ML18040A867
| ML18040A867 | |
| Person / Time | |
|---|---|
| Site: | Susquehanna  |
| Issue date: | 03/31/1988 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML17156A555 | List: |
| References | |
| NUDOCS 8804040191 | |
| Download: ML18040A867 (13) | |
Text
~p,$ RCgy~
~c~
wp
+
O~
Cy C
"o%
Pl Wp*y4 UNITED STATES NUCLEAR REGULATORY COMMISSION WASHINGTON, D. C. 20555 ENCLOSURE SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION RELATING TO LICENSING TOPICAL REPORT PL-NF-87<<001, REV.O
" UALIFICATION OF STEADY STATE CORE PHYSICS METHODS FOR BWR DESIGN AND ANALYSIS" PENNSYLVANIA POWER 5 LIGHT COMPANY SUS UEHANNA, UNITS 1
AND 2 DOCKET NOS. 50-387 AND 50-388
- 1. 0 INTRODUCTION By letter dated March 31, 1987, the Pennsylvania Power and Light Company (the licensee) requested approval of Topical Report PL-NF-87-001, Rev. 0, for the purpose of its use in licensing actions for the Susquehanna Steam Electric Station (SSES)
Units 1 and 2.
The report describes the qualification of the CPM-2 lattice physics and SIMULATE-E three-dimensional nodal core simulator programs for the steady state design and analysis of boiling water reactors (BWRs).
These programs are part of the Advanced Recycle Methodology Program (ARMP) developed by the Electric Power Research Institute (EPRI) for steady state analyses of light water reactors.
Brief descriptions of the CPM-2 and SIMULATE-E programs are presented along with comparisons to measurements from operating BWRs and experimental criticals.
The results of selected PD97 calculations for uniform lattice criticals and single fuel bundles are also presented.
These programs and associated methodologies are used by the licensee for plant operations support, various fuel 'cycle and safety related calculations, and to provide necessary neutronics input data to transient analyses for the two unit Susquehanna Steam Electric Station.
2.0
SUMMARY
OF TOPICAL REPORT The SIMULATE-'E three-dimensional code is used by the licensee to model the coupled neutronic and thermal-hydraulic behavior of the Susquehanna Unit 1 and 2
cores.
The required nuclear data are generated by the CPM-2 program which models the BWR fuel bundle and its environment (by-pass
- channel, cruciform control rod, etc.) in two-dimensions.
- 2. 1 Descri tion of the CPM-2 Pro ram CPM-2 is a modified version of the CPM (Collision, Probability Module) code developed in Sweden by AB Atomenergi/Studsvik for the analysis of PWR and BWR fuel assemblies.
The modeling combines fine group spectrum calculations for sub-regions of the assembly (e.g. fuel pin-cells), with a multigroup transport calculation for a partially homogenized, hetrogeneous assembly in two-dimensional (xy) geometry.
The code is distributed by EPRI, and is identical to the original CPM except for the input module which has been improved to make the program more "user friendly."
Since these modifications (as well as those made by the licensee in their implementation and use of CPM-2) did not affect the neutronics calculations, all the original benchmarking of CPM by EPRI/Studsvik is applicable to CPM-2 as well.
The calculational sequence for a typical BWR assembly involves three basic
- steps, with the spatial and energy detail becoming successively coarser as larger regions of the assembly are considered.
These steps are termed the micro-group, macro-group, and two-dimensional assembly calculations.
Cruciform control rods are treated via a special subroutine, and the depletion of gado-linia bearing fuel pins requires an auxiliary calculation with the MICBURN code
~
2.2 C M-2 IM The accuracy/adequacy of various aspects of. CPM-2 and its models (e.g. nuclear
- data, treatment of control rods and gadolinia) is demonstrated by comparisons to measured results from power reactors and experimental configurations.
Comparisons of eigenvalues (k ff), pin power/fission rate distributions, and
'eff
'sotopic concentrations versus burnup are presented.
Some of these results were generated by the licensee, while others were taken from the EPRI/Studsvik benchmarking of the original version of CPN.
Pin-cell calculations simulating 14 room temperature uniform lattice critical experiments were performed by PPRL to assess the accuracy of the CPN-2 reactivity calculation (based on the measured buckling).
