059K999 (5) | |
Text
^
+
k
.gu tt c
o UNITED STATES
].I
' k-NUCLEAR REGULATORY COMMISSION t,
j.
WASHING TON, D. C. 20555
\\...../
SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION RELATING TO MAINE YANKEE RPS SETPOINT METHODOLOGY USING STATISTICAL COMBINATION OF UNCERTAINTIES MAINE YANKEE ATOMIC POWER COMPANY
.i MAINE YANKEE ATOMIC POWER STATION DOCKET NO.-50-309
1.0 BACKGROUND
By letter dated April 19, 1988 Maine Yankee Atomic Power Company submitted ~
Volumes 1 and 2 of. the' report entitled " Maine Yankee RPS Setpoint Methodology.
Using Statistical Combination of Uncertainties". This report describes an alternative method of statistically combining the uncertainties applicable-to the determination of operating limits for the protection of the fuel from experiencing centerline melting (Volume 1) anc' Departure from Nucleate Boiling (DNB)(Volume 2).
Volume'1 describes the application of the Method of Moments technique'to statistically combine the effects of some of the uncertainties on the power-to-the-fuel-melt limit, Pl. Volume.2 describes the use-of Monte Carlo Simulation for ct,mbining the uncertainties applicable to the determination of operating limits for the protection of the fuel experiencing DNB. This technique consists of propagating many randomly sampled sets of values for each of the uncertainties being statistically combined, through-a system simulation to quantify their effect on the output of interest.
~ Uncertainties c included in the statistical combination continue to be
-applied deterr,istically. The use of this methodology in the setpoint and safety analys"s will result in a reduction in the excessive amount of operating margin utilized to cover allowance for uncertainty. The margin recovered will be used to restore operating margin lost during the transition to 18-month, low fluence fuel cycles.
2.0 EVALUATION
-Volume 1 describes the current methodology for determining the limits to prevent fuel centerline melt. The overall methodology consists of four, phases:
1.
Determinationofa_conservativevaluefortheLinearHeatRate-(LHR)at which incipient fuel centerline melting would occur for each fuel-type'in the core.
900926001e 900920
{DR ADOCK 05000309
' PDC o
IJ.
.r r
4' 2.
Determination of the core power level (P ) at which the heat. rate at the I
3 core " hot spot" would just equal the limiting LHR. This is done for a spectrum of pessible power distribution, which bound those expected to occur during normal operation or A00s. The results are usually presented in the' form of scatter plots of P versus peripheral symmetric offset 3
(I ), referred to as the "P flyspecks."
p y
3.
Determination of the required Symmetric Ofhet Trip Setpoint (SOTS) to prevent operation at a condition where core power level is in excess of P). for the existing power distribution.
4.
Determination of the Synnetric Offset Limiting Condition for Operations (SOLCO)requiredtoensureadequateoverpowermarginfortransientswhich may not result in a protection system trip.
The report describes the uncertainties and allowances which are treated statistically.
and which ones are treated deterministically.
Using the Method of Moments the equation for P with all uncertainties is derived. The uncertainties being combinei statiktically are mostly normally distributed. The functional form of the rela 1. ion between many of these variables in the P equation is multiplicative l
not additive. Thus, there is no assurance that P is a normally-distributed variable,weichisassumedintheequationforthb95/95 tolerance 11miton P.
To provide this assurance and'to cc: firm the appropriateness of using the MbthodofMoments,aMonteCarlosimulationexperimentwasperformedtocombine the uncertainties using the appropriate underlying. input uncertainty distributions to determine-tle distribution of P. The Method of Moments result was always slightly more canservative, thus cbnfirming the acceptability of using the Method.of Moments.
Application of the methodology to analyses supporting future operating cycles will be via the use of the appropriate uncertainties for the cycle. Justifica--
4, tion for' the distribution, mean and standard deviation of each uncertainty will i
be included in the submittal for the first application of this methodology.
The P with all uncertainties will be determined from the nominal P associated with bach " flyspeck" power distribution. Determination of operating limits and.
g trip setpoints from the " flyspeck" data would then proceed as is currently done.
Volume 2 describes the use of the Monte Carlo Simulation technique to statistically combine the uncertainties in the determination of the operating limits for the prevention of DNB. The Monte Carlo of the'" Power-to-the-DNB-Limit" (Pa) process is used to develop the distribution due to the influence of the uncertainties being statistically combined. A "5tatistical DNBR Limit" (SDL) is then derived from the lower 95/95 'olerance limit value for P obtained from the P distribution.
Similarlyusingappro'riateuncertaintiesandopdratingconditionsSDtswillbe a
determined for use in determining'the Thermal Margin / Lcw Pressure trip setpoints, j
for use in determining adequate initial overpower mar DNB during Anticipated Operational Occurrences (A005) gin for the prevention of and for use in determining core-wide fuel failure fractions following the Seizure of a Reactor Coolant Pump Rotor.
Each SDL will replace the current Specified Acceptable Fuel Design Limit (SAFDL) in the applicable safety and setpoint analyses.
