PLA-3566, Forwards Responses to NRC 910215 Request for Addl Info on Statistical Combination of Uncertainties Methodology for Calculating MCPR Operating Limits

Forwards Responses to NRC 910215 Request for Addl Info on Statistical Combination of Uncertainties Methodology for Calculating MCPR Operating Limits

ML18026A238
Person / Time
Site:	Susquehanna
Issue date:	04/23/1991
From:	Keiser H PENNSYLVANIA POWER & LIGHT CO.
To:	Butler W Office of Nuclear Reactor Regulation
References
PLA-3566, TAC-75999, TAC-76000, NUDOCS 9104260077
Download: ML18026A238 (81)

Text

ACCELERATED DI TRIBUTION DEMONS TION SYSTEM E ~l I

REGULATORY INFORMATION DXSTRIBUTION SYSTEM (RIDS)

ACCESSION NBR:9104260077 DOC.DATE: 91/04/23 NOTARIZED: NO FACIL:50-387 Susquehanna Steam Electric Station, Unit 1, Pennsylva 50-388 Susquehanna Steam Electric Station, Unit 2, Pennsylva AUTH.NAME AUTHOR AFFILIATION KEXSER,H.W.

Pennsylvania Power 6 Light Co.

RECIP.NAME RECIPIENT AFFILIATION BUTLER,W.R.

Project Directorate I-2

SUBJECT:

Forwards responses to NRC 910215 request for addi info on statistical combination of uncertainties methodology for calculating MCPR operating limits.

DISTRIBUTION CODE:

AOOID COPIES RECEIVED:LTR

(

ENCL L SIZE:

TITLE: OR Submittal:

General Distribution NOTES:LPDR 1

cy Transcripts.

LPDR 1 cy Transcripts.

DOCKET 05000387 05000388 D

/

05000387 05000388 D

iRECIPIENT ID CODE/NAME PD1-2 LA RALEIGH,J INTERNAL: ACRS NRR/DET/ESGB NRR/DST 8E2 NRR/DST/SICB 7E NUDOCS-ABSTRACT OGC/HDS2 RES/DSIR/EIB EXTERNAL: NRC PDR NOTES COPIES LTTR ENCL 1

1 2

2 6

6

.1 1

1 1

1 1

1 1

1 0

1 1

2 2

RECIPIENT ID CODE/NAME PD1-2 PD NRR/DET/ECMB 9H NRR/DOEA/OTSB11 NRR/DST/SELB 8D NRR/DST/SRXB 8E OC L MB G FX 01 NSIC COPIES LTTR ENCL 1

1 1

1 1

1 1

1 1

1 1

0 1

1 1

1 D

R NOTE TO ALL"RIDS" RECIPIENTS:

PLEASE HELP US TO REDUCE iVASIE! CONTACT THE DOCUMENT CONTROL DESK, ROOM P 1-37 (EXT. 20079) TO ELIMINATEYOUR NAME FROM DISTRIBUTION LISTS FOR DOCUMENTS YOU DON'T NEED!

TOTAL NUMBER OF COPIES REQUIRED:

LTTR 26 ENCL 24 D

D

Pennsylvania Power & Light Company Two North Ninth Street~Allentown, PA 18101-1179 ~ 215/774-5151 Harold W. Keiser Senior Vice President-Nuc/ear 215/7744194 APR 23 )gg)

Director of Nuclear Reactor Regulation Attention:

Dr.

W. R. Butler, Project Director Project Directorate I-2 Division of Reactor Projects U.S. Nuclear Regulatory Commission Washington, D.C.

20555 SUSQUEHANNA STEAM ELECTRIC STATION INITIALRESPONSE TO RAI ON PL-NF-90-001 (SCU qUESTIONS)

PLA-3566 FILE A7-8C A17-2 R41-2 Docket Nos. 50-387 and 50-388

Reference:

Letter, M. C. Thadani to H.

W. Keiser, "Requ'est for Additional Information - Susquehanna Steam Electric Station, Units 1 and 2

(TAC Nos. 75999/76000)",

dated February 15, 1991.

Dear Dr. Butler:

This letter transmits PP&L's responses to questions 1

& 2 of Enclosure 2

(questions on PL-NF-90-001) of the referenced Request for Additional Information.

The responses to the remainder of the questions are in preparation and will be submitted to you in a timely manner.

The responses to questions 1

& 2, provided as Attachment 1 to this letter, provide additional information relating to PP&L's Statistical Combination of Uncertainties (SCU) methodology for calculating MCPR operating limits.

PP&L has elected to utilize the SCU method described. in PL-NF-90-001, because we believe it is the best technical approach available for establishing appropriate, conservative MCPR operating limits for the Susquehanna SES reactors.

To further ensure the validity of PP&L's methods, PP&L contracted S.

Levy, Inc. (SLI) to perform an independent technical assessment of PP&L's SCU methods.

Their conclusion was that the PP&L methods are technically acceptable, conform to the appropriate regulatory requirements, and provide an adequate level of conservatism.

The report from SLI is provided as.

>i04260077 9i0423 PDR

/DOCK 05000387 P

PDR gobi Ifl FILE A7-8C/A17-2/

PLA-3566 R41-2 Dr.

W. R. Butler These two questions regarding PPSL's SCU methodology are being addressed at this time in order to resolve this issue as soon as possible, because we currently plan to submit our proposed reload license amendment for Susquehanna Unit 1 Cycle 7

(U1C7) based on this methodology.

The proposed U1C7 amendment is currently scheduled for submittal to you in December 1991.

Given the time necessary to prepare that analysis (5-6 months),

PPLL requests that a high priority be maintained on the review of these responses.

This will minimize the potential for the UlC7 submittal to be inconsistent with the results of that review.

Any questions on this response should be directed to Hr. R. Sgarro at (215) 774-7916.

Very truly yours, H.

W.

er Attachments cc:

(Document-Control Desk:(on gi.nal:)>

NRC Region I Hr. G. S. Barber, NRC Sr. Resident Inspector - SSES Hr. J. J. Raleigh, NRC Project Manager -

OWFN Hr. K. Desai, NRR/SRXB -

OWFN

ATTACHMENT 1 EXECUTIVE

SUMMARY

OF RESPONSES TO QUESTIONS 1

& 2 The two NRC questions address the Statistical Combination of Uncertainties (SCU) methodology developed by PP&L for calculating MCPR operating limits.

The PP&L SCU method is a conservative, logical extension of currently used, NRC approved SCU methods.

The main difference between PP&L's method and already approved methods is that PP&L's method combines the traditional MCPR safety limit and change in Critical Power Ratio (hCPR) calculations into a unified statistical calculation.

The PP&L method directly calculates the MCPR operating limit and demonstrates compliance with the applicable specified acceptable fuel design limit.

Based upon their initial review, the NRC has not accepted PP&L's SCU methodology.

PP&L is, therefore, providing further technical justification in response to each NRC concern identified in Questions 1 and 2.

The NRC questions imply that the HCPR operating limits produced by PP&L's SCU methods are non-conservative and significantly lower than those produced by existing NRC approved methodologies.

This is not the case due to the comprehensive treatment of uncertainties and the use of conservative assumptions inherent'n PPEL's methodology.

Comparisons of PP&L and Advanced Nuclear Fuels (ANF) generated MCPR operating limits are provided in the detailed responses which follow.

The PP&L calculated HCPR operating limits are comparable to or conservative with respect to the ANF calculated HCPR operating limits.

In fact, for the limiting pressurization

events, the PP&L HCPR operating limits are the same or higher than the ANF generated limits.

The NRC questions indicate that the SCU method of combining the uncertainties in the HCPR safety limit type and hCPR analyses is acceptable if these uncertainties are statistically independent.

While the complete statistical independence of the HCPR safety limit and hCPR uncertainties may be desirable, neither the NRC approved nor PP&L's proposed SCU methodologies require it in order to be valid.

In the detailed responses, these uncertainties are demonstrated to be largely independent, and detailed discussions are provided to demonstrate that minor dependencies are conservatively treated in PP&L's methodology.

Various currently used, NRC approved SCU methods also treat the uncertainties in the two analyses as statistically independent, despite some minor dependencies.

The only difference between PP&L's SCU method and the approved versions is that PP&L utilizes a statistically rigorous method to combine the results of the analyses, while the approved SCU methods simply add them together.

Thus, PP&L believes that our SCU methodology is valid.

The responses to Questions 1

& 2 contain detailed technical discussions designed to fully respond to the concerns raised by the NRC in their initial review of PP&L's SCU methodology.

To further validate PP&L's SCU methodology for establishing MCPR operating limits, S.

Levy, Inc. was contracted to perform an independent technical assessment of the methodology.

SLI's conclusion is that PP&L's SCU methodology is technically sound and conforms to the applicable regulations.

The S.

Levy, Inc. report is included as.

gUESTION 1

There are two significant changes included in the analysis of the rod withdrawal error (RWE) event.

These are:

(I) the statistical treatment of the LPRH failures and (2) the statistical combination of the safety limit uncertainties and the SINULATE-E RWE calculational uncertainties.

In determining the core response to the RWE the hCPR is calculated as a

statistical average over all allowable (within technical specifications)

LPRH failure states.

Consequently, the calculated average hCPR is conservative for cases of low failures and non-conservative in the case of high LPRN failures.

Since this analysis is non-conservative for reactor states which cannot be precluded it is considered unacceptable.

The worst case condition of LPRH failures must be assumed in determining the hCPR resulting from a rod withdrawal error.

The basic assumption of the statistical combination of uncertainties (SCU) method is that the POWERPLEX safety limit uncertainties're independent of the RWE SINULATE-E hCPR,calculational uncertainties.

This assumption allows the statistical combination of the safety limit and SIHULATE-E calculational uncertainties, and'esults 'in a..non'-conservative reduction in the operating limit NCPR for the RWE.

