NOC-AE-12002784, Summary of the South Texas Project Risk-Informed Approach to Resolve Generic Safety Issue (GSI)-191

Summary of the South Texas Project Risk-Informed Approach to Resolve Generic Safety Issue (GSI)-191

ML12023A056
Person / Time
Site:	South Texas
Issue date:	01/11/2012
From:	Crenshaw J South Texas
To:	Document Control Desk, Office of Nuclear Reactor Regulation
References
GSI-191, NOC-AE-12002784, STI: 33215081
Download: ML12023A056 (20)

Text

Nuclear Operating Company South TTas Pmect Electnc Generatin$Staton PO. Bar 289 Wadsworth. Txas 77483 ,

January 11,2012 NOC-AE- 12002784 STI: 33215081 File: G25 U. S. Nuclear Regulatory Commission Attention: Document Control Desk One White Flint North 11555 Rockville Pike Rockville, MID 20852-2738 South Texas Project Units 1 and 2 Docket Nos. STN 50-498, STN 50-499 Summary of the South Texas Project Risk-Informed Approach to Resolve Generic Safety Issue (GSI)-191 A summary of South Texas Project's methodology towards a risk-informed resolution of GSI-191, "Assessment of Debris Accumulation on PWR Sump Performance" is provided in the Enclosure. The summary includes the initial quantification and project results to date.

There are no regulatory commitments in this letter.

Should you have any questions regarding this letter, please contact either Jamie Paul, Licensing, (361) 972-7344, or me at (361) 972-7074.

J.W. Crenshaw VP New Plant Deployment / Special Projects JLP

Enclosure:

Summary of GSI-191 Risk-Informed Closure Pilot Project 2011: Initial Quantification

NOC-AE- 12002784 Page 2 cc:

(paper copy) (electronic copy)

Regional Administrator, Region IV A. H. Gutterman, Esquire U. S. Nuclear Regulatory Commission Morgan, Lewis & Bockius, LLP 612 East Lamar Blvd, Suite 400 Arlington, Texas 76011-4125 Balwant K. Singal John Ragan Senior Project Manager Chris O'Hara U.S. Nuclear Regulatory Commission Jim von Suskil One White Flint North (MS 8B13) NRG South Texas LP 11555 Rockville Pike Rockville, MD 20852 Kevin Polio Senior Resident Inspector Richard Pena U. S. Nuclear Regulatory Commission City Public Service P. O. Box 289, Mail Code: MN116 Wadsworth, TX 77483 C. M. Canady Peter Nemeth City of Austin Crain Caton & James, P.C.

Electric Utility Department 721 Barton Springs Road C. Mele Austin, TX 78704 City of Austin Stewart Bailey Richard A. Ratliff Branch Chief, Safety Issues Resolution Alice Rogers U. S. Nuclear Regulatory Commission Texas Department of State Health Services One White Flint North (MS 011 F01) 11555 Rockville Pike Rockville, MD 20852 Donnie Harrison Balwant K. Singal Branch Chief, PRA Stewart Bailey U. S. Nuclear Regulatory Commission Donnie Harrison One White Flint North (MS 011 F01) U. S. Nuclear Regulatory Commission 11555 Rockville Pike Rockville, MD 20852

Summary of GSI-191 Risk-Informed Closure Pilot Project 2011: Initial Quantification South Texas Project, Wadsworth, TX NOC-AE-12002784 January 11, 2012

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/Industry. GSI-191 Closure Pilot Project Executive Summary The main objective of the Risk-Informed GSI-191 Closure Project is, "Through a risk-informed approach, establish a technical basis that demonstrates that the current design is sufficient to gain NRC approval to close the safety issues related to GSI-191 by the end of 2013." The results presented in this summary are the joint work of STP Nuclear Operating Company (STPNOC) Risk Management, Los Alamos National Laboratory, The University of Texas at Austin, Texas A&M University, Alion Science and Technology, ABS Consulting, ScandPower, Mike Golay, The University of New Mexico, and KnF Consulting Services, LLC.

In the risk-informed approach, STPNOC would seek exemption from certain requirements of 10 CFR 50.46 if the risk associated with the fibrous insulation in STPNOC's containment buildings is not risk significant.

If STPNOC determines this insulation to pose a significant risk, STPNOC is committed to investigating plant modifications including insulation removal and other measures to preserve sufficient margins for nuclear safety.

The 2011 preliminary results show that the change in risk for fibrous insulation in contain-ment is less than 1.OE-06 in core damage frequency (CDF) and less than 1.OE-07 for large early release frequency (LERF), that is, very small per RG 1.174.

