ML20198B755

Draft SE of Topical Rept NEDO-32686,Rev 0, Util Resolution Guidance for ECCS Suction Strainer Blockage

ML20198B755
Person / Time
Issue date:	12/30/1997
From:	NRC (Affiliation Not Assigned)
To:
Shared Package
ML20198B746	List: ML20198B740 ML20198B755
References
RTR-NEDO-32686 IEB-96-003, IEB-96-3, NUDOCS 9801070066
Download: ML20198B755 (133)

Text

. - . . . . - - . ..

.= ,

.c

. 4 DRAFT inom i

SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION RELATED TO NRC BULLETIN 96-03 BOILING WATER REACTOR OWNERS GROUP i TOPICAL REPORT NEDO-32606

" UTILITY RESOLUTION GUIDANCE FOR ECCS SUCTION STRAINER BLOCKAGE" DOCKET NO. PROJ0691 DRAFT

.t p' We# q ~

u

. i DRAFT inom EXECUTIVE

SUMMARY

. . . . . . ..... ..... ..........................ES-1 EG.1 Selection of Breaks for Analysis . . ............................ES-1 ES.2 Insulation Debris Generation by a Postulated Break . . . . . . . . . . . . . . . . . ES 2 ES.3 Other Drywell Debris Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ES-4  :

ES.4 Drywell Debris Transport . . . . ..................................ES5 .

ES.5 Suppression Pool Debris . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ES-7 '

ES.6 Suppression Peol Transport . . ..............................,.ES7 ES.7 - Head Loss across the Strainer . . . . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . ES-8 ES.8 Estimation of Available NPSH for ECCS Pumps . . . . . . . . . . . . . . . . . . . E S-9 ES 9 Other l ssues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ES-10

1.0 INTRODUCTION

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .................... 1 1.1 BAC KG R O U N D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 3 2.0 DI S C U S S I ON . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 3.0 URG GUIDANCE FOR DEMONSTRATING COMPLIANCE WITH 10CFR50.46. . .8 3.1 EVALUATION OF RESOLUTION OPTIONS . . . . . . . . . ..,........... 8 Staff Evaluation for Section 3.1. . . . . . . . . . . ... .. .. 11 3.2 METHODOLOGY FOR SIZING PASSIVE ECCS SUCTION STRAINERS . . 16 3.2.1 Drywell Insulation Debris Sources . . . . . . . . . . . . . . . . . . . . . . . . , , 16

• 3.2.1.1 Pipe Break Locations . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . 16 Staff Evaluation for Section 3.2.1.1. . . . . ... . ......... 17 Conclusions on Section 3.2.1.1 ... ...........,,...... 20 3.2.1.2 Zone of Influence . . . . . . . . . . . . . ....... .. ... .... 21 Staff Evaluation of Section 3.2.1.2 . . ......... ....... 22 Conclusions for Section 3. 2.1. 2 . . . . . . . . . . . . . . . . . . . . . 2 6 3.2.2 Other Drywell Debris Sources . . . . . . . . . . , , . . . . . . . . . . 27

, Staff Evaluation of Section 3.2.2 . . . ...................28 Conclusions on Section 3.2.2 . . . . . . . . . . . . . . . . . . . . . . 31 3.2.3 Drywell Debris Transport . . . . . . . . . . ... ......,,.... ...... 32 Staff Evaluation of Section 3.2.3. ..... ... ...... . 34 Conclusions on Section 3.2.3 . . , , ......... . ..... 36 3.2.4 Suppression Pool Debris ............. . ......... ... .. 39 Staff Evaluation of Section 3.2.4 . . . ..... ... ....,.... 39 3.2.5 Suppression Pool Transport and Settling . . . . . . . . . . . . . . . . . . . , , . 39 Staff Evaluation of Section 3.2 5 . . . . . . . . . . . . . . . . . . . . . . 39 3.2.6 Verification of Adequate ECCS Pump NPSH . . . . . . . . . . . , , . 39 Staff Evaluation of Section 3.2.6 . . . . , . . . . . . . . . . . . . . 40 Conclusions on Section 3.2.6 ... .... ..... ... .... 43

- 3.3 BAC KFLU S H . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3 Staff Evaluation of Section 3.3 . . . . . . . . . . .. ....... 43 3.4 SELF CLEANING STRAINERS . . . . . . . . . . . . ........ .... . . . . . 44 Staif Evaluation of Section 3.4 ......... . . . . . . . . 44 DRAFT

DRAFT 2 w s7 4.0 ADDITIONAL FEATURES WHICH PROVIDE: DEFENSE IN DEPTH . . . . . . . . . . . 44 Staff Evaluation of Section 4.0 . . . . . . . . . . . . . . . . . . . . . 44 5.0 C ON S E RVATI S M . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 6.0 MISCELLANEOUS REVIEW COMMENTS ON BWROG GUIDANCE . . . . . . . . . . . . 45 7.0 OVERALL CONCLUSIONS AND RECOMMENDATIONS . . . . . . . . . . . . . . . . . . . . . 45 8.0 RE FE R EN C E S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 8 9.0 LIST OF RELATED REPORTS AND DOCUMENTS IN THE PUBLIC DOCUMENT ROOM..........................................................52 Appendix A - Calculation to Check Estimates of Bulk Dynamic Pressures Computed by B WR OG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A- 1 Appendix B - Analyses to Verify Values of P for Selected Materials Reported by the BWROG and Suggestions for Development of Scaling Analyses . . . . . . . . , . . . , . B-1 Appendix C - Calculations to Eumine Accuracy of Reported Jet Volumes Bounded by Selected Load isobars. . . . . . . . . . .,. . . . . . . . . . . . . . . . . . . . . . . . . . . .... . . . C-1 Appendix D - Calculations to Examine Zone of Influence Models Proposed for Two-Phase Jets.............................................. . . . . . . . . . D- 1 Appendix E - Calculation to Examine ns URG Guidance on Thin-Bed Effect on Alternate Strainers . . . . . . . . . . . . . . . ...................... .... ..... .. E-1 Appendix F - M'apping of the Zone of influence (ZOI) . . . . . . ... .......... ...... F1 Appendix G - Calculations to Evaluate Methods Recommended by the BWROG to Estimate the Quantity of Fines . . . . . . . . . . . . . . . . . . . ...... . .... . . . . . . . G- 1 Appendix H - Calculations to Examine Accuracy of BWROG Drywell Debris Transport Fa ct o rs . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . ........H-1 Appendix l- Analyses to Examine Accuracy and Applicability of ECCS Strainer Head Loss 1-1 Appendix J - Calculations to Examine Accuracy of the Sludge Generation Factors . . . . . . . J-1 i

DRAFT

+ ..

DRAFT wm Executive Summary l

I

The guidance provided in the Boiling Water Reactor _ Owners Group (BWROG) Utility Resolution

. Guidance document (URG) can be broadly divided into the following areas

• I 1)- , Selection of breaks for strainer blockage analysis 1

12) - insulation debris generation by a postulated break

3) Other than-insulation debris (e.g., concrete dust, paint chips etc.) drywell sources

- 4) Debris transport from the break location in the drywell to the suppression pool .~

i

5)  ; Suppression pool debris (e.g., Wudge and rust flakes) _

J

- 6) Transport of debris in the suppression pool and its accumulation on the strainer  ;

7) Head loss resulting from debris buildup for selected strainer geometries Guidance for Estimating /.vailable NPSH Margin  ;

8)-

9)- Otherissues' . ,

The following sections discuss a brief description of the guidance provided in the URG and the staff evaluation of the guidanz for each area.

E8.1 Selection of Breaks for Analysis The BWR primary system piping varies in diameter from 1.5 inches to 24 inches (or higher).

Therefore, the postulated breaks can be small, medium or large breaks in the main steam lines, ,

recirculation lines or feed water lines. It is difficult to analyze each postulated break. Therafore, ,

a criterion is needed to select tha bounding breaks that maximize head loss across the ECCS Pump Suctions Strainer, The Regulatory Guide 1,82, Rev.2, states that as a minimum, the following postulated break locations should be considered:

e Breaks on the main steam, feed-water and recirculation lines with the largest amount of potential debris with'n the expected zone of influence e Large breaks with two or more different types of debris within the expected zone of influence e- Breaks in areas with the most direct path between the drywell and wetwell, and '

e Large arid medium breaks with the largest potential debris to insulation ratio by weight The guidance provided in the URG reproducesthe RG 1.82, Rev. 2 guidance. However, it states that:

o Plants licensed to current requirements of the Standard Review Pian and MEB 3-1 need not analyze all the break k. cations. Such plants can evaluate only those breaks that are most likely to occur,

• Other plants may use RG 1.82, Rev. 2, guidance or other guidance consistentwith 10 CFR 50.46. Howevericare should be taken to differentiate between pipe break locations used

- for ECCS evaluation and those. that are in the plant-licensing basis.

e - Plants employing attemate strairiers (i.e., strainers with large surface areas and cavities to -

acc9mmodate considerable quantities of, debris without significant increase in head, loss) need not ana' lyze large and medium tireaks with the largest potential debris to insulation ratio by weight.

4 DRAFT ES-1 4

E'*,, -- _<- .c . e- _.. . - - -

..~..m . . - . . - ,. ~, ,,hym _ u,.,, , . - , , . , , .-

. ~ _ . _ _ .

DRAFT em

- Upon reviewing the URG guidance, the staff concluded that:

• RG 1.82, Rev. 2l provides the complete spectrum of breaks that should be analyzed to meet the intent of 10 CFR 50.46. The staff considers SRP section 3.6.2 and MEB 31 to

_ be inappropriate for demonstrating compliance with 10 CFR 50.46.

• Licensees considering a solution option that is based on attemate strainers should experimentally verify that thin-bed effect is not an issue for the strainer design being considered for implementation. If verified, the licensee need not analyze " medium breaks ,

with the largest potential debris to insulation ratio by weight." The large breaks with largest potential particulate debris to insulation ratio by weight shru!d still be analyzed.

The staff has communicated these concems to the BWROG in the past as part !.x the preliminary review comments. The staff has reviewed several licensee submittals to ensure that breaks selected for analysis are consistentwith RG 1.82, Rev. 2 guidance. So far, all licensees (of those providing detailed plant specific submittals to the staff), with the exception of one, used RG 1.82, Rev. 2 guidance to select a set of breaks for analysis. The one licensee who initially relied on MEB 3-1 to select " credible break locat;ons" has since modified the analysis to include other "non-

- credible breaks." The licensee stated that as result of reanalysis a more bounding break location was identified.

ES.2 Insulation Debris Generation by a Postulated Break Postulated breaks in the primary piping would cestroy insulation locatsd in a close region surrounding the separated broken ends due to combined effects of blast wave and jet Impingement. This zone over which the destruction occurs, referred to as the Zone of influence (ZO!), depends strongly on the type of insulation and mode of encapsulation. In addition, the ZOI also depends on the type of break (i.e., main steam line break, recirculation line break or feed-waterline break) as well as other considerations, such as extent of axial and radial separation of the broken ends.

Section 3.2.1.2l" Zone of Influence" provides the BWROG guidance related to mapping a zone of influence around a postulated break location. The BWROG guidance was developed using experimental data obtained from air jet impact testing (AJIT) performed by the BWROG and the results of the associated computational fluid dynamics (CFD) modeling. The guidance provided four options (or methods) for determining the zone of influence over which insulation would be damaged, although not all of the damaged insulation is in a readily transportable form. These methods are as follows:

• Method 1 assumes that entire drywell is the zone of influence and thus all the insulation contained in the drywell would be damaged by the postulated break.

• Method 2 provides the following procedure for determining the zone of influence:

1) Assume that break is a double ended guillotine break with full separation resulting in continuous blowdown from both ends.

2) Determine the lowest dynamic pressure at which' destruction would begin to occur for each insulation material of interest. Estimate these destruction pressures after scaling the Alt Jet impact Testing (AJIT) measured values to the plant and taking into account that destruction pressure is inversely proportional to the target pipe DRAFT ES-2

m

]

q - -

q j

DRAFT pam 1 diameter. t 3)1 For each insulation type, determine the zone of influence by calculating the spatial -!

volume enveloped by a specific damage pressure of interest for a jet expanding in i free space and mapping a spherical zone of equal volume surrounding the break.

The radius of the sphencal zone is highoot for the weakest insulation (i.e., lowest'- .:

o destruction pressure). _

4) The zone of influence for Method 2 is the largest volume sphere resulting from this  ;

- analysis. _

l

5) - Determine tha most limiting quantities of damaged debris by placing the 201 at  !

different locations in the drywell and estimating the volume of each type of insulaticn

- debris' damaged by the jet, i

1. Method 3 is similar to Method 2 except for the following differences: j

1) Method 3 allows the licensee to take credit for break restraints and evaluate axial i and radial offsets consistent with the restraint, l

2) . Method 3 allows the licensee to take credit for single ended blowdown (e.g., steam _ -

line break)

- 3) Method 3 also allows the licensee to map different zones of influences for different 4 insulation materials. ,

e. Method 4 allows the user to directly employ results of CFD analyses in conjunction with air i jet impact testing data to map the zone of influence. Specific guidance was not provided

- how to apply the CFD tools. 4 The staffidentified several concoms and forwarded them in the past at appropriate forums to the BWROG. The majority of the concems related to logic used by the BWROG:

o- To scale AJ!T test data to the actual BWR conditions. The AJIT data was obtained by subjecting real scale inselstion blankets to high pressure air jet originating from a 3 inch diameter nozzle. In a real plant, on the other hand, jets may originate from much larger area pipe breaks. '

• - To account for the impacts of drywe4 congestion and drywell layout on jet expansion.

The Staff has conducted several independent analyses to examine if the URG-guidance is

- appropriate and bounding. Based on these analyses, the stafl concludes that Method 1 is clearly a bounding and conservative. method for evaluating the 201 since it' encompasses the entire drywell, and therefore, bounds the amount of insulation d6bris that may be generated by the break.-

The staff believes that the spherical zones of influence osveloped by using Method 2 would be .

sufficientij 'arge to envelop the entire zone over which destruction would actually occur, This i method is sufficiently conservative, and therefore,is considered acceptable by the staff for use on  ;

- insulations with low P. values. For the _ insulations with high P. values, the staff recommends that the zone of influence be developed based on the " target area averaged pressures" instead of Ljet-center-line pressures. The staff believes that spherical zone of influence developed by using t

.. .. Method 3 is acceptable, with the same comments as for Method 2. Also, the URG does no,t- .

provide guidance on the type of analyses to be undertaken by the' licensees to determine the extent b of axial and radial separations of the broken ends,- Licensees desiring to take credit for limited L separation of the broken ends due to pipe restraints or single ended blowdown should conduct r -

DRAFT ES-3

-]

e w -, ,

g 9 m -- - ,, , - .w, rrw-w e r- w -w .-----e 9,-m--e-- * *e- - - * - - + - w-- --- - - - --

• l DRAFT maw supporting analyses and submit them for further review before their use.

Section 3.2.1.2 of the URG did not provide detailed guidance on Method 4. Therefore, the staff cannot accept Method 4 at this time if a licensee intends to use Method 4, then they should provide the staff with a detailed plant speci5c submittal for review. The submittal should include details on how the licensee intends to validate the CFD code used.

Of the licensee submittals reviewed by the staff to date, all have used Methods 1,2 and 3 for estimating insulation debris generated by postulated breaks. The plants that used Method 3 assumed the breaks to be fully separated and with blowdown from both ends.

ES.3 Other Drywell Debris Sources The URG provided guidance that can be used by individual utilities to identify other sources of debris in the drywell and to estimate the quantities of such debris that should be used in the ECCS strainer analysis. The URG identified the following other sources of debris:

e Dirt / Dust The URG suggests that the licensee assume 150 lbm of dust / dirt for estimating the strainer head loss. This estimate is based on engineenng judgement.

o Other Transient Debris: No specific value is given. The judgement is left to the individual utilities.

• Rust from unoainted steel surfaces: The URG recommended a value of 50 lbm based on engineering judgement.

• Particulate Debris: No specific value is given. The judgement is left to the individual utilitiess e Paints /Coatinos The recommended values are 47 lbm for inorganic zine coatings, 85 lbm for inorganic zine top coated with epoxy, and 71 lbm for 100% epoxy coating. These estimates are based on a Bechtel study.

e Concrete: No specific value is given. The judgement is left to the individual utilities, e Unaualified/Indete.rminatePaint/Coatinas No specific value is given. The judgement is left to the individualutilities. As an alternative, the URG notes that licensees may remove the unqualified or indeterminate coatings, or attempt to qualify them through in-situ qualification No guidance on how in-situ qualification is accomplished is provided.

In addition, the URG provided cautions to remind users that their foreign material exclusion (FME),

housekeeping, and inspection programs must be adequate to assure that the quantities of each of th3se types of materials do not exceed the quantities assumed in their evaluation of ECCS strainer loading.

The staff compared to recommended va!ues with the previous estimates developed as part of the NUREG/CR-0224 study. Based on this comparison, the staff concluded that most of the values suggested in the URG are about the same or larger than NUREG/CR-6224 values. As a result, the staff concludes that use of these values is acceptable; however, the staff notes that the NUREG/CR-6224 study was conducted on a reference Mark I containment. For this reason, the staff believes that it would be prudent for individuallicensees to evaluate the recommendationc of this section for applicabilityto their plant. It would also be useful for the BWROG to provide better guidance related to estimation of dirt / concrete / rust debris sources in the containment as well as DRAFT ES-4 e

4

\  ;

DRAFT wm paint / coatings for plants with unquelified coatings. This concern is especially true for Mark lli containments because they have larger concrete areas, in addition, pctential for dirt / dust accumulalon may be greater in the larger containments if they have significantly larger amounts

~ of horizontalsurfaces. The sensitivityof the head loss calculationsto these numbers may vary with the assumptions and the resolution option selected by the licensee. If the licensee is installing a -

large passive strainer which is based on very conservative assumptions relative to the quantities of fibrous debris and sludge assumed to reach the strainer surface, then that licensee may be able to demonstrate by sensitivity analysis that variations in the quantities of d:rt, oust, rust flakes, etc.

may not significantlyaffect the overall head loss across the strainer. If, however, the licensoe were attempting to lustify their current strainer or use as " realistic" assumptions as possible, then the significance of the values determined in accordance with Section 3.2.2 of the URG escalates significantly.

Consistent with the staff's guidance above, it would also be prudent for licensees to evaluate the conclusions of the Bechtel coatings report to ensure applicabilityof those conclusions to their plant.

Most importantly, as noted above, the staff concludes that licensees should be cautioned to carefully evaluate the potentia! impact of unqualified and indoterminate coatings on ECCS suction strainer head loss. The URG downplays the significance of unqualified / indeterminate coatings relative to the potential for strainer clogging. The staff is concemed because concluding that uaqualified or indeterminate coatings are not a significot contributor to ECCS blockage may lead to inadequate strainer sizing and potentially significant operability concems depending on the findings of the staff's research on the coatingsissue. The staff states that if a licensee is in doubt about the ability of any coating to survive a LOCA, then assuming that all such coatings reach the strainer surface would clearly be the conservative measure.

ES.4 Drywell Debris Transport Insulation debris would be generated in the drywell and then be transported to the suppression pool by the reactor v,essel blowdown and the various mechanisms that can wash debris down to the suppression pool. The guidance provided in RG 1.82, Rev. 2 requires the licensees to assume all damaged debris would be transported to the suppression pool.

Section 3.2.3, entitled, 'Drywell Debris Transport,' documents BWROG guidance on various options for estimating the fraction of the damaged insulation (generated in the drywell) that will be transportedto the suppression pool as a result of these mechanisms. The URG guidance for the fibrous insulation is based on the following research conducted by the BWROG:

• _ The damaged insulation was assumed to fall into three categories: ' fines', "large pieces" and

" blankets'. For each insulation, the relative fractions of the insulation destroyed into each of these size categories (i.e., the size distribution factors) were derived using the AJIT test data.

These fractions were calculated as integral values averaged over the entire zone of influence.

• -The fraction of the mass of ' fines" that would be transported as a result of blowdown and washdown following a main steam line break and recirculation line break was estimated. The estimates are based on small-scale testing undertaken by the BWROG. The URG DRAFT ES-5 l

. m .

c , ,

3 fl W j h

~

DRAFT wm i

recommended. values are 1.0. for. Mark I and Mark lit containments.: For the Mark ll ~  !

containments, however, this fraction is 0.5 and 0.56 for steam line and recirculationline breaks, -

- respectively.'. .

-i

• The floor gratings were identified 'as the ~ major. locations for capture of "large piecesi and ,

blankets." Based on engineering judgement the URG concluded that 76% of this class'of debris generated below the lowest fk,or grating and 6.25% of the debris generated above the lowest grating would be transported to the suppression pool. 1 The guidance related to transport of RMI debris is also similarc Ti.. URG divides the RMI Jebris i h '

into "small pieces (<6.0 in *, "large foils (>6.0 in')' and " Intact Assembly". The ZOl integrated size '

distribution fractions were derived using AJIT test data for each type of RMI. Based on small scale transport ter, ting, the URG guides the licensee to assume that all the smak pieces ~ would be  !

transported to the suppression pool in Mark I and ill containments, whereas only 10% and 5% of  !

- small pieces _would be transported in mark ll containment corresponding to steam line and

- recirculation line break respectively. .

The staff had previously communicated several concems related to scaling of small-scale transport

' test data to BWR conditions, in addition the staff conducted a series of experiments to verify the ,

- accuracy of guidance provided in the URG. These results were shared with the BWROG and individual utilities at appropriate forums Based on the studies the staff concluded that: -

o 3The URG guidance is non-conservative for MARK ll containments; The experimental data 1

- clearly established that Mark ll vents, when they are wet, do capture small debris; However, capture fractions are no more than 15% fer " fines

• with the majority of them deposited on the Hoor. This value is much lower than the 50% proposed by the BWROG. Furthermore, it is likely that the ECCS flow during the recirculation phase will likely resu: pend the majority of this debris and transport it to the vents. It is suggested that URG be revised to change the '

transport factors for Mark ll containments to the same as Mark I and Mark 111 for both fibrous 1 and RMI insulations.

- * - The suggested transport fractions for large debris below the lowest grating lacked adequate ,

_ justification for their use._ Based on the staff's evaluation,the staff believes a transport fraction ,

of 100% should be used for large debris generated below the lowest continuous grating in the drywell. >

e For the other containments, the URG guidance would result in conservative estimates for transport fractions. While applying URG suggested transport factors, care should be taken to ensure that the underlying assumptions are consistent with the particular plant being analyzed

- In particular, the following considerations should be addressed: .

1) The testing done by NRCLand BWROG employed floor gratings with a 4" x -1" x %"

clearancei All the results (e.g., capture efficiencies)were derived assuming that the grating -

occupies 100% of the cross-sectionwith no chance for debris to bypass them. On the other hand, if_the' utility identifies large discontinuities or gaps in gratings that would allow a

. ~ fraction of the debris to pass through them, then the transport factors should be revised Jappropriately.

2) Implicitly, both studies assumed that unthrottled ECCS overflow occurs over a period of no

. longer than 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> following a recirculation line break. If the licensee expects unthrottled DRAFT

- ES-6

[

=.

W7 e 44- pe- w 4 e r+Jm m mt'pg 4-- w e- 9creT-4W-e m 7 e # W t'- t-'w iw sm- 4.+mh s t--wt yi mdv y47 et P g-i<-rn-s=-g----r- e ge C

t o

DRAFT imam ECCS operation for more than 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />, the washdown fractions should be properly revised.

In such a case, the URG assumed 25% washdown fraction for *large pieces" is non-conservative.

ES.5 Suppression Pool Debris During recent surveys, the suppression pools were found to cor.tain various types of non LOCA related debris. This debris would be transported to the strainer and may contribute towards head loss across the ECCS strainer. Section 3.2.4 of the URG ' Suppression Pool Debris" provides .

guidance related to identification of various debris sources and important considerationsthat should i

be used to estimate quantities of such debris.

The URG identified the following sources of suppression pool debris:

• Fibrous debris entrained in the pool volume prior to a LOCA.

• Non-fibrous debris entrained in the pool volume prior to a LOCA.

e The dirt / dust above the pool area which may be washed down during pool swell.

• Other potential debris, such as operational debris and unqualified / indeterminate coatings, i

The BWROG undertook a survey to estimate the quantities of sludge intrained in the suppression pools. Based on this survey, the BWROG recommended a sludge generation rate of 150 lbm per

- year. This sludge generation rate can be coupled with frequency of pool cleaning to estimate the total quantity of sludge for use in the ECCS Strainer blockage analyses.

The URG did not specify generic estimates for any other debris types. However, it has provided a list of considerations that should be taken into account while estimating the quantities of other debris to be used in the ECCS strainer blockage analysis. The URG cautioned the licensees to recognize that if suppression pool debris source terms can not be shown to be controlled at values less than assumedin the strainer sizing calculations, the operability of the ECCS system may be challenged. .

The staff finds no deficienciesin the URG recommendations. The staff reiterates the importance of the FME program to minimize the quantity of other potential debris.

ES.6 Suppression Pool Transport Transport of the debris within the suppression pool has been studied extensively by the staff as part of NUREG/CR-6224 study NRC sponsored experiments were conducted to explore fibrous and RMI debris transportwhen aubjected to chugging and post-chugging periods following a LOCA.

The BWROG relied primarily on these experiments to develop the guidance documented in Section 3.2.5 of the URG, entitled " Suppression Pool Transport."

The URG recommended that no credit be taken for settling of debris in the pool during the high energy phase where the pool undergoes thorough mixing. It is also recommended that all suppressien pool debris be assumed to be resuspended during this phase. Finally, an individual licensee is given option to select no settling in the pool even after the high energy phase terminates DRAFT ES-7 w

D DRAFT 12/30/97 and the pool returns to quiescent conditions or to determine settling rates using methods described in NUREG/CR-6224, Appendix B.

The staff found no deficiencies in the URG recommendations. However, it is pointed out that NUREG/CR-6224, Appendix B provides the required data for selected insulation and particulate types only. Licensees using NUREG/CR-6224, Appendix-B methods for other insulation debris should be cautious about extrapolating the expeeental data and models ES.7 Head Lose across the Strainar The BWROG devoted considerable resources to explore various generic designs for large passive strainers that can be used to replace the existing truncated cone strainers. These designs are: (1) 20-point star strainer, (2) small stacked disk strainer, (3) 60-point star strainer, and (4) large stacked disk strainer. For each design head loss data was obtained for various combinations of fibrous insulation debris, sludge, recipe (paints + concrete chips) and RMI debris. The general characteristics (size, shape etc.) of the debris used are consistent with that used in the NUREG/CR-6224 study. Note that the NUREG/CR-6224 study selected these characteristics to maximize the resulting head loss. Based on the test data, the BWROG developed two correlational methods that can be used to estimate head loss acror ;he strainers.

The first method provides a non-dimensional head loss cualation that can be used to estimate head losa across an attemate strainer design. This correlation was specified to be valid for lower debris loadings, where debris layer theoreticalthicknest,-to-the strainer diameter ratio is 0.15. For higher debris loadings the correlation is not applicable, and the URG recommends strainer specific testing.

The second method provides a six step process for estimating strainer RMI capacity and head loss across RMI debris beds. This method N Mgnizes that RMI beds on strainers reach a saturation thickness beyon,d which flow-induced drag forces are not large enough to retain the RMI pieces on the strainer surface. This saturation thickness was provided as a function of the type of RMI debris.

A correlation is provided to estimate head loss across the RMI debris beds.

For mixed beds (i.e., RMl+ Fiber + Particulate) the URG suggests that RMI does not contribute additionalhead losses in combination with other debris, and that it may actually reduce the head loss. Based on this, the URG states that head loss for mixed beds can be evaluated by ignoring RMI altogether and estimating head loss resulting from the other debris.

The Staff conducted several confirmatory calculations to validate the URG calculational procedures and to examine the applicability of the calculational procedures to the actual plant conditions.

Results of these analyses were shared with the BWROG as part of the preliminary calculations.

Based on these analyses, the head loss correlationwas found to be unreliable and incomplete for plant analyses, and is therefore, unacceptable. The staff strongly recommends that utilities use vendor provided data to qualify strainerdesigns, but not to rely on the URG developed correlatiors DRAFT ES-8

DRAFT inom and calculationalprocedures. m addition, the staff recommends that licencee designs should be hble to accommodate experimentaluncertainties associated with correlations and/or calculational methods developed by the vendors.

Finally, the staff also disagrees with the URG conclusions related to mixed beds. The staff has concluded that the head loss for RMI combined with fibrous debris and corrosion products would be bounded by summing the numerical values of the: 1) the head loss across the strainer with a fibrous debris / corrosion product debris bed only, and 2) the head loss across the strainer for an RMI debris bed only.

ES.8 Estimation of Available NPSH for ECCS Pumps Section 3.2.6 provides BWROG guidance related to evaluation of ECCS Pump NPSH. The important points of this guidance are es follows:

e No credit for containmentoverpressuregrcaterthan atmospheric pressure should be taken in determining available NPSH unless such credit is in conformance with the existing plant licensing basis.

• A range of expected fluid temperatures should be consHered when evaluating NPSH available, unless the plant licensing basis specifies the maximurn expected fluid temperature. If specified, this temperature should be used unless plant licensing basis is changed. The fluid temperature could effect both the available NPSH margin and the head loss across the strainer, e All strainers etn be expected to L>e clean at the start of the postulated LOCA. It is not necessary to essume pre-existing blockage.

o if ECCS blo::kage analysis uses reduced ECCS fiow through the strainers in order to meet NPSH reautrements, care should be taken to ensure that changes in ECCS flow rates are consistent with inputs and assumptions used in the evaluation model required by 10 CFR 50.46 to calculate ECCS cooling performance. Also ensure that appropriate operating and emergency procedures are in place.

Licensees should evaluate applicabilityof head loss correlations closely before they are applied to estimate head loss across the strainers, o The ECCS strainer design should be consistent witn the plant limiting single failure assumptions.

The staff finds no deficiencies in the URG recommendations. The licensees are strongly encouraged to design strainers with performance characteristics that result in additional NPSH margin above the minimum required. Also, it is pointed out that the staff is evaluating calculation of NPSH and plant licensing bases as part of its review of Generic Letter 97-04, " Assurance of Sufficient Net Positive Suction Head for Emergency Core Cooling and Containment Heat Removal Pumps." Additionally, the staff discourages the use of licensing basis changes as resolution options. Conservative licensing basis assumptions provide safety margin for the plant. There is a substantialamount of uncertaintyassociated with the strainercicgging issue, and as a result, the staff does not recommend licensing basis changes.

DRAFT ES-9

0 l

DRAFT wws7 .

l ES.9 Other issues Resolution Options: Section 3.1.3.4 discusses the various resolution options available to licensees.

Nine options are discussed; however, the staff notes that many are partial solutions which would stiillikely require strainer replacement by the licensee. The following provides a brief overview of ,

the " resolution options' discussed in the URG section 3.1.3.4: '

e "Further refinement of fixed debris source terms." This option is simply a more detailed analysis of the containmentfor calculating the amount of debris that could reach the strainers in order to reduce the debris source term for sizing of the streers.

• Replacementof the existing strainers with afternate passive designs such as the stacked disk or star strainer designs, e Install Jacketing to reduce the insulation debris source term.

• ' Reduction in Transient Debris Source Terms

• To justify the reduction in transient debris, licensees may need to implement additional FME controls, housekeeping controls, or higher suppression pool cleaning frequency.

• Change the plar.t licensing basis. One example cited as a potentiallicensing basis change would be the use of a more realistic decay heat curve in lieu of the conservative curve used in the original licensing basis (i.e., ANS 5.1 versus May-Witt). This, in tum, would reduce the calculated suppression pool temperatures following a LOCA. The BWROG does not recommend taking credit for containment overpressure for the calculation of NPSH margin.

• Reevaluation of ECCS suction line penetration loads without reopening the licensing basis for containment loads. The BWROG does not provide any guidance on how this can be done.

The BWRCG also does not recommend reopening the licensing basis for containment loads.

• Partial replacement of fibrous insulation with reflective metallic insulation (RMI).

• Installation of a backflush system.

e installation of self-cleaning strainers.

The staff's concems with these ' Resolution Options

• are as follows:

e The URG states that ' Additional details on the use of jacketing is provided in Section 3.2.1."

No further guidance is provided in Section 3.2.1 other than some information on debris generation. No guidancewas provided on types of bandsin be used, construction of the bands and mounting instructions. In addition, the characterization of the effectiveness of insulation Jacketing on this page is inconsistentwith the test report from the BWROG air jet impact testing (AJIT). The AJiT test report states 'In the case of fibrous insulation materials, the use of jacketing as a means of reducing debris generation does not appear to be effective without the use of an additional banding material which better secures the jacketing to the insulation assembly and the pipe." Wrthout further detailed informatbn on how to apply this option, the benefits achieved, and the technical basis supporting the use of this measure, the staff is unable to make a determination as to the acceptability of this option.

• The URG discusses the taking of credit for FME and housekeeping programs as a justification for creditinglower amounts of transient debris in a plant analysis. The staff has two concerns related to incorporating this option into a licensee's final resolution. First, if a licensee selects an amount of transient debris in their analysis which is too low, then there is potential for operability ccncerns to be raised every time debris is found in the drywell, suppression pool, or wetwell. Second, if a licensee is considering increasing the amount of transient debris in DRAFT ES-10

y  ;

G .

e .

DRAFT nm slEing'their new strainer, the staff believes that housekeeping controls or foreign material

. exclusion (FME) controls are not a substitute for periodic inspections and cleanings of the, strainers or the suppression pool. Given the numerous events reported over the last few years ,

related to FME issues, the industry has not demonstrated that FME controls alone are effectin  ;

in ensuring that materials are not left in the drywell, wetwell, or suppression r.molf As such, the staff believes that regularinspections of the suppression pool and ECCS sucf.on strainers, and

- cleaningslwhen necessary, should be conducted every refueling outage until _ licensees have 4 demonstrated over time the ability to control foreign materials. The staff believes M is more ,

, prudent to add margin when sizing strainers to account for the uncertainty in the effectiveness >

. of housekeeping controls. This, in tum, will minimize the need for operability assessments when araall amounts of foreign material are found in the containment or suppression pool.

e yThe URG notes that a part of a licensee's resolution to the strainer clogging issue could include a licensing basis change. An example is cited where a licensee may wish to use a more .

realistic decay heat curve to reduce the calculate post LOCA suppression pool temperaturer..

The BWROG specifically states that use of credit for containment overpressure is nce recommended.' The staff concurs that additionsi containment overpressure (other than an , l amount already approved by the staff for the existing licensing basis) should not be used as part of the resolution of this issue at all, The staff is evaluating its position on use of-containment overpressure in calculating NPSH margin as part of its review of Generic Letter 197-04, " Assurance of Sufficient Not Positive Suction Head for Emergency Core Cooling and .

Containmer'tHeatRemovalPumps,"datedOctober7,1997. Additionally,thestaffdiscourages tne use of licensing basis changes as resolution options. Conservative licensing basis assumptions provide safety margin for the plant, There is a substantial amount of uncertainty associated with the strainer clogging issue, and as a result, the staff does not recommend licensing basis changes.

• The URG provides a discussion on the potential to reanalyze suction line penetration loads.

Specifically,the URG states that it "may be possible to furtherincrease the size of the attemate strainer by reanalyzing ECCS suction penetration loads and suppression pool structural loads using more pophisticated techniques or through reduction of conservatism in current design basis calculations, but without reopening the licensing basis for containment loads." The staff has concems regarding this statement. Specifically, the hydrodynamic load programs for the Mark I, Mark ll, and Mark 111 programs were very specific in providing generic methodologies for calculation of hydrodynamic loads. Each plant submitted Plant Unique Analysis Reports which provided the specific details of each plant's calculated hydrodynamicloads, including the methodology used, coefficients used, etc. The approved methodologies had their basis in test data. The staff cautions against making changes to hydrodynamic load calculations without testing to demonstrate the validity of the revised calculations. -

~ e The URG discusses the option of reducing the fibrous debris snurce term in the drywell by

partially replacing fibrous insulation with RMI.- The s%*f agrees that this is an appropriate option i for reducing the amount of fibrous debris which c# oe transported to the suppression pool; however, when.using this option for a selected break, the licensee should go back and _

M

. reassess other breaks to ensure that the licensee has not changed which break is the most limiting in terms of NFSH margin, in addition, the licensee must now consider head loss in terms of combined fibrous /RMI debris beds.