Since 'the-POWERPLEX monitoring and the 'SINULATE-E rod block response calculational models employ much of the same nuclear and thermal-hydraulic modeling data and LPRH input, they cannot be considered to be independent.

It is therefore concluded that the SCU method is not applicable to the rod withdrawal event, and the SINULATE-E and safety limit uncertainties must be applied separately.

How will the removal of the SCU methodology and the statistical treatment of the LPRH failures be accommodated in the rod withdrawal methodology?

The POWERPLEX safety limit uncertainties are the POWERPLEX monitoring uncertainties (e.g.,

on bundle power) that are used in the statistical determination of the CPR safety limit.

il fl Q

RESPONSE

1 I.

Overview PP&L developed a Statistical Combination of Uncertainties (SCU) methodology for the Rod Withdrawal Error (RWE) event as described in PL-NF-90-001.

The PP&L SCU methodology for calculating HCPR operating limits for the RWE is a

logical extension of currently used NRC approved vendor and utility SCU

methods, which conforms to the applicable regulations.

In addition, the RWE SCU method requires a very thorough understanding of the design and performance characteristics of the systems and instruments involved in a RWE.

The methodology recognizes and explicitly models the fact that the rod block monitor system's response is dependent on a number of inputs, each of which may have an associated uncertainty.

Results of SCU analyses of the RWE show calculated HCPR operating limits similar to those produced by currently used NRC approved

methods, which, in combination with the rigorous treatment of uncertainties, provides strong evidence of the overall validity of the PP&L SCU method.

The SCU methodology is particularly applicable to the RWE event because it specifically accounts for the wide variability of the many analysis inputs allowed by the Technical Specifications and the RBH system design.

These inputs include:

the number of failed LPRHs, the LPRH failure combination, and the RBH channel assumed to be operable.

In addition, various measurement and calculational uncertainties are specifically accounted for in PP&L's SCU methodology.

Currently used NRC approved RWE methods use a prescriptive approach to select an analysis scenario and do not require all applicable parameters to be selected at their "worst case",. v'alues '(i.e., the LPRH failure combination, RBH system response, RBH trip setpoint, and parameter uncertainties are not all conservatively set in the adverse direction).

In

contrast, the PP&L SCU methodology statistically considers all combinations of important inputs and scenarios.

The RWE is an event which principally affects a localized region of the core.

The limiting HCPR assembly is assumed to be located near the control rod being

(I

withdrawn.

Those assemblies far away from the error rod (i.e.,

two or more control cells away) do not experience significant reductions in HCPR, and therefore would not contribute significantly to the number of fuel rods calculated to be in boiling transition.

In the HCPR SL type calculations used in PP&L's SCU methodology, conservative values of axial power shape, radial power distribution, and local peaking distribution (corresponding to a core wide event) are used.

The SL type calculations used are identical to those used for the analyses of core wide transients.

These assumptions represent a

significant conservatism in that they result in a conservative number of fuel rods calculated to be in boiling transition for the RWE.

In addition to the conservatism inherent in using the power distributions corresponding to a "core wide" event, the PP&L RWE methodology contains significant conservatism as described in PL-NF-90-001.

Specific conservative assumptions include:

1) an unrealistic, conservative control rod pattern, 2) zero xenon concentration, 3) a conservatively high LPRH failure probability, 4) one rod block monitor channel inoperative, and 5) conservative values of applicable uncertainties.

To further validate PP&L's SCU methodology'or, establishing HCPR operating limits, S.

Levy, Inc. was contracted to perform an independent technical assessment of the methodology.

The conclusion of the review is that PP&L's SCU method is technically sound and conforms to the applicable regulations.

The S.

Levy, Inc. report is included as Attachment 2.

The following sections address the issues of: 1) the conservatism of the HCPR operating limits produced by PP&L's SCU methodology,

2) the validity of the statistical treatment of LPRH failures,

3) the conservatisms inherent in n

)1

'll l

li F

I

'I f

PP&L's methodology relative to LPRHs and the RBH system, and 4) the validity of treating the HCPR SL type and hCPR calculations as statistically independent.

II.

HCPR 0 eratin Limits The NRC question states that the HCPR operating limits produced by the PP&L SCU method are non-conservative.

This is not the case because PP8L's SCU method employs a more extensive treatment of uncertainties and the use of more conservative assumptions than currently used NRC approved RWE methodologies.

Currently used NRC approved RWE analysis methods do not explicitly include all the uncertainties which are explicitly included in PP8L's analysis methodology, rather they assume that the effects of these uncertainties are covered by other conservative assumptions.

When comparing the PP8L SCU methodology to currently used NRC approved RWE

methods, the PP&L SCU method produces comparable HCPR operating limits.

The HCPR operating limits 'for the RWE event calculated by both the PP8L SCU method and the NRC approved Advanced Nuclear Fuels Corporation RWE method (currently used for reload analysis on both Susquehanna units) are presented in the following table for Susquehanna Unit 2, Cycles 2

& 5:

Unit C cle U2C5 U2C2 PP&L 1.27 1.26 HCPR OL*

ANF 1.29 1.27

• The RWE event is not the limiting event in establishing the HCPR operating.'limit for.',the Susquehanna units.

These values would be the HCPR OL, if the RWE were the limiting event.

'I i

r'TI I

t As stated previously, the PP&L'SCU method for the RWE event requires the explicit treatment of all relevant uncertainties and produces HCPR operating limits which are comparable to those produced by current NRC approved methodologies.

tl

III. Statistical Treatment of LPRN Failures The statistical treatment of LPRH failures is performed in order to account for the wide variability in RBH response due to the variability of possible LPRH failures consistent with the RBH system design.

The SCU methodology includes consideration of all allowable LPRH failures by statistically considering every LPRH failure combination.

PP8L believes it is necessary to investigate all LPRH failure combinations because no a priori "worst" configuration can be determined due to the complex interaction of the local LPRH responses with the bundle power responses produced by the error rod withdrawal.

Sample analyses performed by PP&L demonstrate that no a priori "worst" LPRN failure configuration can be determined.

, Table 2. 1-2 in PL-NF-90-001 provides the RBM response for all possible LPRM failure'ombinations.

In many cases, but not always, an increase in the number of failed LPRHs would result in a less sensitive RBN response.,

However, analyses, have demo'nstrated that the 4

locations of inoperable LPRHs affect the results more than the number of LPRM failures.

The Channel-A RBH responses for a sample RWE analysis are summarized in Table l-l.

For this sample analysis the rod was withdrawn to rod position 24.

The results shown in Table l-l demonstrate that the difference between the minimum and maximum RBN response increases as the number of LPRH failures increases.

For the case of four LPRH failures per RBM channel, which is the maximum allowed by the RBH system design, the difference between the minimum and maximum RBN response is extremely large.

Therefore, the statistical analysis is used to include the probability of the failure combination corresponding to the minimum RBH response in the analysis.

The probability of the failure combination corresponding to the occurrence of the minimum RBH response, however, is less than 3

X 10

, conservatively assuming a

LPRH failure probability of 0. 15.

For a more realistic failure probability based on plant

data, the probability of this failure combination is approximately 2 x 10 When this low probability is combined with the probability of having a

RWE event for the specific rod pattern and error rod location used in the analysis

I

.t lg d

Pi

and the probability of failure or bypassing of one RBM channel, the total probability of the event is well below the probability used to classify the event as an anticipated operational occurrence.

This type of probability evaluation provides the justification for the NRC acceptance of other statistical methodologies, such as currently used vendor and utility SCU methodologies for pressurization events.

The PP&L SCU method for RWE analysis, in addition to considering all possible LPRH failure combinations and applicable parameter uncertainties, contains a

number of significant conservatisms in the treatment of LPRH failures and the RBN system response.

These conservatisms are discussed in the following section.

IV.

Conservative Treatment of LPRM Failures and the RBM S stem The PP&L SCU methodology contains three major input conservatisms which result in a conservative probability distribution function for the RBM response.

In other words, the calculated probability of having a given low value of RBH response is greatly overestimated.

First, one channel of the RBN is assumed to be out of service.

Second, a conservatively high LPRM failure rate is assumed.

Third, the RBH initialization/normalization process is I>

conservatively evaluated.

N The first major 'input conservatism consists of the assumption that only one RBM channel is available for the event.

Although allowed by Technical Specifications for a short period of time (i.e.,

24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />),

actual Susquehanna RBN channel inoperability rates are very low.

This assumption significantly decreases the RBM system's assumed ability to detect and terminate the RWE.

The second major input conservatism is the use of a high LPRM failure probability.

Based on examining the number of LPRM failures at the end of cycl,e for each Susquehanna unit and cycle, the average number of actual LPRN failures is approximately seven.

The number of LPRH failures ranged from three to ten.

The use of a LPRH failure probability equal to 0. 15 in the RWE analysis corresponds to approximately 26 LPRH failures in the core, which is

almost four times the actual average and 2.6 times the maximum number of observed LPRH failures.

Therefore, the calculated probability distribution for the RBN response is conservative.

The third input conservatism consists of the analysis assumption that the initial RBM response is set equal to the initial power level.

The initial RBM response for a particular rod is determined when the rod is selected by the operator.

The RBH system then initializes its response to the higher of either the average power level from the APRHs or the average local power of the LPRHs.

In reality, the initial RBM response for a highly peaked power

location, as in the case of the analyzed RWE event, would be higher than the initial APRN average power and closer to the trip setpoint.

In the RWE

analysis, however, the initial RBN response is conservatively set equal to the core average power level.

Therefore, the analysis assumption of initializing the RBH response to the initial power level is very conservative, because the analysis allows the error rod to be withdrawn farther than the RBH system actually would allow.

V.

Statistical Inde endence of hCPR and NCPR SL T e Anal ses The NRC question also asserts that statistically combining the HCPR safety limit and hCPR analyses is valid only if the two sets of uncertainties are independent.

In general, when"d'eveloping models and methods, it is desirable to use the best data and correlations available.

This does produce some common assumptions between the HCPR SL"type and "hCPR calculations.

