ML20128H966

Forwards Second Response to RAI Re Integrated Containment Reanalysis Methodology

ML20128H966
Person / Time
Site:	Maine Yankee
Issue date:	10/03/1996
From:	Frizzle C Maine Yankee
To:	NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM)
References
CDF-96-169, MN-96-144, TAC-M94835, NUDOCS 9610100136
Download: ML20128H966 (22)

Text

._ . _ _ . ._ .._. . . ._.

MaineYankee HF t l ABI L E L EC TRICIT Y SINCE 1972 1

l l

329 Bath Road l Charles D. Frizzle Brunswick. Maine 04011 President and Chief Executive Officer (207) 798-4100 I

October 3,1996 MN-96-144 CDF-96-169 l

UNITED STATES NUCLEAR REGULATORY COMMISSION Attention: Document Control Desk Washington, DC 20555

References:

(a) License No. DPR-36 (Docket No. 50-309)

(b) MY Letter to USNRC dated April 12,1996 (MN-96-048) - Integrated Containment Analysis - Methods Summary and Identific". ion of Differences from SRP (c) USNRC Letter to Maine Yankee dated May 31,1996 - Maine Yankee Containment l Reanalysis (TAC NO. M94835)

(d) MY Letter to USNRC dated July 23,1996 (MN-96-102) - Response to RAI Integrated Containment Analysis I

Subject:

Second Response to RAI - Integrated Containment Reanalysis Methodology Gentlemen:

We submitted a summary of the methodology we plan to use to reanalyze the containment pressure response following a LOCA in Reference (b). Your staffissued a request for additional information (RAI) on our proposed methodology in Reference (c). We responded to seven of the thirteen questions posed in your RAI in Reference (d). We also provided a schedule for responding to the remaining questions and completion of the containment reanalysis. The purpose of this letter is to respond to four additional questions posed in your RAI. This letter also provides a revised schedule for responding to the remaining questions and completion of the containment reanalysis.

Our responses to four additional questions are provided in the enclosure. Note that the enclosure also includes the seven responses provided in our July 23,1996 submittal, Reference (d), and the status of the remaining two responses. One of the seven responses provided in Reference (d) has been revised to reflect the new schedule.

Please note that in response to one of your questions and in keeping with our original plan, Reference (b),

we have been assessing steam generator heat transfer in Yankee Atomic Electric Company's (YAEC's) version of RELAP5/ MOD 3. This has been done by comparing RELAPS/ MOD 3 predictions to several tests performed in the FLECIIT-SEASET Steam Generator Separate Effects Test Facility. This assessment indicates that a modification to the methodology may be necessary to ensure a conservative bias in steam generator heat transfer calculations. We are still investigating the differences between code calculations and test data.

Our revised schedule for compleun, which is provided below, reflects the effort necessary to perform additional assessment of steam generator heat transfer as well as the diversion of resources that have been necessary to be responsive to the Integrated Safety Assessment Team (ISAT).

g(

t 9610100136 961003 ,

PDR ADOCK 05000309 ,

p PDR

Mam.eYankee UNITED STATES NUCLEAR REGULATORY COMMISSION MN-96-144 Attention: Document Control Desk Page 2 Schedule for Completing the Integrated Containment Reanalysis Task Description Current Actual / Revised 1 Submit Methods Summary and Identification of 04/12/96 NA Differences from SRP 2 NRC Review of Methods Summary and 05/31/96 NA Differences from SRP t 2a Submit Answers to Remaining RAI Questions 10/01/96 12/01/96 3 Submit Methods and Sample Calculation 10/01/96 01/30/97 4 Submit Analysis Results and Associated Proposed 11/01/96 05/01/97 Technical Specifications We trust that the information in this letter is satisfactory. Please contact me if you have any questions.

Very truly yours, Charles D. Frizzle President and Chief Executive Officer Enclosure c: Mr. IL J. Miller Mr. D. H. Dorman Mr. J. T. Yerokun Mr. Clough Toppan Mr. Patrick J. Doctie Mr. Uldis Vanags Donald Zillman, Esq.

Lawrence J. Chandler, Esq.,

Assistant General Counsel for llearings and Enforcement STATE OF MAINE Then personally appeared before me, C. I). Frizzle, who being duly sworn did state that he is President and Chief Executive Officer of Maine Yankee Atomic Power Company, that he is duly authorized to execute and file the foregoing response in the name and on behalf of Maine Yankee Atomic Power Company, and that the statements therein are true to the best of his knowledge and belief.

' Notary Pubhef l l lhanyMh,SheafIhhs Khamseestehame

8. - _ _ _ _ _ _ _ _ _ _ _

ENCLOSURE l MN-96-144 )

Page1of20 l

l l

l Maine Yankee Response to Request for AdditionalInformation Containment Reanalysis Methodology i

j l

l l

l l

l I

l ENCLOSURE MN-96-144 Page 2 of 20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology Maine Yankee submitted a report to NRC describing the methodology that would be used in the Maine Yankee containment reanalysis in Reference 1. NRC reviewed this report and asked thirteen questions in Reference 2. Maine Yankee submitted a letter responding to seven of these questions in Reference 3. This enclosure repeats six responses, revises one of our previous responses, and responds to four additional questions. The table below summarizes the status of Maine Yankee's responses to the thirteen questions.