Eight of the configurations contained U02 fuel and the fuel for the remaining 6 contained 2.0 weight percent Pu02 in natural uranium.
CPM-2 slightly underpredicted (by about 0.55 k) the keff for the U02 criticals, and overpredicted the multiplication factor for the remaining criticals, resulting in an average k ff of 1.0005 with a standard deviation of 0.0072 considering all criticals.
The accuracy of the CPN-2 calculation of the rod-wise power distribution was evaluated by comparisons to the gamma-scan measurements performed at squad Cities Unit 1 at the end of Cycle 2.
Two 7x7 N02 and three U02 bundles (one Sx8 and two 7x7) were considered in the comparisons.
Burnup and void operating history data were obtained for each bundle-elevation from a SIMULATE-E simulation.
These data were used in CPM-2 bundle calculations to arrive at the CPN-2/SIMULATE-E predicted statepoints corresponding to the measured data.
The comparisons showed generally good agreement between measurement and prediction (average
= 4.0$ ) with CPM-2 tending to overpredict the peak rod power.
The results of the EPRI/Studsvik benchmarking of the original CPM code to uniform lattice criticals, small core critical experiments performed at the KRITZ facility, and measured concentrations of uranium and plutonium isotopes from Yankee and Saxton spent fuel are also presented.
These comparisons show generally reasonable agreement between CPM predicted and measured quantities.
2.3 Oescri tion of SIMULATE-E The EPRI distributed SIMULATE-E three-dimensional coupled neutronics/
thermal-hydraulics core simulator program is used by PP8L in their steady state core analyses.
The thermal-hydraulics calculations use an EPRI developed void correlation and the FIBWR methodology developed by Yankee Atomic Electric
Company.
The methodology employed for the neutronics calculations may be selected by the user from several available options; PP8L uses the Modified Coarse Mesh Diffusion Theory '(PRESTO) option.
Two-group macroscopic cross sections for each fuel type are determined by CPM-2 as a function of fuel
- exposure, void history, moderator, fuel and control conditions, and xenon concentration.
After processing by NORGE-B2, they are input to SIMULATE-E along with radial and axial albedos applied at the core-reflector interfaces.
Normalization of the model to match plant operating data is performed via adjustment of several input data parameters.
Separate models are created at hot operating and cold conditions.
The licensee has made a number of changes to the code, including the ability to calculate the Critical Power Ratio (based on the Advanced Nuclear Fuels Corporation, formerly EXXON Nuclear, XN-3 critical heat flux correlation),
and linear heat generation rate and average planar heat generation rate thermal limits evaluations.
These changes have not resulted in any changes to the basic neutronics or thermal-hydraulics calculations.
2.4 SIMULATE-E uglification The qualification of the SIMULATE-E program is based on simulations of the first two cycles of squad Cities Unit-1 (gC-I) and Peach Bottom Unit-2 (PB-2),
and of the first two-plus and one-plus cycles (i.e., from BOL to approximately early 1987) of Susquehanna Units 1 and 2, respectively.
Comparisons of SIMULATE-E predicted values were made to hot and cold multiplication factors (k ff) and power and flow distributions.
The accuracy of the predicted power distributions was evaluated based on comparisons to TIP detector readings, and to results from gamma-scans.
Power and flow distributions were compared to results from the on-line core monitoring system.
The k ff comparisons for the Susquehanna units considered 257 hot operating condition steady-state statepoints, and 39 (3 local and 36 in-sequence) cold critical statepoints.
These comparisons indicated that the ability of the PPSL SIMULATE-E hot and cold models to predict k ff depends on the core average eff exposure and the gadolinia loading.
There is a hearly constant bias between the hot and cold predictions, with the hot k'ff consistently lower.
Using this eff data, the licensee generates hot and cold cycle-dependent target critical core keff curves for use in the core fol 1 ow, and shutdown margin and control rod worth analyses of individual cycles.
The power distribution comparisons utilized all available TIP sets from both Susquehanna units and considered nodal and axially averaged (radial) quantities.
Asymmetries in the measured data were quantified by considering symmetric nodal or radial TIP readings to provide an estimate of the measurement uncertainties associated with each TIP set.