However, as will be discussed later, the validity of ".the statistical combination does not require that the two sets of uncertainties be completely independent.

The uncertainties in both the HCPR SL and hCPR are highly separable for the RWE as a result of the calculations involved.

The MCPR SL type calculations account for uncertainties in the absolute values of a particular set of parameters, while the hCPR for the RWE event is sensitive to changes in a different set of parameters (e.g.,

RBH response change).

The uncertainties associated with the absolute state condition have been adequately evaluated for POWERPLEX in the safety limit type calculations.

One key assumption of the PP&L RWE analysis methodology effectively establishes the independence of the HCPR SL type and hCPR calculations for the RWE.

This assumption relates to the choice of the initial conditions for the hCPR calculation as opposed to the initial conditions for the HCPR SL type calculations.

The cycle exposure, rod pattern, and xenon concentration assumed for the hCPR calculations are completely different from the cycle

exposure, rod pattern, and xenon concentration assumed for the HCPR SL type calculations.

As a result, the axial, radial, and local pin power distributions used for the RWE hCPR calculation (selected to be conservative for the hCPR calculation) are significantly different from the power distributions used for the MCPR SL type calculations (selected to be conservative for the HCPR SL type calculations same as those used for core wide transients).

This significant conservatism effectively decouples the two analyses; the limiting bundles in the HCPR SL type calculations have no relation to the limiting bundles in the hCPR calculation and, thus, the uncertainties relating to bundle power, bundle flow, axial power distribution, and local peaking factor are uncorrelated.

In addition, absolute independence of the HCPR SL and hCPR uncertainties may be desirable,

however, the validity of the statistical combination of hCPR and HCPR SL type analyses does not require it for the following reasons:

l.

If an uncertainty is common to both calculations, but a variation in that parameter affects the two calculations in opposite 1

directions (e.g,,'higher hCPR in the'transient calculation and fewer fuel rods in boiling transition in the HCPR SL type analysis),'then treating variations',of that parameter-'as if they

'er'e independent'is conservative, since thi's assu'mption will produce more variation in the result than actually occurs.

2.

If a parameter which has an uncertainty that is common to both analyses has only a small effect on either calculation, then treating them as independent does not affect the result and is valid.

f1 t

I f

There is a significant difference between the uncertainty in a

parameter and the uncertainty in the change in that parameter.

The hCPR calculation largely depends on the codes'bilities to calculate changes in key parameters, while the HCPR SL calculation depends on the monitoring system's ability to calculate the absolute value of a parameter.

In general, small changes in a

parameter (comparable to the uncer tainty on that parameter) have a

much smaller effect on the hCPR than on the absolute value of HCPR.

I II 3.

If a parameter which has an"uncertain'ty.,that is common to both analyses has a significant effect on both calculations, and the

.effect of a variation in,that parameter affects both calculations

.in'he same directi'on,'.then it"1s possible to conservatively treat the dependence.

For example, one solution would be to treat the parameter in a conservative, deterministic manner in one of the calculations, thus eliminating the dependence.

For a calculated parameter common to both analyses, if the error in the parameter for the HCPR SL analysis is independent of the error in that parameter for the hCPR analysis, then treating them as independent is correct.

These errors are independent (i.e.,

uncorrelated) if the parameter for the HCPR SL analysis comes from a different source (i.e., different neutronic and thermal hydraulic methodologies) than that parameter for the hCPR analysis.

All common uncertainties in the RWE and HCPR SL analyses fall into one of the above categories.

Appendix l-l to this response describes each uncertainty or set of uncertainties used in the HCPR safety limit type analyses in its relationship to the RWE event analyzed with PP&L's SCU methodology.

The discussion of the individual parameters in the appendix shows that the uncertainties that are related to or common to both calculations have been appropriately considered in the SCU methodology.

I h

1

(

w r

1 l

I l

l

It should be noted that the assumption of independence is also made in all currently used NRC approved vendor and utility SCU methods (GE,

ANF, PECo, TVA, etc.).

Adding a statistically calculated HCPR SL and a statistically calculated hCPR is also a metho'd of combining the uncertainties;

however, the PP8L method is a more statistically rigorous method.

VI.

Conclusion The PP8L SCU method for calculating HCPR operating limits for the RWE event is a logical extension of currently used NRC approved SCU methods, which conforms to the applicable regulations.

The treatment of LPRM failures in the PP&L RWE methodology is conservative.

The HCPR operating limits produced by the PPEL SCU methodology for the RWE event are comparable to operating limits produced by currently used NRC approved RWE methods.

The treatment of individual uncertainties in the HCPR safety limit type and hCPR analyses as statistically independent has been shown either to have no effect on the calculated HCPR oper ating limits, or to be, conservative.

Therefore, the SCU methodology applied to the RWE ev'ent as described in'PL-NF-90-'01alid and conservative.

N j

PPKL believ'es that the SCU methodology for establishing HCPR operating limits for the RWE event presented in PL-NF-90-001 is technically sound and conforms to the applicable regulations.

References l-l XN-NF-734 (P)(A), "Confirmation of XN-3 Critical Power Correlation for 9x9 Fuel Assemblies,"

February 1985.

1-2 XN-NF-80-19 (P)(A), Volume 1 and XN-NF-80-19 (P)(A), Volume 1, Supplements 1 5 2, "Exxon Nuclear Methodology for Boiling Water Reactors Neutronic Methods for Design and Analysis," Harch 1983.

TABLE 1-1 Calculated RBN Response vs.

LPRH Failure Combination Withdrawal to Axial Position 24 (Taken from Table 2.1-2 of PL-NF-90-001)

Condition LPRH Failure Combination Calculated RBN Res onse No LPRHs failed 1

LPRH failed (Hin Response) 1 LPRH failed (Max Response) 2 LPRHs failed (Nin Response) 2 LPRHs failed (Hax Response) 3 LPRMs failed (Min Response) 3 LPRHs failed (Hax Response) 4 LPRHs failed (Hin Response) 4 LPRHs failed (Hax Response) 12 35 41 87 163 110.4 107.0 112.4 105.1 115.2 102.8 119.2 101.0 126.0 "Hin Response" refers to the failure combination that results in the lowest RBH response "Hax Response" refers to the failure combination that results in the highest RBH response b

F K

l I

I h

APPENDIX 1-1 Discussion of Uncertainties in the hCPR and POWERPLEX HCPR SL Analyses for the RWE Event The uncertainties that affect the HCPR Safety Limit (HCPR SL) type calculations are discussed below in relation to the effect each uncertainty has on the hCPR calculation for the RWE transient.

In addition, uncertainties relating to the LPRH response are discussed.

Specifically, the question of whether or not the uncertainty is common to both the HCPR SL type and hCPR calculations and the effect that the assumption of statistical independence has on the conservatism of the resulting.HCPR operating limit are discussed.

Four major justifications for treating the variations in a parameter in the HCPR SL and hCPR analyses as statistically independent are provided in the main text of,the response'to guestion 1'.

It should be noted that the axial, radial, and local pin power distributions used for the RWE dCPR calculation (selected to be conservative for the hCPR calculation) are significantly different from the power distributions used for the HCPR SL type calculations (selected to be conservative for the SL type calculations same as those used for core wide transients).

This significant conservatism effectively decouples the two analyses; the limiting bundles in the HCPR SL type calculations have no relation to the limiting bundles in the hCPR calculation and, thus, the uncertainties relating to bundle power, bundle flow, axial power distribution, and local peaking factor are uncorr elated.

Even if the HCPR SL type and hCPR analyses for the RWE were not largely decoupled by this assumption, treating the uncertainties as independent would be acceptable for the reasons discussed below.

a)

System Parameter Uncertainties (Feedwater Flow Rate, Feedwater Temperature, Core Pressure, Core Flow Rate, and Core Inlet Temperature):

In the PPKL SCU methodology for the RWE event, the uncertainties in these parameters do not have a significant impact on the hCPR calculation, and are not treated statistically in the hCPR calculation.

The initial conditions for.the RWE event are established by setting these parameters to their nominal rated values.

These parameters do not change significantly during the event.

The initial conditions define the core thermal power, core flow, core pressure, and core inlet subcooling.

Sensitivity analyses were performed that demonstrate that 1

1

,1 t

f" W

rated conditions on the 100X rod line, combined with other conservative assumptions (e.g.,

xenon free condition) produce conservative results for the RWE event.

The system parameter uncertainties are considered in the statistical evaluation of the core monitoring system.

XN-3 Correlation Uncertainty:

The XN-3 correlation has been shown to be conservative for ANF 9X9-2 fuel assemblies (Reference l-l), and the XN-3 correlation uncertainty is not included in the RCPR analysis for the'WE event.

This approach is reasonable, since even though small variations in the critical power may affect the absolute value of HCPR, they will not have a significant impact on the change in CPR.

The SCU method employed for the RWE event assures that the XN-3 correlation inputs are appropriately conservative.

Assembly Flow Uncertainty:

I'n the RWE event analysis, the total core flow is held constant.

Although the total core flow does not change, the assembly flows are redistributed when the error rod is withdrawn.

This flow redistribution is caused by the fact that the two phase pressure drop increases as the bundle power increases.

However, the RCPR for the RWE event is affected by the calculated change in assembly flow, while the number of rods in boiling transition in the NCPR SL type analyses is affected by the absolute value of the assembly flow.

Thus, the uncertainties related to assembly flow which affect the two analyses are different.

A sample analysis was performed which showed that the change in assembly flow due to redistribution for a RWE is on the order of IOX, and that change contributes approximately 0.056 to the RCPR; the majority of the hCPR for the RWE comes from the changes in assemblies powers.

A change in assembly flow comparable to the assembly flow uncertainty (2.7N which would imply a 27X error in the change in assembly flow) would only contribute a 0.015.

However, it is expected that the uncertainty in the change in assembly flow would be much less than the uncertainty on the

1 Y

k h

I

absolute value and, thus, would contribute less than a 0.01 to the calculated RCPR.