Number Status Number Status 1 Response scheduled for 12/1/96 8 Response Provided in Reference 3 2 Response Provided in Reference 3 9 kesponse Provided in Reference 3 3 Response Provided Below 10 Response Provided in Reference 3 4 Response Provided Below 11 Reference 3 Response Revised Below 5 Response Provided Below 12 Response Provided in Reference 3 6 Response scheduled for 12/1/96 13 Response Provided in Reference 3 7 Response Provided Below Question 1 RELAP5 and GOTHIC have not been evaluated by the staff and approved for generic use for containment analysis. To assess the conservatism of your proposed methods we request that you benchmark RELAP5 and GOTHIC against methodology that has already been approved for facilities similar to Maine Yankee. For the most severe hot and cold break scenarios calculated for Maine Yankee, the benchmark effort should include graphical comparisons with the results of analyses using approved methods. Include plots of the mass and energy release as a function of time. Discuss the cause of any significant differences in the predicted results.

Response Approach to Question 1 Our planned approach to assess the conservatism or accuracy of various aspects of our model was to perfomi comparisons of analysis results to experimental data rather than to calculations by NRC approved methods. We will continue this assessment. To respond to NRC's request, we have performed a R5M3 calculation of mass & energy release for a LBLOCA at a Westinghouse four loop PWR with a steel-lined reinforced concrete containment, having pome:y similar to Maine i

Yankee's, and are in the process of comparing results to the mass & energy information provided in that plant's FSAR. We have also started to develop a GOTHIC model of that pimt's containment using the same techniques that will be used for Maine Yankee. The mass and ener gy data provided

in the FSAR and the mass and energy data predicted oy R5M3 will be input to the GOTHIC model to assess the conservatism in the YAEC methodology. The resulting comparisons will be plotted and significant differences discussed. Completion of this effort will require more time and is dependent on the completion of our response to Question 6 regarding steam generator heat transfer.

Completion of this response is currently scheduled for December 1,1996.

l

ENCLOSURE MN-96-144 Page 3 of 20 Maine Yankee Response to Request for AdditionalInformation Containment Reanalysis Methodology Question.2 Paragraphs 2.2.2 and 3.5 ofyour April 12,1996, submittal state that measurement uncertainties will not be included in your calculations, but uncertainties will be addressed by assessing their effects against existing margins. Provide your methodology for combining the effects of uncertainties.

Uncertainties must be accounted for in a defensible manner (e.g., statistical method with 95%

probability criterion).

Response to. Question.2 YAEC will use a process similar to that discussed in NUREG/CR-5249 to demonstrate that the calculated peak containment pressure will bound the 95% probability level. The phenomena important to predicting containment pressure, temperature and long term response will be identified and ranked by a panel of experts to detemiine those parameters that should be treated statistically, those that should be bounded and those that can be taken at their nominal value. A fractional factorial design will be used to assess the impact of significant variable parameters on predicted mass and energy releases. Results from each of the R5M3 cases will then be used to define a systematic set ofinputs for a series of GOTHIC cases, which will also include combinations of uncertainties that affect the containment response. The resulting predicted maximum containment pressure will be used to develop a response surface model for the sensitivity of predicted containment pressure to the important inputs. Uncertainties will be propagated through the response surfaces by a Mcnte Carlo method to demonstrate, with 95% confidence, that the maximum calculated containment pressure that bounds the 95% probability value, is less than the design pressure.

This response was provided in MN-96-102, Reference (3).

Question 3 1

Paragraph 4.1.3.1 of your submittal states that the R5M3 code critical flow mcdel (Ransom and Trapp) will be used for break flow computation. This "best estimate" model typically yields lower mass fluxes than the conservative Henry-Fauske (subcooled) and Moody (saturated) models. ,

NUREG/CR-5535,"RELAP5/ MOD 3 Code Manual- Summaries and Reviews ofIndependent Code Assessment Reports," Vol. 7, paragraph 2.3.1.10, discusses break flow under predictions using the Ransom and Trapp model and provides a basis for questioning the Marviken results that you discuss in your submittal. In addition, the Ransom-Trapp model recently has been found to predict flows that are only 10% of the homogeneous flow model. (See minutes of the Penn State Camp meeting dated April 1,1996). Discuss your methods of ensuring that conservative mass flow rates are calculated for containment analyses.

Response to Question 3 The comments contained in NUREG/CR-5535, Vol. 7, paragraph 2.3.1.10 only applied to the first released version of RELAP5/ MOD 3 (i.e., Mod 3.0, also known as 5m5). The same comments are not expected to apply to YAEC's version, especially in light of the interphase drag modifications which affects the break void fraction and slip. To confirm our expectation, one of the assessment cases considered in Oc above referenced source (Marviken Test 10) was assessed using YAEC's version. The results show that, using discharge coefficients of 1.0, the predicted break flow rate agrees very well with experimental data.