Nodal RMS errors tend to be in the 4-6X range, with differences near the middle of cycle and,end of cycle power coastdown in the 6-9X range.
The average nodal and radial RMS errors considering all 82 TIP sets are 5.74 and 2.58 percent, respectively.
The corresponding average asymmetries based on 44 TIP sets are 5.22 and 2.55
- percent, respectively.
four core average axial power distribution and three-bundle flow comparisons are also presented, considering one statepoint per Susquehanna unit/cycle.
These comparisons are made to data produced by the on-line Core Monitoring System (CMS) to demonstrate consistency of the results.
(The GE process computer Pl program was used for the first cycle of both units, with the ANF POWERPLEX CMS used in all subsequent cycles).
These comparisons showed good agreement between the SIMULATE-E and CMS results.
i Comparisons to measured data from the first two cycles of guad Cities Unit 1
(gC-1) were also performed.
In addition to hot reactivity and TIP data similar to that from the Susquehanna units, the gC-1 measurements included 33 cold critical configurations (22 local) from Cycle-l, and bundle gamma scan measurements from the end of cycles (EOC) one and two.
The gC-1 hot critical comparisons showed a similar trend versus exposure to that observed earlier; however, the relatively low gadolinia loading in gC-1 resulted in the absence of the bowl-shaped gadolinia component in the variation.
The large cold critical data base served to augment the earlier analyses.
The (}C-1 cold critical comparisons were used to confirm that there
is no significant bias between SIMULATE-E predictions of k ff for in-sequence eff and local cr itical configurations.
The gC-1 based power distribution comparisons considered 15 TIP sets from Cycle 1 and 13 sets from Cycle 2, along with gamma scan data from 31 and 89 bundles at EOCl and EOC2, respectively.
The nodal and radial RMS differences from the TIP comparisons are roughly twice as large as those observed for the Susquehanna comparisons.
The EOC1 gamma scan data consisted of measuring the axial peak to bundle average La-140 activities and served to benchmark the SIMULATE-E calculation of the axial peaking factor.
The resulting difference was 1%
(
=25) with the agreement for controlled bundles considerably better than for uncontrolled.
The EOC2 gamma scan data is much more extensive and permits comparisons of individual bundle axial La-140 activity distributions, as well as radial, nodal and peak to average comparisons.
Peripheral and mixed oxide bundles were not included in the radial and nodal comparisons and the top and bottom six inches were eliminated from the nodal comparisons.
The peak-to average comparisons resulted in an average difference of about 0.2$
(
=1.5$ )
with a maximum difference of about 4A.
The average standard deviation from the individual bundle gamma scans was 6.35 with more than 855 of the individual bundle
's in the 5-8Ã range.
The standard deviation from the radial and nodal gamma scan comparisons were about 2X and 5.5~, respectively.
The quoted measurement uncertainty for the qamma scans was 3X.
The final qualification of SIMULATE-E presented in the report consists of power distribution, comparisons to TIP measurements and data from the GE P1 process computer for Peach Bottom Unit 2 (PB-2) cycles 1 and/or 2.
The level nf agreement with measured TIP data from these comparisons is treasonable and consistent with that observed earlier.
The purpose of the PB-2 simulations was to generate input for the analysis of the turbine trip tests performed near the end of Cycle 2, including an accurate representation of the initial conditions.
The non-steady state operation that preceded these tests required an accurate modeling of non-equilibrium xenon distributions and concentrations.
Comparisons of the predicted core average axial power distributions just prior to the
three tests (top peaked, middle peaked and slightly bottom peaked) to data from the process computer showed good agreement.
The geometry in the CPM-2 lattice physics code is limited to representing an individual fuel assembly.
In some applications,
- however, a multiple assembly calculation is required, and for these applications PPKL uses the general purpose PD(7 code.
The program solves the few group diffusion theory equation based on the finite difference spatial approximation in one, two, or three dimensions.
While up to five energy groups are permitted (including two ov'erlapping) thermal energy groups, the licensee generally utilizes four groups with a single thermal group.
Microscopic or macroscopic cross section data may be employed; PP8L typically uses macroscopic data from CPM-2 and processed with the COPHIN code.