DRAFT ES-11 4

• 1 ..p._:

a na _ . _ - ~ _ L: _ . A_ u __~ _ ;_ __.,~_,

? ,

l DRAFT imm Active Strainers: Although NRC Bulletin 96-03 allows use of self-cleaning strainer omgns as a resolution option, the BWROG recommends that a self cleaning strainer only be used if resolution with a passive strainer is not viable. The URG cites several concerns including:

e Optimization of clean head loss and torque.

e Delayed start of rotation of the plow / brush assembly following operation with minimum flow through the strainer.

e The effect of debris which passes through the strainer on downstream components. l e Surveillance / maintenance requirements.

The URG guidance on concems to be addressed with an active strainer design is comprehensive.

The staff noted no significant deliciencies in their review of this guidance. However, due to the l concerns cited by the PWROG in the URG relative to the self-cleaning strainer, any licensee wishing to use an active strainerdesign as a resolution option should address the concerns stated in the URG in a submittal to the staff.

Back Flushing: Although NRC Bulletin 96-03 allowed backflushing at a potential resolution option, the 8WROG recommends against use of strainer backflush systems as a primary means of resolving tne ECCD suction strainer clogging issue. The basic concern that the BWROG has is that ,

backflush will be needed early in the accident and frequently within 30 to 60 minutes. The BWROG considers backflush more viable as a defense-in-depthmeasure only. Because of the reliance on optrator action and mechanicalsystems, the staff concurs with the BWROG that backflushis more' effective as a defense in-depth measure only. Section 3.3 provides guidance on the types of considerations a utility should address if designing a backflush system either as a primary means of mitigating an accident or as defense-in-depth measure only. No significant deficiencies were noted by the staff in their review of this section.

4 DRAFT ES-12

- r- - - . . - . - - - m. ._

., . ~ ---.e..,

.-x.-

DRAFT m3aw .

SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION RELATED TO NRC BULLETIN 96-03 BOILING WATER REACTOR OWNERS GROUP TOP 8 CAL REPORT NEDO 32686

" UTILITY RESOLUTION GUIDANCE FOR ECCS SUCTION STRAINER BLOCKAGE" DOCKET NO. PROJ0691 1.0 lNTRODUCTION By letter dated November 20,1996, the Boiling Water Reactor Owners Group (BWROG) submitted NEDO 32686," Utility Resolution Guidance for ECCS Suction Strainer Blockage"(URG)(Refs 1 and

2) for staff review. Additionalinformation regarding the URG drywell transport methodology was submitted by the BWROG to the staff by facsimile dated November 23,1996 (Ref 3). The purpose of the documentwas to provide boiling waterreactor(BWR) licensees with guidance on complying with NRC Bulletin 96-03 (NRCB 96-03), " Potential Plugging of Emergency Core Cooling Suction Strainers by Debris"(Ref 4). Specifically, the major focus of the URG was to provide detailed guidance on performance of plant specificanalyses consistentwith Regulatory Oulde 1.82, Revision 2 (RG 1.82), " Water Sources for Long Term Recirculation Cooling Following a Loss-of-Coolant Accident" (Ref 5). Prior to their November 20,1996 submittal, the BWROG had submitted eight draft sections of the URG to the staff in order to facilitatc the staffs review of the final document.

Four sections were submitted on March 31,1996 (Ref 6), and four more were submitted on May 28, 1996 (Ref 7). By letters dated July 25,1996 (Ref 8), and August 20,1996 (Ref 9), the staff transmitted corpments oi, the draft URG sections to the BWROG. The staff discussed their comments on the draft URG sections with the BWROG in a meeting on August 19,1996. The results of the meeting are describedin a meeting summary dated September 4,1996 (Ref 10). In addition, by facsimile dated December 23,1996, the staff transmitted a request for additional information(RAls) on the final URG to the BWROG (Ref 11). On January 13,1997, the BWROG provided the staff with rrrvey results showing how mar' utilities intended to use the various options prov;ded in the URG (Ref 12). The BWROG then pro..wd their response to the staffs RAls on January 30,1997 (Ref 13),

Prior to the staffs review of the URG, the staff had closely followed the generic effort conducted by the BWROG for resolving the BWR ECC S strainer clogging issue. The staff continuously provided the BWROG with feedback as they conducted their efforts, By letter dated June 13,1994 (Ref 14),

the staff identified initial concems about the resolution under development by the BWROG. The BWROG began conducting testing to support their generic effort, and a draft test plan for testing performance of alternate strainer designs was forwarded to the staff by facsimile dated August 9, t 1994 (Ref 15). The staff responded with comments on the proposed test program in a letter dated September 12,1994 (Ref 16). By letter Jated March 15,1995 (Ref 17), the BWROG submitted their DRAFT 1 ,

DRAFT umw l

final test plan to the staff. In RAI letters dated June 22,1995 (Ref 18), August 21,1995 (Ref 19) and April 22,1996 (Ref 20), the staff identified its concems relative to the test programs and M s  ;

application of the test results. Additional concerns were transmitted to the BWROG via facsim:,e ,

dated August 14,1995 (Ref 21). The BWROG provided a written response to the June 22nd RAI l by letter dated July 17,1995 (Ref 22), and to the April 22,1996 RAI by letter dated July 9,1996 (Ref 23). No formal response to the staff concems in the RAl dated August 21,1995, were received by the staff although the staff concerns were discussedin meetings on May 31,1995, September 28-29,1995, April 4,1996, July 911,1996, and August 19,1996 (see meeting summaries and trip reports dated June 13,1995 (Ref 24), October 6,1995 (Ref 25) Agil 16,1996 (Ref 26), aly 25, 1996 (Ref 27), and September 4,1996 (Ref 10)). The BWROG also provided copies of their '

proposed test matrices by facsimiles dated June 23,1995 (Ref 28), and June 30,1995 (Ref 29). ,

The staff has completed its review of the URG, its supporting documentation, and all relevant  :

documents, and its conclusions are documented in this safety evaluation report (SER). The staff l

found portions of the URG to be acceptable for use in the conduct of plant specific analyses to estimate combined debris loadings for sizing of emergency core cooling system (ECCS) suction strainers. However, several portions of the URG were not accepted by the staff because the methods lacked sufficient guidance, supporting data, or analysis to justify their technical basis.  ;

Each section of the URG is discussed herein, and the basis for the staffs conclusions provided. ,

Licensees desiring to use the portions of the URG not accepte'J by the stashould address the staffs concerns cited herein in plant specific submittals in addition, there ae several portions of the URG where the staff believes that additional clarificationis needed to minimize the potential for misinterpretation of the information pnside'd. The staff has stated its position relative these portions of the URG in the associated section of this SER.

The staff closely reviewed the materials contained in the references cited above as well as other pertinent references. Independent confir natory analyses were per'ormed to identify:

e inadequacie,s in the generic guidance provided in the URG, and e inaccurate or unsubstantiated assumptions which were used to develop the stated guidance.

This SER is formatted to address each part of a plant specific analysis. For this reason, the SER includes sections on each of the following topics: 1) resolution options, 2) pipe break locations, 3) debris generation / zone of influence, 4) other drywell debris sources, 5) drywell debris transport, 6) suppression pool debris 7) suppression pool transport and settling,8) net positive suction head (NPSH) including strainer head loss, 9) backflush,10) self-cluning strainers, and 11) general /overallVRG comments. These sections are also consistent with the sections of the URG ,

on the same topics. Confirmatoryanalyses performed by the t,taff and their results are summarized in Appendices A through J to this SER.

DRAFT 2

DRAFT imm

1.1 BACKGROUND

On Julv 28,1992, an event occurred at Barseback Unit 2, a Swedish BWR, which involved the plugging of two containment vessel spray system (CVSS) suction strainers. The strainers were plugged by mineral woolinsulation that had been dislodged by steam from a pilot operated relief valve that spuriously opened while the reactor was at 3,100 kPa [435 psig) Two of the three strainers on the suction side of the CVSS pumps were in service and became partially plugged with mineralwool. Followingan indication of high differential pressure across both suction strainers 70 minutes into the event, the operators shut down the CVSS pumps and backflushed the strainers.

The Barsebeck event demonstrated that the potentialexists for a pipe break to generate insulation debris and transport a sufficient amount of the debris to the suppression pool to clog the ECCS strainers.

On January 16 and April 14, 1993, two events involving the clogging of ECCS strainers also occurred at the Perry Nuclear Power Plant, a domestic BWR, The first Perry event involved clogging of the suction strainers for the residual heat removal (RHR) pumps by debris in the suppression pool. The second Perry event involved the deposition of filter fibers on these strainera The debris consisted of glass fibers from temporary drywell cooling unit filters that had been inadvertentlydropped into the suppression pool, and corrosion products that had been filtered from the pool by the glass fibers which accumulated on the surface of the strainer. ihe Perry events demonstrated the deleterious effects on strainer pressure drop caused by the filtering of suppression pool particulates (corrosbn products or " sludge") by fibrous materials adhering to the ECCS strainer surfaces. These corrosion products are typically present in varying quantities ir, domestic BWRs.

The sludgb is generated during normal operation, and the amount of sludge present in the pool depends on the frequency of pool cleanings/desludging conducted by the licensee. Separate test programs have been conducted by the BWROG and the staff to quantify this filtering effect.

Based on these events, the NRC issued Bulletin 93-02, " Debris Plugging of Emergency Core Cooling Suction Strainers" (NRCB 93 02), on May 11,1993 (Ref 30)/ The bulletin requested licensees to rer6ove fibrous air filters and cther temporary sources of fibrous material, not designed to withstand a LOCA, from the containment. In addition, licensees were requested to take any immediate cornpensatory measures necessary to ensure the functional capability of the ECCS.

Following these events, the staff performed calculations to assess the vulnerability of each domestic BWR. The results of these calculations showed that the potential existed for the ECCS pumps to lose net positive suction head (NPSH) margin due to clogging of the suction strainers by debris generated during a postulated LOCA. The staff then conducted a detailed study of a reference BWR 4 plant witt 1 Mark I containment. The study results confirmed the results of the earlier staff calculations, ar' sere publishedin NUREG/CR-6224," Parametric Study of the Potential for BWR ECCS Strainer Blockage Due to LOCA Generated Debris,"in October 1995 (Ref 31).

Members of the NRC staff also attended an Organisation for Economic Co-operation and Development / Nuclear Energy Agency (OECD/NEA) workshop on the Barseb8ck incident held in Stockholm, Sweden, on January 26 and 27,1994. Representatives from other countries at this conference discussed actions taken or planned which would prevent or mitigate the consequences DRAFT 3 i

e

t DRAFT www 3 of BWR strainer blockage. Based on the preliminary results of the staffs study, as reinforced by information learned at the OECD/NEA workshop, the staff issued NRCB 93 02, Supplement 1

" Debris Plugging of Emergency Core tooling Suction Strainers," on February 18,1994 (Ref 39).

The purpose of the bulletin supplement was to request that BWR licentees take appropriate interni actions to ensure the reliable operation of the ECCS following a LOCA so that the staff and industry would have sufficient time to develop a permanent resolution. In addition, the bulletin supplement informed licensees of pressurized waterreactors (PWRs) and BWRs of the latest 'qformation on the vulnerabilityof ECCS suction strainers in BWRs and containment sumps in PWRs to clogging during i the recirculation phase of a LOCA.

On September 11,1995, Limerick Unit 1 was being operated at 100-percent power when control room personnel observed alarms and other indications that one safety relief valve (SRV) was open.

Emergencyprocedures were implemented. Attempts to close the valve were unsuccessful, and a manual reactor scram was initiated. Prior to the opening of the SRV, the licensee had been running the "A" loop of suppression pool cooling to remove heat being released into the pool by leaking SRVs. Shortly after the manual scram, and with the SRV still open, the "B" loop of suppression pool cooling was started. Operators continued their attempts to close the SRV and reduce the cooldown rate of the reactor vessel. Approximately30 minutes later, fluctuating motor current and flow were observed on the "A" loop. Cavitationwas believed to be the cause, and the loop was secured. After it was checked, the "A" pump was successfully restarted and no further problems were observeu.

After the cooldown following the blowdown event, a diver was sent into the suppression pool at Unit 1 to inspect the condition of the strainers and the general cleankness of the pool. Both suction '

strainers in the "A" loop of suppression pool cooling were found to be almost entirely envered with a thin " mat ~ of material, consisting mostly of fibers and sludge. The "B" loop suction strainers had a similar covering, but less ofit. Analysis showed that the sludge was primarily iron oxides and the fibers were polymeric in nature. The source of the fibers was not positively identified, but the licensee has determined that the fibers did not originate within the suppression pool, and that no trace of either fiberglass or asbestos was in the fibers.

The Limerick event demonstrated the need to ensure adequate suppression pool cleanliness. In addition, it re emphasized that materials other than fibrous insulation eculd also clog strainers (Perry's strainers were clogged by fibrous filter media), in response to this event, the staff issued NRC Bulletin 95-02," Unexpected Clogging of Residual Heat Removal (RHR) Pump Strainer While Operatingin Suppression Pool Cooling Mode"(NRCB 95-02), on October 17,1995 (Ref 40). The bulletin requested that licensees (1) assess the operability of their ECCS based on the cleanliness of their suppression pool and ECCS strainers, (2) verify the operability of the ECCS through an appropriate pump test and strainer inspect 6vn within 120 days from the date of the bulletin, (3) establish a pool cleaning program, (4) review their foreign material exclusion (FME) practices and

• correct any identified weaknesses, and (5) implement any appropriate additional measures for ensuring the availability of their ECCS, Licensee responses to NRCB 93-02 and its supplement demonstrated that appropriate interim measures had been implemented by licensees to ensure adequate protection of public health and safety, and to allow continued operation until the final actions requested in NRCB 96 03 are  ;

DRAFT 4 ,

- ., , -n. - - - , . , . . , . . , , , , - - , ----,-,.-v-.----e - r- -- ,

t DRAFT $2/3o 87 i l

implemented. In responding to these bulletins, licensees ensured that (1) alternate water sources (both safety and nonsafety-relatedsources) to mitigate a strainer clogging event were availabb, (2) emergency operating procedures (EOPs) provided adequate guidance on mitigating a strainer clogging event, (3) operators were adequately trained to mitigate a strainer clogging event, and (4) loose and temporary fibrous materials stored in containment were removed. Licensee responses to NRCD 95-02 showed that most suppression pools had been cleaned recently, and that those licensees who had not cleaned their suppression pools recently were scheduled to do so dur;ng their upcoming refueling outage. In addition, a generic safety assessment conducted by the BWROG concluded that operators would have adequate time to make use of alternate water sources (25-35 minutes) after initial reflood of the reactor vessel. The staff also notes that the probability of the initiating event is low. For these reasons, continued operation by BWR licensees was allowed until such time as licensees were able to fully implement their resolution to NRCB 96 03. Satisfactory resolution of NRCB 96-03 will ensure that the ECCS can perform its safety function and rninimize the need for operator action to mitigate a LOCA.

The results of the staff study, documentedin NUREG/CR-6224, demonstrate that for the reference plant, there is a high probability that the available NPSH margin for the ECCS pumps will be inadequate followino dislodging of insulation and other debris caused by a LOCA and transport of the debris to 19 ht M strainers, in addition, the study calculated that the loss of NPSH could occur quickly (lea r e % minutes into the event). The study also demonstrated that determining the adequacy of w& Wrgin for an ECCS system is highly plant specific because of the large variations in such plani characteristics as containment type, ECCS flow rates, insulation types, plant layout, plant cleanliness, anJ available NPSH margin.

The Barsebeck event demonstrated that a pipe break can generate and transport sufficient quantities of insulation and other debris to the suppression pool where they can be potentially deposited onto strainer surfaces and cause the ECCS to lose NPSH. The Perry events further demonstrated that fibrous debris combined with corrosion products present in the suppression pool (sludge)can exacerbate the problem. This phenomenon was confirmed in the staff study described above. The effect of filtering sludge from the suppression pool water by fibrous debris deposited on the strainer surfa;e was further confirmed in NRC-sponsored testing conducted at the Alden Research Laboratory which dernonstrated that the pressure drop across the strainer was greatly increased by this filtering effect. The results of the staff sponsored tests are discussed in NUREG/CR 6367,

• Experimental Study of Head Loss and Filtration for LOCA Debris,' dated December 1995 (Ref 41). Additionaltesting sponsored by the NRC at Alden Research Laboratory demonstrated that the energy conveyed to the suppression pool during the high energy phase of a LOCA is sufficient to ensure thet the fibrous debris and sludge are well mixed and evenly distributedin the suppression pool, and can remain suspended for a sufficientlylong period to allow large quantities to be deposited onto the strainer surfaces. As a result, the staff has concluded that this problem is applicable to all domestic BWRs with pressure suppression types of containments.

The basis for the staff's conclusionis as iollows: (1) there do not appear to be any features speci[c to a particular plant, class of plants, or containment type that would mitigate or prevent the generation, the transport to the suppression pool, or the deposition on the ECCS strainers of sufficient material to clog the strainers, and (2) parametric analyses performed in support of the NUREG/CR-6224 study, using parameter ranges which bound most domestic BWRs, failed to find DRAFT 5

s DRAFT maw parameter ranges that would prevent BWRs with other containment types from being susceptible to this problem. In addition, the staff study was conducted on a Mark 1; Barseback had a strainer clogging event and is similar in design to a Mark 11; and Perry, a Mark lil, also had a strainer clogging event.

Section 50.46 of Title 10 of the Code of Federa/ Regulations (10 CFR 50.46)(Ref 42) requires that licensees design their ECCS systems to meet five enteria, one of which is to provide long-term cooling capabikty of sufficient duration following a successful system initiation so that the core temperature shall be maintained at an acceptably low value and deay heat shall be rem /ed for the extended period of time required by the long-lived radioactivityremaining in the core. The ECCS is designed to iceet this criterion, assuming the worst single failure. Experience gained from operating events 1md detailed analysis, as previously discussed, demonstrate that excessive buildup of debris from thermal insulation, corronion products, and other particulates on ECCS pump strainers is highly likely to occur, creating the potentisl for a common-cause failure of the ECCS, which could prevent the ECCS from providing long term cooling following a LOCA. The staff has concluded, therefore, that this issue must be resolved by licensees in order to ensure compliance with the regulations. RG 1.82, Revision 2 provides an ac'aptable method of ensunng compliance with 10 CFR 50.40. The URG was written to provide licenseen with guidance on complying with 10 CFR 50.46 consistent with the guidance in RG 1.82, Revision 2.

Plant specific analyses to resolve this issue are difficult to perform because a substantial number of uncertainties are involved. Examples of these uncertainties include the amount of debris that would be generated by a pipe break for various insulation types; the amount of debris that would be transportedto tne suppression pool; the characteristics of debris reaching the suppression pool (e g., size and shape); and head-loss correlations for various insulation types combined with suppression pool corrosion products, paint chips, dirt, and other particulates. Many of these uncertaintieswould be plant specificoecause of the differencesin plant characteristicssuch as plant layout, insulation types, ECCS flow rates, containment types, plant cleanliness, and NPSH margin.

Testing and analyses were conducted to quantify many of these uncertainties.

The staff closely followed the work of the BWROG to resolve this issue. The BWROG evaluated several potential solutions, and completed testing on three new strainer designs: two passive strainerdesigns and one self-cleaning design. The BWROG effort was consistent with the options proposed in NRCB 96 03 for resolution of the ECCS potentialstrainer clogging issue. The BWROG then developedthe URG to provide utilities with (1) guidance on evaluation of the ECCS potential strainer clogging issue for their plant, (2) a standard industry approach to resolution of the issue that is technically sound, and (3) guidance that is consistent with the requested actions in the bulletin for demonstrating compliance with 10 CFR 50.46. The URG includes guidance on a calculational methodology for performing plant specific evaluations.

The staff noted in NRCB 96-03 that much of the effort and discussion on this issue has focused on the threal caused by fibrous insulation. While the staff recognized th'at fibrous insulation represents the largest source of fibrous materialin the containment, licensees were reminded in NRCB 96-03 that both the Perry and the Limerick events involved other sources of fibrous debris. In determining DRAFT 6

t DRAFT inw their resolution 5 this issue, licensees were reminded to focus on protecting the functional capability of the ECCS from all potential strainer clogging mechanisms.

2.0 DISCUS $1QN The issue of potential strainer blockage is a complex one. Head loss across the suction strainers is not only a function of the amount of debris, but also of the types (e.g., fibrous insulation, paint, reflective metallic insulation, dirt, corrosion products, etc.) and characteristics of the debris (size, shape, etc.). This creates a challenge for the analyst to evaluate the worst case for potential strainer debris loadings. In addition, the analyst must consider the potential for foreign material to be introduced during normal plant evolutions such as refueling and maintenance outages. Plant maintenance practices need to be evaluated, including the maintenance of qualified coatings in the drywell and wetwell. RG 1.82, Revision 2, provided non-prescriptive guidance on performir.g a plant specific analysis (e g., what should be consideredin a plant specific analysis, but not how to perform the calculations). The BWROG developed the URG in order to provide additional details to the individuallicensees on how to conduct the plant specific analysis, in addition, this document would help to provide a consistent response by the Ir-fustry to NRCB 96-03.

The URG is a comprehensive,but complex, document providing: 1) general guidance on resolution eptions, and 2) detailed guidance on performing plant specific analyses to estimate potential worst case debris loadings on ECCS suction strainers during a LOCA. The document consists of one volume providing the analysis guidance (Ref 1) and three volumes of technical support documentation (Ref 2). The technical support volumes include test reports and analyses which form the basis for the methodologies of calculating strainer debris loadings and ECCS pump NPSH.

For performing a plant specific analysis, the URG provides methodologies for: 1) estimating the amounts and types of debris that could be gerierated by a LOCA,2) estimating the amount of the generated debris which could be transported to the suppression pool, 3) estimating the amount of debris that coulgf be entrained on the ECCS suction strainer surfaces,4) determining the head loss caused by the estimated debris accumulation, and 5) calculating the NPSH margin. Various plant factors may affect the level of engineering effort and resources a licensee may wish to utilize in performing a plant specific analysis to ensure adequate sizing of its ECCS suction strainers.

Flexibilityin the generic guidance for performing these analyses is, therefore, desirable for licensees in order to accommodate the plant specific design details affecting the level of detail needed to complete the analysis. For instanco, some plants have more margin in the structuraldesign for the ECCS suppression pool penetrations than others. When evaluating the loads that would be caused by bulk fluid motion across the strainer and its associated suction piping (e.g., standard and acceleration drag forces) during a postulated transient or accident (i.e., hydrodynamicloads), a plant with more structural margin may be able to accommodatelarger suction strainers without the need for significant structural modifications than a plant with a lesser margin. Therefore, the plant with more margin may opt to conduct.a bounding type of, analysis rather than utilize engineering ,

resources in a detailed plant analysis. In order to provide this flexibility to the utiSties, the URG provides multiple methods for performing each part of the plant specificanalysis. This allows plants with more hydrodynamic load margin or more space in their suppression pool for larger strainers to use more conservative, or bounding, approaches which are less detailed and require less DRAFT 7

DRAFT imm licensee resources to perform. Plants with less margin or less space in the pool could opt for methodologies which result in lower calculated debns loadings, but require a higher level of engineering effort by the licensee to conduct the analysis. In providing flexibility, the BWROG opted to not link methodologiesfor calculating each part of a plant specific analysis (e.g., a licensee may use Method 1 for estimating the amount of debris generated, but then use Method 3 for estimating the amount of debris transported from the drywell to the wetwell, etc.). Therefore, individual utilities are provided with the flexiblity to pick and choose the methods from each part of the analysis that they will usk This flexibikty resulted in a more complex staff review because it created many ways a licenseo could apply the URG guidance. In all, the staff estimates that there are 38 different ways to apply the URG. Due to incomplete guidance and inadequate supporting documentation or analysis in several areas, the staff was unable to determine if all of the methodologies, or combination of methodologies, were conservative. Similarly, much of the general guidance on

' resolution options

• also lacked sufficient detail for the staff to review. In order to facilitate the maximum efficiencyin the conduct of its review, the staff primarily focused on the analytical sections of the URG as these sections willlikely be used by licensees to size their strainers, Since the staff lacked sufficient detail and supporting justification on many of the ' resolution options,' these were generally considered unacceptable without additional supporting justification or technical details being provided by a licensee or the BWROG. If individuallicensees desire to make use of any portions (analyticat or resolution options) of the URG that are not accepted by the staff, they should resolve the staffs concerns cited herein in a plant-specific submiital.

3.0 .URG GUIDANCE FOR DEMONSTRATINQAQMPLlANCE WITH 10CFR50.46 3.1 EVALUATION OF RESOLUTION OPTIONS This section of the URG provides guidance to utilities on evaluating the level of detail they should use in analyzing their plant and implementing a resolution in response to NRCB 96-03. The BWROG identifies key factors affecting the complexity and the cost of evaluating strainer performance arid implementing potential resolutions. The section includes four flow diagrams (Figures 1 through 4) which display the process recommended by the BWROG for sizing ECCS suction strainers.

Figure 1, page 8, provides an overview of the plant analysis process. For plants that have more than a minimal amount of fibrous insulation, the URG recommends that a licensee size their strainers as large as possible without violating their penetration hydrodynamic load limits. The strainer performance is then evaluated with the calculated debris loading to ensure that it provides adequate performance for maintaining ECCS pump NPSH margin. For plants with almost all reflective metallic insulation (RMI), Figure 1 provides two suggested methodologies: 1) sizing the strainers based on the head loss when the straineris loaded at the saturationlevel with RMI and/or

2) sizing the strainer based on a calculated expected debris loading. For mostly RMI plants, Figure 1 does not provide a recommendation for one sizing methodology over the other.

Figure 2, page 9, provides a flow diagram showing the process for performing a plant evaluation to determine the amount of debris generated and transported to the suppression pool. The diagram DRAFT a t

t

DRAFT umw also provides references to the appropriate URG and Regulatory Guide 1.82, Revision 2, sections which are related to each step in the analysis. There are two methods depicted on the flow diagram.

One is to specifically perform a detailed analysis of the zone of influence for a pipe break and to calculate the insulation debris generated. In parallel with this zone of influence calculation, the analyst would also evaluate non insulationdebris such as dirt, dust, rust, coatings, etc. Theese two calculations would be inputs into the next part of the analysis shown on Figure 3. An alternate method to the detailed analysis shown on Figure 2 would be to perform a simple bounding calculation which assumes that all of the materials of interest are generated as debris and transported to the suppression pool. If this method is selected, then it would provide the input into the next part of the analysis shown in Figure 3.

Figure 3, page 10, provides the flow diagram for evaluating the amount of debris present in the suppression pool and calculating the total debris loading on the strainer including the debris from the drywell. The totaldebris loading on the strainerthen serves as one input to Figure 4 (page 11) which provides the process for calculating the ECCS pump NPSH.

Section 3.1 also gives some additional overview guidance on estimation of debris sources, establishing the current licensing basis, and resolution options. The URG breaks debris sources down into two categories: fixed and transient. Fixed debris sources are those which must be impacted by the LOCA break jet or blowdown forces to form debris which is transportable to the ECCS suction strainers. Transient debris sources are those which generally result from normal plant operations,and are presentin a transportable form prior to the postulated LOCA. Examples of fixed debris include piping insulation and coatings. Examples of transient debris include dirt, dust and any other loose materials present in the drywell or suppression chamber. Section 3.1 also points to some of the important licensing basis assumptions that need to be identified by each utility.

These considerations include containment overpressure and suppression pool temperatures assumedin the NPSH calculations,and pipe breaklocations. An importantrecommendationby the BWROG states that licensees "should consider, if practicable, use of a strainer that results in acceptable ECCS pump NPSH without reliance on containment pressure." The staff notes that this recommendation is consistent with the recommendations of Regulatory Guide 1.1 (RG 1.1), " Net Positive Suction Head for Emergency Core Cooling and Containment Heat Removal System Pumps * (Ref 57).

In Section 3.1.2, the URG provides some overview guidance for plants with reflective metallic insulation (RMI). The BWROG notes that even plants with mostly RMI will have to demonstrate the adequacy of their strainers or implement an appropriate resolution. This statement is based on the results of testing which demonstrates that only a small amount of fiber combined with corrosion products can result in unacceptable head losses. The staff notes that the Perry and Limerick events also demonstrated this same conclusion.

,. Section S.1,3 provides guidance for plants with significant amounts of fibrous insulation in their containments. The section indicates that small amo'unts of fiber can lead to unacceptable head losses on ECCS systems employing the conical or cylindrical perforated plate strainers typically used prior to the issuance of NRCB 96-03 by many BWRs. The BWROG racommends the

- replacement of existing strainers with one of the attemate designs which have demonstrated better DRAFT 9

1 DRAFT wm performance characteristics. They also recommend the use of the largest straina possible without changing the hcensing basis for hydrodynamicloads. The BWROG believes that staff review of new hydrodynamic load methodologies could delay final resolution of this issue. The rest of Section 3.1.3 and its associated subsections discuss various considerationsthat utilities should keep in mind

{

when considering their final resolution to this issue. These considerations include evaluation of i containment penetrationload margins, plant physical constraints, suction strainertype and size, and  ;

resolution options.

Section 3.1.3.4 discusses the various resolution options availab!: te licensees. Nine opt!;ns are discussed; however, the staff notes that many are partial solutions which would stilllikely require strainer replacement by the licensee. The following provides a brief overview of the

• resolution options

• discussed in the URG section 3.1.3.4:

The first option (Section 3.1.3.4.1) discussed is "further refinement of fixed debris source terms."

This optionis simply a more detailed analysis of the containment for calculating the amount of debris that could reach the strainers in order to reduce the debris source term for sizing of the strainers. A simphfied calculation with bounding assumptions would provide a quicker answer requiring the least amount of engineering effort, but would tend to be very conservative. The more refined analysis discussed here would likely result in a lower debris source term, but would cost the licensee more in engineering time and effort than would using a simpkfied conservalve calculation.

The second option (Section 3.1.3.4.2) discussed is replacement of the existing strainers with attemate passive designs such as the stacked disk or star strainer designs. The concept with this option is to install a strainer which has a higher capacity for debris capture. Specifically, the strainer would have a lower head loss when compared against the existing strainer for the same debris loading. This option is the same as that suggested by the staff in resolution option 1 of NRCB 06-03 (Ref. 4).

The third opfion (Section 3.1.3.4.3) provided is to installJacketing to reduce the insulation debris source term. The URG states that BWROG test data has shown that metal Jacketed insulation generates less debris when exposed to a high pressurejet than its unjacketed counterpart. The jacketing should include an appropriate attachment mechanism to hold the jacketing in place.

The URG also states that use of the jacketing and attachment mech:mism *will significantly reduce the amount of fibrous debris which can be transported to the suppression pool and contribute to strainer blockage

• The URG also notes that the manner in which the jacketing is attached is critical.

The fourth option (Section 3.1.3.4.4) discussed in the URG is

• Reduction in Transient Debris Source Terms." This option is recommended by the BWROG if a small reduction in deb-is would be needed to demonstrate acceptable NPSH margin. To justify the reduction in transient debris, licensees may need to implement additional FME controls, housekeeping controls, or higher suppression pool cleaning frequency.

The fifth option (Secton 3.1.3.4.5) would be to pursue a licensing basis change with the NRC.

DRAFT to

t DRAFT wnw One example cited as a potentiallicensing basis change would be the use of a more realistic decay heat curve in lieu of the conservative curve used in the originallicensing basis (i e., ANS 5.1 versus May Wttt). This, in tum, woth reduce the calculated suppression pool temperaturm following a LOCA improving the calculated NPSH margin. The BWROG does not recommend crediting containment overpressure in calculating NPSH margins.

The sixth option (Secton 3.1.3.4.6) is to pursue reevaluation of ECCS suction line penetration loads without reopening the licensing basis for containment loads. The BWROG does not provide any guidance on how this can be done. The BWROG also does not recommend reopening the licensing basis for containment loads.

The seventh cption (Section 3.1.3.4.7)is partial replacement of fibrous insulation with reflective metallicinsulation (RMI). This option consists of selective replacement of fibrous insulation in the containment to reduce the ultimate amount of fibrous material that can get on the ECCS suction strainers. No further guidance on how to implernent this option is provided in the URG.

The eighth option is the installation of a backflush system. The BWROG does not recommend this option as a primary means of mitigating action because it would likely require operator action early during the accident increasing operator burden, and it would likely require repeated use within 30-60 minutes after the accident. It is recommended as a defense-in-depthmeasure.

The ninth option is installation of self cleaning strainers. lne BWROG notes however, that significant design, qualification, and surveillance issues would need to be resolved prior to implementationof this option. Further guidance on this option is discussedin URG section 3A.

Section 3.1.4 discusses some potential solWions other than these previously discussed. These options have not been fully evaluated by the BWROG, and therefore, would require additional evaluation and testing by licensees desiring to use them. The potential solutions discussed are all attemate strainer designs which are either large capacity passive strainers or strainors with a self-cleaning capabiIity.

Staff Evaluation for Section 3.1: Figure 1, page 8 provides guidance for plants with *miriimal" amounts of fibrous insulation. No definition of what constitutes a minimal amount is provided. This should be clarified by the BWROG to avoid confusion.

Figure 2 on Page 9 of the URG has a discrr.pancy between Section 3.2.1.2.3.1 and its associated step in the flow diagram. The text in the process diagram indicates that in performing a bounding analysis, all the material of interest is assumed to be debris and transported to the suppression pool.

This clearly would be the most conservative method for calcu . ting the amount of debris that could be transported to the suppression pool, and would eliminate the need for a detailed analysis.

, However, the staff notes that the flow diagram for this method cites Section 3.2.1.2.3.1 of the URG.

This section provides more detailed guidance on Method 1 for calculating debris generation. The text of this sectior, on page 37 indicates that the transport factors may be applied to the total amount of debris assumed in the drywell, it is not clear which is the intended URG recommendation for Method 1. It appears, however, that there are two methods being combined into one. It would be DRAFT 11

i DRAFT uom clearer if they were separated. The bounding case would be to assume all the potential debris material in the containment is dislodged and transported to the suppression pool during a LOCA.

A second method would be to assume all of the materialis generated as debris, but that only a fraction of that based on the use of a transport factor reaches the suppression pool. The intent of the URG guidance should be clarified.

In Section 3.1.2, page 14, the BWROG states, 'The head losses which resulted from the tests with combined RMI, fiber, and particulate debris were bounded by those seen in similar tests for the equivalentloading of fiber and particulate without RMI debris." This statementis incorrect. The staff evaluated the test data contained in Volume I of the URG Technical Support Documentation and concluded that the data did not support this conclusion. For example, test run number 15 was conducted with a 20 point star strainer at 2500 gallons per minute (GPM) with 3 pounds (lbs.) of NUKON insulation and 240 lbs. of corrosion products. The resultant head loss across the strainer was 50 inches of water (in. H,0). Test run number 33 was conducted with the same strainer at the similar conditions and debris loadings except for the addition of 88 square feet (ftr) of RMI debris.

Test run number 33 resulted in a head loss across the strainer of 115 in. H2 0. Another example would be shown by the results of test nin numbers J6 and J7 (with pump 2 mixing the tank for test run J7) for a truncated cone type strair. r at 5000 GPM. Still another example is shown in test runs J23 and J24 for the 60 point star strainer. The NRC conducted independent confirmatory testing with fiberglassinsulation, corrosion products, and RMI at the Alden Research Laboratory (ARL)in Holden, Massachusetts. The results of the staff sponsored tests are documentedin ARL test report entitled, " Head Loss of Reflective Metallic Insulation Debris With and Without Fibrous insulation Debrir, and Studge for BWR Suction Strainers," dated May 1996 (Ref 43). After analyzing the results of the ARL tests, the staff has concluded that the head loss for RMI combined with fibrous debris and corrosion products would be bounded by summing the numerical values of the: 1)the head loss across the strainer with a fibrous debris / corrosion product debris bed only, and 2) the head loss across the strainer for an RMI debris bso onif.