'f ENCLOSURE MN-96-144 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology With regard to the latest comment on a coding error (6th CAMP Meeting Minutes, Penn State, April 1,1996), the error only applies when the abrupt area change option is activated at the breakjunction.

Yankee's modeling guidelines specify using the smooth area option with user specified loss coefficients. These guidelines are consistently used in code assessments and plant calculations.

4 IIence the reported coding error does not have any impact on oui assessment results or plant calculations.

$ Jur approach of ensuring that conservative flow rates are calculated for containment analysis is based on separate effect assessment against the Marviken experimental data, as discussed in the Methods Summary Report, Reference (1). As part of our planned code assessment, we have recently obtained results for eleven other Marviken Tests (in addition to Test 10 aiscussed in the submittal), covering the entire range of UD ratios in the Marviken experiment (from 0.3 to 3.7).

Mean discharge coefficients have been determined separately for the subcooled and two-phase portions of the tests. This information is shown below.

Cc.nparison of PMM3Y92 Predictions to Marviken Test Data Test UD h D (m) AT, (K) Cd,ma Cd,,,.,3,,,

10 3.1 .509 <5 -

0.929

~

12 3.0 .300 30 U.947 -

~

13 3.0 .200 30 1.067 -

! 15 3.6 .500 30 0.996 0.987 16 3.6 .500 30 1.022 0.923 18 3.7 .300 30 0.933 0.969 19 3.7 .300 <5 1.019 1.001 20 1.5 .500 <5 -

1.218 21 1.5 .500 30 1.114 0.986 1 22 1.5 .500 50 0.988 1.0'43 23 0.3 .500 <5 -

1.066 24 , 0.3 .500 30 1.326 1.131 The discharge coefficiems in the above table were obtained from the comparison of predicted flow

-ates to measured flow rates as a function of time. The values over each time period (i.e., subcooled and two-phase) were averaged to arrive at the numbers in the table. A statistical analysis of these data inaicates that statistically significant relationships exist between flow discharge coefficients and (UD). While further analysis demonstrates that the linear relationship with (UD) is defined by lines with a common slope, this will not be credited. Instead, the data for the subcuoled and two-phase regimes will be used separately to establish the uncertainty in discharge coefficients for each regime.

Figures 3-la and 3-lb indicate the dependence on (L/D) of the predicted mean dischrge coefficients. The Maine Yankee analysis will determine the limiting location for the break in a particular run of pipe based an the discharge coefficients determined for each side of the break.

E-

ENCLOSURE MN-96-144 Page 5 of 20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology l A sensitivity study was performed for a double-ended hot leg break at Maine Yankee using the l diacharge coefficients shown in Figure 3.2. Since the peak containment pressure for a hot leg break is driven primarily by the break flow during blowdown, a sensitivity study for breaks at other locations is not necessary. As expected, using a high discharge coefficient results in higher break flows initially, but causes a more rapid uecrease in syNm pressure. As a result, the break flow .

during the latter portion of the blowdown is less than it would have been if a lower discharge I coefficient were assumed. The resulting mass and energy data were used as inputs to our GOTHIC l model of the Maine Yankee containment to determine the peak containment pressures. As shown in Figure 3-2, the peak containment pressure increases linearly with increasing discharge coefficient. ,

Note that we used the same discharge coefficient for both sides of the break in de sensitivity study I and therefore over-estimated the im,.act of discharge coefficient on peak pressure because of the discharge coefficient's dependence on L/D (i.e., each side of the break should use a different I dieharge coefficient). We conclude from this sensitivity study that the variation of discharge coefficiem with L/D and associated uncertainty needs to be considered. The Methods Report will describe how these effects will be included in the plant analysis.

This completes our response to this question.

Question 4 l

Paragraph 3.4.4 of NUREG/CR-5535, Vol. 5, written in 1991, states: l Because RELAP5 was developed primarily as a small break LOCA analysis tool, the reflood model has received only limited developmental assessment evaluation and independent application experience. The little experience to date indicates that code time step control feat - .aay not be adequate to handle reflood problems. Also, the reflood model should not be invaed when wall condensation eff: cts are important or when non-condensibles are present.

These statements create concern regarding use of RELAP5 for reilood analysis. Paragraph 4.1.1 of your submittal indicates that the code contains a modified reflood model. It is not clear if the current reflood model is the same one as that to which the above words apply. Provide a basis for confidence in the validity of your R5M3 model for the reflood phase. Include detailed comparison with the FLECHT or FLECHT-SEASET data for the entire range applicable to Maine Yankee.

Provide comparisons of the test conditions to those expected in the Maine Yankee reactor core.