2.6 ILII The PPllL qualification of PDg7 consisted of analyzing the same uniform lattice criticals used in the benchmarking of CPM-2, along with comparisons to CPM-2 assembly calculations for typical controlled and uncontrolled BWR fuel bundles.
The uniform lattice calculations modelled the critical core configurations in one-dimensional cylindrical geometry with an explicit accounting of the radial reflector and a buckling correction to account for axial leakage.
Reasonable agreement was obtained with the CPM-2 calculated keff s, 0.9972 versus 0.9951 and 1.0076 versus 1.0144 for the U02 and mixed oxide lattices, respectively.
The PDQ7 single fuel assembly calculations modelled each pin-cell explicitly, and used shielding factors derived by comparison to CPM-2 results, for gadolinia bearing fuel pins and control rods.
Two separate fuel bundles from the initial core loading of the Susquehanna units were selected for the comparisons.
The results showed generally good agreement between CPM-2 and PDg7 for the bundle k
's and rod-wise power distributions with maximum errors of about 4X and 7~ for uncontrolled and controlled bundles, respectively.
3.0 EVALUAITON The CPM-2 and SIMULATE-E programs were developed by EPRI for the steady state analyses of LWRs.
The licensee plans to use these codes for plant operations support, various fuel cycle and safety related calculations, and to provide necessary neutronics input data to transient analyses for the two BWR units at the Susquehanna Steam Electric Station.
The present review considered the information presented in the topical report and additional information provided by the licensee in a letter dated February 17, 1988.
The review considered the qualification of the FIBWR thermal-hydraulics methodology only in its role as an integral part of the SIMULATE-E program.
The performance of FIBWR as a stand-alone thermal-hydraulics
- code, and the validity/applicability of the ANF XN-3 CHF correlation were considered to be outside the scope of this review.
The methodologies (not including the qualification presented in this report) embodied in the CPN-2 and SIMULATE-E programs have been previously reviewed and found acceptable for steady state nuclear core design analyses of plants other than Susquehanna, and are representative of current practice.
The primary role of CPN-2 within the PPSL calculational sequence for BWR analyses is to provide nuclear data (basically two-group cross sections) to the SIMULATE-E core simulator program.
The benchmarking of SIMULATE-E via comparisons to measurements from operating BWRs therefore serves as the ultimate, though somewhat indirect, qualification of CPM-2.
- However, PPSL and EPRI/Studsvik have performed a number of comparisons to measured data from experimental configurations and operating BWRs to test various aspects of the CPM/CPM-2 neutronics calculation methodology and nuclear data.
Comparisons to uniform lattice cold criticals and KRITZ small core criticals provide an integral test of the ability of CPM-2 to predict reactivity (multiplication factors).
Comparisons to measured rod-wise gamma scan data for selected assemblies from an operating BWR, and to measured rod-wise fission rate distributions from KRITZ experiments, serve as a qualification of the treatment of neutron transport and other aspects of the modelling in the highly heterogeneous environments of real BWR fuel bundles and reactor cores.
Finally, comparisons of calculated uranium and plutonium isotopic concentrations were made to data from the destructive analysis of spent fuel from the Yankee and Saxton reactors.
The level of agreement between CPM-2 calculated and measured quantities is reasonable, and typical of that observed with currently accepted methods.
In addition, CPM-2 tends to overestimate the local peaking factor in an assembly, implying a generally conservative prediction of the linear heat generation rate.
The benchmar king of the SIMULATE-E program consisted of simulations of several cycles of operation.of three BWRs including all available data from PP8L's Susquehanna units starting at beginning of Cycle-1 (BOC1).
The hot reactivity comparisons involved more than five operating cycles (almost 300 statepoints) for cores containing a variety of BWR fuel bundle designs.
The calculated hot k ff exhibited a bias relative to the measured critical eff k ff which was consistent in magnitude with that observed for accepted eff three-dimensional core simulator codes.
The observed variation led to the development of a correlation which is a bowl shaped function of gadolinia loading and a roughly linear function of exposure.
This "target" k ff is used eff to predict the critical core k ff for a particular unit-cycle.
eff The cold critical comparisons considered 47 insequence and 25 local configurations.