In the last paragraph on page 16, the BWROG provides a discussion which implies that utilities should take no' action until the URG review is complete. Although any change to that discussion would be a moot point with the issuance of this SER, the staff emphasizes that licensees should take whatever action they can to resolve this issue as soon as possible. Interim actions which help assure the licensees ability to mitigate a LOCA or a strainer clogging event are encouraged until such time as the licensee is able to fully implement its final resolution for this issue.

Section 3.1.3.1 discusses the evaluation of containmentload penetrationmargins. The URG states that a licensee's evaluation of containment penetration load margin shou ld "also identify any excess conservatism used in the existing structural analysis which could be reduced thr: Ugh the use of altemate analysis methods." The staff has concems regarding this statement. Specifically, tb hydrodynamic load programs for the lark I, Mark 11, and Mark ll1 programs were very specific in

,,providing generic methodologies for c,a ulation of hydrodynamicloads. Each plant submitted Pir '

Unique Analyttis Reports (PUARs)for staff review and approval. The PUARs provided the spe6 details of each plant's calculated hydrodynamicloads, including the methodology used, coefficis ,

used, etc. The staff notes that methodologieswhich were approved by the staff had supporting test data and conservative assumptions as part of the basis for acceptance. One main assumption is DRAFT 12

+

DRAFT mm that the strainers were treated as solid cylinders since testing had not been performed to determine the drag coefficientsfor the perforated strainers. Since the strainers are perforated, the drag on a sohd cylinder would bound the actual drag on the perforated strainer. The staffs review of licensee responses to NRCB 96-03 has shown that many licensees are changing the coefficients used for calculating the standard and acceleration drag coefficients used in calculating the hydrodynamic loads to take credit fur the fact that their new strainers are not solid cylinders. While the staff intuitively agrees that the perforated strainers have lower drag coefficientsthan similarly sized solid cylinders, it has not been demonstrated that the standard and acceleration drag coefficients of a perforated strainer can be accurately determined by analysis alone. The staff cautions that licensees making changes to hydrodynamic load calculation coefficients and/or methodologies should not do so without testing to demonstrate the validity of the revised calculations.

In Section 3.1.3.4 and its subsections, the URG discusses nine

• Resolution Options." Three of the options are consistent with the resolution options proposed in NRCB 96-03: installation of alternate passive design strainers, self cleaning strainers, or backflush. The other options discussed are not, in the opinion of the staff 'resolutionoptions." None of the other six options are sufficientto resolve this issue for any plant. Based on the staffs current knowledge of existing plant strainer designs and the deleterious phenomena which can affect head loss across the strainer and NPSH margin, the staff does not believe that any BWR licensee can justify maintaining their existing strainer design for resolution of the stroiner clogging issue. Rather, the other six ' resolution options

• discussed are better characterizedas potentiallicensee actions which could help licensees reduce the size of th:

strainer needed to resolve the issue. However, the resolution options lack substantial detail on the technical basis for these options and the specific detail information on how to apply each option.

For this reason, a licensee should resolve the staffs concerns for each option in a plant specific submittal prior to using the option to as a part of their resolution for the strainer clogging issue. The staffs specific concerns related to the

• resolution options' discussed in Section 3.1.3.4 are as follows:

In Section 3.1.3.4.2, page 20, the URG states that "When properly applied, the results of the extensive te' sting conducted by the BWROG may be applied to strainers other than those tested by the BWROG." The staff finds that there is insufficient information as tL sow to apply the BWROG test results to other strainers, in addition, the basis for this statement is not provided.

Without further guidance and justification, this statement by itself is not considered acceptable by the staff. For this reason, the staff recommends that licensees use vendor specific data based on the licensee's analyzed conditions as the basis for determining head loss across the strainer.

In Section 3.1.3.4.3, page 20, the URG states that " Additional details on the use of jacketing is provided in Section 3.2.1." No further guidance is provided in Section 3.2.1 other than information on debris generation. Section 3.2.1 also indicates the importance of banding to

, obtalning the desired effect of reducing debris generation; however, no guidance was provided on types of bands to be used, construction of the bands (e g., details as to size, thickness, material, weld versus riveted construction, etc ), and mounting instructions. Due to lack of information as to how a plant is to apply this option, the staff is unable to reach any conclusion regarding this option. In addition, the characterizationof the effectiveness ofinsulationjacketing DRAFT 13

. i I

t  !

l DRAFT wm on this page is inconsistent with the test report from the BWROG air jet impact testing (AJIT) ,

conducted at the Colorado Engineering ExperimentalStation, Inc. (CEESl). The AJIT test report is includedin Tab 3 of Volume 11 of the URG technical support documentation in Section 5.1,

'GeneralTest Conclusions,' page 181, the AJIT report states *In the case of f5ous insulation j materials, the use of jacketing as a means of reducing debris generation does not appear to be  ;

offective without the use of an additionalbanding materialwhich better secures the jacketing to '

the insu'ation assembly and the pipe " The wording in Section 3.1.3.4.3, page 20 of the URG contradicts the test conclusion by stating that " Testing performed by the BWROG (Reference

6) confirms that the fibrous insulation whir.h protected with n,otaljacketing is able to ,,urvive without producing debris at distances much closer to a pipe break than the same fibrous insulation without jacketing.* This statement implies that jacketing by itself provides an improvementwhich is contraryto the test report conclusions. The title of the section

• Installation of Jacketing to Reduce the Insulation Debris Source Term

• also implies that jacketing by itself provides an additional degree of reduction in debris generation. Without further detailed information on how to apply this option and the associated benefits achieved, and the technical basib supporting the use of this measure, the staff is unable to make a d9 termination an to the acceptability of this option.

In Section 3.1.3.4.4, pages 20 and 21, of the URG, the BWROG discusses taking credit for FME and housekeeping programs as a justificatim for crediting lower amounts of transient debris in a plant analysis. The staff has two concems related to incorporatingthis option into a licensee's final resolution. First,if a licensee selects an amount of transient debris in their analysis which is too low, then there is potential for opetability concerns to be raised anytime debris is found in the drywell, suppression pool, or wetwell. Second,if a licensee is considering an increase in transient debris in order to reduce the need for inspections for their new strainer, the staff believes that housekeeping controls or foreign material exclusion (FME) controls are not a substitute for periodic inspections of the drywell, the strainers or the suppression pool. Given the numerous events reported over the last few years related to FME issues, the industry has not demonstrated that FME controls alone are effective in ensuring that materials are not left in the drywell, wetwell, or suppression pool. As such, the staff believes that regular inspections of the suppression pool and ECCS suction strainers, and cleanings when necessary, should be conducted every refueling outage untillicensees have demonstrated over time the ability to control toreign materials. The staff believes it is more prudent to add margin when sizing strainers to account for the uncertainty in the effectiveness of housekeeping controls. This, in turn, will minimize the need for detailed operability assessments when small amounts of foreign material are found in the containment or suppression pool.

In Section 3.1.3.4.5, Page 21, the BWROG notes that a part of a licensee's resolution to the strainer clogging issue could include a licensing basis change. An example is cited where a licensee may wish to use a more realistic decay heat curve to reduce the calculated post-LOCA suppression pool temperatures. The BWROGy specifically states that use of credit for containment overpressurels not recommended? When discussing re' solving NPSH issues in a letter from the Advisory Committee on Reactor Safeguards (ACRS) to the NRC Executive Director for Operations (EDO) dated June 17,1997 (Ref 44), the committee explicitly stated, "We believe that allowing some level of containment overpressure is not an acceptable DRAFT 14

t DRAFT mm corrective action because adequate overpressure may not be present when needed." The staff concurs that additional containment overpressure (other than an amount already approved by the staff for the existing licensing basis) should not be used as part of the resolution of this issue. The staff is evaluating its position on use of containment overpressure in calculating NPSH margin as part of its review of Generic Letter g7-04, ' Assurance of Sufficient Net Positim Suction Head for Em.)rgency Core Cooling and Containment Heat Removal Pumps,' dated October 7,1907 (Ref 45). The staff notes that conservativelicensing basis assumptionsprovide safety margin for the plant. There is a substantial amount of uncertainty associated with the strainer clogging issue, and as a result, the staff does not recommenu licensing basis changes as a ' resolution option.'

In Section 3.1.3.4.6, page 21, the URG provides a discussion on the potential to reanalyze suction line penetration loads. Specifically, the URG states that it *may be possible to further increase the size of the attemate strainer by reanalyzing ECCS suction penetration loads and suppression pool strudural loads using more sophisticated techniques or through reduction of conservatism in current design basis calculatbns, but without reopening the licensing basis for containmentloads.' The staffs concems regarding Section 3.1.3.1 are stated above and apply to this section of the URG also.

Section 3.1.3.4.7 indicates that an option for reducing the fibrous debris source term in the drywell is to partially replace fibrous insulation with RMI. The staff agrees that this is an appropriate option for reducing the amount of fibrous debris which can be transported to the suppression pool; however, when using this option for a selected break, the licensee should go back and reassess other breaks to ensure that the licensee has not changed which break is the most limitingin terms of NPSH margin. In addition, the licensee must now consider head loss in terms of combined fibrous /RMIdebris beds. See the staff evaluation of Section 3.1.2 above.

In Section 3.1.3.4.8, page 22, the BWROG states that backflush is not recommended as a primary m,eans of ensuring ECCS flow to the core. The basis for the BWROG's recommendationis that backflush would place an unnecessary burden on operators early in the accident and would lik.,y require repeated initiations during an accident. The staff concurs with the BWROG that backflush is a more viable option as a defense-in-depth measure. The staff also believes that if used as a primary means of ensuring adequate ECCS flow, it should be combined with the installation of a large capacity passive strainer to maximize the amount of time before backflush initiation would be required.

See the staff evaluation of Section 3.4 of the URG for the staffs evaluation of the use of self-cleaning strainers as a resolution option.

The staff also notes that due to uncertainties associated with any resolution to this issue, a good practice would be to maintain defense-in-depth. The staff strongly encourages the enhancement of attemate water sources (including maintenance of crossover valves), operator training, and emergency operating proceduresto ensure that operators can mitigate any situation involving loss of ECCS flow due to strainer clogging. In its letter to the EDO dat.<d t'ebruary 26,1996 (Ref 46),

the ACRS stated that a diverse means of providing emergency r, ore cooling is desirable because DRAFT is

l I

DRAFT mm of the difficulty in predicting with confidence the amount of debris that would challenge the ECCS strainers. In its response to the ACRS dated March 22,1996 (Ref 47), the staff concurred with the ACRS comment and agreed to T! ace more emphasis on the need to improve, where appropriate, i the procedures to better address core cooling from attemative sources of water.'

3.2 METHODOLOGY FOR SIZING PASSIVE ECCS SUCTION STRAINERS 3.2.1 DrywellInsulation Debris Sources This section provides the BWROG recommended methodologies for calculating LOCA generated debris from piping insulation quantities. The guidance"ovided includes determination of the break locations to be analyzed, the zone of influence (ZOI) wr the break jet, and the destruction factors for piping insulations typically used in domestic BWRs. Section 3.2.2 discusses other potential -

LOCA generated debris sources.

32.1.1 Pipe Break Locations The BWROG guidance related to selection of pipe break locations is provided in Section 3.2.1.1.

The introduction to Section 3.2.1.1 provides general considerations for selecting pipe break locations. It also summarizesimportant guidance provided in 10 CFR 50.46 (Ref 42) and RG 1.82 (Ref 5). Guidancefrom Section 3.6.2 and Branch Technical Position (BTP) MEB 31 of NUREG-0800 (SRP)," Standard Review Plan for the R'eview of Safety Analysis Reports for Nuclear Power Plants"(Ref 48) are also included.

The URG correctly states that 10 CFR 50.46 (Ref 42) explicitly states that "ECCS cooling performanos must be calculated in accordance with an acceptable evaluation model and must be calculated for a number of postulated loss-of coolant accidents of different sizes, locations, and other properties sufficient to provide assurance that the most severe postulated loss-of-coolant accidents are calculated." When analyzing a plant for compliance with 10 CFR 50.46 (Ref 42), this statement provides the analyst with the guidance that must be met when choosing the break number and locations of the breaks to be analyzed.

The URG then provides a discourse on SRP Section 3.6.2 and BTP MEB 3-1 (Ref 48). On page 26, the URG quotes SRP Section 3.6.2 as stating " Acceptable criteria to define postulated pipe rupture locations and configurationsinside containment are specified in Branch Technical Position (BTP) MEB 3-1." The staff notes that this statementis taken completely out of context, and as such, does not convey accurately the position stated in the SRP Section 3.6.2. The statement quoted by the BWROG is preceded in the SRP by

• Specific criteria necessary to meet the relevant requirements of GDC 4 are as follows:' The URG then notes that NUREG 0897, Revision 1,

' Containment Emergency Sump Performance'(Ref 58), states *SRP Section 3.6.2, ' Determination of Rupture Locations and Dynamic Effects Associated with the Postulated Rupture of Piping,'

should be used to identity potential break locations." BTP MEB 31 is then quoted *...The rules of this position are intended to utilize the available piping design information by postulating pipe ruptures at locations having relatively higher potentialfor failure, such that an adequate and practicW DRAFT 16

[

..-. -- ~ - .- - .- -- -

a 1

DRAFT mnw consider the MLOCA when performing their plant analysis, if verified experimentally, the licensee would not need to analyze medium LOCAs with largest potential particulate debris to !nsulation ratio by weight.

Likewise, based on the staff's review of several plant specific submittals detailing licensee design bases for estimating maximum debris loadings on their new strainer design (see Refs 32,33,34, 35,36,37, and 38), it appears unlikely in most cases that large LOCAs with the highest part:culate to fiber ratios by weight will be the limiting break in terms of creating the worst case headloss across the strainers. Review of licensee submittals has shown that when using Method 1 or 2 for calculating the pipe break zone of influence, the amount of fibrous debris calculated is sufficiently conservative to bound the head losses in breaks with the highest ratio of particulates to fibrous debris by weight. However, the staff believes that in certain plant configurations,the potentialexists for debris beds with higher particulate to fiber ratios to have the bounding head losses. One such case may be in plants with large quantities of both Calcium Silicate and fibrous insulation materials.

The difference between the breaks with the largest amounts of fibrous debris and the breaks with the highest particulate to fibrous debris ratios may not be significant enough for the former to bound the latter. Therefore,the staff concludes that licensees should still evaluate large LOCAs with the highest particulate to fibrous debris ratios by weight to ensure that those breaks do not become more limiting in terms of head loss than the breaks that produce the highest volume of fibrous debris.

The staff notes that if the BWROG intends to include generic statements regarding all attemate strainer designs, then it should support the guidance with additional data either covering a wider range of experimental parameters (e.g., fiber volume, sludge to fiber ratios, ditferent strainer geometries, etc.) for all attemate designs that are actively considered by various individual licensees, or demonstratingwhich parameters are key in strainer design to ensure that breaks with high particulate to fibrous debris ratios are not the bounding breaks for the strainer design under consideration.

Ronclusions on Section 3.2.1.1: In surnmary, the staff concludes the following:

1) To be consistent with NRC requirements, the URG should give clear guidance that helps licensees select break sizes and locations that recsonab'y assure acceptable ECCS performance under the most severe conditions. URG Section 3.2.1.11argely, but not completely, succeeds in meeting this objective. The staff's concem deals with the clarity of the guidance and the unnecessary focus on SRP Section 3.6.2. The staff has concluded that use of SRP Section 3.6.2 for determining pipe break locations for demonstrating compliance with 10 CFR 50.46 is an inappropriateuse of the guidance in that section cf the SRP. There are two reasons for the staffs conclusion. First, SRP Section 3.6.2 does not provide guidance on compliance with 10 CFP 146, nor does it orovide acceptance enteria for demonstrating compliance with 10 CFR 50.w. The second reason is that the BWROG has not demoastrated that break locations selected consistent with SRP Section 16.2 would bound the worst case debris generation scenarios, and therefore. meet the intent of 10 CFR 50.46. In fact, in the case of one plant response to NRCB 96-03, it was leamed that the breaks in Sect'on 3.6.2 were att adequate to bound the worst case debris generation scenarios.

DRAFT 20

l DRAFT mww efficiency (entrapment of particuates by the strainer and fibrous debris bed) increases resulting in a significantly higher head loss than in debris beos containing significantly higher quantities of fibrous debris.

In URG Sections 3.2.1.1.1 and 3.2.6.2.3,the BWROG states that a thin-bed effect was not observed for alternate strainers; therefore, licensees that propose to use attemate strainers need not analyze medium breaks or large breaks with the largest potentialparticulate debris to insulation debris ratio by weight. The staff identified coricerns with this statem3nt because it appears that the data provided in the URG is not complete enough to support this concuon. There are three rasons for this. First, all of the data provided is for two types of alternate strainers (star and stacked-disk) only, and it is evident that not all licensees will use only these two attemate strainer designs. The second reason is that other

• alternate

• strainer designs may not have the entical features that eliminate the ' thin bed effect.* The third reason is that even for these attemate strainer designs that were tested, the data set is too limited to support such a broad conclusion.

Because of the staff's concerns, a confirmatory analysis was performed in suppurt of this review.

The objective of the confirmatory analysis was to evaluate the BWROG guidance which states that licensees proposing to use alternate strainers can screen out medium and large breaks that produce the largest potential particulate debris to insulation ratio by welght. The staffs confirmatoryanalysis is documented in Appendix E. The analysis concluded that while the URG Statement that a high head las will not occur with debris beds with the highest particulate to fiber ratio by weight in altemate strainer designs is insufficiently substantiated because the reported data are for two strainer designs (stacked disc and star), and the data address a narrow range of experimental parameters which are insufficient to support the BWROG conclusion. However, the staffs analysis of the existing data does suggest that licensees may screen out medium LOCAs (MLOCAs) if they use the stacked disc strainer number 2, the star strainer, or another alternate strainer design that has large cavities for debris entrapment similar to these strainers. The basis for the staff's conclusion is that the test data provided in the URG do demonstrate that with fiber loadings up to the equivalent amount to fill the crevices in the stacked disk strainer number 2, high head losses were not experienced even with high sludge to fiber mass ratios. The amounts of debris that wculd be required to completely fill the cavities between the disks in the stacked disk number 2 strainer would be approximately10 ft3 This amount of fibrous debris is similar to the amount estimated for a medium break LOCA for the reference plant in NUREG/CR-6224 (Ref 31). Therefore, the staff has concluded that licensees may screen out MLOCAs in performing their plant specific analysis if they intend to install a strainer similar to the stacked disk number 2, star strainer, or another geometrically similar strainer with deep crevices sufficient to hold debris loadings consistent with a MLOCA for their plant. For these types of strainer designs, the MLOCA does not appear to be the bounding break. However, the staff notes that its conclusion is based on limited amounts of test data, estimates of debris loadings for the reference plant MLOCA only, and evaluation of two strainer designs only. Because of this, the staff believes that licensees considering a solution option using an alternate strainer design should experimentally verify (or ensure using existing vendor or BWROG data) that the thin-bed effect is not an issue for the strainer design being considered ulng i debris cuaniities which are consistentwith a MLOCA for their olant. Otherwise, the licensee should DRAFT ig

t DRAFT nm concluded in SRP Section 3.6 2 that an adequate and practical level of protection for compliance with GDC 4 could be provided by physically protecting equipment important to safety from the postulated pipe breaks that have a relatively higher potential for failure (e.g., postulated failures at high stress and fatigue locations). As a result, pipe breaks are analyzed based on pipe stress analysis methodologies similar to that provided in BTP MEB 31 of SRP Section 3.6.2 Whc0 kmonstratino comokante with GDC 4. Licensee safety analysis reports are evaluated against the requirements of SRP Section 3.6.2 to " confirm that requirements for the protection of structures, systems, and components relied upon for safe reactor shutdown or to mitigate the consequences of a postulatedpipe rupture are met.' The only acceptance crite:ia specified in SRP Section 3.6.2 is cumpliance with GDC 4. GDC 4, SRP Section 3.6.2 and BTP MEB 31 are not to be used for ECCS functionaldesign or compliance with 10 CFR 50.46. The staff considers SRP Section 3.6.2 and BTP MEB 31 to be inappropriate for demonstrating compliance with 10 CFR 50.46.

The primary concern in the ECCS strainer clogging issue is the adequacy of ECCS strainer design celative to ensuring that the ECCS can meet the cooling performance requirements of 10 CFR 50.46. The staff notes that no concerns have been identified by the staff during the resolution of this issue relative to physically protecting the ECCS from the dynamic or environmental effects of a LOCA. In order to ensure adequate ECCS cooling capability,10CFR50.46 requires that ECCS cooling performance "must be calculated for a number of postulated loss-of coolant accidents of different sizes, locations, and other propertic* ffficient to provide assurance that the most severe postulated ioss-of-coolant accidents are calculated." The staff notes that the worst breaks from the standpoint of peak clad temperature may not be the worst from the standpoint of NPSH margin. As a result, additional breaks may have to be evaluated. When evaluating ECCS performance for compliance with 10CFR50.46, SRP Sections 6.3 and 15.6.5 are the appropriate sections of the SRP to consider, For instance, the review procedures of 15.6.5 specifically require reviewers to ensure that 'A variety of break locations and the complete spectrum of break sizes were analyzed." This is consistent with 10CFR50.48. RG 1.82 provides the staff's position on what constitutes an appropriate specirum of breaks to be considered when evaluating the potential for strainer clogging.

Regulatory Position 2.3.1.5 of RG 1.82 states, 'As a minimum, the following postulated break locatbns should be considered. (a) Breaks on the main steam, feedwater, and recirculation lines with the largest amount of potential debris within the expected zone of influence, (b) Large breaks with two or more different types of debris within the expected zone of influence,(c) Breaks in areas with the most direct path between the drywell and wetwell, and (d) Medium and large breaks with the largest potentialparticulate debris to insulation ratio by weight." The staff believes that the RG provides the complete scope of breaks needed to meet the intent of 10CFR50.46.

Criterion 4 of Regulatory Position 2.3.1.5 of RG 1.82 states that one group of break locations a licensee should identify as a minimum are

• Medium and Large Breaks with largest potential particulate debris to inselation ratio by weight.' The reason these breaks should be considered is the " thin-bed effect" wMch has been observed in cylindrical and truncated cone strainer tests.

Specifically,it has been observed that high head losses can occur on cylindricalor conical strainers with thin fiber beds and high concentrations of sludge. The reason for this is that with thinner fibrous debris bedt. the bed compresses more tightly to the strainer surface. When this occurs, the filtration DRAFT 18

DRAFT www level of protection may be achievedJ These quotes are provided by the BWROG in the URG to '

form a basis for the use of SRP Section 3.6.2 for defining the pipe break locations which should be evaluated for demonstratingcompliancewith 10 CFR 50.46 (Ref 42). RG 1.82, Revision 2 (Ref 5) guidance for pipe break locations to be analyzed is also provided in Section 3.2.1.1.1 of the URG.

The URG quotes the following guidance out of Section 2.3.1.5 of RG 1.82, Revision 2: .

• As a e Nmum, the following postulated break locations should be considered.

Breaks on the main steam, feeowater, and recirculationlines with the largest amount of potential debris within the expected zone of influence, '

• Large breaks with two or more different types of debris within the expected zone of influence, i e Breaks in areas with the most direct path between the drywell and wetwell, and '

Medium and large breaks with the IMgest potential particulate debris to insulation ratio by weight.'

The BWROG does not consider the fourth critorion dealing with the largest particulate to insulation ratio to be applicableif using an allernate strainer design. The basis to this conclusionis presented in Section 3.2.6.2.3 of the URG, After providing the above information as background, the BWROG guidance for determining pipe break locations is provided in Section 3.2.1.1.2, items 1,2, and 3 of this section present alternatke approaches for selecting pipe break locations. Item 4 cautions licensees about differentiating between pipe break locations used for ECCS evaluations and those which are in the plant licensing basis. Item 5 gives guidance regarding the treatment of pipe breaks inside the bio shield walls.

Item 1 gives pipe break selection criteria for three different licensing basis situations, item 1.a provides pipe break selection cnteria for plants that were licensed with an MEB 31 analysis. Item 1.b provides pipe break location selection enteria for licensees who have not performed a stress analysis consistent with MEB 31, but may desire to perform such an analysis. Item 1.c discusses the pipe break selection enteria for plants who have not identified pipe break locations using an saproved stress analysis technique.

Sjaff Evaluallon for Sution 3.2.1.1: The guidance provided in this section was compared to the corresponding guidance provided in RG 1.82, applicable sections of the SRP, and with i ! CFR 50.46.

The focus by the BWROG on the postulated pipe break locations to be analyzed by a licensee is on breaklocationswhich were enalyzed to demonstrate compliance with 10 CFR 50, Appendix A, General Design Criteria 4 (GDC 4). GDC 4 (Ref 42) requires licensees to protect structures, systems and components important to safety from the dynamic (e.g., pipe whip, direct steam jet impingement, etc.) and environmental (e.g., temperature, pressure, radiological effects) effects of postulated pipe ruptures. GDC 4, therefore, places a requirement on licensees to provide physical ,,

protection (e.g., pipe restraints, physical barriers (walls) or physical separation) for equipment important to safety from the effects of postulated pipe breaks and to ensure that those components are capable of performing their safety function in a post-LOCA environment. Because it is not practicalto physically protect equipmert in the containment from every postulated LOCA, the staff DRAFT 17 i

I i

a DRAFT imm

2) The guidance of 10 CFR 50.46 and RG 1.82, Regulatory Position 2.3.1.5,is adequate to provide licensee'swith suffcient ir.folmation regarding selection of pipe break locations for performing a plant specific analysis relative to debris generation. This guidance more concisely indicates where a licensee should focus their analytical efforts.

3) Licensees may screen out it,OCAs in performing their plant specific analysis if they intend to install a strainer similar to the stacked disk number 2, star strainer, or another geometrically similar strainer with deep crevices sufficient to hold debris loadings consistent with a MLOCA for their plant. However, licensees should still evaluate large breaks with the highest particulde to fiber ratio by weight.

In addition, the staff notes:

1) Each alternative evaluation method presented must provide reasonable assurance that the most severe ECCS suction strainer debris loadings (from drywell insulation sources) be suitably evaluated;

2) Bounding analyses, if used by licensees in lieu of specific pipe break location analyses, need to address all debris species (from insulation sources) which have not been addressed by the specific pipe break location evaluations.

3) The URG guidance on pipe breaks located inside the bio-shield wall is incomplete. This comment was forwarded to BWROG whose response is documented in Reference 11. Citing utility responses documentedin Ref.12, the BWROG argued in reference 13 that breaks inside the bio-shieldwall are not major contributors of debris. And as a result, the BWROG does not believe generic guidance is needed on this issue. However, the staff believes that additional guidance on the analytical considerations to be evaluated by the licensee would be beneficial and consistent with the BWROG's stated goal of providing a consistent industry response to NRCB 96-0,3.

3.2.1.2 Zone of influence URG Section 3.2.1.2, entitled

• Zone of Influence," documents BWROG guidance on various options for selection of a zone of influence model that can be used by an individual utility to estimate the volume of debris generated by a postulated pipe bresk. The URG guidance provided in this section is based on the following supporting analyses:

1) " Zone of Influence as Defined by Computational Fluid Dynamics,' Rev. 3, URG Technical Support Documentation, Vol. II, Tab No.1, Continuum Dynamics Inc., Princeton, NJ.

2) " Air Jet Impact Testing (AJiT) of Fibrous and Reflective Metal Insulation,' Rev. A, URG Technical Support Documentation,Vol. II, Tab No. 3, Continuum Dynamics Inc., Princeton NJ.

3)

• Evaluation for Existence of Blast Waves Following Licens.!,g Basis Double Ended Guillotine Breaks,' URG Technical Support Documentation, Vol.111, Db No.13. O .neral Electric Nuclear Energy, San Jose, CA.

DRAFT 21

t DRAFT mm

4)

• Total Pressure Topography and Zone of Destruction for Steam and Mixture Discharge from Ruptured Pipes,' URG Technical Support Documentation,Vol. Ill, Tab No.14, General Electric Nuclear Energy, San Jose, CA.

The staff reviewed this section along with the supporting analyses listed above to judge completeness and accuracy of the guidance provided. In addition, the staff compared the guidance provided in the URG to that provided in RG 1.82, Revision 2.

Section 3.2.1.2 provides four options (or methods) for selectireg se zone of influence oven which insulation would be damaged by the LOCAjets. Each method reduces the amount of conservatism relative to the previous method, but requires a more rigorous analysis by the licensee. The metnod selected is based on licensee resources and the amount of conservatism the licensee desires to build in to its strainer design. Method 1 is the most conservative method. It bounds the zone of influence (Z01) by assuming that entire drywell is the ZOI. Methods 2 and 3 define a 201 by determining the spatial volume enveloped by a specific damage pressure of interest for a jet expanding in free space and mapping a sphericalzone of influence of equal volume surrounding the break. All the insulation contained within that spherical volume is then assumed to be damaged'.

Severalimportant differences exist between Methods 2 and 3, but they are essentially refinements that reduce the amount of conservatismintroduced by assumptions associatedwith Method 2. For example, using Method 2, the zone of influence would be defined assuming full separation of both ends of a double-endedguillotine break (DEGB). On the other hand, Metlod 3 allows the licensee to take credit for pipe restraints, and to evaluate axial and radial offsets consistent with the restraint Finally, Method 4 allows the user to directly employ the results of the ComputationalFluid Dynamics (CFD) modeling effort to define the zone of influence.

The BWROG guidance allows flexibility for licensees to select a model option that best suits its solution, and use it to construct the zone of influence surrounding each break location selected for analysis. All the insulation contained in nat zone of influence is assumed to be

• damaged.'

Ucensees that wish to take no credit for debris retention in the drywell, can put all the ' damaged

• Insulation into a single category and track it as it is transported to the ECCS strainer via the suppression pool model of interest. On the other hand, licensees wishing to take credit for debris retention in the drywell divide the total

• damaged

• insulation into two bins: a) debris generated above the lowest elevation grating, and b) debris generated below the lowest elevation grating. Such a division is required to facilitate application of drywell transport factors that are dependent on the location of insulation with respect to tru grating. This section does not address the size distribution of the ' damaged' debris. Such informationis provided in Section 3.2.3,'Drywell Debris Transport.'

., 4

' Not all of it is in transponable forrn. This is an important distinction.

DRAFT 22

I DRAFT www  :

Staff Evaluation of Section 3.2.1.2. The staff's confirmatory analyses performed in support of the '

review of this section are documented in appendices A, B, C, D, and F. A summary of key >

conclusions from these confirmatory analysee are as follows:

l

1) The BWROG calculations underestimate bulk dynamic pressures in Mark I containments by a factor of 10 at some locations. However, the staff's analysis concludes that the bulk dynamic pressures would likely be insufficient to damage insulation materials such as those present#y used in operational BWRs, if the insulations are properly installed and well-maintained. The tables in the URG for debris generation and transport assums that insulation is correctly installed and well maintained. Licensees should ensure that the analysis is reflective of the actual plant .

conditions. In addition, the URG calculations for bulk dynamic pressure in a Mark l containmert should be updated to be consistentwith the BWROG's response to staff RAls in Reference 13.

The BWROG's response showed that the bulk dynamic pressures in a Mark 1 and a Mark lli could be 510 times higher than that calculated for a Mark ll due to smaller containment cross sections.

2) The BWROG basis for using thejet center line (JCL) pressure at the distance from the jet nozzle (UD) at which incipience of insulation damage was first obsetved in the air jet impact tests as the insulation's characteristicdamage pressure is inadequately supported by either analysis or 1 experimentaldata, in the staff's opinion, the experimentalevidence from the air jet impact tests seems to contradict the JCL concept. Rather, the staff's review of the data leads to the conclusion that the incipience of damage is related to the totaljet impingement load rather than the local maximum, in addition, this intuitivelymakes more sense. In the staff's opinion,it does not seem logical that a peak pressure at one point on the insulation blanket would be responsible for destruction over the entire blanket where other portions are clearly exposed to lower pressures than the peak. The staff also notes two additional concerns related to use of the JCL. First, the choice of the JCL to characterize damage raises questions related to scaling BWROG airjet impact test results to BWR drywells. And second, interpretation of JCL as the damage pre,ssures, and their subsequent usaga in enveloping the volume over which debris damage occurs, makes the volume estimates for zone of influence non-conservative and in-consistent. For all of the reasons described above, it is recommended that P. values listed in Table 2 of URG be revised with more logical values. The staff believes that a target area averaged pressure or jet impingement load are more technically correct bases for assessing potentialdamage. If this approach is not taken, the BWROG should provide a better rationale for selecting the localjet centerline pressure at the UD location where incipient damage was first observed as the fundamental property that uniquely controls insulation damage.

The staff's concems over the use of JCL pressure can be characterized in the following way.

The BWROG logic of assuming that damage is only related to JCL pressure would lead to the conclusion that an insulation blanket that survived at 10 UD in the airjet impact tests would also

, , , , survive at 10 UD if the 3 inch nozzle is replaced by 12 inch nozzle. H for this conclusion. Similarly, there would be no need for the suggested,owever, the correctionformula(Pb a

= P% *(12*+2t)/ (D+21)) if damage is only related to the JCL (or alternately to the maximum pressure an insulation blanket is subjected to). The BWROG's rationale for extrapolation of AJIT experimentairesults to pipes of different diameters is unclear. The BWROG should develop a DRAFT 23

, , , - , + , . , ,

1 DRAFT ner i

logicat rationale which would also resolve the staffs scaling concern (desenbed in Appendix B) relative to applying the BWROG air jet impact testing results to full size BWR drywells. '

3) The suggested BWROG correction formula for scaling damage pressures measuredin the AJIT facihty for insulation blankets installad on 12' pipe to pipes of different diameters is reasonable. '

The staff notes that this formula also supports choice of a target area averaged pressure in place of the JCL suggested by the BWROG.

4) The CWROG estimates of volume of a freely expanding steamjet that is bounded by a pressure isobar of interest are conservative and well supported. Typically, the BWROG estimates of jet volumes are higher than or equal to ANSI /ANS 58.2 model estimates for both extremes: a)

DEGB with full separation and b) DEGB with limited separation. Therefore, the staff believes that the je'. volumes listed in Table 1 are reasonable.

5) The BWF.OG suggested correction factors that can be used to scale the jet volumes computed for steam breaks to recirculation breaks appear reasonable. Therefore, the staff believes the use of these correction factors to computejet volume bounded by a pressure isobar of interest following a recirculation line break is acceptable.

6) The BWROG choice of mapping a spherical zone of influence surrounding a postulated break of volume equal to the volume contained within a damage pressure surface of interest fnr a freely expandingjet is unsupported either by analyticalmodeling or experimentalevidence. The BWROG's rationale, however, appears logical (although qualitative). Nevertheless, the URG claim that zone of influence defined by this method is " conservative' is unsubstantiated.

7) The ZOI developed using Methods 1,2, and 3, and subsequentiy the volumo of insulation assumed damaged and available for drywelltransport,are very large. For some insulations,the ZOI enveloped between a fourth to a third of the drywell. As an example, the zone of influence computed fqr steel Jacketed NUKON' using Method 2 is a spherical region approximately 20 break diameters in radius, which is much larger than the 7D sphere used in the NUREG/CR-6224 study as well as the NUREG-08g7, Revision 1 guidance. The smallest zone of influence is greater than 10D in radius.

The strengths of URG Section 3.2.1.2 and the associated technicalsupport documentation are that-The method used to develop the zone of influence models !s, for the most part, sogical and is supported by a large number of analytical and experimental studies.

The airjet impact testing provides valuable experimental data related to the destructive nature of expandng high-pressure Jets. Such data were previously unavailable in the public domain, which limited the,, previous understanding of insulation destruction by high energy jets.

e The URG provides important considerationslicensees should be aware of in selecting the zone of influence model and how to track the damaged insulation.

DRAFT 24

DRAFT www The weakness in this guidance is related to the development of the Z01 models. The development of 201 models are based on two hypotheses which are not supported in the URG by either experimental data or analytical studies. These hypotheses are:

e in air jet impact tests, damage suffered by an insulation blanket is related only to the JCL pressure and is independent of the radial pressure distribution or the target area averaged pressure, i

e A spherical zone of influence with a volume equal to the volume of the double-ended conical zone of influence can be used to

• conservatively" bound the complex impacts of drywell congestion and drywelllayout on jet expansions from a DEGB, Two other weaknesses were also noted by the staff: 1) the first is that the supporting documentation does not support a scaling analysis demonstrating how the exparimental damage generated by jets originating from a 3 inch diameter nozzle can be scaled up to BWR drywells where the jet stagnation diameter is as large as 24 inches, and 2) the experimental data for certain insulations is not very comprehensive, since testing was conducted for only a limited number of L/D values. .