Therefore, the assembly flow uncertainty is not an important parameter for the hCPR calculation for the RWE.

As a result, treating the assembly flow uncertainties in the HCPR SL type and hCPR analyses as independent does not affect the calculated operating limit and, thus, is valid.

Radial Bundle Power Uncertainty:

The bundle flow and bundle power are calculated by XTGBWR in the core monitoring system and by SIMULATE-E in the dCPR analysis.

Although the two simulation codes use the same lattice physics methodology to generate a portion of their input, a significant factor that supports the treatmerit of the two calculations as statistically independent is the POWERPLEX UPDATE procedure.

The POWERPLEX UPDATE procedure adjusts the XTGBWR calculated power distribution based on measured LPRM responses to obtain a pseudo-measured, power distribution.

This Pt 5

r adjustment" process 'effectively decouples the,SIMULATE-E and POWERPLEX methodologies.

Therefore, the SIMULATE-E and the POWERPLEX calculated MCPRs may also be considered independent, because of the different bundle powers and flows calculated by the two methodologies.

The, POWERPLEX core monitoring system uses the XTGBWR program with the UPDATE methodology to calculate the bundle power.

As discussed previously, this methodology is very different from that used in SIMULATE-E, in part because the SIMULATE-E methodology does not employ measured LPRH feedback in its power distribution calculation.

Therefore, the error in the POWERPLEX bundle power may be treated as independent of the error in the SIMULATE-E bundle power.

In the RWE event analysis, the RCPR is evaluated using the SIMULATE-E code and is primarily dependent on the change in the radial bundle power during the RWE event.

The uncertainty of concern for the RWE hCPR calculation is, thus, the uncertainty in SIMULATE-E's ability to predict the change in bundle power.

It should be noted that a conservative uncertainty for the change in bundle power is used as described in Section

2. I of PL-NF-90-001.

r b

ff SI

Therefore, since the radial bundle power related uncertainties relevant to the HCPR SL type analyses and the hCPR analyses are not based on the same parameter (i.e., absolute bundle power vs.

change in bundle power),

and the errors in absolute bundle power in SIMULATE-E and POWERPLEX are not correlated, assuming the bundle power errors for the HCPR SL type and hCPR calculations are statistically independent is valid.

e)

Local Pin Power Uncertainty:

Since the HCPR limiting fuel assemblies for the RWE event are unrodded, the relative local pin power distribution does not change significantly in the limiting RCPR bundles.

In addition, the hCPR is insensitive to the initial local pin power distribution provided it does not change significantly during the event.

Therefore, the local pin power uncertainty has a negligible effect on the determination of the hCPR for the RWE event, and the treatment of the errors in local pin power as statistically independent between the two analyses do'es'ot affect the resulting HCPR operating limit and,

hence, is valid.

U f)

Axial Power Uncertainty:

The axial power uncertainty does not significantly contribute to the HCPR SL type analysis results, since HCPR has only a weak dependence on axial power distribution its effect is small enough so that ANF uses a bias in their HCPR safety limit analysis and does not randomly vary this parameter.

In the RWE event, the local axial powers change significantly during the event.

The SIMULATE-E calculated axial power shape is different than that calculated by POWERPLEX because of the measured LPRH feedback used in POWERPLEX but not used in SIMULATE-E.

As a result, the axial power uncertainties applicable to the HCPR SL type and hCPR analyses are independent.

Therefore, treating the axial power distribution errors in the HCPR SL type and hCPR calculations as statistically independent is valid.

g)

LPRM Response Uncertainty:

In the RWE event, the,calculated;RBH response is affected by the LPRH response uncertainty.

A negative error in calculated RBH response (i.e., the calculated RBN response is lower than the "true" RBH response)'auses, the 'error" rod to be withdrawn further and thereby increases the hCPR.

The LPRH response uncertainty, which is derived by comparing the results of code calculations with measured

data, is comprised of two components:

I) the uncertainty in a code's ability to calculate the "true" LPRH response and 2) the LPRN measurement uncertainty.

Due to the method POWERPLEX uses to match the calculated and measured LPRH responses, the calculational components of LPRH uncertainty in SIMULATE-E and POWERPLEX are shown below to be truly independent.

The measurement uncertainty component of the LPRH uncertainty is also discussed below.

It should be noted that the LPRH response uncertainty is only one factor in determining the radial bundle power uncertainty which affects the HCPR SL type calculations.

Since the radial bundle power uncertainty produces a different error in each bundle for the HCPR SL type calculation, and the LPRM uncertainty produces a different error for each LPRM surrounding the error rod in the hCPR calculation, the uncertainties as they affect their respective calculations are really quite different, and thus they may be treated as independent.

Uncertaint in Calculated LPRH Res onse In the POWERPLEX core monitoring system, the calculated power distribution is modified by using measured LPRM responses so that the calculated LPRH response for the core monitoring system (after the UPDATE procedure) is equal to the measured LPRH response

and, hence, the POWERPLEX "calculational" uncertainty is essentially equal to the LPRM measurement uncertainty.

Thus:

1 J

I l

I

'1 t

A

2 2

OLPRM,PPX,calc

+LPRM,measurement For SIMULATE-E, however, the calculational component of the LPRH response uncertainty is independent of the LPRH measurement uncertainty.

This conclusion comes from the fact that the total SIMULATE-E LPRH uncertainty was calculated by comparisons of SIMULATE-E calculations to measured data.

Thus, the total SIMULATE-E LPRH uncertainty can be expressed as:

2

+LPRM, SIM-E 2

+

2 OLPRM, SIM-E,calc OLPRM,measuxement Therefore, the calculational components of LPRH response uncertainty for POWERPLEX and SIMULATE-E are truly independent.

LPRM Measurement Uncertaint In an actual RWE event, the time at which the RBH trip occurs (i.e., rod motion ceases) is affected by the change in the measured responses of the LPRHs surrounding the error rod location.

Also, the POWERPLEX core monitoring system uses the absolute values of LPRH measurements to generate a "pseudo" measured power distribution.

Therefore, for the HCPR SL type analyses (which account for uncertainties pertaining to POWERPLEX), the absolute values of the measured LPRH responses would affect the calculated bundle powers

and, hence, would affect the number of pins calculated to be in boiling transition.

However, as discussed below, the uncertainty in the absolute value of an LPRH measurement has no impact on the hCPR calculation for the RWE, due to the conservative assumption regarding the RBH initialization.

The related uncertainty for the hCPR calculation is the uncertainty in the change in LPRH measurements during the event, which is not the same as the uncertainty in the absolute value of LPRH measurements.

I 1x it C

l f

The LPRN measurement uncertainty, as it applies to the RWE event

analysis, relates to the following three parameters:

1) the initial RBH

response,

2) the RBN response during the event, and 3) the RBN trip setpoint.

The initial RBH response is first discussed.

The RBM system is designed such that the initial RBM response is automatically set equal to the larger of 1) the APRH average power and 2) the local LPRN average power.

For PP8L's RWE analyses, the initial RBH response is conservatively set equal to the APRH average power for all cases.

Therefore, as a result of this conservative assumption pertai'ning to,the RBH system i

initialization, the 'uncertainty on the absolute value of LPRH measurements does not affect, the calculation, of.the initial RBH response.

The second important parameter related to the LPRH measurement uncertainty is the RBH response during the event.

For a RWE event, the RBM system response varies based on the changes in the input LPRH responses and the initial RBN response established by the RBM initialization process.

The PPKL method of calculating the transient RBH response models the fact that the RBH response is proportional to the ratio of the summation of input LPRH responses at the time of interest to the summation of the initial LPRH responses.

Any absolute error in the initial LPRH response will carry through the event and will cancel in the ratio calculation.

This ratio is multiplied by the initial APRM power level to determine the calculated RBH response.

As a

result, the calculated RBN response during a

RWE is independent of the

uncertainty in the absolute value of the LPRH measurements.

The RBN response calculated by the PP8L method described above is equivalent to the actual RBH system response for cases in which the initial average LPRN response is less than the APRH power level.

However, the method is very conservative for cases in which the initial average LPRH response is greater than the APRN power level.

The RBN system response for these cases is equal to the average of the LPRH 19-

kl 1

l f

I

responses.

Given an initial LPRH local average of 103K and a final LPRH local average of 108X, the plant RBH system would calculate a response of 103X and 108K, respectively.

Using PP&L's method to calculate the RBH response with the same LPRH inputs would result in corresponding RBH responses of 100X and 105X.

The PP&L method thus creates a significant conservatism in the RBH response for a case in which the average of the local LPRHs is greater than the APRH power level.

Recognizing the fact that the uncertainty in the change of the LPRH response needs to be considered, PP&L uses a conservative RBH trip setpoint based on the General Electric evaluated drift, accuracy, and calibration uncertainties.

The application of these uncertainties is described in PL-NF-90-001.

0 In conclusion, i,t has been shown that,, while the HCPR SL type analysis is dependent

'on the, uncertainty in the absolute'alue'of LPRH,"

measurements, the RWE hCPR calculation is not dependent on that uncertainty.

Similarly, the uncertainty in the change in LPRH measurements is an important uncertainty in the RWE event hCPR calculation, which is conservatively treated, however, this uncertainty does not affect the MCPR SL type analyses.

In addition, the PP&L RWE analysis methodology essentially decouples the two analyses with regard to the LPRH measurement uncertainty by the use of a conservative initialization process.

1

QUESTION 2

The analysis of the generator load rejection without bypass (GLRWOB) event employs the SCU methodology, and statistically combines the POWERPLEX safety limit uncertainties with the RETRAN/SIMULATE-E hCPR calculational uncertainties.

The statistical combination results in a significant reduction in the calculated HCPR operating limit.

This statistical combination is valid only if the POWERPLEX CPR monitoring and the RETRAN/SIMULATE-E hCPR calculational uncertainties are independent.