The results showed a similar variation in the predicted cold critical k ff to that observed for hot conditions; the cold critical keff "target" for use with SIMULATE-E is therefore obtained bv adding a constant bias to the hot correlation.
In addition, the results showed no significant differences between the k ff for local and in-sequence criticals, thereby eff
demonstrating the ability of SIMULATE-E to perform shut-down margin ca 1 cul ati ons.
The benchmarking of the SIMULATE-E calculation of power distributions considered measured TIP detector readings and gamma
- scans, and data from plant core monitoring systems.
The albedos and other adjustable parameters were determined during model normalization to operating data from Susquehanna Unit 1 Cycles 1
and 2, and remained unchanged for all subsequent simulations.
The comparisons for the 82 TIP sets covering more than three cycles of operation of the two Susquehanna units yielded average nodal and radial RMS differences of 5.7 and 2.6 percent, respectively.
The estimated errors in the l
TIP measurements were determined by considering symmetric detector readings, and were of the same order.
The TIP comparisons for guad Cities and Peach Bottom yielded higher differences, i.e.,
nodal and radial RMS errors considering all TIP sets of about 10 and about 5 percent, respectively, for guad Cities, and somewhat lower for Peach Bottom.
The comparisons to the guad Cities gamma scan measurements at EOC1 and EOC2 further demonstrated the ability of SIMULATF.-E to calculate power distributions.
The axial peak to average was predicted to within about 15 with a standard deviation of 1-2$, and the standard deviations from the radial and nodal comparisons were about 2 and about 5 percent, respectively.
The peripheral bundles were not included in these comparisons, and in addition the top and bottom six inches were not considered in the nodal comparisons.
The quoted uncertainty for the gamma scan measurement is 3.0X.
Comparisons of core average axial power distributions to results from the AE Pl or ANF POWERPLEX core monitoring systems for the Susquehanna units and Peach Bottom Unit-2 (PB-2) near EOC2, though limited, showed good agreement.
The PB-2 comparisons considered the effects of non-equilibrium xenon and included top, middle and bottom peaked axial power distributions.
Three bundle flow distributions from the Susquehanna core monitoring systems were also compared to results generated by SIMULATE-E with generally good agreement.
The power distribution comparisons of SIMULATE-E to measured data showed generally reasonable agreement and were consistent with the levels of agreement observed with accepted methods.
The larger differences observed in the guad Cities and Peach Bottom comparisons are partially due to the SIMULATE-E models not being specifically normalized for these simulations.
The generally good agreement,
- however, provides reasonable confidence that SIMULATE-E can be used for predictive calculations for the Susquehanna units.
The limited comparisons of PD(7 to results from uniform lattice criticals and CPM-2 single assembly calculations showed reasonable agreement.
The comparisons were based on the use of 4 energy group cross sections from CPM-2.
The licensee notes that while it does not intend to perform three-dimensional calculations with PDg7, it may use the program for various two-dimensional analyses including independent verification of calculations, calculations of non-standard configurations such as partially loaded cores, and in the development of future model improvements for SIMULATE-E.
Appropriate qualification by the licensee of the use of PD(7 for configurations larger than multiple bundle arrays is recommended.
4.0 CONCLUSION
S The CPM-2 and SIMULATE-E codes were developed under the sponsorship of the Electric Power Research Institute and are part of the presently recommended procedures for BWR analyses similar to those intended for application to Susquehanna Units 1 and 2.
The benchmarking of the codes by the licensee relative to measurements from operating reactors and experimental configurations resulted in agreement typical of that observed with accepted methods.
The comparisons of PD(7 to results from uniform lattice criticals and CPM-2 single assembly calculations also showed reasonable agreement.
The staff therefore concludes that the CPM-2/SIMULATE-E methodology, and the use of PDg7 for auxiliary calculations represent an acceptable approach for analyses performed by the licensee in support of license applications and operation of the two BWR reactors at the Susquehanna Steam Electric Station.
The staff recommends that appropriate qualification be made by the licensee of the use of PDg7 for configurations larger than multiple bundle arrays, if such configurations are considered for calculation by PD(7.
The staff also recommends continued comparisons of calculated physics parameters with measured data from future physics startup tests and reactor fuel cycles.
Principal Contributor:
D. Fieno