The staff believes that use of the first hypothesis can lead to non-conservative predicted ZOls for insulationswith high P.,, Specifically,this assumption hat the potentialof making zone of influence '

models developed for Darchem DARMET* with Camloc' Strikes and Latches, Transco RMI, Jacketed NUKON* with "Sure-Hold" Bands, Diamond Power MIRROff with *Sure-Hold' Bands, and Calcium Silicate with Aluminum Jacketing insulations considerably smaller than would be calculated with average target area pressure. On the other hand, its (JCL) impact on other types of insulation would be minimal because the P.,,, for the other insulations tested occurred at relatively far distences from the break where the JCL is not significantly different from the average pressure on the target insulation.

The impact 01 de second hypothesis wab not readily clear to the staff. As a result, a confirmatory analysis using a limited CFD model was conducted to demonstrate the effect of the jet interaction with structures and piping in the drywell. The analysis demonstrated the diffusion of the break jet as it interacts with structure and piping. The staff concluded based on this analysis that the sphericalIOls developed using methods 2 or 3 in Section 3.2.1.2 of the URG are conservative and acceptable. The basis for the staff's conclusionis that the jet emanating from a broken pipe williose energy and diffuse with distance and its interaction with surrounding pipes and structures, it will

- likely diffuse into a high velocity flow within 5 10 UD from the nozzle as it interacts with various structures. The staff notes also that if the jet were not interacting with the surrounding structures (i.e., it was on a path which allowed free expansion),then debris would not likely be generated. The staff concurs with the URG recommended use of a spherical model as the best means for accounting for the impact of drywell congestion, drywell teructural interactions, and the dynamic effects of pipe separation. The staff did' identify one additional $oncern related to the use of Method

3. Method 3 allows the licensee to ' determine whether the break results in a single jet (such as a steam line break) or doublejet (such as for a recirculationline break)." The staff points out that until the MSIVs close (about 0.3 0.:., s) fluid is expelled from both sides. The jet may not be as

-DRAFT 25 l

1 . . . . -- . -. . - . .

DRAFT mm energetic due to the effect of the flow restrainer limiting the flow (i.e., choking occurs in the restrainer). Licensees should pay close attention to accounting for these phenomena if they wish to take credrt for a single jet following a MSLB The staff recommends that utihties not take credit for a single jet in the case of a main steam line break inside the containment. Since this is a design basis accident, licensee assumptions should be consistent with past design basis scenarios in their Updated Final Safety Analysis Report (UFSAR).

With regard to the limited data sets for certain types of insulations, the staff notes that in the AJIT testing, the exact location at which damage first occurred was not pwperly explored for certa;,, types of insulation. For example,in the case of stainless steelJacketed NUKON*, damage was reported at 50 UD with 12% of insulation destroyedinto fines and 29% into largcr pieces. But the BWROG has not explored damage beyond 50 llD. Similarly,in the case of unjacketed NUKON'significant damage occurred at 60 l/D and no damage at 119 L/D. No data points were reported in between.

in view of such lack of data, some of the conclusions stated in URG become questionable, especially considering that this damage information is later on used to estimate fraction of fines generated. Another deficiency of AJIT testing is that for selected insulation blankets, the JCL pressures at which damage was first noted may be overestimatedin the URG. This would lead to non-conservativeZOls. For example,in the case of DPSC Mirror with Sure-Hold bands, the URG reported a damage pressure of 100 psid. But using the same experimentaldata and same NPARC computer code results (See URG page 40, Figure 9), the value was estimated to be 150 psid in the staff's confirmatory analysis. Several such discrepancieswere found, and all of them are annotated in Table 1 of Appendix B in this SER by an "" in all cases, the estimated damage pressures are lower than the URG values. The staff recognizesthat the BWROG's choice of limiting the number of airjet tests is resource related. As a result, the staff believes that utilities having materials where the data are kmited should conduct additionaltesting for those materials unless they use a bounding approach to those materials.

EQaduzions for Section 3.2.14 Overall,the staff concludes that Method 1 is clearly a bounding and conservative method for evaluating the ZOI since it encompasses the entire drywell, and therefore, bounds the amount of insJlation debris that may be generated by the break. The staff believes that the spherical 2Ols developed by using Methods 2 and 3 would be sufficiently large to envelop the entire zone over which destructionwould actually occur. These two methods are then sufficiently conservative to compensate for the weaknesses noted above, and therefore, are considered acceptable by the staff for use o.) insulations with low P., values. For the insulations noted above with high P., values, t'ie staff recommends that the ZOI be developed based on the

• target area averaged pressures"instead of JCL pressures.

Section 3.2.1.2 of the URG did not provide detailed guidance on Method 4. This method would use an unnamed CFD code to determinethe 201. Because of the lack of detail provided on this method, however, the staff cannot at this time accept Method 4. Any licensee desiring to use method 4

,, sh,ould provide a plant specific submittal to the staff which includes the details of the analysis and .,

how the code will be benchmarked. The staff believes that CFD code's are useful in obtaining insights, such as in the staffs evaluation of the interaction of a breakjet with surrounding piping and structures. It has not been demonstrated, however, that a CFD code can accurately predict the specific zone of influence for a pipe break or the amount of debris transport to the suppression pool.

DRAFT 26

l DRAFT mw The purpose of Mr* hod 4 is to reduce the conservatismin the calculation of a ZOI. Because of the uncertainties involved in a LOCA and the plant specific nature of debris generation and transport in a LOCA, this method would require a much higher level of review in order for the staff to accept it. However, at the present time, the BWROG has not provided sufficient detail for the staff to reach any specific conclusions re!ative to the adequacy of using a CFD model for the purpose of determining the 201 for a pipe break. As & result, Method 4 is not considered acceptable by the staff without further detailed justification on a plant specific basis.

3.2.2 Other Drywell Debris Sources The focus of this section is on quantificationof sources of debris in the drywell other than that from pipe insulation. The introduction to Section 3.2.2 reviews guidance provided in Regulatory Guide 1.82, Rev. 2, regarding such materials. The URG breaks the other drywelldebris sources down into three types of potentialdebris: transient debris, fixed debris, and latent debris. Transient debris is defined as non permanent olant material brought into the drywell, typically during an outage.

Examples of transient debris melude tools, rags, and temporary filters. Fixed debris is debris which is part of the plant, and which only becomes debris during a LOCA. Paints and coatings which are delaminated from the coated surface by direct steam impingement from a pipe break is an example of fixed debris. Latent debns is debris which appears after prolonged exposure to a LOCA environment such as an unqualified coating which is not directly impinged by the LOCA break Jet, but which subsequently falls due to prolonged exposure to the temperature, pressure, and radiatim of the post LOCA environment.

Cautions are provided in the URG to remind users that their foreign material exclusion (FME),

housekeeping,and inspection programs must be adequate to assure that the quantities of each of these types of materials do not exceed the quantities assumedin their evaluation of ECCS strainer loading.

The URG used the following supporting analysis in developing the guidance for this section:

'Performanceof ContainmentCoatings during a Loss of Coolant Accident,"Bechtel Corporatiort Vol. Ill, Tab 12, of the URG Technical Support Documentntion.

The URG provides the following recommendations for "other drywell debris sources" when performing a plant specific analysis for sizing the strainer:

1) Dirt / Dust: The URG suggests that the licensee assume a value of 150 lbm in the strainer head loss correirHon. This value is based orsngineeringjudgement. The URG states that this is believed to be a conservative value and includes consideration for cor' crete dust generated by breakjet impingement on the containment boundaries. If a lower value is ustd

• , by a licensee, the URQ states that they should document the basis for the value used including any programmit'ic co'ntrols which would help maintain the value within limits.

2) Other transient debris: No specific values are provided in the URG for other transient debris.

This judgement is left up to the individual licensees based on their own baseline DRAFT 27

DRAFT mm assumptions. The URG states that inspection frequency should be based on the amount of NPSH margin. Smaller NPSH margins requiring, in the BWROG's opinion, greater inspection frequency.

3) Rust from unpainted steel surfaces: The URG recommends a value of 50 lbm based on engineering judgement. Again, the URG states that this is believed to be a conservative value and includes consideration for

• unpainted steel surfaces in the drywell, the main vents /downcomers,and those in the suppression chamber above the pool level which may be swept as a result of LOCA-induced swell. If a lower value is used by a licensee, the URG states that the licensee snould document the basis for the value used including any programmatic controls which would help maintain the value within limits.

4) Particulate debris sources: No specific guidance is given. Licensees are cautioned that other sources of fired particulate debris may be present in the drywell on a plant-specific basis.

5) Paints / Coatings: Based on the Bechtel report cited above (Ref 51), the URG recommends the following

• bounding" values for coatings be assumed as available for transport to the suppression pool. The recommended values are 47 lbm for inorganic zine coatingt., 85 lbm for inorganiczine topcoated with epoxy, and 71 lbm for 100% epexy coating. The coatings are assumed to be in the ZOI of the steam jet from the break.

6) Concrete: No specific recommendations are given other than to evaluate the potential for concrete debris if the Feensee is not using the value for dirt / dust noted above.

7) Unqualified / Indeterminate Paint /Coatiny: No specific guidance is provided other than to determine the quantity that can be avoitable to transport to the suppression pool. As an attemative, the URG notes that licensees may remove the unqualified or indeterminate coatings, or attempt to qualify them through in-situ qualification, No guidance on how in-situ qualificalion is accomplished is provided.

8) Other Latent Material: The licensee is cautioned to consider the possibility of other latent materials to become debris from exposure to the LOCA environment. An example of the kinds of materiais that might be of concem in this category include adhesive backed tags which have en unqualified adhesive.

Staff Evaluation of Section 3.2.2: Although much of the guidance in this section is general, the guidance appears adequate for cautioning licensees about considerations needed for an adequate

'ECCS analysis. However, the s'.aff believes the guidance in this, section should be used with caution, especialiy when considering plant operation relative to strainer design. For example, the URG states pn page 50 that ' Licensees should recognize that the rigor of FME programs should be

. adequate to ensure that the transient debris sourca term used as a design input value in the strainer sizing calculationsis not exceeded. Consequently, licensees should considerthe trade-off between operationalflexibility and strainer size when establishng the values of drywell debris considered to be available for transport to the suppression pool." Wnile this statement is accurate as far as the DRAFT as

1 1

DRAFT wnw

• iCCS suction is concerned,it lays open a potentiallylarger problem with FME FME is not lust an ECCS suction strainer issus. The FME problem also deals with materials being left in piping systems and system components such as valves (See Refs 49,50,59,60 and 61). Licensees should also consider other potential problems caused by poor or inadequate FME programs, notjust  ;

the potentialimpact on suction strainers. The staff believes that FME should always be maintained at the highest level possible when the potential exists to impact a safety system. Providing margin in strainer size to allow for breakdowns in FME controls is prudent to avoid operability concerns; ,

however,it ir no way relieves the licenset, of their responsibilities in foreign material control. As a l

result, the staff considers the following statement on page St *,he URG to be unacceptable,' Note that where use of an adequatelyconservativevalue is ass .hc for a partiet ardebris species (e.g.,

dirt / dust), periodic inspection are not necessary." It is the vu opin%n that periodic inspections are Alway 1 necessary regardless of any baseline assumptions used in the ECCS analysis. The interval between such inspections should only be adjusted based on actuallicensee performance and the amount of conservatism in the base assumption. Similarly, the staff notes the same problem with the following statement from Page M of the URG: "A larger NPSH margin should require less frequent inspection / audit than a smaller margin

• NPSH margin is not the only characteristic of significance when evaluating inspection and maintenance intervals. If licensee performance does not demonstrate effective FME controls, then inspection intervals must be increased. In addition, the potential for introduction of debris since the last inspection should also be a consideration in determining in=pection scope and interval.

The staff notes that licensees should make every effort to ensure that the guidance in Section of 3.2.2, or any other section of the URG, is applicable to their plant before implementingthat guidance on their plant.

The strengths of Section 3.2.2 are that it reviews important considerations licensees should be aware of in assessing debris sources other than pipe insulation. It also provides some quantitative suggestionsfor the amounts of debris that may exist in drywells and suppression chambers. The discussions emphasize that forciqn material exclusion, housekeeping, and inspection programs must be capable of keeping debos quantities within the limits assumed in their evaluation of the ECCS strainer loading from such cebris.

However, the staff believes that Section 3.2.2 has some teen.3alshortcomings. First, some of the guidance is too vague to be used in a consistent generic manner. Second, important assumptions made regarding debris quantities are not supported by technical analysis or plant observation.

While the staff does not believe these judgements are inappropriate,there is still concem on the lack of adequate justification for the assumptions recommended. Based on this, the staff offers the following specific comments:

Dirt / dust: The discussion in Section 3.2.2.1 on dirt and dust states that only a small fraction of l

dirt / dust not direc"y. exposed to the LOCAjet or impacted by break leakage flow would be expected to be transported by the break fTow or due to sprays if containment sprays are used. The statement that only a small fraction might be transported to the suppression poolis not supported by test data or actual observation. Consideration should be given to the fact that steam / air velocities may be high thioudut the drywel DRAFT 29

i. ,

1 i

. DRAFT w3ow .

. following a large break 1.OCAr Also, steam con:lensation and run-off on structures d

could contribute.to entraining dirt / dust in the runoff. Once reaching the floor of the  !

drywell, the turbulence in the water pool on the floor may be sufficient for the dirt to = '

remain suspended and transported to the suppression pool, espwcially in the case of a recirculationline break where water flowing out the break can provide a substantial j

amour.t of turbulence in the water pool on the drywell floor. The staff believes that  !

conditions in the post-LOCA environment would be conducive to transport of dirt / dust -

to the suppression pool.

' A value of 150 lbm of dirt / dust is suggested as the value licensees could use to

• conservativery~ eddioss this type debris originating in both the drywell and from the suppression chamber areas above the suppression pool whoie dirt and dust could be entrained by pool swell. The URG also states that this quantity includes concrete that would be eroded by the LOCA jet. This quantity of 150 lbm is termed a judgment and *

-is not supported by any analysis or investiga+ ions.- The staff believes it would be prudent for utilities to verify this number threugh sampling. ' It would he relatively straightforward for the BWROG or an individual licensee to take samples of dirt and dust from drywell and suppression pool areas, determine typical weight-per-unit-area on collecting surfaces, and calculate a bounding value for all applicable surfaces. An i

allowance would have to be added for concrete dust generated by LOCA forces. The current approach is simply a judgement, and should be evaluated in greater detail to ensure that it is conservative. in addition, consideration should be given to containment size, and the potential for additional dirt and dust occumulation over the operating life of the plant. While the staff does not believe that use of the URG 150 >

lbm assumption is inappropriate, the staff does believe that it would be prudent for '

individuallicensees to conduct a more thorough evaluation.

Rust Flake: The BWROG states that 50 lbm of rust flakes in the strainer head loss evaluation conservatively addresses the amount of rust which may be removed arom unpainted steel surfaces and transported to the suppression pool, it includes surfaces in the  ;

- drywell, the vents /downcomers,and the suppression chamber above the suppression L

^

pool. No basis is provided for the 50 lbm recommendation other than engineering judgment. This presents the same concem to the staff for this suggestion as for the dirt / dust value noted above. Again, the staff believes that a sound technical basis could be developed by sampling applicable drywell and suppression chamber areas combined with walkdowns to establish typical quantities of rust per unit area. This i

' information could be obtained and used to develop total plant estimates. The URG

- states that it is expected that only a small fraction of the rust would become detached from steel surfaces during a LOCA event.' This statement is unsupported by data or technicalevaluations. As with the dirt and dust discussion noted above, the staff does

' ' not tplieve that use of the URG 50 lbm assumption is inappropriate. However, the ~

~

staff does-believe that licensees should conduct a more thoroughTvaiuation to determine its applicability to their plants.

DRAFT 30

l. -

i s

. . _ . . _ _ _ _ u .,-a. , , , , ,, ,. , . . .;-..,,- _,,...~m, .._ ,.m. ,, , , ,,,, ...-,m. ,. .c,-p ,, . , ,

DRAFT imm Coatings: Estimates of quantities of paints and coatings that may become debris due to LOCA forces are provided fur three typea of quakfied coating / paint materials. The values -

presented in Table 3 on page 58 are based on a study performed by Bechtel for the BWROG (Ref 51) and are for qualified coatings delaminated by direct jet impingemert from a pipe break. The staff has not identified any current concems relative to the conclusionsin Reference 51. However, the staff points out that the issue of coatings and their potentialimpact on the ECCS is an issue currently under staff review, and is the subject of a genericletter expected to be issued in early 1998. As a result, the conclusions of Reference 51 are still under staff review.

The guidance for unqualified coatings, on the other hand, is incomplete and unsupported. For example, ca page 61, the URG states, "After exposure to the LOCA environment, but only after containment pressure is reduced, there is a possibility that indeterminate / unqualified coatings may detach from the surface to which they were applied and become a debris source." The staff is particularly concemed because there is no basis for this statement, and in t;is staff's opinion, it is entirely inaccurate.

In fact, there is ne evidence that indeterminate or unqualified coatings would be latent debris at all. For instance,if the coatings have lost adhesion over time due to improper application or lack of qualification for the environment, then there is no reason to assume that the turbulent environment of the reactor vessel blowdown could not cause the coating to detach and transport to the suppression pool. On page 62, the URG continues, ' Dependent on several plant-specific factors, it may be possible to show that the failure of indeterminate /unqualifiedcoatings would not ocCJr until late enough in the LOCA progression that there is no transport mechanism available to transport the failed coating from the drywell to the wetwell." No discussion on wh9t the " plant-specific factors' are or the basis for the statement is provided. The staff is very concerr.ad that this part of the URG downplays the potential significance of unqualified / indeterminate coatings greatly.

Wh'ile not explicitly stated, the URG as currently written tends to lead licensees away from accounting for this potential debris source by only discussing why these coatings might not have an impact on the ECCS. In actuality,the potentialsignificance of these coatings depends on the amount of such coatings in the containment, the licensee's NPSH margin, how well these coatings were applied, and the coating performance characteristics in both normal operation and LOCA environments. In light of recent qualified coating events and the lack of substantive technical basis for the BWROG guidance regarding unqualified and indeterminate coatings, the staff does not have enough information to evaluate the adequacy of',he BWROG guidance. However, the staff considers it prudent to address unqualified / indeterminate coatings in the sizing of the new strainers installed in response to NRCB 96-03. This would help to avoid potentialoperabilityconcerns and hardware modifications (e.g., resizing of the suction strainers) in the future if clogging concems are raised as part of the staff re' view of coatings related issues. In addition, the licensee should also consider the potential impact of improperly applied or maintained " qualified' tastings for the same reason.

DRAFT 31 W

. l DRAFT imoro7 C.onclusions on Section 3.2.2: Considerable time was devoted to the development of the NUREG/CR-6224 (Ref 31) study to develop conservative estimates of dirt / dust and rust. Where provided, the values suggested in the URG are about the same or larger than NUREG/CR-6224 values. As a result, the s:aff concludes that use of these values is acceptable; however, the staff notes that the NUREG/CR-6224 study was conducted on a reference Mark l containment. For this )

reason and the additional reasons cited above, the staff believes that it would be prudent for l individuallicensees to evaluate the recommendations of this section for applicability to their plant. 1 It would also be useful for the BWROG to provide better guidance rel0ted to estimation of dirt / concrete / rust debris sources in the containment as well as paint / coatings for plants with unqualifiedcoatings. This concern is especially true for Mark Ill containments because they have larger concrete areas, in addition, potentialfor dirt / dust accumulation may be greater in the larger containmentsif they have significantlylarger amounts of horizontal surfaces. The sensitivity of the head loss calculations to these numbers may vary with the assumptions and the resolut!on option selected by the licensee. If the licensee is installing a large passive strainerwhich is based on very censervative assumptions relative to the quantities of fibrous debris and sludge assumed to reach the strainer surface, then that licensee may be able to demonstrate by sensitivity analysis that variationsin the quantities of dirt, dust, rust flakes, etc. may not significantly affect the overall head loss across the strainer. If, however, the licensee were attempting to justify their current strainer or use as " realistic" assumptions as possible, then the significance of the values determined in accordance with Section 3.2.2 of the URG escalates significantly. ,

Consistent with the stafi's guidance above, it would also be prudent for licensees to evalucte the conclusions of the Bechtel coatings report to ensure applicabilityof those conclusions to their plant.

Most importantly,as noted above, the staff concludesthat licensees should be cautioned to carefully evaluate the potential impact of unqualified and indeterminate coatings on ECCS suction strainer head loss, if in doubt, assuming the coatings reach the strainer surface would clearly be the cerservative meaeure. This would provide the licensee with less risk relative to the coatings issue, at4 could lead to additional margin in the strainer design if URG statements w1ich minimize the pctentialimpactof unqualified protective coatings are supported by the results of t1e staff's coatings review.

3.2.3 Drywell Debris Transport Section 3.2.3, entitled, 'Drywell Debris Transport,' documents BWROG guidance on various options for estimating the fraction of the damaged insulation (generated in the drywell) that will be transported to the suppression pool as a result of the reactor vessel blowdown and the various mechanisms that can wash debris down to the suppression pool. The guidance provided in this section is based on the following supporting analyses:

1) " Testing of Debris Trar) sport Through Downcomers/Ventsinto the Wetwell," Revision 1, URG

, ~

Technical Support Documentation,Vol. II, Tab No. 2, Continuuni Dynamics, lnc., Princeton, NJ.

DRAFT 32 l

I

DRAFT maom 2)l 'BWR Drywell Floor Flow Modeling Following Pipe Break Loss-of-Coolant Accident," URG Technical Support Documentation, Vol. Ill, Tab No.1, General Electric Nuclear Energy, San Jose, CA

3) " Summary of Debris Washcown Experience,'URG TechnicW Support Documentation, Vol.

Ill, Tab No.17, BWROG The guidance in this section and the associated supporting analyses were reviewad for completeness and accuracy. In addition, the guidance provided in Section 3.2.3 of the URG was evaluated against the guidance of RG 1.82, Revision 2 (i.e., transport fraction = 1), the results of the BWROG sponsored research, and the preliminary insights from NRC sponsored research 1 conducted by Science and Engineering Associates, Inc. (SEA).

The BWROG developed the following methodology to determine the amount of debris that

' transports from the drywell to the wetwell:

1) For each insulation type tested at the AJIT facility, the fractions of insulation contained in the zone of influence that would be destroyed (destruction factor) was estimated. The debris was then divided into three distinct size categories based on the test data: ' fines," 'large pieces," and ' blankets." These fractions are calculated as integral values applied over the entire zone-of influencevolume. The fractions were computed for severalinsulation types making use of experimentaldata obtained from the air let impact tests. In addition, the URG estimates assume that for NUKON, all insulation contained within 3UD of the break is destroyed into fines. It is not clear, however, that this assumption was made for all debris types.

For Fines:

2) The fraction of the mass of ' fines' that would be transported as a result of blowdown

. following a steam line break and a recirculation (water) line break was estimated. Different estimates were developed for Mark I, ll and lli containments. These estimates are based on small-scale tests of downcomer geometries which were sponsored by BWROG and conducted by CDI.

. 3) The fraction of the mass of fines that would be transported as a result of washdown that follows blowdown was also estimated. For the recirculationline break, the ECCS runout flow was used as the basis, and for the steam line break, spray flow was used as the basis.

These estimates are based on a) small-scale experimental data from BWROG tests, b) washdown test data obtained from previous sources and 3) one dimensional (1-D) and three dimensional (3-D) modeling of water flow on the drywell floors. Different estimates were developed for Mark I, ll and lit containments.

~

4) Th"e sum of these fractions (transport factor) provides the total fraction of fin'es '

to suppression pool. Multiplicationof this transport factor by the estimated fraction of fines in the damaged insulation provides a total mass of fines from the damaged insulation that would be transported by blowdown and washdown to the suppression pool.

DRAFT 33

.t f

DRAFT umm For large'piecesi

- 5)1 - No direct transport is assumed for the pieces located above the lowest grating. The URG considers that they can only be transported after erosion.

' - 6) ' A fraction was estimated for *large pieces *' located below the lowest grating that would be transported during blowdown,- and for the amount that would be eroded and transported during the washdown phase that follows the reactor vessel blowdown. .!

~ For Blankets:

7) According to the BWROG, no transport is possible.

The individual destruction and transport fractions were used to develop. combined

destruction /transportfactors for each insulation type as a function of the containment type and the location of the debris. Tables 5 and 6, pages 83 and 84 of the URG, list these fractions for several  ;

insulations.

In the case of insulation debris, individual licensees may choose to use a drywell transport factor of 1.0, the value listed in the Regulatory Guide 1.82, Rev. 2 (Ref 5), or the values listed in Tables 5 and 6, to estimate the fraction of insulation debris (contained in the Zone of Influence) reaching the suppression pool. For other drywell debris, a drywell transport factor of 1.0 was recommended.

The URG-states that use of other trar. sport values by licensees should be supported by an evaluation which justifies their bases.

Staff Evaluation of Section 3.2.3: A confirmatory analysis (see Appendix H of this report) was e conducted in evalusting this section. The analysis results are as follows:

Drywell To Suppression Pool Transport Factors: The experimentaldata for drywell transport

. is based on small scale tests which have not been properly scaled. The staff forwarded its i scaling concems about the BWROG testing at the time of their inception by letter (Ref. 20).

None of the staff comments appear to have been resolved. Finally, our analysis (See Appendix H) suggests that flow rates and duration of flows simulated in BWROG testing are not prototypicalof conditions that exist in BWR drywells following a LOCA. For example P.s few velocities simulated in these tests are about 50% of the flow velocities expected -

- following a postulated 24 inch diametc. am line break. As a result, these tests may have provided erroneous estimates of drywell transport. It was not clear to the staff in evaluating the BWROG test prograt, that the test results would be reasonable, conservative, or non-conservativeif scaled to a thil-size plant. The BWROG claims that use of these test results would conservatively bound the drywell transport fraction is inadequately substantiated.

From the staff's research, thelwo of the primary keys to debris trans5o'rt are debris size

distribution and location. Testing conducted by the NRC in support of its confirmatory

research in the area of debris transport has shown that floor gratings typically used in BWR

- plants are very effective at limiting transport of larger debris sizes. However, these gratings

, DRAFT 34 4

-s er N -e~- n + ^N. ,..w---0,-e ,, , , .nww,,, ,n n ,~m.

n-, ..,---c

1 DRAFT imois7 are not very effective at limiting transport of fine debris. The restits of the staffs study and testing will be documented in NUREG/CR-6369,'Drywell Debris T '

sport Study," which is scheduled to be published in February 1998. The study's findingi, Jad to the conclusions that calculation of fine debris is critical in evaluating the amount of debris that could

• ultimately transport to the suppression pool during a LOCA. In addition, calculation of debris ,

which is generated below the lowest continuous grating in the drywell is also critical. Large debris generated above the lowest grating should not be discounted if the lowest floor i grating is not continuous or has substantialopenings in it through which larger debris could transport, in this case, an appropriate fraction of large debris from above the grating should be considered as transportable.

Another result obtained from this study is that the staff sponsored tests of Mark 11 downcomer geometry found no basis to conclude that the transport fraction for a Mark 11 would be different from that of a Mark I or Mark lli containment. Therefore, the staff finds the transport fractions for Mark ll containmentslisted on pages 75 and 80 of the URG to be unacceptable. Fractions are also given on the same pages for fine fibrous debris transport and RMI debris transport in Mark l's and Mark lil's. The values given for Mark l's and Ill's assume that 100% of the fine debris will transport to the suppression pool. These fractions are consistent with the staffs transport study and RG 1.82, Revision 2 (Ref 5), and are, therefore, considered acceptable. The staff believes that these same fractions should also be used for Mark ll's.

On page 76 of the URG, guidance is given on transport fractions to be used for large fibrous debris (not whole or partial blankets which are considered as not being likely to transport) generated below the lowest grating. The values given are 70% for Mark l's and Mark lil's and 30% for Mark ll's. - As noted above, the staffs confirmatory testing in support of its drywell transport study has found no basis for different transport fractions for Mark ll's.

Based on the resulta of the drywell transport study, the staff concludes that the transport fractions,used should be the same for all three containment types in addition, the drywell transport study findings concluded that the transport fraction should be between 90% and 100% for all three containment types for larger debris below the lowest grating. Intuitively, this is reasonable as the gratings (which are ast umed to prevent large debris from above to transport downward to the downcomers)wouid trap the debris down low providing a high potential for tranrport to the suppression pool during blowdown in addition, break flow i (recirculationline break) or containment sprays may provide suffe 9 agitation in the water pool on the drywell floor to washdown large debris that wa r,ut transported during i blowdown. The staff concludes that insufficienttechnicaljustificaton has been provided for assuming a transport fraction less than 100% for large debris generated below the lowest drywell grating.

" , , As discussedin Appendix H., the staff believes the that the guidance for assuming erosion of large fibrous debris'is adequate provided unthrottled ECCS flow does not continue for

- more than 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />. If it does continue for more than three hours, than the licensee should determine an appropriate fraction to assume. Erosion of NUKON was shown to be linear in the NRC tests, so scaling of erosion of NI'KON can be performed easily. Similar testing i

DRAFT 35 1

..y...

' ' " ~

t .S

'l DRAFT- Lwwe7 could be conducted for other insulation types, if ne'cessary, to determine an appropriate -

fraction to assume for erosion.

Destruction Factors:' The BWROG calculated the destruction factors provided in the URG . -

1 from their AJIT test results. The test results and associated data _ analysis are documented : ,

in Reference 52; The destruction factors are based on the calculation of rl.; q is defined . (

as the fraction of debris generated that would be generated in a form that would have a low - '

transport efficiency (e.g.,'large pieces or blankets). Tests were conducted at several' -

i different distances from the jet nozzle (x/D) to determine the pressure at which debris would 0

. be generated, the amount of debris generated and the rough size distribution of the debris. ,

. q. is then calculated for each x/D test result and then integrated to give an average q ,

(0,,,,,,,,3)value over the entire ZOI where damage would be expected to occur. For each insulation type and mode of encapsulation, the BWROG provided a set of destruction factors in the L.;G that can be used to estimate what fraction of the insulation contained in the ZOI-becomes destroyed into " fines". In principle, these values were derived based on the '

- following methodology: '

1) .- Full-scale insulation blankets were mounted on a targe; pipe located normal to the. ,

jet center line. In almost all cases, the blankets were mounted such that the latches on the steeljacketing (if Jacketing is used) were arranged to face the nozzle atd the seams in the blaaket were arranged to face away from the nozzle.- Thir, resulted in -

a conservative situation because the protective steel Jacket is quickly blown off, '

exposing the canvass covered blanket to the jet. At the same time, because the blanket seam is away from the nozzle, tne blanket was maintained in position'for a longer duration.

2) After exposure to approximately 5 seconds of blowdown (although the valve was closed at 5 second mark, the blowdown continued at least for 3-4 more seconds),

the amount of debris generated and the rough estimate of the debris size distributien was measured..

3) The size and amount measurement were carried out for different UD values. The fraction of debris not destroyed into fines (q.) were then plottedis a function of UD and were integrated over the entire ZOI to estimate the q,,,,,,,,3 for that insulation

- and mode of encapsulation.

4). . To account for unforeseen phenomena, the q numbers were further reduced byJ assuming that all the insulation contained in a spherical region within a radius of 3 I -

pipe diameters from the break is destroyed into fines.

. . . a .-

1 DRAFT .se 1-L i-

.,e-- w -a, ,, , , ,m . <m 6 mm -+.,+,,o-e , , e e.- v -- , .- +-;-v--

2 DRAFT 52noro7 ,

The staff reviewed BWROG derivation process and the overallresults. In addition, the staff conducted additional studies (both experimentaland analytical). The following are the staff's key findings:

The blanket arrangement used in the BWROG testing is highly conservative. In the NRC-sponsored testing, the amount of debris generated is substantially less when the latches were arranged to face away from the nozzle or when the blanket seems were located facing the nozzle. In the former case, the steel jacketed remained for longerduration and protected the blanket. In the later case, the blanket was stripped Vf the target in such a short time that very little damage was observed.

The targets and structures located in the jet pathway would provide considerable protection to the blankets. In the NRC sponsored experiments, artificial means (high strength steel bands) had to be used to hold the blankets in place to maximize debris generation. Otherwise, no more than 25-30% of the insulation blankets were destroyed by the jets.

• The BWROG approach has the following conservatisms:

1) Not all the targets in a BWR drywel are normal to the jet centerline in fact, the staff survey suggests that majority of the piping (>65%) would be located parallel to the jet flow., in such a case, only a small portion of the blanket would be subj;cted to high dynamic pressures, unlike the BWROG experiments where most of the blanket was subjected to jet flow.

2) Not all the targets have steelJacket latches facing away from the nozzle and the blanket seams facing away from the nozzle. The vendor survey has indicates that no order is followed for installation of steel Jackets or the

, insu!ation blankets (although accessibility consideration often plays a role).

3) The blowdown does not continue at high pressures as tested in the BWROG studies.

4) Finally, the structures offer considerable protection. A CFD calculation was performed to quantify the impact of structuralimpediments, in these studies more realistic structura! elements were used to demonstrate how the jets are diffused by structures.

Based on this the staff concluded that, the approach followed by the BWROG would yield conservative estimates for most insulations. This can be demonstrated by the following analysis.

For steelJackgted NUKON insulation, Method 2 yields a spherica: zone of influence ppproximately 11 pipe diameters in radius. In the reference plant from NUREG/CR-6224, such a'ZOI would include approximately 900 ft) of insulation. Using a r), value of 0.78, this translates into a fines of approximately 200 ft3. This value is slightly lower than the total volume of NUKON insulation contained ir' a spherical ZOI with a radius of 7D in the reference plant used in the staff study DRAFT 37

DRAFT urw 97 (NUREG/CR-6224). The CFD calculations, conducted by the staff, clearly demonstratedthat given the structural congestion, high velocity flows (sonic flows) do not continue beyond a spherical region with a radius of SD. Thus, the staff concludes that BWROG suggested numbers are probably conservative. However, the staff identified that rl, values were obtained for several insulations based on a very limited set of experimentaldata. Examoles are Temp-Mat, K-wooland some of the RMI. In these cases,less than 5 data points were used to derive the q,. The staff believes this to be an inacequate amount of data points on which to calculate q,. Therefore, the staff suggests that Ibensees with these insulations determine rl, values as follows:

1) Assume that all insulation contained in 3D spherical region becomes destroyed into fines, and

2) Assume that rest of the insulation would be destroyed such that local q, is equal to the lowest rl, value measured in the experiments.

The strengths of this section are:

e It follows a logical path for estimating drywell transport factor. This path is somewhat similar to the one used in NUREG/CR-6224 (Ref 31) and in the staff's Drywell Debris Transport Study.

e it reviews important considerations licensees should be aware of in estimating the drywell transport factor.

e The transport fractions for fibrous debris above the lowest grating provided on Pages 75 and 80 for Mark I and Mark Ill containments are considered to be conservative and appropriate for use.

its major weakness is that the test data obtained from the BWROG sponsored drywell trar. sport tests were not obtained over the range of experimentalvariables that would support scaling the data for full size drywells. Similarly, the choice of experimentaiset-up does not scale appropriatelyto the BWR drywells.

Conclusione, on Section 3.2.3: In summary, the staff concludes the following:

1) The staff finds the transport fractions for Mark 11 containments listed on pages 75 and 80 of the URG to be unacceptable. Fractions are also given on the same pages for fine fibrous debris transport and RMI debris transport in Mark I's and Mark lil's. The values given for Mark l's and Ill's assume that 100% of the fine debris will transport to the suppression pool.

The staff concludes that the values for the Mark I's and Ill's for fine debris are acceptable and that these same fractions should also be used for Mark ll's.

2) On page page 76 of the URG, the guidance on transport fractions to be used for large fibrous debris (not whole or partial blankets) generated below the lowest grating. The values given are 70% for Mark I's and Mark Ill's, and 30% for Mark ll's. The staff concludes that DRAFT 38 l

l DRAFT urm l insufficient technical justification has been provided for assuming a transport fraction less than 100% for large debris generated below the lowest drywell grating, and that the transport fractions used should be the same for all three containment types.