The POWERPLEX safety limit and RETRAN/SIMULATE-E uncertainties are considered to be dependent because of (1) the common nuclear and thermal-hydraulic modeling data used to represent the

reactor, (2) the similarity of the POWERPLEX and SIHULATE-E calculational methods and (3) the adjustments made to RETRAN/SIMULATE-E such as the water density (kinetics parameter) correction.

While it is recognized that this interdependence does not result in a perfect correlation of the uncertainties, it is concluded that they cannot be considered to be independent as assumed in the SCU methodology.

It is noteworthy that in the review of NEDO-24154 (Reference 21 of PL-NF-90-001) the staff determined that a

5X probability of exceeding the CPR saFety limit is acceptable.

That is, the acceptable calculational uncertainty was determined to be a 95X probability/95X confidence level value.

The SCU methodology does not provide this assurance.

Based on the above, it is concluded that the SCU method is not acceptable for application to the overpressurization transients.

How will the removal of the SCU methodology be accommodated in the GLRWOB transient operating limit HCPR calculation?

l k

hl t

l

)r 1

C I

k j

RESPONSE

2 I.

Overview The PP&L Statistical Combination of Uncertainties (SCU) methodology for calculating HCPR operating limits for thepressurization transients is a

logical extension of currently used NRC approved vendor and utility SCU

methods, which conforms to the applicable regulations.

PP&L's,SCU methodology combines the traditional hCPR and HCPR Safety Limit calculations into a unified statistical calculation.

The goal of the PP&L SCU method is to calculate a

HCPR operating limit such that the applicable Specified Acceptable Fuel Design Limit (SAFDL) will not be violated during normal operation or Anticipated Operational Occurrences (AOOs).

The applicable SAFDL is discussed in the Standard Review Plan (SRP) Section 4.4 and is found in the Susquehanna FSAR (Section 4.4.1.1):

"Specifically the Hinimum Critical Power Ratio (HCPR) operating limit is specified such that at least 99.9 percent of the fuel rods in the core are not expected to experience boiling transition during the most severe moderate (Per Regulatory Guide 1.70 Revision 2) frequency transient events" Similarly, as stated in the SER on GETAB (Reference 2-1):

"the proposed design basis (i.e.,

more than 99.9X of the fuel rods in the core would be expected to avoid a boiling transition caused by single operator errors or equipment malfunctions) is acceptable when applied to core-wide transients such as a turbine-trip" The PP&L SCU method calculates the HCPR operating limit (HCPR OL) to demonstrate compliance with this design basis.

The approach used by GE (5X chance of exceeding the HCPR SL), while a conservative

approach, is not required by the General Design Criteria (GDC) or the SRP.

The appropriate SAFDL is 99.9X of the rods expected to avoid boiling transition.

The PP&L SCU method calculates a

HCPR OL consistent with this SAFDL.

k,

+

The PP&L SCU method for the analysis of pressurization events contains

~

~

~

~

~

~

~

significant conservatism, as discuss'ed in PL-NF-90-001.

The conservative assumptions include:

1)

A conservative scram insertion versus time curve is used.

2)

Conservative (technical specification maximum) reactor trip and "I

recirculation'ump trip delay times are used.

II y

3)

The events are typically~analyzed at end-of-cycle/all-rods-out conditions.

4)

The most limiting core flow is assumed.

5)

Core power is conservatively treated; the limiting value is used for the generator load rejection, and the feedwater controller failure is analyzed as a function of power level.

6)

The RETRAN code uncertainty used was established at the 95/

confidence level.

To further validate PP&L's SCU methodology for establishing HCPR operating limits, S.

Levy, Inc.

was contracted to perform an independent technical assessment of the methodology.

The conclusion of SLI's review is that PP&L's SCU method is technically sound and conforms to the applicable regulations.

The S.

Levy, Inc. report is included as Attachment 2.

The following sections address the issues of 1) the conservatism of the HCPR operating limits produced by PP&L's SCU methodology, and 2) the validity of treating the HCPR SL type and hCPR calculations as statistically independent.

II.

Conservatism of HCPR 0 eratin Limits The NRC question implies that the HCPR operating limits produced by the PP&L SCU method are significantly lower than those produced by currently used NRC approved methods.

This is not the case, as shown below, because PP&L's methodology contains a more extensive treatment of uncertainties and more conservative assumptions than are employed in currently used NRC approved methodologies.

As a result, it is highly unlikely that PP&L's SCU methodology would ever result in HCPR operating limits that are significantly lower than those that would be produced by the currently used NRC approved methods.

For

example, the Susquehanna Unit 2 Cycle 5 HCPR operating limits for the limiting pressurization transients calculated by the PP&L licensing analysis methods were either comparable or 'conservative,wi,th respect to those generated by ANF using their licensing methods as shown in the following table:

EVENT I

f N

PP&L OL ANF OL GLRWOB with RPT 1.32 1.32 FWCF at 80/100 FWCF at 65/100 FWCF at 40/100 1.34

• 1.41 1.55 1.28 1.30 1.32

• Interpolated between 84K and 65X power It should be noted that the ANF calculations assumed technical specification maximum scram times, and the PP&L calculations assumed scram insertion times based on plant data (i.e., faster scram than used in the ANF calculations).

If the calculations were to be performed with consistent scram times, the PP&L calculated HCPR operating limits would be even more conservative relative to the ANF calculated limits.

Thus, the PP&L SCU methodology treated as a whole does not produce less conservative HCPR operating limits than the ANF methods currently used for reload licensing analyses on the Susquehanna units.

III. Statistical Inde endence of dCPR and HCPR SL T e Anal ses The NRC question also asserts that statistically combining the HCPR Safety Limit and hCPR analyses is valid only if the two sets of uncertainties are independent.

In general, when developing models and methods, it is desirable to use the best data and correlations available.

This can lead to some common assumptions between the MCPR SL type and hCPR calculations.

However, as discussed later, either the uncertainties are independent, the uncertainties have a negligible effect, or the commonality is conservatively treated.

While the complete independence of the HCPR Safety Limit and hCPR uncertainties may be desirable, the validity of the statistical combination does not require it for the following reasons:

If an uncertainty is common to both calculations, but a variation in that parameter affects the two calculations in opposite directions (e.g.,

higher hCPR and fewer fuel rods in boiling transition in the HCPR SL type analysis),

then treating variations of that parameter as if they were independent is conservative, since this assumption will produce more variation in the result than actually occurs.

2.

If a parameter which has an uncertainty that is common to both analyses has only a small effect on either calculation, then treating variations of that parameter as if they were independent does not affect the result and is valid.

There is a significant difference between the uncertainty in a parameter and the uncertainty in the change in that parameter.

The hCPR calculation largely depends on the codes'bilities to calculate changes in key parameters, while the HCPR SL calculation depends on the monitoring system's ability to calculate the absolute value of a parameter.

In general, small changes in a

parameter (comparable to the uncertainty on that parameter) have a

much smaller effect on the hCPR than on the absolute value of HCPR.

1 I I

II i

I

3.

If a parameter, whose uncertainty is common to both calculations has a significant effect on both calculations, and the effect of a variation in that parameter affects both calculations in the same

'direction; then it is still possible to conservatively tr'eat the K

dependence.

Core power isthe only'arameter 'that'alls into this category and it was identified as such in PL-NF-90-001.

One solution to this situation is to exclude power from the statistical analyses of hCPR (i.e.,

use the value of core power that produces the highest calculated hCPR).

This approach is planned to be used for licensing analyses using PPKL methods.

Appendix 2-1 to this response discusses each uncertainty that is a contributor to the HCPR SL type calculations.

It is shown that the common uncertainties fall into one of the above described categories and, thus, treating them as statistically independent yields conservative results.

It should be noted that the assumption of independence is also made in al.l currently used NRC approved vendor and utility SCU methods (GE,

ANF, PECo, TVA, etc.).

Adding a

statistically calculated HCPR SL value and a statistically calculated hCPR is also a method of combining the uncertainties;

however, the PP&L method is a

more statistically rigorous method.

For example, no currently used NRC approved methods recognize that power is a

common uncertainty to both the HCPR SL type and hCPR calculations for the GLRWOB.

The PP8L SCU method,

however, recognizes this fact and conservatively treats the dependence (Appendix B of PL-NF-90-001).

Studies by vendors and other utilities indicate that the majority of the RETRAN/SIHULATE-E RCPR code uncertainty is due to uncertainty in the Doppler,

void, and scram reactivities.

These neutronics uncertainties are not part of the HCPR SL type analyses.

Thus, the uncertainties that have the greatest impact on the hCPR analyses are not uncertainties that directly impact the HCPR safety limit type analyses.

Finally, the RETRAN cross section adjustment process is needed so that the 1-D RETRAN model predicts the same changes in cross sections in response to a rapid pressure increase as the detailed 3-D neutronics model and is a

necessary part of the 3-D to 1-D collapsing process.

The cross section adjustment process does not affect the initial state point cross sections.

The adjustment only affects.the change in a cross section as a function of change in moderator density.

Thus, the cross section adjustment process has no effect on.the MCPR SL type calculations and only, affects the hCPR I,

calculations.

IV.

Conclusion The PP&L SCU method for calculating HCPR operating limits for pressurization events is a logical extension of currently used NRC approved SCU methods, which conforms to the applicable regulations.

The HCPR operating limits produced by PP&L's SCU method are comparable to or conservative with respect to operating limits produced by currently used NRC approved methods.

PP&L's SCU methodology contains a more extensive treatment of uncertainties and more conservative assumptions than are employed in currently used NRC approved methodologies.

As a result, it is highly unlikely that PP&L's SCU methodology would ever result in HCPR operating limits that are significantly lower than those that would be produced by currently used NRC approved methods.

The treatment of individual uncertainties in the HCPR SL type and hCPR analyses as statistically independent has been shown either to have no effect on the calculated HCPR operating limit or to be conservative.