L i._
q
.. L The values for the distribution parameters of the uncertainties used in the o
L report were representative values. The actual values and the justification for their use as well as the actual SDLs derived for use in safety related calculations will-be included in the reload submittal for the first operating cycle to which the methodology is to be applied.
'The current methodology for determining the DNB limits has been previously reviewed and approved.
Each of the four analytical steps are discussed 'and the modifications necessary for the use of statistical combination of uncertainties are explained.
1 Thefirststepistodetermineaconservative(95/95)valuefortheSAFDLon DNBR for the combination of CHF correlation and computer code used to determine DNBRs. Sir,ce the SDL will-replace the DNBR SAFDL, the SDL will be derived using the same CHF correlation (YAEC-1), the same computer code (COBRA-3c) and the same core modeling technique. The uncertainty in the YAEC-1 prediction of-
- CHF ;;; included as one of the uncertainties being statistically combined in the derivation of the SCL.
The second step in the derivation of the operating limits for each cycle is the determination of the core power level at which the core " hot spot" would have a minimum DNBR (MDNBR) equal to the DNBR SAFDL for a spectrum of possible cycle specific power distributions. The only. change to this step is the_ substitution of the SDL-for the DNBR SAFDL as the convergence limit for the power level iteration. The COBRA model input used in the process would be more repre-sentative of nominal conditions since the uncertainties will have been statisti-cally. combined in the derivation of the SDL.
t I
The. third step is the derisation of the required TM/LP setpoints to prevent core operation at power levels k excess of the most limiting value of P Of d
I any power distribution whose corresponding I matches the current excore symmetricoffsetmeasurement,fortheexistiRgthermal-hydraulicconditions.
Again the.only changes are the substitution of the SDL for the current DNBR SAFDL and the use of the nominal COBRA model input.
l
'The final' step in the determination of the operating limits required to prevent the occurrence of DNB is the )rocess of verifying that the combination of the Synnetric Offset (S0) LC0 Tec1nical Specification _and Power Dependent CEA Insertion Limit (PDIL) provide adequate initial overpower margin for transients which may not. result in a protection system trip or whose RCS flow conditions are different from that assumed in the setpoint analysis (i.e., Loss of Flow events).
In this step the modification for use of the statistical combination of uncertainties will be the substitution of an appropriate SDL for the DNBR SAFDL as the lower limit for allowable MDNBR and use of the nominal COBRA' input.
-The Monte Carlo simulation technique places no restrictions on the independence of the effects of the unc etainties included in the process. The only criteria is that the uncertainty possess a significant random nature. Several uncertainties L
which were previously excluded from explicit consideration in the deterministic methodology, due to acknowledgement of. the large amount of excess conservatism inherent in the deterministic combination of uncertainties, were explicitly 2
- .+
+ 5 l
considered in performing the statistical combination. The uncertainties that will remain deterministically applied were those that lack a significant random component and those for which no data was available for their probability distribution functions.
In order to derive SDLs which will be independent of cycle, the licensee has selected power shapes that will maximize the offects of the-uncertainties.
Nominal hot full power operating conditions were selected for the base Monte t
Carlo case. To derive an SDL applicable' to transients whose flow conditions are less than 100% design flow, a separate simulation was conducted with a RCS flow rate of 67% of the design flow rate.
The results of the Monte Carlo simulation experiments are the output distribu-tions of P. The distributions were tested for normality and then the tolerance d
factor for a normal, distribution was applied in order to find the lower 95/95 tolerance limit value for P.
This P was then used to determine the SDL.
Sensitivity studies were peYformed to es.timate the effect of various parameters.
d The studies showed very little variation in the SDL value over the range of operating. conditions.
The SDL derived from the flat axial power shape was
-considerably more-limiting than the SDLs derived from-the FSAR design shapes.
3.0 CONCLUSION
S The licensee has presented a methodology using statistical combination of uncertainties. The Method of Moments technique will be used to determine the operating limits for the protection of the fuel from experiencing centerline melting. The Monte Carlo simulation will be used to determine the Departure from Nucleate Boiling limits.
We have reviewed the reports submitted by the licensee.
Based on this review, as outlined in the Evaluation section, we find this methodology acceptable for use. The values for the uncertainties used in the report were representative values. The actual values ~and the justification for their use as well as the actual-SDLs derived for use in safety related calculations will be included in the reload submittal for the first operating cycle to which the methodology is to be applied.
i
jy l ; * -g. --..:7,.
, L. -;
Distribution:
[ Docket:Filei,?-
i-. NRC' &' 1.ocal.. PDRs -
PDI-3 R/F:
S. Varga-R.-Wessman-t iM..Rushbrook'
_E. l.eeds:
' l 4,
Y. Nerses 0GC - 15 G18 E. Jordan - MNBB 3206 o
M. Chatterton - 8 E23 T. Collins - 8 E23 R. Jones - 8 E23 ACRS'(10)-P137
'J. Johnson'- Region I I
o
- i
'I f
I 5
f
(