3) The staff believer, the that the guidance for assuming erosion of large fibrous debris is adequate provided unthrottled ECCS flow does not continue for more than 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />. if it does continue for more than three hours, then the licensee should determine an appropriate fraction to assume (NUKON only). Similar testing could be conducted for other insulation types, if necessary, to cebrmine an appropriate fraction to assume for erosion.

4) The approach followed by the BWROG for determining destruction factors would yield conservative estimates for most insulations and is acceptable. However, the staff Uentified that rl, values were obtained for several insulations based on a very limited set of exnerimental data. Examples are Temp-Mat, K wool and some of the RMI types of insulations. In these cases, less than 5 data points were used to derive the q,. The staff beliaves this to be an inadequate amount of data points on which to calculate rio. Therefore, the staff suggests that licensees with these insulations determine n, values as follows:

a) Assume that all insulation contained in 3D spMrical region becomes destroyed into fines, and b) Assume that rest cMhe insulation would be destroyed such that local rl, is equal to the lowest rle value measured in the experiments.

3.2.4 Suppression Pool Debris The focus of this section is on sources of debris which are present in the suppression pool prior to the. occurrence of a LOCA, Both transient debris and sludge are potentially present in the suppression pool at any given time. This section of the URG also reviews the LOCA-g 1erated debris and tran'sient debris, both of which were discussed in prior sections of the URG. The guidance in part was based on extensive BWROG survey of suppression pools of selected operating US BWRs (Refs. 54 and 55).

Staff Evaluation of Section 3.2.4: A single analysis was performed in support of this review, its objective was to evaluate the adequacy of the BWROG recommended sludge volumes and their tractability. This analysis is documentedin Appendix J. The staff concluded based on the analysis that the BWROG interpretation of survey information is acceptable and the URG guidance in this section is acceptable..

3.2.5 Supprassion Pool Transport and Settling The DWROG recommended that no creilit should be giVen for settling of debris in the pool during

• the high energy phase where the pool undergoes chugging and/or condensation oscillations. 4 also recommended that all the suppression pool debris should be assumed to be resuspended dr g this phase. Finally, an individuallicensee is given an option to select no settling in the pool even DRAFT 3g L

i L _ _ _ _ _ _ _ _ . . _ _ - _ .

l DRAFT imes7

, during later low energy phase or to estimate fraction settling making use of NUREG/CR-6224, Appendix-B data and results.

Staff Evaluation of Section 3.2.5: The strengths of URG 3.2.5 are that it reviews important considerationslicensees should be aware of in estimating quantity of debris settling in the pool. The majority of this information was previously reviewed and found acceptable by the staff. However, the Staff notes that NUREG/CR-6224, Appendix B provides the required data for selected insulation and particulate types only. Licensees using NUREG/CR-6224, Appendix-B methods for other insulation debris should be cautious about extrapolating the experimental data and models.

3.2.6 Verification of Adequate ECCS Pump NPSH The BWROG guidance related to calculating strainer head losses is providsd a Section 3.2.6.2.3 of URG, " Strainer Head Loss Calculations", This guidance is based on Reference 56. The guidance specificallyrefers to Appendices A and B of this report for methodology and calculational procedures. Appendix A appties to passive strainer head loss predictions with fibrous debris and Appendix B applies to RMI debris. The calculationalprocedures described in Appendices A and B were utilized to determine their ability to predict the head losses associated with a number of the tests used to develop the procedures. The following sections summarize the staff's findings.

URG Section 3.2.6.2.3 summarizes important guidance in the application of Appendices A and B of Reference 56. The technical support document, CDI Reoort number CDI-95-09, Revision 4, entitled, " Testing of Alternate Strainers with Insulation Fiber and Other Debris," describes the strainer head loss tests performed in the development of the calculational procedures presented in Appendices A and B, including a description of the test fecility, the testing procedures, data reduction, and test results. The procedures in Aopendices A and B are recommended for use by an individualutility to estimate head loss acrov, r le strainer as a fur.:'in of the quantity and type of debris reaching the strainer and the flow velocity. A dimensionless head loss equation was provided for this estimate.

Staff Evaluation of Section 3.2.6: The strainer head loss was predicted for a number of the strainer head loss tests using the procedures outlined in Appendices A and B and compared with the corresponding experimental head losses. The staff notes that the test results presented in Reference 56 are the only data provided to the staff by the BWROG, and are for the stacked disk and star strainer types.

The objectives of the staff's analysis were 1) to validate the calculationalprocedures by determinirg the ability of those procedures to predict the test results from which the procedureswere developed, and 2) to evaluate the applicability of the procedures to actual plant situations. This analysis is documentedin Appendix l of this safety evaluation. In general, the results of the staff's analysis led to the conclusionthat,the approach outlined in Appendices A and B of Reference 56 was unreliable and incomplete.

DRAFT 40

DRAFT wxm  :

The URG contains valuable and useful data for predicting strainer head losses. The staffs review, '

however, uncovered several concerns regarding the quality and applicability of this data.- These concerns are as follows:

I t

1)- The URG model was developed from test data where a true steady state condition was not generally achieved, implying the debris loadings used in the model development were somewhat different then the actual loadings on the strainers.

2)- The URG model was developed for limited ranges of data, i.e., fiber and particulate debris ,

loadings, velocities, and strainer design, and using the model beyond these limitations is especially risky because the models are non-mechanistic in nature. For example: +

• The model was developed for lower debris loadings where the amount of fibrous debris was only large enough to fill the troughs in the strainer. - Extending the correlation- to higher loadings is inapprcpriate without additional substantive experimental or analytical work, e- The modelwas developed based primarily on NUKON' data. Its extension to other fibers is inappropriete without additional substantive experimental or analytical work

3) The URG model has been shown to under-predict in many cases the experimental data used to develop the model; therefore, each head loss prediction should be reviewed in detal relative to the applicable data to ensure a valid prediction. In many situations, such as the 60-Point Star strainer, a significant safety factor should be factored into the prediction.

4) The bump-up factor used to account for the miscellaneous debris was developed with limited data (Gravity Head Loss Tests) and in many situations, the bump-up factors have been shown to severely over predict the head loss. While this over-prediction is conservative, it can lead,to severe design impacts as a result of over-predicted head loss.

5) Some of the head loss test runs (e.g. -J6) were conducted with thin fiber beds and large amounts of particulate debris. It was assumed that all particulate debris was captured by the debris bed, which is incorrect. The amount of particulate debris that was captured in the debris bed is unknown (i.e., underestimated). The use of this data to develop a correlation may underpredict the impact of particulated debris and lead to erroneous head loss predictions.

6) Two types of RMI debris were used in the URG strainerdebris bed head loss tests, i.e., two different size distributons of the 2.5 mil stainless steel foils, and each size distribution was -

used over a different period of time. Thus, the tests for a specific strainer design are

, generally valid for only one type and size distrioution of RMI debris. ' For examplec the stacked disk tests (stacked-disk section of the self-cleaning strainer) only used the RMI debris obtained from Diamond Power. Therefore, the URG model only applies to the stacked-disk strainer for this one type and one size distribution of RMI debris.

DRAFT 41

& ~

-t aa

~

n DRAFT man

-7)  :

The'URG report did not provide model validation, parameter sensitivity, or uncertainty analysis _ leaving open the question of whether or not a particular combination of input parameters could cause the URG model to severely under-predict head bst Based on this analysis, the staff has the following specific comments on this section of the URG:

1) - The stan is concemed that applying the URG strainer head loss model by blindly plugging numbers into_a etv*haak step-by step procedure as outlined in the report may lead to erroneous _ head loss _ predictions. Rather, each head loss prediction must be carefully anchored into head loss data to ensure that it is reasonable and conservative. The BWROG was provided this comment in Reference 11. The BWROG's response (Ref.13) -

did not address the overall head loss prediction, but instead it relied on accuracy of the bump-up factors. The sta# does not believe the comment has been resolved. The staff believes that resolution of this issue is vital to judging whether_ the BWROG correlation (if used by any licensee)is conservative,best-estimate or low. Resolution of this concem will also need to address the confidence level of the data (i.e., repeatability and uncertainties).

2) The head loss predictions should use the same NUKON' debris properties that were emploved _in developing the URG model, i.e., densities, and diameters because the models are not mechanisticenough to account for the effects of varying these properties The fiber density of 2.4 lbm#t* was used in the URG study which is the number generally assumed for the intact fiber insulation, sometimes referred to as the as-fabricated density. However, there are data showing the density _of actual-debris is considerably reduced by the-destruction process. For example, fibrous debris used in the PP&L sponsored tests conducted at ARL had an actual density of 1.3 lbm#t'. if the user were to use this density rather than the as-fabricated density, the user would obtain a different fiber spacing distance l which would result in a different head loss than if the user actually used the as-fabricated density that is inherent in the non-mechanistic URG model. Since the basis of the URG correlation is not mechanistic, the procedure should be modified to use more empirical paramelers rather than subjecting licensees to the challenging task of estimating more complex parameters, if not resolved, the result is likely that licensee use of the correlation will be inconsistent.

3). The staff is concemed that applying the URG head loss prediction models to plant conditiors that differ markedly from those conditions tested could lead to erroneous head loss predictions. Use of selected dimensionless numbers creates the impression that the equation can be generalized. No analytical evidence to that effect has been presented to the staff. As a result, any extrapolations from the conditions tested should be carefully 4 evaluated. At the inception of the attemate strainer test program, the staff cautioned the BWROG that their approach for development of a head loss correlation did not have a sound

,,, basis for extrapolatingthe correlation beyond the conditions for which it was developed. The

. BWRDG responded that its testing was designed to be " functional testing" rather than a C means for developing a correlation. However, URG attempts to use the data compiled and correlation developed in a more generalized form. For example, the data was obtained for

= conditions that simulate lower fiber loading on the strainers. The correlation then should be DRAFT 42

-r

DRAFT uras7 1 applicable, at best, to those conditions. The URG and its appendices do not state any such limitations. The staff concludes that these limitations should be clearly stated, and that generalized use of the correlation beyond the original range of testing is unacceptable.

4) The experimental data as well as the correlation presented for fibrous beds was obtained for low debris loading on the strainer surface. It is the staffs opinion that the equation and the procedures are not applicable for thick combination fibrous /particulatedebris beds. The staff has been unable to identify any rationale supporting use of the URG correlation for thick fibrous / particulate debris beds. As a result, the staff recommends that licensees utilize vendor specific test data demonstrating the strainer head loss for various debris loadings up to and including the lice .see's limiting debris load.

5) The experimental data is neither complete nor was it supported by sufficient quantitative analytica! reasoning to support the URG statement that the thin-bed effect is not a concem for alternate strainers. The staffs analysis of the BWROG's data suggests that for altemate strainers tested, small fiber loads coupled with large quantities will not result in very high pressure. In that context,it is reasonable to state that medium breaks will not result in large head losses that are not bounded by large breaks, but a statement that no ' thin-bed effect' was observed is incorrect.

6) in calculating NPSH, licensees should ensure that their calculations are consistent with their licensing basis. For instance, typically, no operator action is credited for the first 10 minutes during a postulated LOCA. Therefore, NPSH should be evaluated at runout flow until the plant licensing basis allows otherwise.

7) Reference 31 in the URG is considered as an unacceptable methodology by the staff for determining minimum NPSH. The staff finds this methodology unacceptable because it incorrectly assumes that head loss across the strainer will be governed by laminar flow.

Conclusions on Section 3.2.6: The staff concludes that the URG head loss correlation is unreliable and the guidance is inadequate to prevent inconsistent application of the correlation.

Based on this, the staff recommends use of vendor test data to demonstrate the head loss used for calculation of NPSH margin. The test data should include debris combinations up to and including the worst case debris loadings calculated for the plant. If a scaled strainer model is used for conducting the head loss tests, then the licensee should ensure that they bsve an adequate technical basis for applying the vendor data to the full size strainer.

DRAFT 43

x 1-DRAFT . won  :

i

< 3.3 - BACKFLUSH 4

-- -Section 3.3 provides the BWROG's guidance on backflushing of strainers.- In general, the URG; L guidance can be summarized as followw:

e The BWROG recommends'against use of strainer back/ lush systems as a primary means  !

of resolving the ECCS strainer clogging issue because:

J

-1); - The timeframe during whwh backflush would be required is significantlyless than 30  ;

- minutes placing a hardship on the operators and raising human factors concems. '

2) For many plants, it is anticipated that backflushing would need to be repeated within :

30 to 60 minutes, which may not be feasible depending on the type of design used for the backflush. .

e- The BWROG also notes that the staff believes that backflush by itself would probably not i be sufficient to resolve the issue. This is because the staff concurs with the issues cited by the BWROG above.

e . The BWROG does believe that backflush may be a viable defense-in-depth measure for -

plants with the capability to perform backflush.

Section 3.3 also gives a detailed list of design' considerations for any licensea desiring to implement -

a backflush system.

Staff Evaluation of Section 3.3: The staff reviewed the guioance providedin Section 3.3 and has  !

concluded that it completely addresses the types of considerations a licensee must address in implementinga backflush system. In addition, the staff agrees with the BWROG that backflush is <

better used as a defense-in-depth measure rather than the primary means of mitigating a LOCA.

Therefore, the staff finds the guidance in this section to be acceptable.

3A SELF CLEANING STRAINERS This section describes the design considerations that must be addressed in implementing a self -

cleanMg strainerdesign to resolve the strainer issue. This section correctly identifies some of the

- technicaldifficultiesin developing an adequate self-cleaning strainer design to resolve the strainer -

issue, including: 1) optimizing the clean head loss and torque,2) problems with startup of the plow  ;

_which sweeps the strainer surface after debris has accumulated on the strainer surface during low flow conditions,3) the effect of debris which gets chopped up by the plow on downstream ECCS

- components,and 4) surveillance /maintenancerequirements. Because of these issues, the BWROG

. only recommends this resolution option ,if a passive strainer, solution is not viable.

4 DRAFT 44 e

)s.. . . . . , , . y, y - . , -,, .,.._,...~_--m,., 4

DRAFT wnw Staff Evaluation of Section 3.4: The staff identified no deficiencies in this section. Because it relies on mechanical moving parts, the staff agrees with the BWROG that a self-cleaning streber design is less desirable than a passive strainer design for resolution of the ECCS suction strainer issue.

4.0 @DITIONAL FEATURES WHICH PROVIDE DEFENSE IN DEPTH This section provides a discussion on the other plant features and operator training which provides defensein depth. A key statement in this section relative to availability of attemate water sources reads as follows:

" System alignment to make use of alternate water sources may require manipulation of valves or other equipment that is infrequently operated. The valves and other equipment used to provide alternate water sources for injection in support of the EOPs should be considered for inclusion in the scope of the Maintenance Rule. It is expected that most if not all of the valves and other equipment will already be encompassed by the plant maintenance program. However, licensees should review the alternate water sources available as described in the plant EOPs and ensure that all valves and other infrequently operated equipment necessary to accomplish injection from these attemate water sources are appropriately addressed by the plant maintenance program."

Staff Evaluation of Section 4.0: The staff reviewed this section and identified no deficiencies. The staff concurs with the statement cited above relative to including attemate water source components in the scope of the maintenance program, and strongly encourages licensees to do so. This is an excellent way to ensure defense-indepth by ensuring alternate water sources are likely to be available should they be called upon during an accident.

5.0 CONSERVATISM

~

This section describes the BWROG's position on the amounts of conservatism in their analysis methods described in Section 3.0 of the URG. The staff findings ictative to the information in Section 5.0 are included in the staff evaluation of the appropriate parts of Section 3.0.

6.0 MISCELLANEOUS REVIEW COMMENTS ON BWROG GUIDANCE The staff has the following general comments on the URG.

1)

References:

A) URG, Page #1, Line #3 B) URG Technical Documentation, Vol. Ill, Tab 16, Pipe Break Probabilities

,, C) NUREG/CR-6224, Appendix-A, BWR Pipe Break Frequencies ,

The BWROG states that " probability of a DEGB in a BWR is extremely low (on the order of 1E-9 to 1E 12)" While the staff agrees qualitatively with the statement (i.e., that it is extremelylow),it disagreeswith 1E-9 to 1E-12 numbers. Both the NUREG/CR-6224 study DRAFT 45

l DRAFT urm  !

I conducted by SEA, as well as 90% of the IPEs (including BWR IPEs), estimated these numbers to be several orders of magnitude higher,

2) The terms bounding and conservative are uced frequently throughout the document, but are i not clearly defined anywhere in the document. The final version of the URG should provide the definitions of these terms. In addition, where these terms are used in the body of the  !

text, the basis for stating that the individual method described is " bounding

• or " conservative l

should be provided. The staff does not believe that terms like bounding and conservative should be used to describe an assumption made strictly on engineering judgement unless a technical basis for the statement can be provided.

ZA OVERALL CONCLUSIONS AND RECOMMENDATIONS l

The overall strengths of URG are that it reviews important considerationslicensees should be aware of in estimating quantity of debris generated, transported through drywell, and wetwell to the strainer, and the head loss resulting from the debris bed buildup. The overall framework suggested for ECCC Suction Strainer evaluation is very similar to that developed during NUREG/CR-6224 study. The BWROG obtained valuable data that was lacking at the time of NUREG/CR-6224 study, including data for debris generation and debris transport. Some of the analyses conducted by BWROG are technically sound and based on good engineering judgment.

On the other hand, in several cases BWROG guidance was either unsubstantiated or developed i using assumptions that lacked sound analyticalor experimentalbasis. In a few cases, the BWROG guidance can be shown to result in erroneous estimates of the head loss, while in most cases their l

guidance could not be judged as either " conservative" or "non-conservative." This lack of confidence arose from inconsistent mMeling assumptions used by BWROG or use of experimentd data without sufficient technical justification for scaling to drywell operating conditions.

The staff has found that more defensible analyses can be used by the BWROG to inte pret the same data and sstablish a basis on which to conclude that the URG guidance is reasonable and/or

" conservative." Where possible, the staff has proposed some attematives that may be used by the BWROG to improve their analyses. The BWROG may wish to explore these alternatives. The staff notes that suggesting altemativesis a byproductof the staffs review, not its objective. The following are the staffs overall conclusions categorized by phenomena:

Selection of breaks: The URG should clearly state that a sufficient number of breaks should be evaluated to ensure that the most limiting breaks are analyzed. Possible locations of the breaks should include pipe tections or welds in the area of the drywell where highest density of fibrous insulation is installed. Breaks analyzed in support of strainer sizing should meet the requirements of 10CFR50 46, and therefore, should not be limited to breaks analyzed for compliance with GDC 4 (i.e., MEB 3-1 break analysis). This will determine the limiting break relative to ensuring adequate ECCS NPSH margin. The staff ha's also concluded that it is reasonable for licensees to screen out medium breaks in their strainer analysis if they plan to use attemate strainer designs with deep crevices for debris capture such as stacked disk or star strainers provided that strainer design has a crevice capacity which is large enough to handle debris loadings consistentwith a MLOCA for that DRAFT 46

O DRAFT 12rms7 plant. Licensees should have strainer test data which supports their conclusion that debris loadings consistent with a MLOCA for their plant does not cause more limiting head losses for the strainer design being used than debris loadings consistent with a large break.

Debris Generation: The staff's findings are summarized as follows:

• The P , factors in Table 1 should be revised and provided in terms of jet average pressure instead of JCL, especially for insulations having high P.,,, factors.

• The methodology used ic develop destruction factors for calculation of fine debris was found to be acceptable.

• Methods 1 and 2 are reasonable and appear to provide conservative bounds to the volume of debris generated.

• Method 3's application should be used with caution on a plant-specific basis since it allows the utility to take credit for limited separation (restrained vs unrestrained) and single-jet vs.

doublejet. Licensees should considerthat in the case of a main steamline break, flow would be from both ends of the pipe break until the MSIVs are closed.

• Insufficient information was provided on Method 4 for the staff to reach a conclusion regarding the acceptability of the methodology proposed.

• Damage to calcium sT.cate should be treated as erosion as demonstrated by Swedish experiments. Table 2 should revised to alter the failure criterion.

• For insulations with limited debris generation test data, the staff believes that licensees should conduct additior.at testing for those debris types unless a bounding approach is used for those materials.

Other Sources of Debris: These debris sources should be evaluated on a plant-specific basis.

When generic values are used, licensees should eva'uate the applicability of the generic values to to their plant. The staff suggests that licensees should assume all unqualified and indeterminate coatings are transported to the strainer surface to avoid potential operability problems resulting from the ongoing NF3C and industry efforts related to potential clogging of ECCS strainers by coating debris.

Drywell Debris Transport: The staff concludes that use of a 1.0 transport fraction is acceptable for 1) RMI,2) fine fibrous debris, and 3) large fibrous debris below the lowest grating in the drywell in addition, the staff has found no evidence that transport fractions for Mark 11 containments should be any different than for Mark I's and Mark lil's; therefore, the staff concludes that the same transport fractions should be used for all three containment types.

Suppression Pool Transport: The staff found this section to be acceptable with no comments.

Head Loss: The staff's conclusions are summarized as follows:

i The staff agrees with the BWROG that the " thin b5d effect's" are not likely to be an is's'u e for l the attemate strainer designs tested.

l I

l DRAFT 47 l

DRAFT was7 e

The generic applicabilityof the URG head loss correlation for fibrous bed is not aceptable because:

1) The correlation is not mechanistic.

2) The data obtained were for lower debris loadings than would be expected in many plants.

3) The RMI conclusions are specific to the strainer design tested and the type of RMI used in the test. No basis was provided for extrapolating to other RMI types and strainer designs.

4) The correlation is speific to NUKON* fibers.

As a result of the staff's findings relative to the URG head loss correlation, the stati recommends th.d licensees use test data to support the head loss used in their plant analysis.

• - The RMI correlation and methodology appear reasonable.

The NPSH evaluation describedin Volume ll of the URG Technical Support Documentatiort Tab 15, is not acceptable since it addresses viscous loss only. For RMI, this would yield erroneous results.

e- The Bump-up factors provide ' conservative means' for extending fiber correlation to other debris, but can create severe design impacts if used.

Resolution Options:

Use of self-cleaning strainersis discouraged unless the licensee cannot reach adequate resolution with a passive strainer design.

Backflush is recommended only as a defense-in-depth measure rather than a primary method of mitigating a LOCA.

DRAFT 48

___ _m__._-_--------------=------- - --

i DRAFT imow

8.0 REFERENCES

1) NEDO-32686, Rev. O, " Utility Resolution Guidance for ECCS Suction Strainer Blockage,"

BWROG, November 1996.

2) " Technical Support Documentation Utility Resolution Guidance for ECCS Suetior, Strainer Blockage," (3 Volumes), BWROG, November 1996.

3) Facsimile Transmittal from Mr. Tom Green to Mr. Rob Elliott,

• Summary information Regarding URG Drywell Transport Methodology," dated November 25,1996.

4) NRC Bulletin 96 03 (NRCB 96-03), "PotentialPlugging of Emergency Core Cooling Suction Strainers by Debris," May 6,1996.

5) Regulatory Guide 1.82, Rev. 2, " Water Sources for Long Term Recirculation Cooling Following a Loss of Coolant Accident," May 1996.

6) Draft Utility Resolution Guidance Document Sections 3.1.4, "Backflush;" 3.2.1.1,

• Postulated Break Locations;" 3.2.2.2,

• Suppression Pool Transport and Settling;" and 3.2.3.4, *ECCS Pump NPSH Calculations." Dated March 31,1996.

7) Draft Utility Resolution Guidance Document Sections 3.1, " Evaluation of Resolution Options;"

3.2.2, "Other Drywell Debris Sources;" 3.2.4. " Suppression Pcol Debris Sources;" and 3.4

'Self-Cleaning Strainer." Dated May 28,1996.

8) Letter from Mr. Carl Berlinger to Mr. Rocky Sgarro, " Comments on Draft Utility Resolution Guidance Sections 3.1.4, 3.2.1.1, 3.2.2.2, and 3.2.3.4," dated July 25,1996.

9) Letter from Mr. Carl Berlinger to Mr. Rocky Sgarro, ' Comments on Draft Utility Resolution Guidance Sections 3.1,3.2.2,3.2.4,3.4," dated August 20,1996. Accession number 9608260002.

10) Memorandum from Mr. Robert B. Elliott to Mr. Carl H. Berlinger, " Summary of August 19, 1996, Meeting with the Boiling Water Reactor Owners Group (BWROG) to Discuss issues Related to the Potential Loss of Emergency Core Cooling System (ECCS) Capability Due to Clogging of the Suction Strainers by Debra Generated During a Postulated Loss-of-Coolant Accident (LOCA)." dated September 4,1996. Accession number 9609100214.

11) Facsimile Transmittal from Mr. Michael L. Marshall, Jr. to Mr. Tom Green, " Transmittal of NRC Staff's initial Comments on URG," dated December 23,1996.

12) Letter from Mr. Tom Green to Mr. Rob Elliott,' Preliminary Design Considerations for ECCS Suction Strainer Debris Loading and Head Loss," dated January 13,1997.

13) Leiter from Mr. Tom Green to Mr. Rob Elliott, *BWR Owner's Group Response to NRC Comments and Questions Regarding NEDO-32686 Revision 0, ' Utility Resolution Guidance for ECCS Suction Strainer Blockage'," dated January 30,1997,

14) Letter from Mr. Martin J. Virgilio to Mr. Robert Pineli dated June 13,1994. No subject line.

Accession number 9407010102.

15) Facsimile transmittal from Mr. J. H. alunchausen to Mr. Al Serkiz, " Plan for Testing Pipe insulation Debris Generation Due to Simulated Pipe Breaks," dated August 9,1994.

16) Letter from Mr. Gary Holahan to Mr. Robert Pinelli dated September 12,1994. No subject line. ,

17) *BWR Owners' Group Program to T6st Alt 6mative ECCS SuetioriStrainers,' Letter fro'm Mr.

T.A. Green to Mr. Aleck W. Serkiz dated March 15,1995.

DRAFT 49

~

~ . - . - -. .

DRAFT inom I

18) " Request for Additional Information (RAI) Regarding the Strainer Test Program Being I Conducted by the BWROG,' Letter from Mr. M. D. Lynch to Mr. R. Sgarro dated June 22, 1995. Accession number 9507060112.

19) Letter from Mr. M. D. Lynch to Mr. Rocky Sgarro, *Second Request for Additional  !

Information Regarding the Strainer Test Program Being Conducted by the BWROG (TAC  :

No. M86925)," dated August 21,1995. Accession number 9508240384. l

20) - Letter from Mr. Carl Berlinger to Mr. Rocky Sgarro,

• Request for Additional Information )

Regarding the Drywell Transport Program Being Conducted by the BWROG,' dated April 22,1996. Accession number 9608260002.

21) Facsimile Transmittal from Mr. Michael L. Marshall, Jr. to Mr. Alan Bilanin of Continuum Dynamics, Inc., " Transmittal of Questions and Concerns Regarding the BWROG Strainer Tests," Dated August 14,1995.

22) Letter from Mr. T. A. Green to Mr. Rob Elliott,

• Closure of BWR Owners' Group Response to NRC ' Request for Additional Information Regarding the Strainer Test Program Being Conducted by the BWROG,' dated July 17,1995.

23) Letter from Mr. T. A. Green to Mr. Robert B. Elliott, " Transmittal of BWR Owners' Group ECCS Suction Strainer Committee Response to ' Request for Additional Information Regarding the DrywellTransport Program Being conducted by the BWROG,"' dated July 9, 1996.

24) Memorandum from Mr. Robert B. Elliott to Mr. Carl H. Berlinger," Summary of May 31,1995 Meeting with the Boiling Water Reactor Owners Group (BWROG) to Discuss issues RelatM to the Potential Loss of D.egency Core Cooling System (ECCS) Capability Due to Clogg6 0 of the Suction Strainers by Debris Generatcd During a Postulated Loss-of-Coolant Accident (LOCA)," dated June 13,1995.

25) Memorandum from Mr. Michael Marshall, Jr. to Mr. Charles Z. Serpen and Mr. Carl H.

Berlinger," Trip Report and Meeting Summary of September 27 and 28,1995 Public Meeting with Boiling Water Reactor Owners Group in San Jose, CA Regarding the Resolution of the Boiling Water Reactor Suction Strainer Blockage issue," dated October 6,1995.

26) Memorandum from Mr. Michael Marshall, Jr. to Mr. Charles Z. Serpan and Mr. Ctrl H.

Berlinger," Summary of April 4,1996 Public Meeting Between the NRC Staff and BWRCG Representativesto Discuss BWROG URG Document Conceming the BWR Suction Strainer Debris Blockage issue and the BWROG Planned Drywell Transport Test," dated April 16, 1996.

27) Memorandum From Mr. Michael L. Marshall, Jr. and Mr. Robert B. Elliott to Mr. Charles Z.

Serpan and Mr. Carl H. Berlinger,

• Trip Report July 9 through 11,1996:- Observation of BWROG Air Mt Tests," dated July 25, 1996. Accession numbers 9608070185 and 9603120157,

28) Facsimile Transmittal from Mr. Tom Green to Mr. Rob Elliott, "BWROG ECCS Suction StrainerCommittee Draft Test Matrices for Forthcoming Altemate Strainer Testing," dated June 23,1995.

29) Facsimile Transmittal f,om Mr. Tom Green to Mr. Rob Elliott, "BWROG ECCS S,uction Strainer Committee Draft Test Matrices for Forthcoming Altsmate Strainer Testirig," dated June 30,1995.

30) NRC Bulletin 93-02 (NRCB 03-02), " Debris Plugging of Emergency Core Cooling Suction Strainers," May 11,1993.

DRAFT 50

DRAFT wm

31) NUREG/CR-6224," Parametric Study of the Potentialfor BWR ECCS Strainer Blockage Due to LOCA Generated Debris," dated October 1995.

32) Letter from Mr. Keith R. Jury of Carolina Power and Light Company to the USNRC,

" Supplemental Response to NRC Bulletin 96-03, ' Potential Plugging of Emergency Core Cooling Suction Strainers by Debris in Boiling-Water Reactors,' dated June 5,1997.

Accession number 9706110277.

33) Letter from H.L. Sumner, Jr. of Southern Nuclear Operating Company to the USNRC,

" Proposed Criteria for ECCS Strainer Design," dated March 25,1997. Accession number 3704010557,

34) Letter from Ngoc B. Le to H.L. Sumner, Jr., " Safety Evaluation Related to NRC Bulletin 96-03 " Potential Plugging of Emergency Core Cooling Suction Strainers by Debris in Boiling-Water Reactors' - Edwin I. Hatch Nuclear Plant, Units 1 and 2," dated June 17,1997.

Accession number 9706200164.

35) Letter from E.C. Simpson of Public Service Electric and Gas Company to the USNRC,

• Proposed Resolution Approach - NRC Bulletin 96-03 Potential Plugging of ECCS Suction Strainers by Debns," dated May 20,1997. Accession number 9705290189.

36) Letter from David H. Jaffe to Leon Eliason, " Safety Evaluation for Hope Creek Generating Station - NRC Bulatin 96-03,' dated October 31,1997. Accession number 9711180235.

37) Letter from G.A. Hunger, Jr. of PECO Energy to the USNRC, " Request for License Amendments Associated with ECCS Pump Suction Strainer Plant Modification," dated May 5,1997. Accession number 9705150005.

38) Memorandum from Carl H. Berlingerto John F. Stolz, " Safety Evaluation for Peach Bottom Atomic Power Station, Units 2 and 3, Relative to the Licensee's Response to NRC Bulletin 96-03," dated September 25,1997. Accession number 971006^349.

39) NRC Bulletin 93-02, Supplement 1, " Debris Plugging of Emergency Core Cooling Suction Strainers," dated February 18,1994.

40) NRC Bulletin 95-02, " Unexpected Clogging of Residual Heat Removal (RHR) Pump Strainer While Opercting in Suppression Pool Cooling Mode," dated October 17,1995.

41) D.V. Rao ano F.J. Souto, NUREG/CR-6367,"Experimentalstudy of Head Loss and Filtration for LOCA Debris," dated December 1995.

42) Title 10, Part 50, Code of FederalRegulations.

43) Alan B. Johnson, Mahadevan Padmanabhan, and George E. Hecker. Alden Research Laboratory Test Report number 92-96/M787F, entit!ed " Head Loss of Reflective Metallic insulation Debris with and without Fibrous insulation Debris and Sludge for BWR Suction Strainers," dated May 1996.

44) Letter from Mr. R.L. Seale to Mr. L. Joseph Callan, " Proposed Final Generic Letter,

' Assurance of Sufficient Net Positive Suction Head for Emergency Core Cooling and Containment Heat Removal Pumps,'" dated June 17,1997.

45) Generic Letter 97-04, " Assurance of Sufficient Net Positive Suction Head for Emergency Core Cooling and Containment Heat Removal Pumps," dated October 7,1997.

46)

Letter from Dr.* Thomas S. Kress to Mr. James M. Taylor," Proposed Final NRC Bulletin 96-XX, ' Potential Plugging of Emergency Core Cooling Suction Strainers by Debris in Boili6g Water Reactors,' and an Associated Draft Revision 2 of Regulatory Guide 1.82, ' Water Sources for Long-Term Recirculation Cooling Following a Loss-of-Coolant Accident.'" dated February 26,1996.

DRAFT 51

}

DRAFT mao /97

47) Letter from Mr. James M. Taylor to Dr. Thomas S. Kress,

• Proposed Final NRC Bulletin E XX, ' Potential Plugging of Emergency Core Cooling Suction Strainers by Debris in Boiling Water Reactors,' and an Associated Draft Revision 2 of Regulatory Guide 1.82, ' Water -

- Sources for Long Term RecirculationCooling Following a Loss of CoolantAccident,'" dated March 22,1996.

48) NUREG 0800, Rev. 4, " Standard Review Plan." October 1985.

49) - I!RC Information Notice 94 57, ' Debris in Centainment and the Residual Heat Removal System," dated August 12,1994.

50) NRC information Notice E59, 'Eolential Degradation of Post Loss-of-Coolant Recirculation Capability as a Result of Debris," dated October 30,1996;

51)

• Performance of Containment Coatings during a Loss of Coclant Accident,' dated November 10, 1994, Bechtel Corporation, Volume lil, Tab 12, of the URG Technical Support Documentation.

52) CDI Report 96-05, Revision 1," Testing of Debris Transport Through DowncomersNentsinto the Wetwell," Continuum Dynamics incorporated October 1996, URG Tenhnical Support Documentation, Volume 11, Tab No. 2.

53) Not Used.

54) BWROG Letter OG95 388161, Attachment 4, *BWR Owners Group Suppression Pool Sludge Generation Rate Data," BWROG, June 1995.

55) BWROG- Letter OG96-321161, Attachment 2, " Suppression Pool Sludge Particle Distribution Data - Average Distribution Calculation," September 1994.

56) Continuum Dynamics Report 95-09 Revision 4, " Testing of Alternate Strainers with Insulation Fiber and Other Debris," URG Technica! Support Documentation, Volume 1.

57) Regulatory Guide 1.1, " Net Positive Suction Head for Emergency Core Cooling and Containment Heat Removal System Pumps' dated November 2,1970.

58) NUREG 0897, Revision 1, " Containment Emer0ency Sump Performance,"

59) NRC Information Notice 92-85, ' Potential Failures of Emergency Core Cooling Systems Caused by foreign Material Blockage " dated December 23,1992.

60) NRC Information Notice 89 77, ' Debris in Containment Emergency Sumps and incorrect Screen Configurations," dated November 21,1989.

61) NRC Information Notice 88 67, ' Pump Wear and Foreign Objects in Plant Piping Systems,"

dated November 16,1988.

DRAFT 52

DRAFT inw 9.0 LIST OF RELATED REPORTS AND DOCUMENTSIN THE PUBLIC DOCUMENT ROOM The following table provides a list of related test reports and documents relative to the ECCS clogging issue that are available in the NRC public document room.

m.

Technical Documents in Public Document Room Document Accession Number 1 Zigier, G., et. al.,

• Experimental Investigation of Head Loss and 9608050121 Sedimentation Characteristics of Reflective Metallic Insulation Debris," SEA NO. 95-970-01-A:2, Science and Engineering 9608050125 Associates, May 1996.