PP&L believes that the SCU methodology for establishing HCPR operating limits for pressurization events presented in PL-NF-90-001 is technically sound and conforms to the applicable regulations.

References 2-1 "General Electric BWR Thermal Analysis Basis (GETAB): Data, Correlation and Design Application", NED0-10958-A, January 1977.

2-2 "gualification of the One-Dimensional Core Transient Model for Boiling Water Reactors",

NEDE-24154-P, Volumes I, II, and III.

2-3 "Methods for Performing BWR Reload Safety Evaluations",

PECo-FHS-0006-A, June 15,1990.

'PPENDIX 2-1 Discussion of Uncertainties in HCPR SL and hCPR Analyses for Pressurization Events t

The uncertainties that affect the HCPR Safety Limit (HCPR SL) type calculations are discussed below in relation to the effect each uncertainty has on the hCPR calculation for a pressurization transient.

Specifically, the question of whether or not the uncertainty is common to both calculations and the effect that the assumption of independence has on the conservatism of the resulting HCPR operating limit are discussed.

Three major justifications for treating variations in a parameter in the HCPR SL and hCPR analyses as statistically independent are provided in the main text of the response to Ouestion 2.

NON-NEUTRONICS UNCERTAINTIES a)

Feedwater Flow Rate 1.76X In the HCPR SL type calculations, the feedwater flow affects calculated core power.

Core power is conservatively treated in the transient analyses (i.e., the worst power level is used for the GLRWOB or the HCPR OL is defined as a function of power for the FWCF).

Also, core power is not used in the statistical hCPR calculations

and, hence, its uncertainty is not common to both HCPR SL and hCPR analyses.

Uncertainty in the feed flow itself does not affect the

hCPR, since initial feed flow in RETRAN is determined via a heat balance based on core power.

Thus, a different feed flow would imply a different core power for the RETRAN transient analyses.

Therefore, feedwater flow rate uncertainty does not affect the transient hCPR calculation, and assuming independence in the statistical analyses is valid.

This conclusion is supported by reference to various NRC approved transient analysis

methods, in which feedwater flow rate uncertainty is not included in the calculation of the system model code uncertainty (e.g.,

References 2-2 and 2-3).

I h

1 V

Bl

b)

Feedwater Tem erature

.76K = 3 De rees F i:-

In the HCPR SL type calculations, the feedwater temperature affects calculated core power.

Core power is conservatively treated in the transient analyses (i.e., the worst power level is used for the GLRWOB or the HCPR OL is defined as a function of power for the FWCF).

Also, core power is not used in the statistical hCPR calculations

and, hence, its uncertainty is not common to both SL and hCPR analyses.

Uncertainty in the feedwater temperature itself has no effect on the dCPR for a GLRWOB (see Table 3. 1-2 of PL-NF-90-001).

The uncertainty in feedwater temperature has only a minimal impact on hCPR for the FWCF (on the order of.001; see Table 3.2-2 of PL-NF-90-001).

For the FWCF

event, the small effect of the feedwater temperature uncertainty is included in the RETRAN code uncertainty as shown in Table 3.2-3.

Since the impact of feedwater temperature is negligible for the dCPR calculations, treating the uncertainty in feedwater temperature in the HCPR SL type and hCPR calculations as statistically independent does not affect the results

and, hence, is valid.

It should be noted that both the feedwater flow and feedwater temperature uncertainties used for Susquehanna HCPR safety limit type calculations were derived in GETAB (Reference 2-1) based on the assumption of 2 measurement sensors for each measurement.

The Susquehanna

units, however, have 3 sensors for each measurement.

The derivation of these uncertainties using the same approach used in GETAB and crediting the three sensors would reduce these uncertainties (i.e.,

the uncertainties used in the analysis would be multiplied by a factor of the square root of 2/3).

c)

Core Pressure

.5X = 5 si Core pressure variations of 5 psi have a negligible impact on hCPR calculations for pressurization events as demonstrated by appropriate analyses (Tables 3. 1-2 and 3.2-2 of PL-NF-90-001).

Therefor e, treating the pressure variations in the SL and hCPR calculations as statistically independent does not affect the result and is valid.

d)

C~F1

.5K In both the GLRWOB and FWCF events, a higher core flow produces a higher calculated hCPR.

For the HCPR SL type analyses,

however, a higher core flow produces fewer fuel rods in boiling transition.

Therefore, variations in core flow affect the HCPR SL type and hCPR calculations in opposite directions, and assuming that variations in core flow in the SL and hCPR calculations are statistically independent is conservative.

e)

Core Inlet Tem erature 0.2X = 1.0 De ree F

The core inlet temperature used in RETRAN calculations is determined by a heat balance calculation using the design carryunder and user specified values of core power, core flow, and pressure regulator pressure.

Thus, core inlet temperature is determined in the RETRAN analyses by user specified input.

Any uncertainty introduced by this assumption is included in the RETRAN RCPR code uncertainty, which is based on Peach Bottom turbine trip test analyses using this initialization process and is defined at the 95X confidence level.

Therefore, the measurement uncertainty in core inlet temperature does not affect the transient hCPR calculations.

This conclusion is supported by reference to various NRC approved transient analysis

methods, in which core inlet temperature uncertainty is not included in the calculation of the system model code uncertainty (e.g.,

References 2-2 and 2-3).

e No currently licensed method explicitly treats this uncertainty in transient hCPR analyses.

Rather, the conservatism of the critical power correlation for transient analysis is demonstrated by comparisons of calculated and measured CHF for a set of transient CHF tests.

The p

N

conservatism of PPEL's use of,the~XN-3 correlation for transients is demonstrated'y c'omparisons'f calculated and measured times to boiling transition for transient CHF tests (see Appendix B of PL-NF-89-005).

i I

A g)

Assembl Flow 2.7X

'ncluding the assembly flow uncertainty in the transient analysis of pressurization events would have negligible impact on the I-D neutronics data used by the RETRAN system model, since the effects of small variations in the individual assembly flows would tend to balance each other (some assembly flows would be overpredicted, while others would be underpredicted).

As a result, the effect of the assembly flow uncertainty on the hCPR calculation for a pressurization event is negligible, and assuming that the uncertainties are independent does not affect the results of the calculation.

Analysis has demonstrated that small differences in the RETRAN hot bundle flow rate due to uncertainties have a small effect on dCPR for pressurization events.

All currently used NRC approved methods also assume that hCPR is insensitive to small changes in initial bundle flow. It should also be noted that the RETRAN hot bundle model uses significantly different two-phase flow correlations than used by POWERPLEX.

Thus, treating the assembly flow uncertainties in the HCPR SL type and hCPR calculations as statistically independent does not affect the calculated HCPR operating limit and,

hence, is valid.

NEUTRONICS UNCERTAINTIES Uncertainties in the three local neutronics parameters would have negligible impact on the RETRAN system model response to pressurization events since there are 764 bundles in the Susquehanna

cores, and the effects of small local variations will balance each other in calculating the I-D neutronics data which will determine the core response.

It should also be noted that the use of a "pseudo hot bundle" in the analysis of pressurization events effectively decouples the HCPR SL type and hot bundle analyses.

The inputs to the RETRAN "pseudo hot bundle" calculations (bundle power, axial power distribution, gap conductance, etc.)

are chosen so that the calculated RCPR bounds the expected RCPRs for the actual HCPR limiting bundles in the core.

Thus, the RETRAN "pseudo hot bundle" has no relation to the actual HCPR limiting bundles in the core, and the uncertainties on radial bundle power, bundle flow, axial power distribution, and local peaking factor are uncorrelated with those uncertainties in the HCPR SL type calculations.

h)

Radial Bundle Power:

As stated

above, the radial bundle power uncertainty has a negligible impact on the I-D neutronics data used by the RETRAN system model, since the effects of small variations in the individual bundle powers would tend to balance each other (some bundle powers would be overpredicted, while others would be underpredicted).

As a result, the effect on the 4CPR calculation for a pressurization event is negligible, and treating the uncertainties as independent does not affect the results of the calculation,

and, hence, is valid.

The PPEL 4CPR methodology for pressurization transients (as well as currently used NRC approved vendor and utility methodologies) utilizes a "pseudo-hot bundle" method.

The hot bundle analyzed is selected so that the calculated 4CPR bounds the expected behavior for actual hot bundles in the core.

The 4CPR method treats bundle power as an iteration variable to produce minimum CHFR =

1.0, as described in Appendix 8 of PL-NF-89-005 (identical to the approved ANF 4CPR methodology).

Thus, the uncertainty in SIHULATE-E's ability to calculate bundle power is not relevant to the hot bundle methodology.

1 f

E i)

Local Pin Power':

0 Calculations performed in determining the RETRAN code uncertainty (PL-NF-89-005) demonstrated that local peaking factor does not affect calculated 4CPR for the pressurization transients.

There are two forms

~I

I r4 I

I Il

~

II

'I

of the XN-3 correlation one for internal rods in a bundle and one for peripheral rods.

The internal form of XN-3 is limiting for pressurization

events, and local peaking factor does not affect calculated hCPR due to the mathematical form of the XN-3 correlation.

Thus, since local peaking factor does not affect the calculated

hCPR, treating the uncertainties as statistically independent does not affect the results

and, hence, is valid.

j)

Axial Power:

The axial power uncertainty does not significantly contribute to the SL type analysis results, since HCPR has only a weak dependence on Axial Power Distribution (APD) its effect is small enough so that ANF uses a bias in their HCPR SL analysis and does not randomly vary this parameter.

PP&L methods assume the hot bundle and core average APDs are the

same, in conformance to approved ANF methodology.

Since limiting fuel bundles at Susquehanna have more bottom peaked APDs than the core average APD, assuming that the normalized hot bundle power versus time is the same as the core average normalized power versus time represents a significant conservatism, due to the more rapid power reduction due to scram for bottom peaked APDs.