Response to Que:; tion 4 Activation of the reflood model in RELAP5/ MOD 3 invokes a '. o-dimensional wall heat conduction solution insteaa of the standard one-dimensional radial conducten solution. The heat transfer models on the fluid side are the same whether or not the reflood option is activated. We have successfully applied this reflood option throughout our code development effort. The statement in our Methods Summary Report, Reference (1) submitta' " ding a modified reflood model refers to the differences between RELAP5/ MOD 2 and ' u. 5/ MOD 3. We have not made any modifications to the 2-D heat conduction scheme to which the reflood option really refers.

. l I

1 ENCLOSURE ,

MN 96-144 Page 6 of 20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology The statement quoted from the RELAP5 User Manual pertains to RELAP5/ MOD 2 and should have been removed from the manual for RELAP5/ MOD 3. This issue arose for MOD 2 because it used a separate renood heat transfer package that did not consider wall condensation and non-condensible c h'e cts. Ir ~ 'D3, the heat transfer models has e becu collected into one heat transfer package that is invoked or renood as well as non-renood situations. This package accounts for wall condensation tnd non-condensible effects. INEL plans to remove this statement in a future update I to the RELAPS/ MOD 3 manual.

The predictive capability of R5M3Y92 for the reDood process has been assessed against four of the FLECHT-SEASET forced reDood tests (. Reference 4), as discussed in the Methods Summary Repon, Reference (1) submittal. As shown below, the Dooding rates range from 0.81 to 3.01 in/sec.

FLECIIT-SEASET Tests Assessed FLECHT-SEASET Test Flooding Rate (in/sec, 1

3:302 3.01 1 1

312d3 1.50 31504 0.97 31805 0.81 Figure 4-1 compares the predicted peak clad temperatures (PCTs) to the measured PCTs for the four tests. Note that PCTs at each measured location are included in this comparison. The assessment ~

shows that, on average, R5M3Y92 slightly over-predicts the PCT. Although this tendency is conservative for assessing the ECCS performance (e.g., predicting PCTs), it is slightly non-conservative for predicting the transfer of stored energy into the coolaat for assessing containment response to a large break LOCA. However, the impact of this bias in the model on the predicted peak containment pressure is considered negligible in comparison to other mechanisms affecting the transfer of energy to the coolant. It is expected that the code modification discussed in our response to Question 5 (modification of the CHF correlation) will compensate for any non-conservatism in the prediction of PCT.

It is also important for mass ar.a energy calculations to predict quenching of the core at the right time as this ensures that the sensible energy stored in the fuel rod is deposited in the coolant at the appropriate time. Figure 4-2 compares the predicted quench times to measured quench times for the four tests. As shown, some of the quench times are predicted earlier than measured, while others are predicted later than measured. A detailed review of the data indicates that the code predicts quenching to occur early when the reHood rate is greater than approximately 1.5 irdsec. These points are indicated by squares on Figure 4-2. From this assessment, we can say that R5M3Y92 conservatively predicts removing energy from the core when the reHood rate is greater than 1.5 in/sec.

l I

ENCLOSURE MN-96-144 Page 7 of 20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology To ensure that R5M3Y92 will predict core quenching conservatively in plant applications, the core inlet flow for a typical double-ended cold leg pump suction break (with the failure ofone emergency )

diesel generator) at Maine Yankee was examined. Since the code predicts a fluctuating inlet flow, the inlet flow was integrated such that mean flooding rates could be determined. This showed that there was an initial insurge of coolant with an average flooding rate of about 16 in/sec. After the i initial influx of water into the core, the average flooding rate decreased to between 2.0 and 2.5 in/sec. '

The core was completely quenched by 110 seconds. Since the flooding rates are reasonably high during the time the core is being quenched, it is concluded that the code will predict the transfer of .

heat from the core to the coolant conservatively.

As shown in Figures 4-6 thrcugh 4-9 of the Methods Summary Report, Reference (1), our assessments of the FLECHT-SEASET tests show that R5M3Y92 predicts a higher carryover fraction (ratio of mass flow rate out to mass flow rate in) than was measured experimentally. Thus, the code is conservative from this point of view.

This completes our response to this question.

Question 5 RELAP5 critical heat flux correlations for blowdown heat traufer are primarily intended for fuel / clad response analysis. For containment analysis (where increased core cooling is conservative) these correlations may be nonconservative. Provide a discussion of measures you will take to preclude non-conservative core heat transfer modeling during blowdown. Evaluate the sensitivity of containment pressure to delayed DNBR during the blowdown.

Response to Question 5 The code version being proposed for containment pressure analysis (R5M3Y92) has been assessed against several steady state CHF tests (Columbia tests, GE 9-Rod tests and ORNL THTF tests) in bundle geometry. These assessment results were not included in Reference 1. As pointed out in the NRC question, the results show that the code under-predicted the CHF value for the above mentioned tests (i.e., the code predicted DNB to occur earlier than experimentally measured). As a result, the code will under-predict the rate of heat transfer from the structures to the coolant, which is an acceptable bias when predicting peak cladding temperatures.