2 Murthy, P., Padmanabhan, M., and Hecker, G.,

• Tests of 9511010046 Particle Settling Velocity in Still Water," Alden Research Laboratory, Inc., Holden, MA, June 1994.

3 Murthy, P., et. al., ' Head Loss of Fibrous NUKON Insulation 9511010046 Debris and Sludge for BWR Suction Strainers," 124-95/M787F, Alden Research Laboratory, tra, Holden, MA, June 1995.

4 Johnson, A., Padmanabhan, M., and Hecker, G., "NUKON" 9511010046 Insulation and Sludge Settling Folicwing a LOCA in a BWR Suppression Pool,' 114-95/M787F, Alden Research Laboratory, Inc., Holden, MA, June 1995.

5 Hoffmann, D. and Knapp, A., 'RMI Debris Generation Testing: 9511010046 Pilot Test with a Target Bobbin of Diamond Power Panels,"

NT34/95/e32, Siemens AG, Karlstein, Germany, July 1995.

6 Memorandum from E. Cramer of SFA to M. Marshall of 9511010046 USNRC,

Subject:

Review of Existing Knowledge Base related to Generation of Iron Oxide Sludge in BWRs, September 21, 1994.

7 Meme andum from G. Zigler to A. Serkiz and M. Marshall of 9511010046 USN <C,

Subject:

RMI Debris Size Distribution, June 9,1995.

8 Video of RMI Debris Generation test in Karlstein, Germany. 9511010046 9 Williams, D., " Measurements on the Sink Rate and Submersion 9412080219 Time for Fibrous Insulation," ITR-93-02N, Transco Products, .

May 22,1993. 9404070203 DRAFT 53

DRAFT m3oe97 Technical Documents in Public Document Room Document - Accession Number 10 Williams, D., "The Effects of pH and Thermal Exposure on the 9412080219 Head Loss of Series A Shreds, NRC Guide 1.82,"ITR-92-3aN, Transco Products, May 18,1992. 9404070203 11 Williams, D., " Experimental Measureme.its on the 9412080219

' i Characteristics of Flow Transport, Pressure Drop, and Jet '

Impact on Thermal Insulation, NRC Guide 1.82," lTR-92-03N, 9404070203 Transco Products, May 18,1992.

12 Williams, D.," Postulations of the Range of Fibrous insulation 9412080219 l Debris Size Generated by High Energy Jet impact,"ITR 93-01N, Transco Products, April 16,1993. 9404070203  !

13 " Air Blast Destructive Testing of NUKON@ Insulation 9404070203  !

Simulation of a Pipe Break LOCA," Performance Contracting i inc., October 1993.

l 14 " Head Loss Tests with Blast Generated NUKON@ Insulation 9404070203 Debris," Performance Contracting, Inc., October 1993.

15 "Karlshamn Tests 1992. Test Report. Steam Blast on insulated 9410130310 Objects," ABB Atom, RVE 92-205, November 30,1992.

16 "KKL - Specific ECCS Strainer Plugging Analysis According to 94101b310 Reg. Guide 01.82, Rev,1, for a Loss of Coolant Accident,"

KKL. i 17 "The Marviken Full Scale Contair, ment Experiment," MXA 9410130310 206, September 1973.

18 "OECD/NEA Workshop on the Barsebeck Strainer Incident in 9410130310 Stockholm January 26-27,1994 Documentation."

19 " Report Conceming the Quantity of insulation Which Was Not 9410130310 Washed Down in Connection with the 314 Event," Sydkraft, PBM 921123, November 26,1992.

20 "Swedish N-Utilities Explain BWR Emergency Core Cooling 9410130310 Problem," ENS Nuc Net, News No 358/92, September 18, 1992.

21 Anderson, " Guaranteed Emergency Core and Containment 9410130310 Cooling - Laboratory Experiments with Insulation on Strainers,"

ABB Atom, RVD 92-193, December 1,1992.

DRAFT 54

DRAFT imam Technical Documents in Public Document Room Document Accession Number 22 Blomqvist and Dellby, "Barseback 1 & 2, Oskarshamn 1 & 2, 9410130310 Ringhals 1. Report from Tests Conceming the Effect of a Steam Jet on Caposil Insulation at Karlshamn, Carried Out Between April 22-23,1993 and May 6,1993." SDC 93-1174.

~

23 Bystedt, " Thermal Insulation on New Steam Genorators and E,10130310 primary Pipes. Influence on Safety During Normal Operation and in the Case of Malfunction," August 2,1989.

24 Fax 6n, "Barseback 1 and 2, Oskarshamn 1 and 2 - Strainers in 9410130310 System 322 and 323. Results from Blowdown Experiments in a Test Rig," ABB Atom, RVA 92-340, November 27,1992.

25 Fredell, "Karshamn Tests 1992. Steam Blast on InNtated 9410130310 Objects, Logbook," ABB Atom, RVE 92-202.

26 Greis, " Pressure Drop in Connection with Fluid Flow Through 9410130310 Insulation Deposited on Strainers," ABB Atom, RPD 92-109 December 23,1992.

27 Hallstem, " Guaranteed Emergency Core and Containment 9410130310

, Cooling. Laboratory Experiments Concerning insulation," ABB Atom, RVD 92-192, November 30,1992.

28 Henriksson, "Ringhals 1 Pressure Drop on Screens from 9410130310 Thermal Insulation Debris," Vattenfall Utveckling AB, VU-S93:BS March 30,1993.

29 Henriksson, "Ringhals 1 - Strainer System 322/322," Vattenfall 9410130310 Utveckling AB, VU-S92:56 December 18,1992.

30 Henriksson and Norling, "Barseback 1-2, Oskarshamn 12, 9410130310 Ringhals 1 - Filter Systems 322/323. Coverage and Back Rinsing Tests with Fiber Insulation and Newtherm, Scale 1:2.

Status Report February 6,1993," Vattenfall Development Company, VU-S93:B14, June 22,1993.

31 Henriksson, "Barseback 1 and 2, Oskarshamn 1 and 2 - 9410130310 System 322/323 Testing of Permanent Strainers," Vattenfall

- Utveckling AB, VU-s93:B12, May 28,1993..

-r -

32 Hyverinen, " Summary of STUK's Research Activities 9410130310 Conceming Strainer Clogging," STUK.

DRAFT ss e

DRAFT imw Technical Documents in Public Document Room Document Accession Number 33 Molander, Arnessen, and Jansson, " Steam Jet Dislodgement 9410130310 Tests of Thermal Insulating Material," Studsvik Material, M.

93/24, March 1,1993.

34 Murthy and Hecker, " Head Loss with Blast GeneraMd NUKON 9410130310 Insulat3n Debris Mixed with iron Oxide Particulate < 7L, April 1994.

35 Nystrom," Evaluation of Transport Velocity for NUKON 0410130310 Insulation Base Woal at Elevated Temperature and pH," ARL, May 1991.

36 Nystrom," Investigation of the effect of pH on Head Loss of 10130 10 NUKON Insulation Base Wool," ARL, April 1991.

37 Nystrom, "Nukon insulation Head Loss Tests," ARL, October 9410130310 1989.

38 Pennino and Heckhr, " Head Loss Tests with Blast Generated 9410130310 NL.lKON insulation Debris," ARL, October 1993.

39 Perssson, " Downward Transport and Sedimenta' ion of 9410130310 insulation Material and the Build-Up of Pressure Loss in the Suction Filters," OKG, March 12,1992.

40 Steen and Dufva," Review Report Concerning SKl's Handling of 9410130310 the So-Called Filter Affair."

41 Tarkrea and Amesson, " Steam Jet Dislodgement Tests of 9410130310 ThermalInsulating Material of Type Newtherm 1000 and Caposil HT1," Studsvik Material, M-93/41, April 7,1993.

42 Tarkpea and Arnesson,

• Steam Jet Dislodgement Tests of Two 9410130310 Thermol Insulating Materials," Studsvik Material, M 93/60, May 24,1993.

43 Wilde, " Strainer Test with Fiber Insulation and Reactor Tank 9410130310 Insulation, Results from the Small Model," Vattenfall

_ Development Company, VU-S 93;B16, June 21,1993.

,; ,, 44 , Vpigt and Jerrebo.." Testing of Straines Performance in the 94101303,10 .. ...

Condensation Pool at Bvt 1," Sydkraft. '

DRAFT se

DRAFT nr.m Technical Documents in Public Document Room Document Accession Number 45 Wilhelmsson and Tinoco, "Forsmark 1 & 2 Strainer System 9410130310 37,7/373 " VU S 93 B8 April 26 1993  :

Principal contributors:

Rob Elliott, NRR Rich Lobel, NRR Kerri Kavanagh, NRR Michael Marshall, RES Al Serkiz, RES D.V. Rao, SEA 4

Frank Sciacca, SEA Clint Shaf..t, SEA DRAFT 57

- - , , - --- - - , , - - , - - , - - , - - - - - - , w- , & .----r---n-,-, - , . -,

1 DRAFT um i Appendix A j i

Calculation to Check Estimates of Bulk Dynamic Pressures Computed by BWROG

References:

A.1 URG Section 3.2.1.2.3, Pags 36, Line 1 i A.2 SEA 96 3105-010 A:2, Debris Drywell Transport Study, Draft Phase 1 Letter Report, 1996 A.3 BWROG Response to NRC Comments and Questions Regarding NEDO 32686, Rev. O Problem Definition: In Reference A.1, the URG states the following:

• The bulk dyr:amic pressures in the drywell far from the break are lower than 0.02 psi.

e The licensee need not worry about damage to insulation located outside the jet's zone of influence.

The following calculations were carried out to evaluate 'bese URG conclusions:

LOCA Information: I

1) Discharge from both ends of a pipe break will reach a maximum flow (M) of 8000 lbm/s, assuming 24 inch double ended guillotine break with full separation (Ref. A.2). A higher flow rate is calculated if containment atmospheric pressure is considered.

2) Instantaneous containment pressure will be 15 40 psi. It takes several seconds for the containment pressure to buildup to 40 psi. The density of the steam and air mixture in the initiPl stages of accident (p)is 0.08 lbm/fi3 (Ref. A.2).

GeometricalInformation: t The following geometrical information was compiled for an average size Mrk l containment (Ref.

A.2). Realistic equipment congestion was used to estimate flow velocities and dynamic pressures.

Kation Location _ Total Area (ft'L Poros_ity_ Not Flow Area (ft'L Neck 750 0.8 000

_ Upper Grating _ , iB00 0.6> m 1080 .

t.ower Gratina 1550 0.57 (= 0.6) 930 DRAFT A1 4

w - -w, - - - 1- ---- -g e

DRAFT uom Velocity and Dynamic Pressure. l Assuming that a quasi-steady state break flow occupies 100% of net flow area (A,..) available for flow, the estimated flow velocities (V.) and dynamic pressures (P. )can be calculated as:

V,= M/(pA,..)

P,= M'/(2,_ par, ,g,)

Msed on these assumptions, the bulk dynamic pressures estimsted are:

ElfYAtlon Locatlan Dynamic _Palant

. _ _ _ (Psi)_ _ _ _

Neck 0.24 Upper Grating 0.07 Lower Grating 0.10 These pressures are probably not large enough to inflict damage on properly installed jacketed fibrous insulation (such as those tested by the BWROG), but are, in the staff's opinion, suihciently large enough to inflict damage on improperly installed insulation (e g., cut to-fit partially exposed insulation blanket pieces some utilities may use to insu%te hard to-fit components such as valves) or not well maintained insubtion blankets (e.g., partially torn insulation blankets that some utilities may not have reid s sd in the past). Note that in the Barseback 2 incident, pressures in the same range generated sub.Jntial debris from aged unjacketed mineral wool blankets.

Conclusion:

The BWROG underestimated bulk dynamic pressures for at least certain types of containments (e.g., Mark l). They should revise these calculations to be representative of all types of containments, it is also recommended hat URG be modified to reflect concems raised above instead of minimizing the potential for debrit generation in the regions far from the break.

Such a caution will emphastze to the licensee the importance of proper installation and maintenance of insulation blankets (especially at hard to reach places). Licensee should also be cautioned regarding the potential for generation of miscellaneous fibrous debris due to these elevated dynamic pressures that exist far from the break.

Resolution: The BWROG addressed this comment in Reference A.3 above. The staff concludes that the BWROG response cited above (Ref. A.3) adequately addresses this issue.

DRAFT A-2

DRAFT tmm Appendix B  !

l Analyses to Verify Values of P , for Selected Materials Reported by the BWROG and l

Suggestions for Development of Scaling Analyses l

References:

l B.1 URG Section 3.2.1.2.3.2, Page 38, Une 13 and Page 46, Table 2.

i B.2 URG, Vol. II Tab 3. CDI report No. 95-06, Air Jet Impact Testing of Fibrous and Reflective Metallic insulation.

B.3 URG, Vol. II, Tab 1, CD1 report No. 96-01, Zone of influence as Defined by Computational Fluid Dynamics, Rev. 3 B.4 ANSl/ANS 58.2 Jet Model, ANSI,1988, Design Basis for Protection of Ught Water Nuclear Power Plants Against Effects of Postulated Rupture.

s Problem Definition: In tha URG, the BWROG provided a table of destruction pressures for selected insulation blankets (References B.1,8.2 and B.3). This table, reproduced here for convenience, lists P , values for insulation blankets installed on 12 diameter piping.

Material Pm(psi)_

Darchem DARMET* 190*

Transco RMI 190*

Jacketed NUKON' with "Sure Hold' Bands 190*

Diamond Power MIRROR

• with "Sure Hold" 190*

Calcium Silicate with Aluminum Jacketing 160*

K Wool 40 i Temp Mat with stainless steel wire retainer 17 Knaupf' 10 Jacketed NUKON' with Standard Bands 10 Unjacketed NUKON' 10 Koolphen K' 6 ,

Diamond Power MIRROR' with Standard 4 Bands Min-K 4

.  ;.- .in gddition, for insulation types where the ,P,.., values are marked by '*'in the table above, the , , ,

. BWROG suggested that the P,.., values for other pipe sizes can be estimated using the following relationship:

P% = P",,,,(12 inches /D)

DRAFT B-1

.n-~ ,.e.. - --

I DRAFT mnw I The DWROG derived the P.,, listed in the table above using the following:

1) The Air Jet impact Tests (AJIT) were ccnducted at the Colorado Engineering Experimental Station, Inc. (CEESl) facility using a 3 inch diameter jet nozzle and 12 inch diameter target pipe to detarmine the farthest location at which a given insulation blanket will be damaged and dislodged from the target pipe. The results of these tests are documented in Reference B.2. The location determined by the tests is defined as ,

(UD),,,.  !

2) The NPARC bascu Computational Fluid Dynamics (CFD) calculations were used to l determine the jet-centerline pressure associated with (UD),... This pressure is listed in Reference 0.1 as P ..

3) The correction for aula listed above is suggested to scale CEESI results to different target pipe sizes.

The staff notes that no scaling issues were addressed in the URG relative to the applicability of test data using 3 inch diameter nozzle from the CEESI facility to a main steamline break (MSLB) in a full size plant.

Several of the BWROGs AJIT tests were witnessed by the staff and its contractor. In addition, the documentation provided by the BWROG was carefully reviewed The following paragraphs summarizes the confirmatory analyses performed and the results of those analyses:

CEESI Test Facility sind Experimental Procedure: The experimental program carr' d out at CEESI appeared reasonable and tractable according to the NRC QA program. The experimental data provided in Reference B.2 was reviewed to determine the maximum distance from the jet (UD) beyond which no roticeable damage was inflicted on the target. Column 2 of Table 1 lists these numbers. The stati made the following observations:

1) In the majority of the tests, the target consisted of a single piece of blanket (24 inches long) mounted on a 12 inch diameter target pipe. This allows for formation of a at 1:

orototyoicallow pressure region on both ends of the insulation blanket leading to air flow from the high pressure mid region to both ends. Such an air flow would inflict more damage than would be expected in a BWR plant, where the insulation blankets are continuous. As a result, the staff believes that the CEESI tests probably overestimate damage, especially in case of fibrous insulation.

2) For some insulation types, the exact location at which damage first occurred was not properly explored. For example, in the case of stainless steet jacketed NUKON',

' , damage was reported at 50 L/D with 12% of insulation destroyed into fines and 29% into large'r pieces. But the URG has not explored damage beyond 50 UD. SI'rfillarly, in the case of unjacketed FJKON'significant damage occurred at 60 UD and no damage at 119 UD. No data points were reported in between. In vicu of such a lack of data, some DRAFT B.2

DRAFT imm i i

of the conchsions stated in URG become questionable; especially considering that this damage information is later used to estimate fraction of insulation transported.

BWROG Derivation of Damage Pressures: The BWROG amployed the NPARC code results presented in Figure 9 on page 18 of Reference B.3 to estimate the jet center line pressures i corresponding to (UD) where damage was first recorded. These jet center line pressures '

were then interpreted by the BWROG as the damage pressures. Based on a detailed review the staff has identified the following deficiencies in the methodology used by BWROG:

1) The URG may overestimate the jet center line pressure at which damage first occurs. l Although this is non-conservative, the staff considers this to be a minor deficiency. It was I only noted for selected insulation blankets. For example, in the case of DPSC Mirror with Sure-Hold bands, URG reported a damage pressure d 190 psid. The staffs contractor, Science and Engineering Associater, Inc (SEA), used the same experimental data and same NPARC code results in Figure 9 of Reference B.3 to estir.. ate this value to be 150 psid. Several such discrepancies were found and all of them are annotated in Table 1 by

in all cases, SEA estimated damage pressures are lower than the URG values. The staff recognizes that interpretation of experimental data is often subjective, which may explain these discrepancies. Nevertheless, the BWROG should explain the discrepancy.

2) The staffs main concern relates to the simplifying assumption made by BEROG to equate the jet center line pressure to the damage pressure. The rationalu for such an assumption is unclear and not supported by experimental evidence documented in Reference B.2. In fact, this assumption is contrary to assumptions made in Appendix A, {

i Page 105 of Reference B.2. It is strongly suggosted that BWROG eLplore other alternatives to characterize damage, rather than linking it to the jet center line pressure.

One such an alternative is to use jet impingement load (force) as a scalable variable for characterizing damage, at least in the case of selected insulations. The following analyse,s were performed to explore this attemative, a) The CEESI experiments clearly suggest that in the case of jacketed insulation, damage occurs only after the outer casings on the insulation blankets (e.g.,

stainless steel sheaths) ere separated from the blanket. Usually, the outer casings are secured by bands or latches of various types, ranging from simple latch and strike devices to 'Sure Hold *' bands with modified 'CamLoc strikes.

The BWROG notes on page 182 of Reference B.2 that a typical failure mode of these devices was deformation (or straightening) of the J Hook or destruction of the latch mechanism, in both cases, separation of the outer casing occurs when the load on the insulation blanket exceeds the structural strengths of the bands.

Since the latch mechanisms or handing used were not typically located at the jet

, centerline in the AJIT tests, and they were failing, the staff believes that the banding or latches were failing at a pressure lowe'r than the jet ceIderline.

Consequently, the staff concludes that damage should be related to the total impingement force to which the blanket is subjected rather than the maximum DRAFT B-3

l

?

i DRAFT imm pressure over a smalllocalized region'. The staff believes that the experimental evidence supports this in more cases than usage of jet center line pressures. In cases where the P. is low, then use of the jet centerline pressure would not significantly impact the calculated ZOI since the average pressure applied on the target insulation would not be significantly different from the average pressure en i the target.

b) Ti > difference in the choice of jet center line pressure versus total force can be  :

quantitatively enlained by considering the radial distribution of pressures in an '

expanding jet illustrated in Figure B.1. As evident from this figure, at each UD jet pressures are highest at the jet center-line and decrease rapidly with radial dis +1nce (R/D). For analysis purposes, Figure B.1 uses a linear profile as suggested by the ANSI /ANS 58 2 Jet Model(Ref. B.4) instead of a more realistic profile derived from more CFD calculations (e.g., Figure 11 of Ref. B.3). Similar analyses may conducted using more acct. rate radial profiles than those used in this study. Nevertheless, the net jet impingement load on the blanket is the sum of multiples of the local stagnation pressure in each radial zone and the area of that radial zone intersecting the blanket. As an alternate, one may choose to relate damage to target area averaged pressure, which is essentially the ratio of jet impingement load on the target to the target area. Figure B.2 compares maximum and target area averaged pressures as a function of axial distance from the nozzle exit plane. A blanket length of 24' was assumed to estimate these target area averaged pressures. From Figure B.2 it can be seen that at an UD of 10, the maximum pressure (or the JCL Pressure) is 100 psid versus target area averaged pressure of 40 psid, which is a large difference. For a particular insulation, if the first reported damage occurred at a UD of 10, the URG shows an estimated damage pressure of 100 psid instead of the more realistic damage pressure of 40 psid.

This deficiency 'could have considerable impact on the URG defined zone-of-influence models.

This impact can be demonstrated by considering pressure isobars in an expanding jet (see Figure B.3). As shown in this figure, the 100 psid isobar includes much lower volume than the 40 psid isobar . As a result, the volume of insulation destroyed would be higher if it were estimated based on drywell volume bounded by a 40 psid isobar as compared to that bounded by100 psid. As noted above, the staff believes this comment is only significant on insulation with high P., values (e g., where JCL pressure is significantly higher than average pressure on the target). The staff recommends that the BWROG address this issue in detail, and where needed, update the damage pressure table provided in the URG. NPARC runs can be used effectively to provide a better estimate of target area averaged damage pressures than those estimated in this calculation using a simplified 1-D jet expansion model 3 For example, would a 4" nozzle generate the same amount of debris as compared to 3".

• For a nozzle size of 24", the isobars in Figure B.3 very closely represent ' target area averaged pressures' at that location, assuming the target to be 24" long.

DRAFT B-4

I

- i 1

DRAFT tam Table B.1. Results of CEESI Testing for Each Insulation Material.  !

Insulation Material (UD) URG JCL' Pressures Dest.* Damage Estir.tated in Pressure Conflimatory (psi) Analysis Darchem DARMET* 5.0 190 190 Transco RMI' 5.0 190 190 Jacketed NUKON'with Sure-Hold * >11' 190 150*  !

Bands DPSC MIRROR' with Sure-Hold' 8.5 190 150*

Band Calcium Silicate with Aluminum 7 160 150*

Jacketing K Wool 15 40 40 Temp-Mat with stainless steel wire 30 17 17 retainer Knaupf' 60 10 10 Jacketed NUKON' with Standard >SO 10 6*

Bands '

Unjacketed NUKON' >60 ; < 119 10 6*

Koolphen K' >80' 6 4' DPSC MIRROR'with Standard 99 4 4 Bands Min-K >100' 4 <4 Scalabil!ty of Results to BWR Drywells: In the CEESI experiments, a 3 inch diameter nozzle was used to sin)ulate a broken oipe, and a 12 inch diameter target pipe was used to simulate a target pipe (Ref. B.2). This raises the following concems related to sca: ability of the experimental results:

1) How can damage pressures obtained for 12' target pipe be used to estimate damage pressures for a .$ aller or a larger pipe? Since allinsulation blankets are attached to

• The maximum distance at which damage or dislodgment is reported for the insulation type of interest.

• The jet center line pressure corresponding to (l>D).. Pjel was extracted from the CFD results of Ref 3.

~ . ' Several different types of Transco RMI insulanon was tested. However, the results presented.here are for TPI 0.024 in. sheath solid end S/S with latch and strike closures. According to the URO this insulation appears to be the most commonly found Transco RMI in BWRs.

' Exact number not known since testing did not go beyond the values listed for which significant damage was evident. '

DRAFT B-5

- - . , .,..r ,. , . . - - _ . - - .m , _- , - . -

a -

g- .

DRAFT imm  !

pipes of different diameters by identical straps., their failure occurs at same target load i irrespective of the target pipe diameter (i e., one should look for load conservation).  !

Therefore, the following equation holds:  !

FD,,, = F'8 ,

where, FD.,, = is destruction load for a blanket installed on a pipe of diameter D F'8 = is destruction load calculated based on CEESI expt. Data for 12' pipe.

The force-pressure relationship is: F% = PD*,,, A p ,,

where, P ",,, = is averaged jet damage pressure (psi)

A p ,, = is area of jet target interaction (in 8).

For target pipes fully immersed in the jet, it can be easily shown that A y..,, = k(D+2t)L k = a constant of proportionality D = Outer diameter of the target pipe (in) t = insulation blanket thickness (in)

L = Length of the insulation blanket (in)

Substituting these variables into the force conservation equation above: i k'(D+2t)*L* PD",,, = k'(12*+2t)*L* P'8 %,

or PD",,, = P'8",,, *(12'+2t)/ (D+2t)

A first order approximation (within i 25%) to the above equation is:

PD",,, = P'8",,, *(12'l D) 1 This is the correction formula suggested by the BWROG. The staff believes it is adequate and can be used for certain types of insulation (i.e., steel encapsulated insulations, RMI etc.).

2)' How can darhiage pressures obtainedIJsing a 3 inch diameter nozile be scaled for

• breaks associated with 22 inch diameter pipes or larger? This issue is pictorially illustrated in Figure B.4. As shown in Figure B.4, in CEESI testing the blar.t,4t is not completely encompassed by the high pressure region of the expanding jet, especially for DRAFT s-s

DRAFT mm4 UD values less than 5. Evidence to that effect can be found in some of the still pictures taken of damaged insulation cassettes (e.g., Figure 4-42 and Figure 4 43, Ref. B.2) where damage occurred over a limited region. In BWRs, the broken pipes can be as large, if not larger than the target pipes. In such cases, high pressure regions of the jet would completely encompasses the insulation blanket, subjecting the entire front and side surfaces to high pressure. This raises questions related to scalability of CEESI results to BWRs. At least in part this issue can be addressed if the damage is thought to be proportional to the totalload. It can be explained as follows.

Consider that in a FWR a postulated LOCA involves a fully separated DEGB of a 24' MSLB. Now assume that an analyst is interested in quantifying the damage caused by the ensuing steam jet impingement on a 24' diameter target pipe and located about 20 ft from the broken pipe (10 UD). As a first step, one can use the ANSI /ANS 58.2 model to predict totalimpingement load on a blanket located on that target pipe. For a 24'long blanket, this load is calculated to be approximately 23,000 lbf, with an average pressure of approximately 40 psi (note that load is calcul&ted assuming the medium to be steam).

Correspondingly, in CEESI air jet tests the insulation blanket located on 12' target pipe also located 10 D (or 30 inches) from tne nozzle would be subjected to an average pressure of 40 psi (see Figure 2). The net force it is subjected to however is only 11, 520 lbf, not 23,000 lbf computed for BWR case above. As a result, care should be taken to ensure that results from CEESI should not be interpreted in terms of UD values, but instead in terms of the impingement load.

On the other hand, in CEESI tests an insulation blanket located on the same 12' target pipe but SD from the nozzle would be et, cted to a total force of about 22,752 lbf. A case can be made that damage observed at 5 UD in CEESI air tests is approximately equal to damage expected at 10 UD following a 24' MSLB.

Fluid Medium of Experimentation: Air used in CEESI testing has obviously different characteristics from high temperature steam released following a LOCA. Once again this difference can be accounted for if care is taken to think in terms of totalload but not UD.

Through NPARC results (see Figure 9, Ref. 3), the BWROG argued that air based test results

" conservatively bound

• the damage expected in case of steam. The staff concurs that compressed air can be used to create high pressure fields at a given UD comparable to a steam jet of same pressure.

DRAFT B-7

l_ _ _ . . _ _ _ . _ _ _ _ _ _ _ _ _ _ _  ;

. l.

. i 1

i I .

l  ?

j l Nozzle (O.D=3") f 200 h 4

n 160 -

g-0 l l l ANSI /ANS 58.2 Based 2 g 120 -

$g  :  : Axial Distance 2g -

LJDn3 I Er 80 ' : *

. + L/D=5 <

l g $,  ;  ; -G= L/D=7 l l

i ae 40

. t/D=10

. gar. L/D=20 J' .

0 '''' 'T N - -- ^

0 1 2 3 4 5 6 l

l Normalized Radial Distance (R/D) l'igure B.I. Radial Distribution of the Stagnation Pressure in the Jet.

The actual pressure profile could be differen; from the linear profile assumed in the ANSI /ANS.58.2 Model.

e l

1 l

1 1

1

.....--_........,_.-_..___-.,_.-_:.~_-.-_--.m_ . _ _ _ . , _ - - _ - _ - . _ . ___ _ __ _________________ _ _ _ __ _ ___________________.___ _._

e' o

e 200

• ^

g S Average Pressure Blanket is Subiected

$ 160 -

_ Mosemum Pressure Stanket #s Subjected to 1  :

g 120 -

I s so -

e e -

1

[ 40 -

a:

' ' 16 0 ' ' ' ' ' ' '

0 2 4 6 0 10 12 14 16 18 20 Aaial Distance from the Nonle Exit (UD)

Figure 11.2. Comparison of Ascrage (SEA Suggested) and .%!asimum (IlWROG URG) Pressures on a 24" long and 3" thick blanket installed on 12" target s ersus its adal location.

7 ANSI /ANS 68.2 Jet Model 6 ' 'u

_ Impingiment Pressure Isobars '.' .

l ' ~ A '

g3 steam from 1000 psi

,,A;,,.A.A A

g . . . Ai g4 .;. A.A g A 3

A AA;,A.A g3 -.'..' . . . . . ' , .'  :. ...;. a 100 psid 3  :

o 40 psid fj W g c.'

]21 :

7 A Jet-Boundary

i cogDo o- -

a . . . .

4

... 552s. i i i o

0 i -

J. 5 10 15 20 Axial Distance (UD) l'igure 11.3. I'ressure Isobars for Steam Jet Expanding from a DEGil with Full Separation.

Each Isobar llounds the Volume of Jet in which Impingement Pressures are liigher than the Value Indicated in the Legend

INWD,3 l n-A. 3" Air Jet Impinging on a Target (Configuration used in CEESI tests)

=

',.. \, Jet Boundary

• • ..'..*,,.+

n .

Jet Center Line lRo/D = 0.5 l Target Pipe 3,) '

\

N

'N N '

, \0 \

e *

\

, gi' i gQed'O ,g #* y\pC*

4% W ggCs5CI ........... \

Figure 4. Schematic illustration of CEESI Testing Versus BWR Drywell Steam Jet Impingement on Targets.

l 1

DRAFT wnw Appendix c i

Calculations to Examine Accuracy of Reported Jet Volumes Bounded by Selected Load Isobars.

References:

C.1 URG Section 3.2.1.2.3.2, Page #38, Line 13 also Page 46. Table 2.

C.2 URG, Vol. II, Tab 3, CDI report No. 95-06, Air Jet impact Testing of Fibrous and Reflective Metallic Insulation.

C.3 URG, Vol.11. Tab 1, CDI report No. 96-01, Zone of Influence as Defined by Computational Fluid Dynamics, Rev. 3 C.4 ANSI /ANS 58.2 Jet Model', ANSl,1988, Design Basis for Protection of Light Water Nuclear Power Plants Against Effects of Postulated Rupture.

C.5 EPRI NP-4302, Two Phase Jet Modeling and Data Comparison,1986.

Problem Definitiort in Ref. C.1 and C.3, the BWROG reported volumes of jet regions enveloped ,

by selected Isobars (2,4,6,10,17,40,160, and 100 psid). These volumes were computed as a function of radial and axial off sets. The calculations used the NPARC code with thermodynamic properties of steam.

The following calculationswere performed to evaluate the adequacy of the BWROG estimates using the ANSI /ANS-58.2Model(Ref. C.4). These calculationsare confirmatory,and intended as a check of the BWROG calculations presentedin Reference 2. The staff believes that the NPARC computer code used in the URG to model steam line breaks is a more capable method to model steam jets than the ANSI /ANS 58.2Model. However,the ANSI /ANS 58.2Model has been thoroughly bench-marked with experimental data, and its accuracy (and limitations) are well established (Ref. C.5).

Furthermore,the ANSI model also provides a good bounding estimate for recirculation line breaks which cannot be modeled with single phase CFD codes such as NPARC.

Volumes of Jet with Pressures Higher than a Prescribed Damage Pressure: Figure C.1 presents impingementload isobars for a postulated DEGB with full separation in a main steam line and a recirculationline break. Figure C.2 presents similar plots for a DEGB with limited separation (no radial separation and an axial separation of 0.25D). These isobars were drawn based on ANSI /ANS 58.2 Model. Careful examination of these figures shows the following trends:

1) The jets originating from Main Steam Line Breaks tend to remain focused around the jet centerlines for longer distances. For example, Jet Center Line Pressure is higher than 1.5 bars (or 22 psid) at distances beyond 20 L/D. Note also that this value compares favorably with NPARC results for total pressure plotted in Figuro 9 of Ref. C.3. On the other hand, the radial extent to which the 22 psid isobar spreads is about 3D.

DRAFT C.1 f

y ,

l l

DRAFT we l

2) The jets originating from recirculationline breaks d:ffuse te low pressures more rapidly than those from main steam line breaks. For example in the case of recirculationlN break, the ,

22 psid isobar extends only up to about 7 UD, instead of 20 UD in the case of main steam l line break.

3) The main steam line results (especially jet center line pressures) are in reasonable agreement with the NPARC results documented in Reference C.2. '

The following conclusions can be drawr, regarding breaks with limited separation:

1) The main steam I;ne breaks once again remain focused and go far beyond a recirculation break of similar magnitude.

2) In general, the jet expansion from a limited separation break is more modest in extent as compared to a full separation.

The jet volumes enveloped by cach isobar were calculated. Figure C.3 presents these estimates for the following cases:

1) A postulated main steam line break as predicted by ANSI /ANS 58.2 Model.

2) A postulated recirculationline break as predicted by ANSl/ANS-58.2 Model. Note that these values are applicable only to the sub-cooled blowdown phase.

3) The volume of a postulated spherical model (such as that used in NUREG/CR-6224), with increments in radius from 3UD to 15 UD.

4) A postulated main steam line break as preuicted by NPARC in the URG (Table 1, Page 46, Ref. C.1).

5) A postulated recirculationline break as predicted by NPARC in the URG (Table 1, Page 47, Note 5, Ref. C.1).

The comparison presented in Figure C.3 suggests the following:

1) For impingement pressures larger than 40 psid, the volumes enveloped by the jet predicted in the URG are more conservativethan the ANSI /ANS 58.2 predictions. As explainedin Ref.

C.3, this is expected since the ANSl/ANS 58.2 model does not model shock dynamics and radial pressure distributions mechanistically. This comparison presents a good check of URG calculations.

2) For imping'ement ~ pressures between 25 and 4 psid, URG results'and NPARC results are very close to each other. Once again, this comparison represents a good check for URG calculations.

DRAFT C-2

l- i DRAFT ins

3) At pressures below 4 psid, the URG volumes are lower than the ANSI Model by up to a factor of 4. This resun is not surprising because neither modells very effedive at these low pressures. The staff notes that very few insulation types are significantly damaged at these low pressures.

4) The muniplicationfactors provided for recirculation breaks appeer to perform reasonably well for all pressures.

5) Overall, it appears that 17 psid isobar bounds a volume equal to that of a sphere with a radius equal to 7 IJD. Similarly, a SUD sphere was found to adequately bound the isobars with pressures larger than 150 psid.

Figure C.4 presents a similar comparison for DEGB with no radial separation and an axial separation of 0.25 D. Once a0ain good agreement was observed between the ANSI /ANS 58.2 results and NPARC resuRs at low pressures, while at high pressures the NPARC resuns were conservative.

As a result of these comparisons, the staff concludes that the URG predicted volumes are conservative or reasonable in the pressure range of interest, depending on the impingement load.

Their use, if properly justified, is acceptable.

..~ .

DRAFT C-3

- , - - ,.-,e,, .

, , - - , ,- --~ .~-

• l I

7 00 - I s .-

M p 400-

1. 3 00 2 00 -- -

g on 1 0 00 ~- --G -S 0 00 2 00 4 00 6 00 8 00 10 00 12 00 14 00 16 00 18 00 20 00 UD bBouncary +P.7 6 -G-P.2 5 P.15 lA. Main SteamlJne Break. Full Separation l 12 00 to DO , --

8 00 g.*-

4=-

2 00 ~

, c oc --

.......... m 0 00 5 00 10 00 15 00 20 00 25 00 LD

~

-G- Bounaay p.7 5 --B** P.2 5 - P.1 s lB. Recirculation Line Break. Full Separation. Sub cooled blowdown l Figure C.I. Jet Impingement Loads Computed Using ANSI /ANS 58.2 Model for DEGil with Full Separation.

e

A. Main Steam Line Break (0.25 D Separation) n; - - . _

a: _ -

,a _

ue -

t .A .