Thus, since one of the calculations (the HCPR SL type calculation) is insensitive to the axial power uncertainty, and the hCPR calculation contains significant conservatisms in regard to this parameter, treating the axial power uncertainties as statistically independent is valid.

ms0725i.crl:el

ATTACHMENT 2 S.

Levy, Inc. Technical Assessment of PP&L's MCPR Operating Limit Methodology I

K k'

7

S. LSIV INCORPORATED 3425 S. Bast:om Flvenue Campbell, CR 95008-7006 USFI 408 / 377-4870 FflX408 / 371-6804 April 9, 1991 Mr. C. R. Lehmann Two North Ninth Street Pennsylvania Power 8 Light Company Allentown, PA 18101

SUBJECT:

Assessment of PP&L's Operating Limit Minimum Critical Power Ratio Methodology.

Dear Chet:

Attached for your use is S. Levy Incorporated's (SLI) report, "Assessment of Pennsylvania Power & Light Company's Methodology for Establishing the Operating Limit Minimum Critical Power Ratio." This report was prepared by Dr. D. L. Fischer, Dr.

S. Levy, and myself. The assessment was based on the information contained in three Pennsylvania Power 8 Light Company reports:

(1) PL-NF-90-001, "Application of Reactor Analysis Methods for BWR Design and Analysis;"

(2) PL-NF-89-005, "Qualification of Transient Analysis Methods for BWR Design and Analysis;" and (3)

PL-NF-90-005, "Susquehanna SES Unit 2, Cycle 5 - Reload Summary Report."

From an overall standpoint, SLI has concluded that the PP&L methodology to establish the. thermal margins is technically acceptable.

Application of the PP&L reactor analysis methodology in the reload fuel design and analysis process produces an operating limit MCPR that is comparable to those produced by fuel supplier and other utility methodologies that have previously been approved by the NRC.

Further, the applicable regulatory requirements have been satisfied, and implementation of this methodology by PP8L is considered appropriate.'f you have any questions related to this report, please contact me.

Sincerely, Concurrence by, R. E. Enge, Manager Systems, Core and Analysis Methodology S. Levy Incorporated S. Levy, President S. Levy Incorporated

0 0

ASSESSMENT OF PENNSYLVANIAPOWER 8. LIGHTCOMPANY'S METHODOLOGYFOR ESTABLISHINGTHE OPERATING LIMITMINIMUMCRITICALPOWER RATIO by R. E. Engel D. L. Fischer and S..Levy S. Levy Incorporated 3425 S. Bascom Avenue Campbell, California 95008

J 0

Ap l~

l ISA I

k>

1.0 INTRODUCTION

AND

SUMMARY

S. Levy Incorporated (SLI) has performed an assessment of Pennsylvania Power &

Light Company's (PP&L) application of its reload analysis methodology documented in the report PL-NF-90-001, "Application of Reactor Analysis Methods for BWR Design and Analysis." As a part of the assessment, the report PL-NF-89-005, "Qualification of Transient Analysis Methods for BWR Design and Analysis," was reviewed with respect to the determination of model uncertainties.

These PP&L reports provide the primary source of documentation of the PP8L reload analysis methodology that is currently undergoing review by the Nuclear Regulatory Commission (NRC). Also reviewed was the report PL-NF-90-005, "Susquehanna SES Unit 2, Cycle 5 - Reload Summary

'eport" which contains the results of the application of the methodology to a specific reload analysis.

The purpose of this report is to provide SLI's assessment of the technical adequacy of the methodology proposed by PP&L for establishing the thermal margin required for plant operation.

From an overall standpoint, the PP&L report PL-NF-90-001, "Application of Reactor Analysis Methods for BWR Design and Analysis," is a very high quality report that, is technically well conceived.

The methodology is a logical extension of current industry practice, including the use of statistical analysis techniques currently employed in the safety analysis process.

The applicable regulatory requirements have been satisfied, the statistical approach to establish the thermal margins is technically acceptable, and an adequate level of conservatism exists in the methodology.

Implementation of this methodology is considered appropriate.

The primary figure of merit for thermal margin is the operating limit minimum critical power ratio (MCPR).

The methodology used to establish the operating limit MCPR includes consideration of analyses and evaluations that are required in the reload fuel design and safety analysis process.

The results of the analyses performed to establish the operating limit MCPR are provided in the reload summary report.

PP&L has developed a comprehensive statistical analysis process to generate the operating limit MCPR considering both the specified acceptable fuel design limit (SAFDL) associated with fuel cladding overheating and the change in critical power ratio (CPR) conservatively calculated for anticipated operational occurrences.

The PP&L methodology closely parallels the approaches taken by reload fuel suppliers and other utilities in their approved reload fuel design and analysis methodologies, as they are used to establish the operating limit MCPR.

The PPBL methodology includes.a more comprehensive treatment of uncertainties and employs more sophisticated statistical analysis techniques.

Based on the SLI assessment, it has been concluded that the overall methodology used to establish the MCPR operating limit is technically acceptable and that an adequate level of conservatism exists in the methodology, This report contains three additional sections.

Section 2.0 summarizes the current regulatory requirements and guidelines that provide the regulatory bases for

establishing the required thermal margin, which is reflected in the development of the

~

~

~

~

~

'perating limit MCPR. Section 3.0 describes the PP&L methodology used to establish the operating limit MCPR. The relationship between the regulatory requiiements and guidelines and the PP8L thermal margin methodology, including the SLI evaluation of the PPBL approach to establish the operating limit MCPR, is also provided.

The conclusions drawn as a part of the assessment of the PP8L thermal margin analysis methodology are given in Section 4.0.

4

~

'th t

I' IA I

I'l jq Q

2.0 REGULATORY REQUIREMENTS AND GUIDELINES The basic objective of reload fuel design and analysis methodology is to demonstrate that a proposed reload fuel design and core operating configuration can be operated safely, has adequate thermal margin, and is in accordance with the current regulatory requirements and guidelines.

This is generally accomplished by demonstrating compliance with the NRC Regulations, considering the guidance provided by specific Regulatory Guides and the Standard Review Plan (SRP).

The specific regulatory requirements and guidance that are used to establish the thermal margin, as defined

'by the operating limit MCPR in the reload fuel design and analysis process, are discussed in more detail below.

2.1

~R Regulations are the statutory requirements placed on nuclear power plants.

Regulations place requirements on all phases of nuclear power plant design, construction, and operation.

With respect to thermal margin, General Design Criterion (GDC) 10 provides the basic regulatory requirement.

For fuel cladding integrity, GDC-10 requires that the reactor core and associated coolant, control, and protection systems shall be designed with appropriate margin to assure that SAFDLs are not exceeded during any condition of normal operation, including the effects of anticipated operational occurrences.

Anticipated operational occurrences are defined in the GDC as those conditions of normal operation which ar' expected to occur one or more times during the life of the plant.

These events are to include but not-be limited to loss of power to all recirculation pumps, tripping of the turbine generator set, isolation of the main condenser, and loss of all offsite power.

Guidance in establishing the specific limits associated with SAFDLs is provided in the SRP.

2.2 Regulatory Guides are issued by the NRC to describe and make publicly available acceptable methods of implementing specific parts of the Regulations, to delineate techniques used in evaluating specific problems, postulated accidents, or to provide guidance to applicants, The Regulatory Guides provide guidance related to the specific events that should be considered in the reload fuel design and safety analysis process; however, the specific methodology that is to be used to demonstrate acceptable thermal margin is not specifically addressed.

2.3 The SRP, documented in NUREG-0800, is prepared and issued for the guidance of NRC Staff Reviewers in performing safety reviews of applications to construct or operate nuclear power plants.

The principal purpose of the SRP is to assure the quality and uniformity of reviews and to present a well defined base for the evaluation

~

I of proposed changes in the scope and requirements of reviews.

It is also the purpose of the SRP to make information about regulatory matters widely available and to irriprove communication and understanding of the NRC Staff review process by interested members of the public and nuclear power industry.

I With respect to the SAFDL for fuel cladding overheating, SRP 4.4 provides the applicable guidance.

For boiling water reactors, the SAFDL is to assure that greater than 99.9% of the fuel rods are not expected to experience boiling transition.

It should be noted that there is a difference between the examples of acceptable approaches identified in the SRP and the PP&L methodology for reload fuel design and analysis related to the treatment of the SAFDL.

In the example currently applicable to BWRs, it is stated that for CPR correlations, the minimum value of CPR is to be established such that at least 99.9% of the fuel rods in the core would not be expected to experience boiling transition during normal operation or anticip'ated operational occurrences.

In current fuel supplier and utility applications, a statistically derived MCPR value which, if not exceeded, willassure compliance with the SAFDL is incorporated into the plant technical specifications as the fuel cladding integrity safety limit.

In the PP8L approach, the SAFDL that greater than 99.9% of the rods not be expected to experience boiling transition becomes the fuel cladding integrity safety limit. The significance of this difference is discussed in more detail in Section 3.0, with respect to the development of the operating limit MCPR.

'I

,1 k'

3.0 OPERATING LIMITMINIMUMCRITICALPOWER RATIO The regulatory requirements and guidelines related to reload fuel design and analysis, as they are applied to establish the operating limit MCPR are described in Section 2.0.

With respect to the thermal margin requirements, as reflected by the operating limit MCPR, the regulatory requirements and guidance are primarily focused on assuring that the SAFDL will not be exceeded.

The operating limit MCPR resulting from the reload fuel design and safety analysis process is implemented through the technical specifications.

Anticipated operational occurrences are analyzed to demonstrate that the SAFDL associated with cladding overheating has a very low probability of being exceeded.

The specific values for the operating limit MCPR are established such that there is a high probability with a high confidence that the SAFDL for the fuel cladding overheating (fuel cladding integrity safety limit) will not be exceeded for the limiting anticipated operational occurrence, considering the entire power/flow map.