We estimated the effect of the uncertainty associated with the CHF prediction using the steady state CHF test data discussed above and a modified version of R5M3Y92. This version of the code permits the user to multiply the CHF correlation by an input factor. An input factor of 1.384 was determined which bounded the test data such that the code predicted DNB to occur later than all test data. The modified version of the code, which predicts DNB to occur when the local heat flux is greater than 1.384 times the predicted critical heat flux value, generally predicts DNB to occur at higher elevations than observed, and in five of the twelve cases, did not predict DNB to occur.

Figure 5-1 shows how the modified version of the code predicts the CHF data relative to the unmodified version of the code.

1 4

ENCLOSURE MN-96-144 Page 8 of 20 2 Maine Yankee l Response to Request for Additional Information l Containment Reanalysis Methodology i The modified version of R5M3Y92 was then used to determine the mass and energy release for a

double-ended break in the hot leg using our Maine Yankee model. These results where then input to the GOTHIC model of the Maine Yankee containment. As expected, delaying the onset of DNB resulted in slightly more energy being transferred to the coolant during blowdown and delayed the l'

end of blowdown slightly. When these results were input to GOTHIC, the predicted containment peak pressure increased by only 0.05 psi.

Although the prediction of CHF has a relatively small effect on peak contamment pressure, we plan

to use the modified version of the code to gredict mass and energy releases to ensure that the code j has a conservative bias in this area.

This completes our response to this question.

Question 6 Typical modeling practice for both PCT and containment mass and energy analysis is to assume very j high heat transfer coefficients for reverse heat transfer from secondary coolant to primary coolant.

Paragraph 6.4 of your submittal indicates that "first principles" will be used to determine the fluid conditions leaving the S/G. Please explain in greater detail. Justify that your proposal is conservative for containment analysis.

Response Approach to Question 6 As indicated in Section 4.1.4 of the Methods Summary Report (YAEC-1932), we had planned to perform additional assessment in the area of steam generator heat transfer to ensure that R5M3Y92 has a conservative bias in predicting the rate of heat transfer between the primary and secondary coolant. YAEC has compared R5M3Y92 predictions to the data from the FLECHT-SEASET Steam Generator Separate Effects Tests for this assessment. This assessment against data from one test is indicating tnat the code may under-predict the rate of heat transfer between the primary and secondary coolant in the test facility. As it is our intent to ensure a conservative bias in our prediction of the heat transfer between the primary and secondary coolant, it v*ill be necessary to perform additional assessment of our methodology in this area.

Completion of this response is currently scheduled for December 1,1996.

Question _7 During the reflood period, stratified flow of steam and subcooled water will occur within the reactor coolant system. These conditions may lead to steam condensation. Also, steam quenching will occur at the ECCS injection points. These effects act to reduce the containment atmosphere steam heat load. Provide comparisons to experimental data to support your predictions for steam condensation. These comparisons should encompass the entire range ofinjection and steam flow rates involved.

1 ENCLOSURE l MN-96-1.4 Page 9 0f 20 ,

l Maine Yankee Response to Request for Additional Infomiation Containment Reanalysis Methodology j l

Response to Question _7 Section 4.1.3.4 of the Methods Summary Report, Reference (1) discussed the assessmer . >f the ECCMIX component of the R5M3Y92 that was accomplished by comparing the code's prediction  !

of the Westinghouse 1/3-Scale Steam Water Mixing Tests (Reference 5). Figures 4-10 and 4-11 of

)

the Methods Summary Report, Reference (1) show the comparison of data and prediction for two  !

of the eight tests. Figure 7-1 compares the predicted and measured liquid temperatmes leaving the test section for all eight test. As shown, the code predicts a lower liquid temperature at the exit of

)

the test section than was measured for each test. The conclusion of this assessment is that the code

. under-predicts the interfacial condensation rate. This is conservative for mass and energy i

calculations because the code will condense less steam in the presence of subcooled water than is expected to actually occur (e.g., at the ECC injection point, ia the downcomer annulus, etc.), which results in a higher containment atmosphere stnm heat load.

I Figure 7-2 compares the test conditions and predicted plant conditions at the injection location. The l comparison is made on a per unit area basis. The figure shows steam flow per unit area upstream j of the injection location versus ECC injection flow per unit area at the injection location. The plant information was obtaind from a RELAP5Y92 calculation for a double-ended pump discharge break at the location of the safety injection nozzle. Representative conditions in the unbroken loop at the location of the safety injection nozzle (i.e., safety injection flow rate and upstreun steam flow rate) l were sampled during and afler accumulator injection. As shown, the predicted ECC flows are in line l with the test conditions. Although some of the predicted steam flow rates into the injection location l are larger than the test conditions contained in our assessment, it is ourjudgement that, since this code calculates the mixing of steam and water mechanistically, the assessment conclusions are valid at the higher steam flow rates. That is, the code under-predicts the interfacial condensation rate.

This completes our response to this question.

Question 8 Section 4.1.2 of your submittal discusses Yankee Atomic modifications to R5M3. Are these modifications unique to the Yankee version of RELAP5 Mod 3.l? To what extent do these modificatiens reflect modeling changes that have been implemented or approved in other codes and

code applications.

l ENCLOSURE MN-96-144 I Page 10 of 20 Maine Yankee Response to Request for AdditionalInformation Containment Reanalysis Methodology Response to Question 8 The YAEC version of RELAP5/ MOD 3 originates from Version 80 of the RELAP5/ MOD 3 code.