.e- ...__ . . . . , . . . . . . . . . . . . . . . . _ -

040 1 03 2 40 3 00 4 00 $00 8t0 ?M tr0 900 10t0 R t.S I

• • G- E omsr/ P =7 5 -#- P Q $ P 6 5 B. Recirculation Line Break

$ CO (0.25 D Separation) 4504 - '

4C0 -

150 - -

)0- --

N 210 -- -_

2:C-g'

y 10' b

-a, r1

? --

000 Y1nrr 6 -

100  ;;0 300 400 100 600 700 800 1;;,0 900

> Lt)

• B owthfy -O- P a7 5 -G- P 2 5 -O-- P 4 5 Figure C.2. Jet Impingernent Loads Computed Using ANSI /ANS 58.2 Stodel for DEGil with I.imited Separation.

O

l l l

{ 10" 5 Steam Break. ANSI Steam Break . URG

' 15UD 10 68 5 $ e recal ode 10UD #

\

~

7UD l 10 n

/>UD -

d

- _gjl g _ __

3

' , 0-

. Il 5 _ _ _

i

rbId.". 190 160 40 25 Sal 17 m _ _lH 6 2 Isobar Pressure (psid)

Figure C.3. Jet Impingement Load Isobar 5olumes for Fully Separated Steam Line and Recirculation Line llr aks. Ilases: ANSI /ANS 58.2 and NPARC/URG

, 10" j E Steam Dreak. ANSI i E Steam Dreak.URG Spherical Model

/

E10" 2UD

!huilTl14.}--

}

190 160 40 25 Isobar Pressure (bar)

Figure C.4. Jet Impingement Load Isobar Volumes for Limited Separation 17 6 2 Steam Line lireaks. Ilases: ANSl/ANS 58.2 and NPAPPJURG

' l j

.i J

1 .

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

- . n_ .- . - - . a ~ - _- .. .-- . __ >

l 1

DRAFT imm ,

Annendix D Calculations to Examine Zone of influence Models Proposed for Two Phase Jets

References:

D.1 URG Section 3.2.1.2.3.2, Page 46, Table 2, Page 47, Note 5.

D.2 URG Technical Support Documentation, Volume ll, Tab 1, CDI report No. 96 01, ' Zone of Influence as Defined by Computational Fluid Dynamics," Rev. 3. '

D.3 URG Technical Support Documentation, Volume Ill, Tab 14, GE Report # DRF A74-00004, " Total Pressure Topography and Zone of Destruction for Steam and Mixture Discharge from Ruptured Pipes," September 1996.

D.4 ANSI /ANS 58.2 Jet Model, ANSI,1988, ' Design Basis for Protection of Light Water Nuclear Power Plants Against Effects of Postulated Rupture."

D.5 EPRI NP-4362, 'Two Phase Jet Modeling and Data Comparison," 1986.

D.6 NUREG/CR 2913,'Two Phase Jet Loads." Sandia National Laboratories,1983.

D.7 SEA 96 3105-010-A:2,

• Debris Drywell Transport Study,' Draft Phase 1 Letter Report, 1996 Problem Definitiort in Ref. D.1, ths BWROG proposed the following correction factors to account for reduction in jet volumes corresponding to the recirculationline break, as a function of the target load:

Insulation Destruction Correction Factor for Pressurejp_s_l) _ _ Recirc. Line Break 0-20 10 20-30 0.9 30-40 0.8 40 50 0.7 50-60 0.5

>60 0.4 The basis for development of this table is documented in References D.2 and D.3. The following calculations were performed to evaluate the the appropriateness of these factors. In these calculations, ANSI /ANS 58.2 Model(Ref. D 4)was again used as a basis for the calculations. The .

results are as follows:

Jet Volumes Corresponding to Subcooled Water Discharge: Immediately following a LOCA, sub-cooledwater exits the break and expands into the containment. Past computer runs using the RELAP code (Ref. D.7) suggest that sub cooled blowdown occurs over a period of 510 seconds during which exit quality of the break jet increases from 0 to 0.02 while the vessel pressure decreases from 1000 to 700 psi. Fig Jres C 1 and C.2 hhow expandingjet isobars for flashing sub-cooled water discharged during thl: initial stage of a postulated DF.GB in a socirculation line.

Figures C.3 and C.4 compare the volumes of jet bounded by each irapingement load isobar with DRAFT D-1

. _ . . -- , , . -,-_m.,..,

DRAFT mm steam break and with as with URG predictions provided in Reference C.1. Reference C.1 predictions are based on the information presented in the table above. As shown in these figures, during subcooled water blowdown, the corresponding Jets tend to be less energetic. For recirculationline break jets, the staffs confirmatory analysis estimated volume ratios are consistent with the fractions listed above.

Based on this, the staff concludes that the values presented in Appendix A to Reference D.3,

" Volume of influence for Saturated Steam versus Saturated Water,' Rev. B are reasonable. The staff also notes that past debris generation data reported by European investigators also supports this finding (Ref. B.5).

Jet Volumes Corresponding to Two-Phase Mixtures: After initial blowdown, the exit quality increases steadily; thereafter,the blowdown at the exit plane consists of two-phase mixtures (Ref.

D 7), Jet impingement loads' associated with such flows were experimentally measured in the Marveklan tests (Ref. D.5). Moody has proposed a scaling scheme which justifies usage of Marveklan test data (Ref. D.5) to BWRs. The scaling criterion is acceptable, and as such that data has been widely used for similar purposes before (e.g., References C.5 and C.6). However, the URG does not explain in a tractable manner how the Marveklan test data is used to corroborate the correction factor table presented above. As a result, the following analyses was performed:

The pressures reported in the Marveklan tests can be tabulated as shown in Table D.1. A plot of jet center line pressures (in column 9) versus UD is presented in Figure D 1 for three types of blowdowns: sub-cooled liquid, steam / liquid mixture and steam. A review of Figure D.1 suggests the following:

1) Stagnation pressures near the nozzle would be higher for subcooled uteaks over a narrow radial range.

2) Far frorn nozzle, for UD > 4, the stagnation pressure is higher for steam line breaks as compared to both subcooled and saturated liquid breaks.

3) However, mixture and steam line breaks look somewhat similar thereafter.

These trends are well established and have been explained using the K-FIX code in Reference D.5, and using the CSQ code in Reference D,6. Therefore, Figure D.1 can be used to conclude that mixture blowdown at a given stagnation pressure is no more severe than steam blowdown from the same stagnation pressure. Coupled with the fact that actual vessel stagnation pressure falls to about 700 psia prior to onset of two phase mixtures, this finding was used to compute the load isobars for mixturejet expansion during blowdown, Figure D.2 illustrates the isobars. Comparison of isobars presented in Figure D.2 with those presented in Figure B.1 would clearly establish that pressures

,, , associated with two-phase blowdowrj are considerably lower than steam h, lowdown. ,

Further calculations were performed to corroborate the correction factors listed above. These calculations support the conclusion that the correction factors cited above from theURG do conservatively bound the jet volumes. Therefore, the sta4 concludes that use of these correction DRAFT D-2 m- _ . , , - , . . _ , _ , _ . - ,

l l

r DRAFT inw j factors is acceptable.

Table D.1. A summary of Marveklan Test Data for Blowdown. [Ref. C.6)  :

Test UD Time Thermodynamic Vessel Pressure JCL Ratio of

-- _._(_s_) _ % Erit.~ 4_T_CCl Range Average _(MPa) _ JCl/ Stag.

-10 ~ 1.2 0 29 Sub-Cool 0-23~

4.38.1

~ ~

3.7 ~

3.7 1 2Si-45~ MEuie x>0.TO 3.i25 2s ~-1 12 0.4 52 55__ Steam x = 1.0 _1.s1.1 1.3_ 0.78 6Te

_8 2 0-25 Sub-Cool _ 0-23 4.3 3.2 3.75 1.875 0.5 25-44 Mixture x>0.10 3.2 2.9 _ 3.05 0.61 0.2 20iIi

^

49 55_ Steam x = 1.0_ 1:55_ 6T465 ~~673 ,

4 ' 0-25

~

7 Sub-Cool 0-23 4412 f8~ 6 58 0.1 25-43 Miture x xiO;iO 3.22.'s 3 0.375 0.125 48:55 Siesm x = i~0' 2Ti'il 1's. 0.28 0.175 DRAFT 04

_m_ + - - - - '

4 *

Analysis of Marveklan Data BWROWURG 1 -S

~

~ $ Subcooled m 0.8

$ Mixture

_- saturated (U _

e .

0.6 :

O O 0.4 .: #

0.2 h + o o

0~ '

1 1.5 2 2.5 3 3.5 4 UD Ratio 7 DD

-- *pg&*r 8..c0

>:0 -

2

0 . -

w0 -- e w ooo *88 .... ..-. -

0 20 2. 400 iN 400 .000 ;2 00 14 00 16 t0 18 00 20 CO

!.t

+ B c ning' -O- P m7 5 -.- P '2 5 + P al 5 171gure D.2. Jet Impingement Load Isobars for Two Phase Mixture Jets that Result During 1, ate Stages of a Recirculation lireak LOCA.

,-n.

o' DRAFT imors7 Appendix E Calculation to Examine the URG Guidance on Thin Bed Effect on Alternate Strainers

References:

E.1 Section 3.1.3, Page 16. Line 7 E.2 Section 3.2.1.1.1, Page 28, Line 12 E.3 Section 3.2.6.2.3, Page 117, Line 24 E.4 URG Technical Support Documentation, Volume I, Tab 2, ' Testing of Attemate Strainers with Insulation Fiber and Other Debris," Appendix l Problem Definition: In the URG, References E.1, E 2, and E.3, the URG states that:

• The thin-bed effect is an issue for semi-conical strainers, but not for stacked disc stralners or any other 'altemate strainers," and e

Therefore, licensees that propose to use altemate strainer designs need not analyze medium LOCAs. because they are not limiting relative to debris generation and transport.

As a result, a series of data-comparisonanalyses were performed to evaluate the completeness and accuracy of the URG statements regarding thin bed effects.

Results: The ability of thin fiber beds to trigge: large pressure drops when combined with particulates (e.g., corrosion products, dirt, etc.) has been demonstrated by both the Perry and Limerick events. Studies by the NRC, European regulatory agencies and the BWROG have focuced on characterizingthe thin-bed effect. Figure E.1 provides a sketch of the ' thin-bed effect*

as it relates to head loss across the strainer. As shown in this figure, at very low thickness (<1/8*)

the strainer surface is partially covered by fibrous shreds. Corresponding to this situation, the nead loss across the straineris low. With a slight increase in the thickness, the straineris more uniformly occupied and starts to filter out some of the particulate debris passing through the debris bed.

However, as experimentally evidenced, these beds break-down or undergo reconfiguration under the influence of the larger head losses caused by the particulate filtration (See NUREG/CR-6224, Appendix E). Tnis trend continues until the bed gains the strength necessary to sustain head losses. Thereafter,there is a rapid increase in the head loss as the thin fiber bed continue to filter out a large fraction of the particulate debris. The resulting debris is characterized by a very dense, relatively impervious, particulate bed formed by a relatively small number of fibers. At higher fibrous debris bed thicknesses, this trend reverses because the debirs bed starts to become more porous.

The thin-bed effect has been widely observed for flat-plate, cylin&ical and semi-conical (truncated cone) strainers. Table E.1 shows the URG data for semi-conical strainers demonstrating the

. . effectiveness of thin-beds to introduce large head losses sAs can be readily seen from this table, high head losses across the strainer is possible at very low insulation loading when coupled with '

large corrosion product (sludge) mass. This data confirms the high head losses reported in the Limerick and Perry events.

DRAFT E-1

/

DRAFT tmom i Proving existence of or non-existence of a thin-bed effect in stacked disc strainers is more complex.

Figure E.2 illustrates the debris bed buildup on a stacked disc strainer surface. When the cavities of the strainer are partially filled, a substantial portion of the strainer surface is still open for particulate flow without being filtered. For these situations, the BWROG experimental data (see Table E.2) has shown that the thin 4)ed effect is not an issue. Even for the case when the cavities are nearly full, there has not been a drastic increase in head loss, in spite of large sludge to fiber mass ratios. However, this does not absolutely rule out occurrence of the thin bed effect in stage 3 of Figure E. The BWROG has not reported any experimental data supporting their conclusion of 'no thin bed effect' during this stage, as all their data reported in the URG is for stages 1 and 2 of Figure E.2. Although test PS provides data for the case when the insulation goes beyond the

. cavities, it is one single data point and corresponds to a rather thick insulation layer (3 inch theoretical thickness).

Table E.1. Experimental Head Loss Data for Truncated Cones Test ID insulation Bed Sludge Mass Head

__ . _._(Ibm) _(inch) libm) _(#) (in-0 0.5 0.14 500 1000 425 7 1 0.28 60 60 385 8

3___ 0.83 16 5 >500 Strainer Surface Area = 18 sq. ft_

Strainer Flow Rate = 5000 GPM

+

(

DRAFT E2

..~m. - -- - , , .- - , -, , - . , ,

I

?

l DRAFT umm  !

Table 2. BWROG Head Loss Data for Stacked Dise Strainer.

Test ID Strainer Type Cavity insulation % Cavity Sludge Head

__ (ft3Cpml__,_ ___ Am]_(in water) 21 Stacked Disc #1 3.53 1 11.8 180 23 22 Stacked Disc #1 3.53 3 35.4 180 65 22 Stacked DiscTW1 3.53 3 35.4 240 72 22 iStickedDiscW1 iS3 4 ~ ~ W.l! NO 'f56- '

23 . Stacked Disc #1

~

3.53 6 70.8 240 350

-P2 stacEed Disc #2 10 17 70.8 85 7 P3 Stacked Disc #2 - 10 25 104.2 100 10 .

P4 :StackedDisc #2- 10 3 12.5 100 0 PS Stacked Disc #2 10 50 208.3 100 28 r in Ni the insulation goes beyond the cavities. The rest of the insulation forms a uniform layer 3 inch thickness.

Additionslexploratory calculations were done to evaluate the question, 'should the licensees who propose to use alternate strainer design (e.g., stacked disc strainer) analyze MLOCA?' Calculatiors performed for NUREG/CR 6224 for a MLOCA postulatedin the reference plant (BWR4, Mark 1) with an assumed drywell transport factor of 1.0 for transportable debris, transports an insulation volume of 24 ft' (or mass of 10 lbm). The BWROG data clearly suggests that both stacked disc #1 and stacked disc #2 can accommodate such fibrous loads coupled with sludge to mass ratios of up to 30 without noticeable increase in head loss (see data for test P4). Therefore, the staff believes that MLOCAs can be screened out if stacked disc strainers are used.

Based on the above analyses, the staff has concluded that:

1) The BWROG claim that thin-bed is rot observed in attemate strainors is unsubstantiated.

The reported data is for two strainer designs (stacked disc and star). In addition, the data addresses a narrow range of experimental parameters. The BWROG should provide addrtional data covering a wider range of experimental parameters (e.g., fiber volume and sludge to fiber ratios) for all attemate designs proposed for to provide generic guidance such as that stated in the URG.

2) The existing dats, however, suggests that licensee may screen out MLOCAs if they use stacked disc strainer 2, star strainers or other such strainers with large cavity (crevice) capacities for debris builduo, which prevent even debris bed buildup across the strainer surface at low fibrous debris loads.

DRAFT E3 i

me em ir- e m m -=w r* r--

DRAFT ins Apnendix F r

Mapping of the Zone of Influence (Z01)

References:

F.1 URG, Section 3.2.1.3.2,

• Method i - Teqet Based Analysis Using Umiting Size Zone of influence," BWROG,- 1996. .

F.2 CDI Report No. 96-06,

• Air Jet impact Testing of Fibrous and Reflective Metallic  ;

Insulation,' URG Technical Support Documentation, Vol.11. Tab #3. '

Problem Definition: In the URG [Ref. F.1) the BWROG recommended that:

• A spherical zone of Mfluence be used to define the geometrical region overwhich damage to the insulation blankets would occur, and e For fibrous insulation (e.g., Nukon) tLs reginn is as large as 10 break-diameters in radius.

Neither of these assumptions were fully substantiated by the BWROG. Furthermore, the BWROG stated that these assumptions would yleid conservative estimates of the ZOI since the assumptions ignore presence of pipes and other structuralimpediments. A series of computationalfluid dynamic simulations of the jet expansion with and without structures were conducted to verify adequacy of the BWROG recommendations.

Methodology: In the URG, experiments were conducted to determine the failure pressure for Nukon insulation cassettes; both encapsulated and non-encapsulated. These experiments reveal that damage is possible when the local total pressure exceeds 10 psi. Considering that downstream of shock front, total pressure is made up entirely of the dynamic pressure, the local flow velocity that would be capable of inflicting damage can be calculated as follows:

P., = % p V2 /g, Where, p is gas (air) density at near atmospheric conditions that exist downstream of shock (#/ft3)

V is gas velocity (ft/s) g, is conversion factor (32.12 lbf/lbm)

DRAFT F.1 F

w- , +.,.ve - ,-. , ay -4 w

i 3

DRAFT w xvs7 - l After appropriata unit ccaversions, the velocity corresponding to damage pressure is .

.V2 = (10)(32.12)(144)(2)/(0.067')

V = 1175 ft/s (or 358 m/s) .

Thus damage is possible whenever the jet velocity exceeds 358 m/s, although more conservatively damage can be assumed for velocities in excess of 300 m/s.

The general method followed in tha present calculation is as follows:

e Conduct CFD calculations to exarnine velocity fields in a freely expanding jet. These calculations can be used to provide a reference case and also to compare the results with the BWROG results in Ref. G.1 and G.2.

e Examine the impact of adding structures in the pathway of the jet, up to the congestion levels typical of BWR drywells.

Results and Discussions Calculation M1: What are the sphericalregion dimensions recommended by the BWROG7 For Steel Jacketed NUKON';

P , for Nukon'in Table 2 = 10 psi Extract A from Table 1 (as predicted by step #3): A = 4708 Volume of the 201: V = 4708 D*%,, ,

D%,, = 24 in. (or 2 ft.)

V = 37,664 ft*; which is about 25% of the drywell volume, in terms of equivalent sphere: R' = % (4708-6/n)D*; R= 10.4 D ,,

The radius of thb sphericai region is 10.4 x break-diameter. These are large spheres compared to NUREG/CR-6224 and any other volumes previously used. This is used as reference case for judging conservative nature of BWROG recommendations.

Calculation #2: Do the BWROG predicted pressure fields for freely expandingjet appear reasonable?

To examine this, air jet originating from a 20-cm diameter nozzle was allowed to expand freely into a cylinder 400 cm in diameter (20-D) and 400 cm in length (20 D). The inlet to the nozzle was air at a pressure of 600 psi and sonic velocity (300 m/s). The expanded jet flow fields are calculated usirig the following computational details:

The dersity was derived from the CFD calculations conducted as pan of this study. This taken into account combined effecu of lower than atmospheric pressure and lower than atmospheric temperatures m thejet.

DRAFT F-2 4

9 e, - ~ m-.-

1 I

l DRAFT urm l

AR = 3 cm; AZ = 2.5 cm; A0 = 45" J 4

Typical time step = 10 to 2.5x 10 4seconds (lower time step initially)

- Typical Execution time = 12-14 hrs CPU for 0.15 seconds of real time The results of the CFD simulations are illustrated in Figure F.2a. As shown in these figures, freely expandingJets resemble semi-conicairegions as described in NUREG-0897. The static pressure approaches atmospheric pressure with 1-2 break diameters and remains steady thereafter. The local fluid velocity increases initially reaching values as high as 850 m/s. ,

However, this reduces to slightly above 300 m/s at a distance of aboue 10 nozzle diameters and remains fairly constant up to and well bepnd 20 D. This confirms the behavior described in the URG, Thus it is likely that high flow velocities would be maintained at very long distances for a freely spandingjet. Figure F.2b shows flow velocity" isomers

• for a freely expanding jet.

The dynamic pressures calculated in the present study are slightly below those in Figure 9 of Reference 2.

Calculation #3: What is the impact of adding a simple structure such as a grating?

To examine the impact of a single grating-like structure, a baffle plate is artificially introduced into the calculation domain. The structure was denoted to be with an area blockage of 20%,

and pressure drop characteristics given by AP = 0.125 (pV2fg,) w This case is shown in Figure F.3a, with the predicted flow fields shown in Figure 3b. As shown here, presence of a single grating can considerably alter the flow fields, such that down stream of the gratin 0 flow velocities are reduced well below 300 m/s necessary to inflict damage. Thus gratings can provide considerable " shadowing" effect. However, it is not clear, if the gratings can withstand such loads, in this context, note that gratings are not load-qualified structures.

Calculations #4: What is the impact of pipes and I-beams arranged in a manner similar to those commonly encountered in BWR drywsils?

To examine this issue a total of five structureswere introducedin the pathway of the expanding jet. As shown Figure F.4a, thase structures included:

e A 20 cm diameter pipe located in the centerline of the break jet to simulate the effect of other end of the broken pipe.

e Three pipes, each 20 cm in fameterand located in the pathway of the jet off set from the jet center line considerhbly.

• ' One pipe anchored on to a 1-beam; the pipe is 20 cm in diameter and I-beam is 20 cm in height.

' AR, AZ, and A0_ radial, axial and angular nodal lengths used in the discretization.

• The value of momentum loss coetTicient (0.125) was calculated from hand-calculations using approximations. This value if at all under predicts pressure drop and thus conservative.

DRAFT F-3

DRAFT m3o<s7 At any given axial location, flow area averaged blockage is less than 40%. The degree of congestion is consistent with the values identif;ed from a survey of BWR drywell:i. To minimize computationaleffort, the pipes were not modeled in detail". The reaults of the CFD simulation are shown in Figure F.4b. As shown in this figure, higher than 300 m/s velocities were observed at distances clnse to the break opening. However, farther from the break, flow velocitiesdecreased considerably. Resulting velocities are well below the velocities required to inflict damage to the insulation blankets.

To further confirm this finding, the following variations were adderi-e A cavity was added to create a narrow region that closely resembles the neck region of the drywell. This case was created to analyze breaks postulated in the neck region of the drywell.

• The boundary conditions were altered to simulate Mark ll drywells.

In all these cases, pipes and I-beams were found to have resulted in dispersing air jet over a substantially larger cross-section. The resulting flow velocities outside a spherical region 7D in radius were considerably lower than the 300 m/s.

Conclusions:

A series of CFD simulationswere carried out to examine the impact of drywell structures on the zone of influence. it is not the intent of these calculationsto address this isste comprehensively, instead, the focus had been to obtain order of magnitude estimates into jet dispersion when subjected to drywell structures. As a result, the results of the calculations described above should not be interpretad as 'best-estimate" predictions.They were conducted for insights only.

These calculations suggest that usage of a spherical zone of influence is appropriate, since it tends to better approximate the jet expansion in a congested drywell volume. They suggest that presence of structures, such as pipes,1-beams and gratings would sufficiently disperse the jet. As a result, the jet velocities outside the sphericalzone of influence with a radius of 7D are lower than thosh required to inflict damage to the insulation blankets. Thus, use of a 10D sphere by the BWROG in the sample problem is conservative and appears reasonable.

Therefore, the zone of influence a i estimated in Calculation #1 above appears to be sufficientV conservative, it should also be recognized that conclusions reached are subjected to the following assumptions:

e Levels of congestion (" piping forest") is limited to a narrow region closer to the break.

Addition of structures beyond this region would further reduce velocities.

The structures were not modeled exactly using CAD drawings of the plants.

" An additional computation suggested that this is not a major issue.

DRAFT F-4

e 1

f l \

l l

1 l

l i

4 4

l

s. I l .

.: . s

, . ~*

, t g a' ' '

i

• N -

i

.' ..'" .

• tr

$ 1 .

l -

i l

i - <

! i ,

g l j

I 4  ; .

- l l

i l b '

l .

. l I ,

I

Figure 2a. Velocity Fields in a Freely expandingjet. The Nozzle diameter is 20 cm. The volume of the

! computational area: 400 cm Diameter Cylinder,400 cm in length. The time step 104 seconds. Reponed flow fields for a quasi-steady state.

i a

4 i

1 l

l j -

e s , - *

' " ~ ~ '

_.s, .,

g i

(

s, i #

s s

q  ? $3 '

'x

\

\

, a. . -

m ( .. 3 - %

kN ,, , - _. . ..

o P

4 we Y Figure 2b. Velocity contours in a freely e.cpandingjet. The associated dynamic pressure can be calculated 2

as P = % pV /gc. At a velocity of 380 m/s, the dynamic pressure is 10 psi. This calculation confirms that for distances beyond 20 D, the jet is fairly flat and it covers the blanket completely.

L___ _ _ _ _ _ __..____ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ ._ _ _ _ _ _ . . _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ __

l 1

l I

l l

I I

1 l

i .

i x

., r a

\ '

d J

s

, d f \

I -

s . x g s

' \

\

T . 4 .

' a

.i.a -

s .

Figure 3a. Geometry used to simulate the efTect of a single grating on the expansion of a airjet. The piping region was not used in these calculations. The Baffle Plate Approximation was used to simulate the floor grating. The bafile plate was assumed to have a blockage of 20% and a non-linear head loss coeflicient of 0.125. These calculations were performed mainly to support PIRT.

r--- -

l 1

. l l

1 I

i 1

l \

l l

l I

I

. 4 i

J l

......n. .-. _ . . _ . . . - - - -

~

,g .y - - - . . . . - -

l r i

, I

$ e .

p

. s * *

=

, (

e  ;

J ,

l C

l a

s .

. 1 i .-

l

\'

I l

Figure 3b. Flow selo:ity fields and contours high lighting the impact of adding a bafile plate in thejet l expansion pathway. As shown here, the flow velocities donstream of the bafile plate are less than 300 m/s. The velocities upstream are much larger. Comparison with Figure 2a would reveal that a single grating can substantially alterjet expr.sion.

1 1

l l

t i F f T 'ii' tl'lV 'T H ' cl ' Ml:.: ; N T E .Ili:' l' ' Til h -\ '2 I

. :. y1 l M

\.. .'

g N . \. \ .

s s,

\

i 4

9

, , __-a-----,.

.. 8 l '.

c 1.

E.

Figure 4a. A geometry used to study the impact of structural impediments to jet expansion. A total of five structures were added. The maximum blockage was 40%. These structures were assumed to have symmetry aroundjG center line. The flow out'et was assumed to be on the sides, to simulate a Mark I geometry.

a

l i I '.'T I)i' .\ l ll l N ' '- '

'T i; ' < ' l ' < E' . N

';

: .A  : '. ; 1 'I

-- r-

• T li( .l('E lY ' Ell W.\'.'

4e .

/ \ .'s

' g' , l . ,

h' 5 -

E .

t g

8 - -

e 9

s

• l 1 '

l

'J :

Figure 4b. Flow 6 elds conesponding to geometrical case shown in Figure 4a. As shown here, majority of the Dow area is localiieo with velocities much lower than 300 m/s.

l l i

k l

A: >!'l, TION I 4' C VITY ON ET :'Ci' CT 3:N 4,

I

• 4 i

,.g ; j El I

'e as i

• I

)

! I #

1

.s s

s .

h T

1. .

u Figure 4c. An axial cavity was used to geometrical arrangement shown in Figure 4a to examine flow fields for a break postulated in the neck region of a BWR drywell. Otherwise, same geometries were used.

l l

l l '

l l

l g ww m, r yp.-r-ye-wwwggy-w rw.-w.m"N9"'N "4-'

FN""'"-~ '- "' ' " * ' '

• l i

l

! A:>DITION ::F' l' A'. ITT f)N 1ET ::I::'l1:(~ :(:N '

l i ] Nill uma -

I _ _ , . m.

l I

s'(>,s ._..

s i

j o

1 1 g ,

. x ,

\ N c

,i . '

\ '

1 Figure 4b. Flow fields corresponding to geometrical case shown in Figure 4a. The flow outlet in this case is through the end to simulate a Mark 11 drywell.

I l

i e

i e

3 n

l l

l ll 1

I i

l

l i

s.

\ . E'

[ NT~ E i

t _ ,

g'B '

i 3 . i.

'i, A ,.

i:

i .

{ } .\ .

.i i .

l

• l, l

1 4 .

n )

i 1

se I \

l pa; - - <,

, -- 1,u, - - , ,

q epg w g;..

u,- ; ,w m c i li Figure 4b. Flow fields corresponding to geometrical case shown in Figure 4a. The flow outlet in this case

! is through the sides to simulate a Mark I dryw ell.

I i

f 1

i A

4 4

d j

i A

i l

DRAFT m3om Appendix G

- Calculations to Evaluate Methods Recommended by the BWROG to Estimate the Quantity of Fines -

References 4

G.1 -- Table 4,- Page 71, URG. _

G.2

.. CDI Report 96-05, " Testing of Debris Transport through DowncomersNents into the -

Wetwell,' URG Technical Support Documentation, Voi, ll, . Tab 2.

- G.3 WBE Report No. 796-001, " Evaluation for Existence of Blast Waves Following LL icensing Basis Double Ended Guillotine Pipe Breaks," Science and Engineering

- Associates,- Inc.,1996.

'GA- - CDI Report 96-06, " Air Jet impact Testing of Fibrous and Reflective Metallic insulation,* URG Technical Support Documentation, Vol. II - Tab 3.

G.5 Appendix-8, " Analyses to _ Verify Values of P. for Selected Materials Reported by BWROG and Suggestions for Development of Scaling Analyses," SEA 97-M06-A:1, 1997.

Objective: In Reference G.1, the BWROG recommended a set of " destruction factors" that  ;

can be used by individual licensees to estimate the volume of fine debris generated by a postulated LOCA These destruction factors were derived based on the experimental data obtained at the AJIT, and were provided for each insulation as a volume averaged quantity to facilitate ease of use by licensees. A series of analyses were conducted to examine acceptability of the recommended values, with particular emphasis on determining whether or not the destruction factors would result in either " conservative" or " reasonable" estimates for the quantity of fines produced. The focus of the analyses was production of fines from Nukon blankets, although the cunclusions reached are probably valid for other insulation types also.

The staff's confirmatory anolyses consisted of the following steps:

• Analysis of experimentaldata reported by the BWROG and its comparison with the data from NRC sponsored testing.

e Incorporation of effects of other phenomena (e.g.: jet deflection by various structures).

Analysis of Air Jet impact Testing (AJIT) Data on Size Distribution: Table G.1 presents percentage of fines generated (by mass) for Nukon blankets as a function of the target distance i i from the jet nozzle (L/D), blanket jacket status and blanket seam arrangement. Tbs fraction of insulation not destroyed, termed as h by the BWROG, is plotted in Figure G.1 as a function

- of L/D for Nukon. This figure is i<:lentical to Figure G.1 of Appendix E in Reference G.2, with 1

the exception that an additional ~ data point obtained for Nukon with Sure-Hold -bands is included.. Also shown in the figure is the linear fit for the data proposed by the BWROG. A close examination of the data shows that:

DRAFT G-1 T.

E _.____.__.__...____._.________b.m_I_.m -.___E_..

_ .__._..'__._ ..__.___i;..____,_____.;

l-

. 3 DRAFT uras7 1, irrespective of the type of encapsulation, the fraction of the fines contained in the debris is only a functio,1 of the distance from the jet nozzle as long as the jet is energetic enough to peel of insulaton immediately_ upon impingement. For example, insulation with no Jacket at SUD resulted in the same debris fraction as that which was covered with stainless steelJacketing secured on the target pipe by "Sure-Hold' bands.

2. The fraction of fines is a strong function of the location of the blanket seam with resped to the jet nozzle. One data point obtained for seam arrangement described as 3 o' clock (i.e., the seam directlyin front of the nozzle) yielded largest h, or least number of fines.

3. The BWROG's linear regression is poor fit for the data, and fails to exp ain all the trends. The data suggests that fraction of fines initially increases as a function of UD until reaching a maximum at 30 UD. It then decreases, thereafteras a function of UD.

. Table G.1. AJIT Experimental data for NUKON7 Blankets.

_ [ Fines (%) is percentage of fines by mass. g, of Ref. G.2 is 100-Fines (%)]

Test Jacketed Soam UD Fines 9, [Ref. G.2] Blankets

(%)

5-2 No 9 5 25.4 74.6 48.8 31-1 Y. "Sure-Hold" 9 7 22.0 78.0 70.0 3-1 Y. Standard 9 20 46.3 53.7 0.0 1 No 3 20 7.1 92.9 35.6 5-3 No 9 30 60.0 40.0 1 31.3 2-1 Y. Standard 9 50 25.3 74.7 I 32.6

_3-2 Y. Standard 9 50 11.9 88.1 59.1 5-6 No 9 50 28.0 72.0 33.5 5-5 No 9 60.5 6.3 93.7 30.3 6-2 -

No 9 80 2.0 98.0 92.0 6 No 9 115.5 0.0 100.0 100.0 y No 9 _119 0.0 l 100.0 100.0 Comparison with NRC Sponsored Testing: As part of drywelldebris transport study (DDTS),

the NRC conducted a series of debris generation /transporttests at the same CEESI facility that was used by the BWROG for the AJIT. Although the intent of the NRC tests was not to

_ generate data regarding debris generation, a by-product of these tests included insights into

- debris generation. The exploratory tests employed a 3.75 inch diameternozzle, instead of the 3 inch diameter nozzle used in the AJIT tests described above. The following findings of NRC tests are in qualitative agreement with the AJIT test data:

• The direction of seam with respect to the jet is the key factor that determines the quantity of fines generated by a breakjet. If the seams are directly facing the jet, then the blanket tended to be blown away or be destroyed into very few large pieces. Very DRAFT G-2

O DRAFT m3m little (if at all) fine debris was generated. This confirms the results if test data 5-1. The BWROG having conducted most of the tests with the seam m the 9 O' clock position maximized potential for debris generation. And thus the data is conservative.

Only rarely did the quantity of debris generated exceeded 50% of the target insulation, irrespective of where the blanket is placed (i.e., UD value) and its seam alignment.

Additional banding (which is non-prototypical)and end plates were required to keep the blanket in place in order to generate fines in excess of 50%. In spite of these engineered features, the maximum fraction of blanket destroyed into fines is 75%. Thus a value of 50% forms a realistic upper bound for the destruction factor. It is very unlikely that higher than 50% is possible.

The destruction factors (or the percentage of fines) vary considerablywith distance from the jet nozzle. Increasing the L/D initially increased the destruction factors, until they reached a maximum. Thereafter,they decreased with distance. This is consistentwith the data plotted in Figure G.1.

Although a linear fit of the data is an easy solution,it is not consistent with the trends exhibited by the data. The data exhibit the following three distinct characteristics:

1) Minimum Fine Debris Generation Reached at 0 UD: As can be seen from Figure G.1, g, reaches a maximum of 80% as UD approachec zero. This implies that about 20%

of the debris would be fines. However, the staff believes that this result can be attributedto the 3 inch diameter nozzle size used in AJIT. !nitially, for UD < 10, airjets from nozzle tend to remain focused. For a blanket located at 5 UD, Figure G.2 illustrates the radial pressure distribution the exposed target blanket (Refs. G.3, G 4, and G.5). In addition, Figure G.3 [Ref G.7)illustratesthe peak and everage pressures on the target blanket. The average pressure suggests that the blanket remains on the target pipe only for a short interval. During this short time, destruction occurs in the localized region of the blanket which is subjected to high pressures. As evident from Figure G.2, high pressures (> 50 psi) are confined to a circular area in the middle of the blanket That is approximately 2.5-3.0 nozzle diameters (D )in radius. If one were to postulate that all the insulation contained in this high pressure area would be pu'varizedinto fines, the fraction of blanket destroyed into fines can be calculated as:

Fines (%) = 1-9, = (Area over which high pressure jet impinges)/(Total Blanket Surface Area) where, Total Blanket Surface Area = n(D, + t ).L.,

Area over which high pressure jet impinges = n(2.5D% )2 D, = Diameter of the target pipe (12 in.)

t = blanket thickness (3 in.)