Anticipated operational occurrences are generally characterized by the nuclear system parameter variation which poses the most significant challenge to the fuel or reactor coolant pressure boundary capabilities; These parameter variations can generally be characterized in eight categories:

(1) decrease in core coolant temperature; (2) increase in reactor pressure; (3) decrease in reactor coolant flowrate; (4) reactivity and power distribution anomalies; (5) increase in reactor coolant inventory; (6) decrease in reactor coolant inventory; (7) increase in reactor coolant flow; (8) increase in reactor coolant temperature.

Based on an evaluation of the events in each of these categories, it has been determined that the analysis of the following seven events establishes the operating limit MCPR, providing all other safety analysis constraints are satisfied (e.g., initial MCPR assumed in the loss of coolant accident analysis):

(1) generator load rejection without bypass; (2) feedwater controller failure; (3) loss of feedwater heating; (4) control rod withdrawal error - power operation; (5) recirculation flow controller failure - increasing flow; (6) rotated fuel assembly loading error; and (7) mislocated fuel assembly loading error.

In the PP&L reload fuel design and analysis methodology, each of these events is evaluated to establish the operating limit MCPR.

In the reload safety analysis process, it is necessary to demonstrate that it is highly unlikely that the calculated event consequences will exceed the SAFDLs considering the uncertainties in the analysis process.

These include uncertainties related to:

(1) model and model input; (2) operating state; and (3) instrument measurement.

Current industry practice for establishing the operating limit MCPR is to add the change in CPR for the limiting anticipated operational occurrence to the fuel cladding integrity safety limit MCPR to assure that greater than 99.9% of the fuel rods in the core would not be expected to experience boiling transition. The change in CPR is typically selected such that there is at least a 95% probability that the change in CPR will not be exceeded with a confidence level assessed to be about 95%.

In this approach, generally 1 to 3 highly sensitive parameters are identified and parametric analyses performed, consistent with an experimental design, for the limiting events, to establish an event unique response surface that can be used to determine the probability of the change in CPR.

This response surface is statistically combined, using Monte Carlo techniques, with the model uncertainties and the response surface fitting error to establish the probability of a given change in CPR for the event being evaluated.

Sufficient Monte Carlo samples are run to establish the probability distribution with a statistically high confidence level such that the 95th percentile is used to represent the 95% probability of not being exceeded.

In typical applications, the confidence level is inferred by the selection of conservative inputs for the non-statistically treated parameters and the selection of the parameter uncertainties for the statistical inputs.

The PP8L process for establishing the MCPR operating limit employs a more comprehensive statistical treatment of the number of rods not expected to experience boiling transition for the limiting anticipated operational occurrences.

In the PP&L approach, the same general process is used to establish the probability distribution for the change in CPR.

This probability distribution is then statistically combined, using Monte Carlo techniques, with the probability distributions for the number of rods 'not expected to experience boiling transition to define an operating limit MCPR such that greater than 99.9% of the fuel rods in the core would not be expected to experience boiling transition for the event.

Based on the number of Monte Carlo samples, the results of the safety limit type calculations are adjusted to assure a 95% confidence level for the sampling process, In developing the probability distributions for the number of rods not expected to experience boiling transition, the Advanced Nuclear Fuels Corporation (ANF) safety limit methodology is used, and the same uncertainties are incorporated.

Thus, the same conservatism that exists in the current ANF methodology is retained in the PP8 L methodology.

The PP8L approach satisfies the requirements of the Regulations (GDC-10) which state that the reactor core and associated coolant, control, and protection systems shall be designed with appropriate margin to assure that specified acceptable fuel design limits (SAFDLs) are not exceeded during any condition of normal operation, including the effects of anticipated operational occurrences.

This approach also satisfies the regulatory guidance of the SRP which states that a value of MCPR is to be established such that at least 99.9% of the fuel rods in the core would not be expected to experience boiling transition during normal operation or anticipated operational occurrences.

Therefore, the PP8L approach is considered a valid method.

Based on the evaluation of the application of the PP8L reactor analysis methodology with the specific uncertainties considered, it has been assured that the actual operating limit established using this process provides adequate margin for uncertainties in the analysis process and operational uncertainties.

It should be noted that the PP8L approach does not lend itself to the identification of a specific MCPR value that can be translated into a fuel cladding integrity safety limit.

From an overall technical standpoint, there is no benefit to having a specific MCPR value for the fuel cladding integrity safety limit rather than using the SAFDL as the safety limit.

The reason for this situation is that conformance to the safety limit is required to be calculationally based, consistent with the event signature being evaluated.

The same is true for the PP8L approach, which uses a fuel cladding, integrity safety limit of greater than 99.9% of the fuel rods in the core not expected to experience boiling transition.

The only difference is in the extent of the calculations required to demonstrate compliance.

The PP&L approach used to establish the MCPR operating limit does provide adequate margin.

This is demonstrated by the value established for the MCPR operating limit. Application of the ANF methodology and the PP8 L methodology to the same reload resulted in essentially the same operating limit MCPR being established.

This situation results primarily because PP&L has made a more comprehensive treatment of uncertainties and included a larger magnitude of uncertainties than ANF.

As a result, a more sophisticated treatment of uncertainties was used to establish an appropriate operating limit MCPR.

Even though the PP8L approach includes a highly sophisticated treatment of uncertainties, it still retains a substantial amount of conservatism.

This conservatism is discussed in more detail below with respect to the events evaluated.

Also, the operating limit MCPR resulting from the application of PP&L's methodology is greater than has been justified for other plants of similar design.

The PP8L statistical methodology is applied to pressurization events (i.e., generator load rejection without bypass and feedwater controller failure) and to the control rod

~

~

~

~

~

withdrawal error.

Previous safety analysis approaches have used statistical methodology in the evaluation of pressurization events only.

In the PP&L methodology for the evaluation of pressurization

events, a substantial amount of conservatism exists.

The sources of conservatism include the following:

(1) The system model is developed on a best estimate basis; however, the remainder of the plant model and the selection of the plant operating state uses conservative inputs.

(2) The uncertainties in the system model are based on a statistical comparison to the results of the Peach Bottom turbine trip tests.

Because there are only three data points, the uncertainties at a high probability and confidence level consistent with the safety analysis process are very large.

It should be noted that in other applications, sensitivity studies have been performed to reduce the magnitude of these uncertainties.

As a result, the PP8L methodology incorporates a larger model uncertainty than other applications, even though the demonstrated prediction capability of the Peach Bottom test data and Susquehanna transient data is comparable to other applications.

(3) In the analysis process to establish the operating limit MCPR, the change in CPR for the hot channel is applied to all the fuel assemblies in the core.

This

approach, in effect, decreases the CPR for the rods in the core more than would be anticipated using a more rigorous evaluation of each of the fuel assemblies in the calculation of the change in CPR.

(4)

The average scram time for the control rods in the core used in the statistical analysis is conservative when compared to actual plant experience.

(5) The conservatism in the ANF MCPR safety limit methodology regarding the power distributions and CPR distribution of the fuel rods in the core is retained.

As with the evaluation of the pressurization events, the evaluation of the control rod withdrawal error contains a substantial amount of conservatism.

The sources of conservatism include the following:

(1) The selection of the control rod pattern used in the control rod withdrawal error analysis is very conservative.

The control rod pattern selected is not realistic for use in actual plant operation because it would cause an excessive power peaking that would result in other operating constraints.

This conservatism is sufficiently large such that, in other methodologies, it is assumed that it dominates all other unceWainties, and therefore, additional conservatism associated with the other uncertainties is not required.

The PPB L methodology explicitly includes the effects of these uncertainties.

(2) The uncertainties assumed in the analysis process are based on the operating state prediction uncertainties.

Because of the rod block monitor initialization

process, it would be expected that the uncertainties, which are based on a change in the operating state, are substantially less than assumed in the analysis.

In addition, the uncertainty in the predicted rod block monitor response is conservatively based on the minimum number of allowable operable local power range monitors.

(3) A conservatively high local power range monitor failure rate is used in the analysis, and a rod b'lock monitor channel is arbitrarily assumed to be bypassed in each Monte Carlo trial ~ These assumptions provide additional conservatism in the overall result.

(4) The hot channel and MCPR safety limit conservatisms (Numbers (3) and (5) for the pressurization

events, above) apply to the control rod withdrawal error.

Based on previous analyses performed by others, the application of the safety limit power distribution used for core wide events is even more conservative for localized events, such as the control rod withdrawal error.

The NRC has not allowed any explicit credit for this conservatism; however, it should be recognized in the overall assessment of the methodology.

As described above, the statistical analysis approach taken by PP&L to establish the MCPR operating limit satisfies the regulatory requirements and still retains a

substantial amount of conservatism.

Therefore, the PPSL approach is considered technically acceptable.

4.0 CONCLUSION

S In the assessment of PP8 L's application of reactor analysis methods, as they are used to establish the operating limit MCPR, a number of conclusions were reached.

These conclusions were based on the information contained in the following reports:

(1) PL-NF-90-001, "Application of Reactor Analysis Methods for BWR Design and Analysis;"

(2) PL-NF-89-005, "Qualification of Transient Analysis Methods for BWR Design 'and Analysis;" and (3) PL-NF-90-005, "Susquehanna SES Unit 2, Cycle 5 - Reload Summary Report." The conclusions are summarized below:

(1) The PP8L reactor analysis methods used to establish the operating limit MCPR are in conformance with the NRC's Regulations, conform to the Regulatory Guides and SRP, and contain a substantial amount of conservatism.

(2) The basic thermal margin methodology is consistent with current industry practice that has been reviewed and approved in other applications.

(3) The statistical analysis approach taken by PP&L to establish the MCPR operating limit includes a comprehensive treatment of uncertainties, employs technically acceptable statistical analysis techniques, and provides an adequate level of conservatism.

(4) Application of the PP&L reactor analysis methodology in the reload fuel design and analysis process produces an operating limit MCPR that is comparable to those produced by fuel supplier and other utility methodologies that have previously been approved by the NRC.

(5) Implementation of this methodology by PP8L is appropriate.