Version 80 was a beta testing version prior to the official release of Mod 3.1. In addition, YAEC <

reviewed the INEL modifications made to the code prior to release ofMod3.1. This is to clarify the introductory statement concerning R5M3Y92 in Section 4.1.2 of our April 12th submittal. All the modifications described in Section 4.1.2 of the April 12th submittal are unique to R5M3Y92. Some of the modifications are motivated by our previous code development effort with the RELAP5 series of codes, such as the vertical interphase drag modification described in Section 4.1.2.1, and the characteristic length modification described in Section 4.1.2.3. Finally, the April 12th submittel unintentionally omitted a discussion on the wall condensation model, which in R5M3Y92 is based  !

on the original Version 80 model instead of the new model implemented in Version 3.1. Recent code assessment pcrformed at INEL (Presentation by Gary Johnson at the CAMP Meeting, Espoo, Finland,1995) has shown that the Version 3.1 model (using Nusselt/UCB for laminar / turbulent condensation) is inferior to the Version 80 model (using Nusselt/ Shah for laminar / turbulent steam only condensation).

In general, YAEC has continuously tracked modifications made by INEL to the RELAP5 code through the current released version, MOD 3 Version 3.2. The modifications are evaluated continuously for incorporation into the YAEC version. '

l This response was provided in MN-96-102, Reference (3).

Question 9 The Standard Review Plan (SRP) recommends that decay heat be calculated using Branch Technical Position (BTP) ASB 9-2. The BTP is based on the 1971 version of ANS Standard 5.1 with a 20 percent margin for uncertainty. You indicate your intent to use the 1979 version of ANS 5.1 with no margin, but with a 20 uncertainty considered in the overall uncertainty assessment. Use of the 1979 standard is acceptable. Uncertainties must be accounted for in a defensible manner. (See item 2 above). The 1979 ANS 5.1 standard permits certain user-supplied options. These are the actinide production multiplier (R-factor), the fission product activation factor (G-factor), the burnup factor (Si), the power history, and the fraction of fission products from each of three fissile elements.

Discuss how these options will be considered.

ENCLOSURE MN-96-144 Page11 of20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology Response to Question 9 With regard to the treatment of uncertainties, please see the response to Questicn 2.

Inputs to the ANS 1979 decay heat stan: lard will be chosen such that conservative fission product and actinide decay powers are calculated. The power history will correspond to continuous full power operation for the maximum expected time for which a fuel assembly will exist in the core.

The fission fractions for Um, Pum, and yn will be selected such that theM contribution is bounded. The proposed method is to base the fractions on core average conditions at beginning of cycle and to assume a bounding set ofconstant fission fractions throughout the assumed full power operation time. Similarly, the actinide production factor will be conservatively maximized by using end of cycle conditions to derive the value. If preliminary results indicate that the long term environmental qualification limits are challenged, the variation with respect to burnup will be accounted for in a conservative manner.

The neutron capture correction factor will be calculated using equation 11 of the standard for 4

shutdown times less than 10 seconds. The number of fissions per initial fissile atom (9') is an input to this equation. A conservative value of unity will be used. For shutdown times greater than 104 seconds the G_ values specified in Table 10 of the standard will be used. In addition, we have reviewed NRC Information Notice 96-39," Estimates of Decay Heat Using ANS 5.1 Decay Standard May Vary Significantly", dated July 5,1996. We have concluded that our application of the decay heat standard is not affected by the issues raised in the Information Notice.

This response was provided in MN-96-102, Reference (3).

ENCLOSURE MN-96-144 Page 12 of 20 Maine Yankee Response to Request for AdditionalInformation Containment Reanalysis Methodology l

Question 10 1 1

Provide a basis for the assumption that the limiting break with single failure can be identified by examining single failure effects for only two break locations. If double-ended guillotine breaks are found not limiting, what analysis will be perfonned to confirm that the limiting break is identified?

l Response to Question 10 Subsequent to submitting the Methods Summary Report, we have decided to investigate three break locations instead of two. Specifically, we will investigate breaks in the hot leg, pump suction and pump discharge piping. Our approach will be to determine limiting conditions for each of the three break locations with regard to offsite power availability and break size. Once that is determined, a single failure study will be performed for each break location assuming the limiting conditions.

For each break location, we will assume a 200% double-ended break and investigate the effect of loss of offsite power to determine whether the effect of running the reactor coolant pumps is more limiting than the delays in starting safety injection and containment spray that are associated with starting the emergency diesel generators. Our preliminary work suggests that the loss of offsite power cases will be limiting. Although we expect the 200% double-ended break to be the limiting break size with regard to containment peak pressure, we will rerun the limiting cases from the ,

previous study assuming a 100% slot break to confirm that the largest break is limiting. '

The limiting single failure will then be determined for each of the cold leg break locations assuming the limiting assumptions with regard to break size and offsite power availability. Table 10-1 shows the single failures that will be considered for each of these three events in addition to the cases with no single failure. Please note that it may not be necessary to perform code calculations to analyze all single failures listed in Table 10-1 because some of them are obviously less limiting than others.