L4 = Length of the blanket (24 in.) ,

D = Nozzle Diameter (3 in.)

After substituting for all the values, the total fraction of insulation in fines can be estimated to be about 16-22.5% (depending on the choice of 2.5D. or 3.0D ).

DRAFT G-3 i

1 9

_ ia

. DRAFT: umm -- - -

At 20 UD, it can be shown that the predicted percentage of Fines using the

same formulae above would be close to 50%, because the jet area envelopes the entire front surface of the blanket.

The implication of this analysis is that the .20% minimum value for fine debris generation -

at 5 UD is a reflection of 3 inch diameternonle used in these tests. Ifin the AJIT tests,--

the nonle diamdr had been increased, the percentage of fines produced would probably increase over this region. This was observed in the NRC tests where a 3.75 .

inch diameter noule was used.-

2)- Maximum Reached At 30 UD; As can be seen from Figure G.3, the average pressum

- on the target blanket decreases with distance from the jet nonie, reaching about 15 *u psig at 30 UD. As UD increases, the blanket residence time, which is inversely 9 proportional to the avensge pressure, increases steadily. Larger residence times would -

imply that there is more time for the high pressure to penetrate the blanket and cause damage to the blanket. Apparently, at a point where a combination of peak pressures greaterthan 17 psid and average pressures on the order of 13 psid occurs, maximum target blanket deetruction takes place. This suggesto that if the target blanket is exposed to an average pressure of approximately 17 psid, maximum target blanket ~

destruction is pc it in addition, this destruction may result in debris generation which is about L. " wiebris. Note:- the 60% fine debris generation is only slightly

-larger than 50% m. t irection of insulauon that is actually exposed to the jet with the -

remaining 50% being shadowed by the target pipe). '

- 3) Decreasein Fines (%)at UDs > 30: As the target distance from the jct nonle increasas -

beyond 30 UD, the peak and average pressures the targd is exposed to decmases.

-Although the residence time for the target insulation blanket also increases with distance from the jet nonie, beyond 30 UD the lower the destructive force of the jet leads to less debris generation. Ultimately, a point is reached where the peak pressum '

- cannot penetrate the outer jacketing, preventing the jet from dama0i ng the insulation.

As the peak and average pressures continue to reduce, therefore, the Fines (%) also -

decreases.

Conclusions of the Comparinost ' Analysis of the experimentaldata and its comparison with NRC sponsored test data suggests that use of a linear fit to derive volume averaged destructico factors does not properly scale the experimentaldata to drywell conditions. Nevertheless, the method employed has the following conservatisms': 4

e All the data included in the analysis was obtained for a condition in which blanket seams are arranged in the 9 O' Clock position. NRC sponsored testing found that seam :

- position alone can sig,nificantly impact debris genera,tk r..

i e- .The method completely igrores the impact of targets present in the path of an expanding jet.

DRAFT G-4

-1 ,

.. l

.._.._i.._..._

r DRAFT imm Quantification of Conservatisms in the BWROG Recommended Values: In order to evaluate the magnitude of conservatism involved in the BWROG recommended values for debris generation, a series of calculations were performed using the NUREG/CR-6224 (Ref.

31) reference plant as the basis for the case study. In the reference plant, Nukon is used as the insulation on all primary piping. The ZOI computed for that plant (also see Appendix F) using Method 2 of the URG is a spherical region 10.2 break diamaters (D)in radius.

For a reefreulalon line break postulated in the most congested region of the drywell, the total quantity of insulation contained in a 10.2 D sphere is approximately 750 ft3. The volume averaged destruction factor suggested

• y the BWROG for Nukon is 0.23. Total quantity of fines generated is 172 ft3.

An alternative method for calculating the debris generated would be to assume that :

e 100% of the insulation would be destroyed in a spherical region 3D in radius due to the blast effects from the postulated double-ended guillotine break. This conservatively bounds the possible effects of the blast wave. As discussedin Reference G.3, the blast effects would most likely be bounded by a spherical zone 3D in radius. For the reference plant the volume of debris contained in a 3D sphere is 35 ft3.

e 50% of the insulation would be destroyed between 3D and the outer boundary where the dynamic pressures fall well below 10 psi. As discussed in Appendix-F due to the prwence of structures, the dynamic pressures subside considerably within the 7D sphere. For the reference plant the quantity of insulation contained between 3D and 7D is 10133. Using a destruction factor of 0.5, the quantity of fines can be estimated as 9' ' ft3.

The totalquantity of fines generated would be 130 ft3. This value is about the same order of magnitude as that recommended by the BWROG. The latter number still has the following conservatisms:'

e It does not take into consideration orientation of seam with respect to the jet direction2 ,

and It does not gwe credit for less than 50% damage observed at lower impingement pressures.

Based on this analysis, it is concluded that destruction factor recommended by the BWROG would result in reasonable estimates for the fines.

i 1

l DRAFT G-5 1

l

\  :

1 I

100 - -

\

. e ... . ' '

i I

80

-~~'Og

,,,,.-6 8WROG Fit

,/

60 g\

.=

e /

F in all cases encapsulations were blown of 40 -

immediately. Debris from jet on Nukon

.. O No Jacket; 9 O' Clock 20 - g ec !acket; 9 O Clock y no 3,ck,t; 3 a.' Clock

+ Sure-Hold; 9 O* Clock 0 ' - - ' ' ' ' ' ' ' ' '- '-

0 20 40 60 80 100 120 Target Length to Diameter Ratio (UD)

F.gure G.I. Dependence of 71, on IJD Observed in AJIT[Ref. G.6) 1 l

l 1

I I

I

~

lNonte (0.D=3") 200 8

_ l ,

160 7 *

,E -( -

ANSI /ANS-58.2 Based 120 -

Axial Distance ES  : - LJD=3 EE 80 1 . + L/D=5 To To  : -

-e-LJD=7 o

"I

. UD=10 40 - '

-er- L/D=20 d --

^

2 m.

O

~ '''''''v- _

^

^

0 1 2 3 4 5 6

' Normalir J Radial Distance (R/D)

Figure G.2. Radial Distribution of the Stagnation Pressure in the jet. The Actual Pressure Profile Could he .

Different from the Linear Profile Assumed in the ANSI /ANS 58.2 Model.

l 4

4 4

~

-100 80 - - ' - - ' * ** - - - * -

I h60

. 15 - - - - - - - - *- - -

1. Percentage of Blanket into Fines -

h -O O Percentage of Blanket in the Jet Area into Fines

,,,s

a. 40

. . .. ..........x.. ..... . ........ ......... ...

s.s**~

I ' ~~~~-- ._. ..

_9 20 -- - -

g- -

P - -

- e

.- e O_

O 10 20 30 40 50 60 70 80 Jet Average Pressure (psi)

Figure G.3. Percentage of Blanket and Blanket in the Jet Area Destroyed into Fines as a Function of Jet Pressure. Analysis of AJIT Data [Ref. G.6],

e 1-

___._ ---- - - " ^ -

DRAFT noom Appendix H '

Calculations to Examine Accuracy of BWROG Drywell Debris Transport Factors

References:

H.1 Page 75, URG H.2 CDI Report No 96-05," Testing of Debris Transport through DowncomersNents into the Wetwell," URG Technical Support Documentation, Voi, 11, Tab 2, November,1996 H.3 SEA 96 3105-010-A:2, *Drywell Debris Transport Study, Phase i Draft Letter Report," Science and Enginering Associates, Inc., September 1996.

H.4 SEA 97-3105-A:1, - "Drywell Debris Transport lesting, Program Review / Planning," Presented to USNRC, Jan 1997.

H.S Draft NUREG/CR-6369, Drywell Debris Transport Study Problem Definition: The URG guidance asst.mes that only 50% of the fines would be transported to the suppression pool for a Mark ll containment during steam blowdown. In addition, transport fractions were also provided for Mark l's and Mark lil's. The transport factors for Mark l's and ll's were based on small scale tests of downcomer/ vent geometries.

The Mark lli factors were based on the more limiting conclusions drawn for Mark I's. The following activities were performed to evaluate the basis for the URG guidance:

1) Four tests simulating steam flow in the lower region of the drywell of a Mark 11 were conducted. The tests simulated Mark ll geometry where steam flow would enter the downcomers.

2) Calculationswere performed to estimate the steam velocities that resulted in the BWROG sponsored testing documented in Reference H.2.

3) The staff conducted a detailed Drywell Debris Transport Study (DDTS) which included testing and analysis to determine the key factors that affect debris transport from the drywell to the wetwell.

The staff's findings are summarized in the following sections:

Confirmatory Experiments: Experiments were conducted as part of both separate effects tests and integrated tests to examine the impact of vent / floor arrangement on the debris transport. In both cases, the vents were designed to simulate Mark ll containment flows as close as possible, both at the vent entrance as well as at distances sufficienti/ farther from the entrance (CFD and, MELCOR results were used to design the vent geometry). , Tunnel approach velocities were close to 25 ft/s and vent velocities close to 150 ft/s. In these tests

.(Ref. H.4):

DRAFT g.1  ;

l j

7 1

h-

+ ,

a DRAFT wmw 1

..1)7Total insulation captured-as ' percentage of insulation arriving at the vent is 813% for  :

classes 2-6. ; Classes 2 and 6 are markedly larger than the insulation shreds used in the j

- BWROG tests. These values in that sense are expected to be larger than what would be - 6 for smaller debrisc

2) The capture percentageincreased to 35% for classes 6+, Note that 6+ debris are typically

. larger than clearances in the gratings. For such debris gratings are found to be much better

- influence Large del N were deposited on the back wall equivalent to Mark 11 floor -

Based on these results, the staff concludes that an inadequate basis exists for the 50% credit 1

- recommended by BWROG in URG.

l

- Scaling Uncertainties Associated with BWROG Testing: In the BWROG soonsored steam

. tests conducted by CD1 (Ref. H.2), the blowdown was typically for 2.45 seconds from a 0.25" orifice. Flows scaled according to the vent /downcomerflows are equivalent to those from a 17

' inch diameter pipe break. In addition, the same flows scaled from the barrel diameter of the test facility to full scale are equivalent to a _1214 inch pipe break LOCA. Thus the velocities used are significantly lower (50-60% lower) than the velocities anticipated in the containment following a main steamline or recirculation line break LOCA. As calculated in Reference H.3, 3

. debris settling velocities are strong functions of the fluid velocity. No scaling rationale was provided in Reference H.2 as to how such a set of experiments can be extrapolated to BWR

. containments. These comments were forwarded to the BWROG. The comments do not a appear to have been addressed by the BWROG.

- Comparison with the Guidance Provided in the URG: The major finding of the BWROG experimental program was that Mark ll vents can retain up to 50% of the debris during both main steam lins and recirculationline breaks. Speedic experimentswere conducted to as part of the DDTS at conditionsjudged to be typical of a main steam line break. The data from the NRC sponsored testing conducted in support of the DDTS study suggest that wet vent entrances do capiture small debris. However, capture fractions are no more than 15% for small pieces with the majonty of the debris deposited on the floor. This value is much lower than the 50% proposed by the BWROG. Furthermore, it is likely that the ECCS flow from the break during the iscirculation line break would likely resuspend the majority of this debris and ,

transport it to the vents. Based on these results, it is concluded that URG guidance is nonconservative for a MARK ll containment.

For the other containments, the BWROG/URG guidance listed on page 75 of the URG is consistant with the upper bound estimates of this study, and consistently higher than the study's center estimate. While applying URG suggested transport factors, care should be taken to ensure that the_ underlying assumptions are consistent with the particular plant being

, P.dalyzed. In particular, the following considerations should be addressed:

~

1) ' All testing done by NRC and BWROG employed floor gratings with a 4" x 1 %" clearance.

- All the results (e.g., capture efficiencies)were derived assuming that the grating occupies 100% of the cross-sectionwith no chance for debris to bypass them. On the other hand, 1

DRAFT . H2 e

- w--.i- -,w-<-ie.e e - .- e ,-wu- * -ee- - ,- -n- .- - +ywe-,r-r-=. o -~~-t>w- y. Per --= o r r.-

DRAFT uws7 if the utility identifies large discontinuities or gaps in gratings that would allow a fraction of the debris to pass through them, then the transport factors should be revised -

appropriately,

2) Implicitly, both studies assumed that unthrottled ECCS conditions exist for a maximum of 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> following a recirculation line break. If the licensee expects unthrottled ECCS operation for more 'han 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />, the washdown fractions should be preoerly revised, in such a case, the URG-assumed 25% washdown fraction is non-conservative. Instead, the washdown factor should be scaled up.

Conclusions:

The staff concludes that BWROG provided transport fractions for Mark !!

containments on page 75 are not adequately supported. Maik lis should use similar fractions as Mark is.- However, the fractions provided for Mark l's and lil's based on transport of fines are reasonable.

DRAFT H-3

7 ,

)

DRAFT unow - ,

.. l

. Appendix I  !

-_ Analyses to Examine Accuracy and Applicability of ECCS Streiner Head Loss:

l C  :

References:

i

" 1,1. URG Technical Support Documentation,Vol. I, Tab 2, ' Testing of Altemate Strainers with

_- insulation Fiber and Other Debris,' _CDI Report No. 95 f l.21 NUREG/CR-6370, BLOCKAGE User's Manual, Science and Engineering Associates, Inc. -

I.3 NUREG/CR-6224, 1 Parametric. Study of the Potential for BWR ECCS Strainer Blockage. q

- Due to LOCA Generated Debris.

[ Problem Definition:. As part of the URG Technical Documentation, the BWROG provided L= results of their heaJ _ loss testing program conducted by EPRI at the NDE test facility in

Charlotte,i NC, and descriptions of the various cook-book approaches that can be used to

estimate' head loss induced by altemate and truncated cone strainers. This analysis was _

conducted to venfy the adequacy of trieir proposed head loss estimation approaches.

~ Results of the Review The URG for ECCS Suction Strainer Blockage (Volume I, November 1998) was prepared by the BWR Owners Group ECCS Suction Strainer Commtee for the .

purpose of providing the members of the BWR Owners Group wit 'es and methods for-F resolution of the ECCS suction strainer blockago issue. This docui..c was reviewed, and computations were performed to determine if the URG_ proposed approach provided.

' reasonable estimates of strainer head loss. Specifically, the URG models were applied to simulating the experimentaldata for a number of the tests for which data were included in the report. . Time-dependent behavior was simulated by incorporating the URG models into a

[ . modified version of BLOCKAGE 2.5.

J J The debris bed head losses predicted by the URG models for selected test runs were compared to the experimentu values in Tables 1'and 2 for fiber / particulate and RMI beds, .

respectively. Test runs were selected that indicated the head losses were approaching a steady state val'ie. The staff notes that one important eriticism of the URG tests is that i

reasonable steady state conditions were not achieved in significant number of the test runs.

This is important because the quantities of debris on the strainer was deduced by taking the

' known quantities of debris introduced into the water tank and assuming that essentially all of

. the' debris was deposited onto the strainer when steady state head loss was achieved.

Therefore, the debris bed composition was relatively unknown during transient conditions. It is believed that pool turbulence was sufficiently high enough to prevent debris from settling

= onto the bottom of the tank. - The results of four time-dependent simulations are shown in

Figures.1' through 4, for tests P5, J28R, 4, and J6, respectively. These results lead to the I

m- following conclusionsr -

- Stacked-Disk (PCl) Prototype No. 2 Strainer. Fiber /particulatesimulationswere performed 1 for tests _P3, P5, and P8 for a range of flow rates, as shown in Table 1.- The URG model DRAFT i-1

= . . . .- - -. - - . . - -- . , . . -.

DRAFT m3om appears to do either a reasonable job or a conservative job of predicting strainer debris bed head loss for the ranges of conditions tested. The ranges of conditions tested include:

• iiber loadings range to 50 lbs.,

e corrosion products to fiber mass ratios to 10, e approach velocities to 0.5 fps (based on the circumscribed area),

e no miscellaneous debris.

Note that the head loss will be overpredicted when the strainer surface area is not completely covered (as was obviously the case in test P4), and that miscellaneous debris was not considered in test;ng this type of strainer. There was insufficient data (see Table 6-5, URG Technical Support Documentation, Volume 1) to determine with certainty that the strainer area was completely covered for the simulated test runs, however it is likely that the coverage was complete, or nearly complete, since the average fiber debris bed thickness across the entire strainer area was 0.74,1.5, and 0.30 inches for tests P3, PS, and P8, respectively.

The time-dependent debris head loss for test P5 predicted by the URG modelis shown in Figure i compared to the head loss predicted by the NUREG/CR-6224 correlation and to the experimentaltest data. The strainer flow rate in the test was nominally 5000 GPM; however, it was varied from 2500 to 10000 GPM starting at about 70 minutes (see test data on Page C-102 of URG Volume 1). The URG model did a good job of predicting the test. It overpredictedearlyin time due to incomplete coverage of the strainer. Apparently debris on the strainerwas shifted as a result of varying the flow rate because the head loss associated with 5000 GPM was less after the 5000 GPM flow was resumed than before the flow rate was varied. The shifting of debris by varying the flow rate was, of course, not predicted by the models.

The NUREG/CR-6224 correlation did not perform well when attempting to predict test PS, as its predicted head loss at 5000 GPM was only about a third of the experimentalhead loss. Its poor performance is attributed to the fact that it was developed for a strainer with a uniform debris bed thickness and a uniform approach velocity as would be expected for the truncated cone strainers. For stacked-disk and star strainers, the debris is apparently driven preferentially into their crevices resulting in distinctly non-uniform debris distributions and approach velocities. The NUREG/CR-6224 correlation may either require modification or the development of approoriate input guidance for it to be reliably applied to these non-uniform strainers.

Apparently,the only stacked-disk strainer design tested with RMI debris was the stacked-disk portion of the self-cleaning strainer (SCS-SDP). One of these tests, Test A1, was simulated over a range of strainer flow rates as shown in Table 2. In all simulations performed for this strainer, the URG* models conservatively overpredicted the debris bed head losses by more +

than 100% *

• Stacked Disk (PCI) Prototype No.1 Strainer Fiber /particulatesimulationswere performed I for tests 20 and 22 at three flow rates, as shown in Table 1. These results are similar to those DRAFT l-2 l

DRAFT moS for prototype 2, however there was less test data to examine which is reflected in the URG plot for non-dimensionalhead loss for the first prototype (see Figure 6-8, Page 70, URG, Volume 1). This curve was drawn with very limited data resulting in a straight line fit which does not seem likely to be the case.

60-Point Star Strainer. The URG modelgenerally underpredcted the 60-Point Star strainer by 20 to 30%, as shewn in Table 1, whenever the strainer surface area was completely covered. (Coverage 'vas determined by comparing the calculated nondimensional thickness associat6d with each debris loading with the required minimum nondimensional thickness reported.) The tfRG model overpredicted two tests, J22 and J23, because the strainers were not completely covered. The ranges of conditions tested include:

e fiber loadings range to 25 lbs.,

e corrosion products to fiber mass ratios to 25, e

approach velocities to 0.5 fps (based on the circumscribed area),

e either zero or full recipe miscellaneous debris.

Note that the miscellaneous debris increased the experimentalhead losses for tests J28R and J23 by 24% and 116%, respectively and the URG model bump up factors appeared to compensate appropriately.

Figure 2 shows the time-dependent debris head loss for test J28R predicted by the URG model versus the head loss predicted by the NUREG/CR-6224 correlation, and the experimentaltest data. The strainer flow rate in the test was nominally 5000 GPM; however, it was varied from 2500 to 5000 GPM starting around the 40 minute point in the test i,see test data on Page C-188 of URG TechnicalSupport Documentation, Volume 1). The URG model underpredictedthe test by 20 to 25% throughout the test. The NUREG/CR-6224 correlation did not perform well when attemptng to predict test J28R, as its predicted head loss at 5000 GPM was only about 15% of the experimental head loss. Again, its poor performance must be attributed to the fact that it was developed for a strainerwith a uniform debris bed thickness and a uniform approach velocity, which is clearly not the case with this strainer design.

The URG model can be used to predict fiber / particulate debris bed strainer head losses for the 60-Point Star strainer if a reasonable safety factor is applied to compensate for the relatively uniform under predictions shown herein, and provided the conditions of the calculations are within the ranges of conditions listed above.

The 60-Point Star strainer was tested with RMI debris and one of these tests, J20, was simulated with the URG model for a range of flow rates as shown in Table 2. In all of these simulations, the URG models overpredicted the debris bed head losses by 15 to 40%.

~

20-Point Star Straineri Fewer test runs were simulated for the '$0-Point Staistrainer than for the 60-Point Star strainer and all of the simulations except one overpredicted the experimental head loss. Test run 11, which had fiber debris but no particulate, was underpredictedby 47%. Two of the overpredictionswere due to incomplete coverage on the DRAFT l-3

DRAFT mm strainer. The test run which involved Kaowool rather than NUKON' was extremely overpredicted by a factor of 259 which appears to be related to the fineness of its fibers (-2.5 microns in diameter) relative to the NUKON@ fibers (~7.1 microns) for which the model was developed. In fact, all of the test runs checked involving Kaowool were way overpredicted (i.e., by factors of 46,259, and 14 for test runs J12,35, and J30, respectively) It is likely that the validity of the URG model as applied to the 20-Point Star strainer is similar to its validity as applied the 60-Point Star strainer, however the uncertainty is higher.

Truncated Cone Strainer The URG model was found to be less reliable for simulating tast runs using the truncated cone strainer. Test runs for this strainer design tended towards lower fiber loadings with higher particulate to fiber mass ratios. Three of the runs (6, 7, JS) simulated had incomplete coverage of the strainer and two of the runs (6, J6) had mass ratios larger than the upper limit of 120 (see Figure 6-6, Page 68, URG Technical Support Documentation, Volume 1) for which the model was developed. Using the URG model beyond this upper limit of 120 is non-conservative because the head loss beyond this limit increases by some unknown amount.

The primary reason Run J6 was overpredicted by the URG model was an excessively large and unrealistic bump-up Factor of 4.2. The bump-up factor accounts for the head loss contribution of the miscellaneous debris where the head loss without miscellaneous debris is multiplied by the bump-up factorto get the total head loss. The reason the bump-up factor is so large is the URG assumption that the fiber-to-particulatemass ratio used in calculating the bump-up factor must be limited to 4 whereas the actual mass ratio for this test was 360 which means that miscellaneous debris contribution over powers the equation. This assumption was designed to guarantee that the head loss predictions would be conservative but under certain conditions, it can render the URG model unusable.

The time-dependent head loss for Run 4 was predicted using both the URG model and the NUREG/CR-6224 correlation as shown in Figure 3. In this situation, the NUREG/CR-6224 correlation does an excellentjob, whereas the URG model overpredictsthe head loss by about 40%. It was not surprising that the NUREG/CR-6224 correlation performed well for this test run because the correlation was specifically developed for the truncated cone strainer. Note that in Run 4, the strainer was completely covered with fiber debris and there was no particulate debris.

Another test, Run J6, was also simulated but with distinctly different results. In the J6 simulation, the NUREG/CR-6224 (Standard Simulation) grossly overpredicted the head loss, as shown in Figure 4, whereas the URG model simply overpredicted the head loss by a factor of about 2. The reason that the NUREG/CR-6244 correlation performed so poorly was that BLOCKAGE was predicting that most of the particulate debris was on the strainer when in i

realityit most certainlywas not. A second s mulation(Realistic Simulation) where only ab,out ,

5% of the particulate debris was allowed to sccumulate on the strainer shows a much better agreementwith the experiment. If all of the 190 lbs. of corrosion products were deposited onto 2

this 18-ft strainer,the entire surface area would have been coveredwith sludge to a depth of 1.8 inches, whereas the thickness of fiber debris (unccmpressed) was only 0.15 inches, i.e.,

DRAFT i-4

DRAFT imots7 it is highly unlikely that this thin layer of fibers could hold that much sludge in place, and if it did, the head loss would have been excessive as predicted by the NUREG/CR-6224 correlation (standard). The conclusion must be that the fiber bed became saturated with particulate debris, thereafter as much particulate passed through the fiber as was deposited onto the fibe s. The NUREG/CR-6224 Realistic solution is considered the correct simulation of the experiment. Therefore, it must be further concluded tha;in at least one of the URG experiments,the assumption that nearly all of the debris added to the system will be located on the strainer when the head loss approaches steady state is in error.

Using the URG model to simulate the truncated cone strainer given the test conditions for which the URG model was developed has been shown to be unreliable.

Conclusions The URG model, when used in the cookbook approach outlined in Appendices A and 8 of the URG Technical Support Documentation, Volume I, was in general found unreliable and incomplete for a number of reasons, such as:

1) The URG model was developed for limited ranges of data, i.e., fiber and particuiate debris loadings, velocities, and strainer design, and using the model beyond these limitations is especially risky because the models are non-mechanistic in nature.

2) The URG model has been shown to under predict in many cases the experimental data used to develop the model; therefore, each head loss prediction should be reviewed in detail relative to the applicable data to ensure a valid prediction, in many situations, such as the 60-Point Star strainer, a significant safety factor should be factored into the prediction.

3) The bump-up factor used to account for the miscellaneous debris was developed with limited data (Gravity Head Loss Tests) and in many situations, the bump-up factor has been shown to severely overpredict the head loss. While this overprediction is conservative, it can render the use of the model impractical, especially with plants having low NPSH margins.

4) Some of the test runs (e.g. J6) simply overpowered a thin layer of fibers with massive amounts of particulate debris so that the strainerdebris apparently become saturated with particulate. Thus, the actual particulate debris loading on the strainer which was used to develop the model was not known. Instead, the total system inventory was assumed to be deposited onto the strainer when it likely was not.

5) Debris such as Duct Tape, Tie Wraps, and PlasticTags that was included in the recipe -

of miscellaneousdebris was not included in the modeling effort. It is possible that this

• debris had only a minor impact on the overall head loss, but,the report did not discuss this issue. Nor did the report discuss what quantities of these debris types would be needed to significantly affect head loss.

DRAFT l-5

-DRAFT ucom

6) Extending the URG model to other types of debris, such as Kaowool rather than j NUKON', could be risky because the URG model was essentially developed for the  ;

NUKON' fibers. This review showed that the UD3 model tended to seriously overpredict the head loss associated with test runs using the Kaowool. First of all, the overprediction may render the URG model unusable for predicting head loss associated with Kaowool. Secondly, the URG model could seriously underpredict the head loss for some other type of fiber, i.e., in the opposite direction from the Kaowool.

l This could be the case for a fiber with thicker strands than NUKON*. l

7) The same NUKON' debris properties should be used as were employed in developing the URG model, i.e., densities, and diameters. The fiber density of 2.4 lbm/ft* was used in the URG study which is the number generally assumed for the intact fiber insulation, sometimes referred to as the as-fabricated density, but there is data showing that the density of actual debris is considerably reduced by the destruction process. For example, fibrous debris used in the PP&L sponsored tests conducted at ARL had an actual density of 1.3 lbm/ft8 . If the user were to use this density rather

~ than the as-fabricated density, the user would obtain a different fiber spacing distance  !

whicle would result in a different head lots than if the user actually used the as-fabricated density that is inherent in the non-mechanistic URG model. )

8) All of the test runs were conducted using a mass of corrosion products with the same representative size distribution (i.e.,83,11, and 6% for size ranges represented by 2.5,  ;

7.5, and 42.5 microns, respectively)and the URG model has not been shown valid for i alternate size distributions. For example, one plant (designated as Plant J in URG '

Volume Ill) found thot 27% of its particles were in the 10-75 microns range compared to the 6% of the representativestudy. It is likely that had the size distribution of Plant J been used in Run J6 rather than the representative distribution that the head loss would have been greater simply because fewer of the particles would have passed through,the fiber bed than was apparent from the simulations performed for this review. In addition, the NUREG/CR-6224 head loss correlation has shown that the head loss is a function of the particulate specific surface area which is in tum a function of the particle diameters.

9) Two types of RMI debris were used in the URG strainer debris bed head loss tests, i.e., two different size distributions of the 2.5 mil stainless steel foils, and each size distribution was used over a different period of time. Thus, the tests for a specific strainerdesign are generally valid for only one type and size distribution of RMI debris.

For example, the stacked disk tests (stacked-disk section of the self-cleaning strainer) only used the RMI debris obtained from Diamond Power since all of these tests were

- conducted in August 1995. Therefore, the URG model, as it applies to the stacked-disk strainer, was de,veloped only for this one type and one size distribution of RMI. debris.

Furthermore, it must be assumed that the self cleaning strainer 1s representative of stacked-disk strainers in general, since sufficient dimensions were not provided to determine if the size of the disks were the same as the two PCI prototypes, i.e., the spacing between the disks could well impact RMI debris loading.

DRAFT i-6

l DRAFT ino/s7- ,

10) The URG nxxiel was developed from test data where a true steady state condition was not generally achieved, implyirq the debris loadings used in the model development

- were somewhat different than the actual loadings on the strainers.  ;

11) The URG report did not' provide model validation, parameter sensitivity, or uncertainty analysis leaving open the question of whether or not a particular combination of input j parameters could cause the URG model to seriously underpredict head loss. l In conclusion, the URG model must be used with care, not by blindly plugging numbers into ,

a cookbook step-by. step procedure as outlined in the report. Rather, each head loss l prediction must be anchored into head loss data to ensure that it is reasonable and  ;

conservative. For all of the reasons cited t.bove, the staff recommends that vendor specific test data be used to demonstrate the head loss assumed in a licensee NPSH calculations.

l i

1 f

t DRAFT l.7 I

-- - , , - , , - - .,,.--....n- s,,,- .,,-n - . - - . , , ~, --e, .

r-.c - - - . -

16 , , , , , , , , , , , ,

Experimental 14 - -------

URG Moctel -

NJREG/CR-6224 m

-.t 12 - .

!s.

C3 o, y PCI-2 Strainer- * .

1 53 Lbs. NUKON

! i 1O - I

• 100 lbs. Por-Iculate m -

?

=s Na Recipe

\

M o 8 -

L . .

a_ I O

y C

== 6 - i

• • O

u. .

o _ . _ . _ . _ . - . _ . - . _ . . J.; -------

- 4

,s 6 -

! i ,  ;

./ '

e' s i i 2 -

.i k I s -

o e _ _ _ - _ - ,! -

/ ',_______________,t' i i

i i

O ' ' t O 1 2 ~

3 4 5 6 Time (103sec)

Figure 1: Debris Bed Head Loss for Run ?5 e

9 s '-

o::_ toc =s _nmonna*"1ga"um 1 1 3 1 2 O 2 4 6 8 0 2 4 8 0 0

0 1 1 00 26 - -

00 50 - ,

% l P - -

R lbo F

i b i -

g 0 e s .s n u .

- c . t r

i p P N ,

5 e aU S NUE e

t

- r<a t

J R x

- Cr R Gp 2 ' ,/ s cN E e

'- ul S G Mir ,

D - a t

r /C dom t

e a e e 1 i

n R e nt l

b - .

e -

r . '

/ r 6 a ,

i /' 2 l

s 0 - .

2

., 4 B T i - -

e '

/

d m _

e _ e.

H (

1 - .

e 1 . > - '

a 0 -

d 3 5 -

s "

L o o c '

s )

s ',

2 .

f - .

o . '

r 0 _

R -

u '

n _ .

J 2 .

2 ,- s

.i..5*s't.i;(

8 . ' .

R 5 _ s ,s _!..!

ls i5.;i;.*.,!  !.;  : .

si,,

it.,.4:

': ;II, ,! !

l r

3 0

4O , , , , , , , . > >

Expe-Imental 35 ------- / i NJREG/CR-6224 e -

/

URG Model e s

s

/ g e s JE 30 -

' i -

3 o i 3: Triancolod Cone 4 a .~.,s 3

1 ,

e5 -

8 '_bs. NUKON ' -

V N: Particulat e < 1. '5 -

a

,- g i L.

s.. e

=> I[ ,-- e ,

s -

!? 20 -

M e"

- __-------------, t i -

c- - \ ,

- i i i

-o 15 -

< s, p c= o e

/ i 8 I i.

- e i

• G- ,4 s .  ;.-

2: 10 -

- .-.-.-.-....-..----.1 s O i < -

,e -

,. - i.n. ' j /

ss

, ./

5 1(i i .'

L_

i!

si s

d.i .

O ' ' ' ' ' ' ' ' ' '

O 1 2 3 4 5 Time (103sec)

Figure 3: Debris Bed Head Loss for Run 4

• e

\ll1 (ll l 9

O -

or2 e3 ~E33E v5y3m 1 1 2 2 3 3 4 5 0 5 0 5 0 5 0 0

o 0 0 0 0 0 0 0 0 0 -

22.r 5

% 05u T

0 r - -

R lc -

i F . .

e lbo t -

g 5 c b .s e - - ,

u pi .s N

d - -

r e e PUC NNUE ,

1 ~/ aKa r On UURx R R Gp 4 - t c No i

E E e

_ / u GG Mir ,

0 .

- om l

D . a -

e / t e SRde a ale tn e

, t b - / ,

r n l i a i 1 - d s l s . , - / c t i

- / r c d ,

B T 5

& e i

m

/

d e H (1 2 f

e a 0

- / ,

d 3 0 -

9 s -

L e - .

o )

c - 7 s 2 s . , .

f 5 . f ,

o r , . / .

R 3 -

u . , -

- /

n .

0 -

O J - /

6 ,

- =

3 -

5 4

0

! , \\l\!iI\ ,)

J DRAFT nom Appendix 4 Calculations to Examine Accuracy of the Sludge Generation Factors

References:

J.1 Section 3.2.4.3, URG J.2 BWROG Letter OG94-661 161, *BWR Owner's Group ECCS Suction Strainer Committee Suppression Pool Sludge Particle Size Districation," BWROG Letter w USNRC,1994 J.3 BWROG Letter OG96-321 161 Attachment 2, ' Suppression Pool Sludge Particle Distribution Data Average Distribution Calculation," BWROG Letter to USNRC, 1996.

J.4 BWROG Letter OG95 388161, Attachment 4, *BWR Owners Group Suppression Pool Sludge Generation Rate Data," BWRCG Letter to USNRr,1996.

Problem Definition: In the URG, the BWROG provided guidance regarding the quantity of suppression pool corrosion products (sludge) that should be used in utility analysis of strainer blockage. A value of 300 lbm/ year was recommended as the sludge generation rate in lieu of establishing plant specific sludge generation rate. The basis for the generation rate and characterization were provided in Ref. J.2 through J.3. These documents were evaluated to determine the adequacy of the basis for BWROG conclusion.

Results and Discussions: The BWROG survey information is provided in Figure J.1. As shown here, in all the plants (Mark I,11 and lil), generation rates are far less than 200 lbm/yr, with one exception. The staffis aware from the NUREG/CR-6224 study that reference plant estimates were very conservative and were based on wet sludge mass (not the dried sludge). Based on this, the staff concludes that the BWROG recommended value of 300 lbm/ year is reasonable and sufficiently conservative. Also, the URG provides comprehensive guidance for those utilities that do not wish to use the 300 lbm/yr number.

'll should be noted that the BWROG incorporated the blast effects by assuming that all the debris located in a spherical zone 3 nozzle-dameters in radius would all be destroyed into fines.

1. Discussions with the insulati0n vendors and contractors suggest that no particular arrangement is followed to align the seams in a particular orientation. They ara randomly arranged with respect any postulated break.

DRAFT J1

i s'

4 I

O e

l*

1400 - 'Ver Conservative mMass of Sludge in The Bounding Pool at Time of Survey Estimation (Ibm) i i Sludge Generation Rate

_ (Ibm /yr) i 800 * . . . . . Sludge Mass of ,50 lbm is used in the Base Case i

! 600 f o . . .b . E L. -. . E_ _ _ E i = s s l gsE s s s s s EJ si i

s-s!ig'g-[y "E

g y *a-gy s!

- ys

?!VEE u-1:7 }!isvt:i ~ItIE 2 ': i "

.t r. di . :I 'EE*E t-r;-'~d }ri11 j y *}f 5

5

[-

. n 1

e >

l Figure J.1. Sludge inventory and Sludge Generation Rate for

, Selected Plants l

n J

i

)

1 a

I

\

l j

--_--.____.__-_w -.__w.--.- ,e--.-- __._-,--wm.-..v-,,_.-y

..my--.-. ._,,m.wr,_ m,,,