The limiting single failure will also be determined for the hot leg break location if consideration of single failures will affect the calculated peak containment pressure. Our prelim nary work indicates that, for hot leg breaks, the peak pressure occurs before any of the mitigating equipment is activred.

This response was provided in MN-96-102, Reference (3).

Question.11 Should the staff decide to perform independent confirmatory analyses using its CONTAIN code, certain information will be needed:

l (a) The input listing for the RELAP5 and GOTHIC models used to calculate mass and energy release and containment pressure for Maine Yankee. This input should be provided in text l

and in electronic form. Indicate key assuiaptions such as single failures.

ENCLOSURE MN-96-144 Page 13 of 20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology (b) Noding diagrams for the RELAP and GOTHIC models.

(c) For the most severe hot and cold breaks calculated for Maine Yankee - tables showing the total steam-water mass and energy in the core, reactor system, steam generators secondary and containment initially and during each phase of the accident including the blowdown, reflood and post reflood periods.

Please indicate when this information can be provided.

Response to Question 11 This information can be provided afler the submittal of the plant specific analysis that is currently scheduled for March 1,1997.

This revises our response to this question.

Question 12 Paragraph 4.2 of your April 12,1996 submittal provides a brief description of your intent to use GOTHIC for the entire post-RAS phase of the analysis by transferring the residual heat remaining in the RELAP analysis at the end of the injection phase and transferring it to GOTHIC heat i structures. Explain how this will be done so as to account for froth entrainment rising into and l boiling within the S/Gs. Justify that any steam generator heat not removed from the steam generators during reflood period is added to the containment in a conservative manner in the post l reflood period.  !

Response to Question 12 As indicated in our April 12th submittal, the post-RAS mass and energy release will be calculated within GOTHIC. A GOTHIC control volume representing the reactor vessel up to the loop nozzle elevation will be used for this purpose.

For the cold leg break, the volume will contain the break, the High Pressure Safety Injection (HPSI) flow path, and a heater component (note that the LPSI system is not active post-RAS). The heater component will use a specified heat transfer rate which contains decay heat, RCS metal heat, and the secondary fluid and metal heat. The heat addition will be calculated using the stored heat present at the time of RAS from the R5M3Y92 calculation. The final temperature of the metal and secondary fluid will be selected such that it is belcw the saturation temperature corresponding to the predicted containment pressure at 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. The heat transfer rate will be based upon a conservative j extrapolation of the rate observed at RAS in the R5M3Y92 calculation. This bounds the heat transfer from the steam generators because any "fioth" effects are expected to be less in the post-RAS period as decay heat is lower. The injection flow rate will be calculated as the heat addition rate divided by the latent heat (i.e., enough injection flow will be provided to remove all heat by boiling). The remaining HPSI flow will be assumed to spill directly to the pool region.

For the hot leg break, all HPSI will be injected into the RCS volume and GOTHIC will calculate the fluid conditions at the break based on the inventory at RAS, the heat addition, and the HPSI flow.

. . . - . .- ~ . . - - _ - - - _ - . . . -. _- -

ENCLOSURE MN-96-144 Page 14 of 20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology The heat addition for hot leg breaks will include decay heat and any sensible primary side metal heat that has not already been removed. Removal of heat from the steam generator secondary fluid and metal will not ' oe included as the flow rate through the steam generators is expected to be negligible compared to cold leg breaks.

The temperature of HPSI injection will be calculated by GOTHIC based on the calculated pool region temperature and the performance of the RHR and Component Cooling Water heat exchangers.

Conservative values for heat exchanger fouling, the service water flow rate, and inlet temperature will be selected.

This response was provided in MN-96-102, Reference (3).

Question 13 Describe and justify any modifications made to the GOTHIC code described in the GOTHIC manual (EPRI Report RP3048-1).

Response to Question 13 Yankee has made no modifications to the GOTHIC code.

This response was provided in MN-96-102, Reference (3).

1 l

' ENCLOSURE MN-96-144 Page 15 of 20 Maine Yankee Response to Request for AdditionalInformation Containment Reanalysis Methodology REFERENCES ,

1. MY Letter to USNRC dated April 12,1996 (MN-96 048) - Integrated Containment Analysis - Methods Summary and Identification of Differences from SRP.

.2. USNRC Letter to Maine Yankee dated May 31,1996 - Maine Yankee Containment Reanalysis (TAC NO. M94835).

3. MY Letter to USNRC dated July 23,1996 (MN-96-102) - Response to RAI Integrated Containment Analysis.

4. "PWR FLECHT-SEASET Unblocked Bundle, Forced and Gravity Reflood Task Data Report," Volume 1, Report EPRI NP-1459, September 1981.

5. " Mixing of Emergency Core Cooling Water with Steam: %-Scale Test and Summary,"

Report EPRI 294-2. June 1975.

1 j

i

~

ENCLOSURE -

MN-96-144 ,

Page 16 of 20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology TABLE 10.1 - MATRIX OF SINGL E FAILURES CONSIDERED WITII AND WITHOUT LOSE OF OFFSITE POWER SINGLE FAILURES CONSIDERED oft-Site AC Diesel IIPSI Pump LPSI Pump Spray Pump Spray IIcader Spray -IIPSI Consequences Generator Valve Valve lost fail I no failure no failure no failure no failure no failure minimum safety injection & minimum spray no fa.!are fail 1 degraded safety injection & maximum spray no failure fail 1 degraded safety injection & maximum spray no failure fail I degraded safety injection after RAS & minimum spray no failure fail i "

maximum safety injection & minimum spray no failure fail I degraded safety injection & increased spray after RAS not lost N/A fail I no failure no failure no failure no failure degraded safety injection & maximum spray no failure fail 1 degraded safety injection & maximum spray no failure fai! ! maximum safety injection & minimum spray no failure fail 1 maximum safety injection & minimum spray no failure fail I degraded safety injection & increased spray atter RAS

S ENCLOSURE MN-96-144 Page 17 of 20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology Figure 3-1a Subcooled Discharge Coefficient vs UD 1.5 e Subcooled Data I'4 -

Linear Regression E

• o 1.3

.E g 1.2 8 N Y 1.1 1N 2 N +

10 o k

• 0.9 arviken Test Data Comparisonp 4 -

08 0.0 0.5 1.0 1.5 2.0 2.5 3.0 35 4.0 Break UD Figure 3-1b Two-Phase Discharge Coefficient vs UD 1.5 l l

, Tw o-phase Data Linear Regression 1.3 i

• o 1.2 E

8 e u 3 ,3  %

5 1.0 N i 2

C

+

l Ny- +

0.9 . MaNken Test Data Comparison i 08 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Break UD

l. .

ENCLOSURE MN-96-144 le Page 18 of 20 l Maine Yankee l Response to Request for Additional Information l Containment Reanalysis Methodology Figure 3-2 Peak Containment Pressure vs Discharge Coefficient 50.75 l

50 65 l g e

_$ 50.55 e

5

[ 50.45 2 Maine Yankee Peak Containment 50 35 i Pressure for Double-Ended Hot Leg _

Break 50.25 l l I I 0.80 0.85 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30 Discharge Coefficient (Cd)

Figure 4-1 Measured Versus Predicted PCTs 2,500 a 3.01 in/sec 2,250 -

o 1.50 in/see V 2,000 _ a 0.97 in/sec ga x 0.81 in/sec /

$ 1,750 of 1,500 'X -

3 O/

ti /

j 1,250 6-

/*

a 1,000 o_a /

~

750 KE6At4EASET Reflood Tests O

500 -

500 750 1,000 1,250 1,500 1,750 2,000 2,250 2,500 Measured PCT (F)

{

. . . - ._. . - _ - _ - _ = _ - _ - _ . - . . - _ _ .-- .

ENCLOSURE MN-96-144 o Page 19 0f 20 Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology Figure 4-2 Measured Versus Predicted Quench Times 800 ,

_a 3.01 in/sec 4 700.- o 1.5 in/sec 3 0.97 in/sec f600

~

x 0.81 in/sec A

l 500 F /

E 400 U

300 ,

U O 5 200 FFLECHT-SEASET Reflood Tests p 5 / *

\

100 4 V' e O b*5- .

0 100 200 300 400 500 600 700 800 Measured Quench Time (sec)

Figure 5-1 Comparison of CHF Data Predictions 12 ,

E C 10 -/

g

',x

• d f8 E

6 x [- O u

] Ci

.2 4 O 1

/ O Base R5M3 Prediction f x Modified R5M3 Prediction 0

0 2 4 6 8 10 12 Measured CHF Location (ft)

ENCLOSURE MN-96-144

, Page 20 0f 20 l

Maine Yankee Response to Request for Additional Information Containment Reanalysis Methodology Figure 7-1 Comparison of Liquid Effluent Temperatures i

I l 300 .,

p l - 280 260 -

240 -

4 h220 l

k= 200 -

e 1 180 E 160 8 140 Westinghouse 1/3-Scale Tests _

120 100 100 120 140 160 180 200 220 240 260 280 300 Measured Liquid Effluent Temperature (F)

Figure 7-2 Comparison of Test and Predicted Plant Conditions I

35 l l eW1/3 Scale Mixing Data 30 x ~

x Maine Yankee Conditions

~ 25 . x N

[ 20 "X W y x M x S 15 .

X xX j M Yx x xx x e X M__j

$ 10 ._sLa a M

• N x xx 5 _sk 4**

• 0 0 500 1,000 1,50; 2,000 2,500 3,000 3,500 4,000 4,500 l ECC flow (Ib/sec/ft2) l l