This result represents the uncertainties of more than 20 input parameters and the complementary execution of the physics-based (CASA Grande) model, thermal-hydraulics model, and PRA models. Although previous realistic testing has shown that chemicals are unlikely to affect the head loss in STPNOC debris beds (sump strainers and fuel assemblies), the Pilot Project has used an initial methodology that adds pessimistic head-loss estimates from chemicals. Including these estimates is believed to fully addresses NRC concerns raised in pre-licensing meetings related to sump chemistry.

This preliminary assessment of the CDF and LERF gives us confidence that the issues asso-ciated with fibrous insulation in the STPNOC reactor containment buildings will be further shown to be non-risk significant with adequate defense-in-depth and safety margins such that STPNOC will be able to provide a sufficient basis for GSI-191 closure by the end of calendar year 2013.

STPNOC will expand upon the project's technical contributions by including the uncertainties of more than 50 input parameters and the seamless integration of CASA Grande, a new jet formation model, un-certainty propagation in the thermal-hydraulics models, and PRA analysis that takes into account detailed operational conditions. The resulting framework will provide STPNOC the ability to assess future issues on risk-informed basis as they may arise.

The methodologies and results of the pilot project in 2011 are presented in the following documents:

analysis of results from the physical process solver, CASA Grande and RELAP5 thermal-hydraulic analyses (Letellier, 2011); LOCA Frequency analysis (Fleming et al., 2011); Uncertainty quantification methodologies and illustrative examples (Popova and Galenko, 2011); Jet formation research (Schneider et al., 2011); and Chemical effects research and experimental design (Sande et al., 2011).

i

Enclosure NOC- AE- 12002784 STP Nuclear Operating Company's NRC/Industry CSI-191 Closure Pilot Project Contents 1 Introduction 1 1.1 Previous efforts .......... ............................................ 1 1.2 Risk-informed ........... ............................................. 1 1.3 Pilot Project .......... ............................................. 2 2 Major assumptions 2 3 Findings in 2011 3 3.1 Chemical effects ............... ........................................... 4 3.2 Strainer and downstream performance ......... ............................... 5 3.3 Destruction zone ............... ........................................... 5 4 CASA Grande 7 5 Plant configuration 7 6 PRA 8 7 Uncertainty Quantification (UQ) 10 8 LOCA frequency 11 9 Quality Assurance 11 10 Licensing 11 11 Conclusions 13 List of Figures 1 Illustration of the core damage (CDF, ACDF) risk associated with as-designed fibrous insu-lation in the STPNOC containment on the Regulatory Guide 1.174 decision-making Region map ........... ................................................... 4 2 Illustration showing how decreasing decay heat removal flow requirement over time reduces cooling flow requirements that would result in significant reductions in head loss across debris beds. Experiments of in-core performance with chemical effects show significantly greater tolerance (higher mass loadings) for fully loaded debris beds at lower flow rates ........... 6 3 Illustration of a CASA Grande (complementary) cumulative distribution and the method used to develop input for the PRA from it........................................ 8 4 High level illustration of the inputs and outputs overlaid on a schematic diagram of the PRA event tree structure. The event tree structure, with the exception of the long term cooling top event, already exists in the typical nuclear power plant PRA ..................... 9 5 Uncertainty quantification for complex computer models ..................... 10 6 Illustration of the major elements of the STPNOC quality assurance plan for risk-informed closure of GSI-191 .......... ........................................... 12 ii

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/Industr. GS1-191 Closure Pilot Project 1 Introduction limiting that it would indicate no design would be ac-ceptable, especially for in-vessel effects (Baier, 2011).

The purpose of this document is to summarize Even the presence of tramp debris (regardless of the the initial quantification performed for the risk- insulation design) may not be tolerated. Unless com-informed closure path in STPNOC's NRC/Industry promised by incorporating the time-evolution of acci-risk-informed GSI-191 pilot project (Pilot Project). dent phenomena and accommodation of realistic be-In this document, significant findings from Pilot havior, it may be difficult to provide a satisfactory Project work in calendar year 2011 are summarized closure path using a deterministic approach.

and the overall program approach is described. While the risk-informed approach is not new to investiga-1.2 Risk-informed tion of GSI-191 closure, the approach in the Pilot Project is more comprehensive than previously con- In keeping with the agency's commitment to move to-templated. The closure process developed is gener- wards increasing risk informed regulation, the NRC ically applicable with site-specific information sup- directed the staff (Vietti-Cook, 2010) to consider plied. Each primary area of investigation in the Pilot GSI-191 resolution strategies including an Option 2:

Project is summarized in this document.

"The staff should take the time needed to consider all options to a risk-informed, 1.1 Previous efforts safety conscious resolution to GSI-191.

While they have not fully resolved this is-Since 2001 GSI-191: Assessment of Debris Accumu-sue, the measures taken thus far in response lation on PWR Sump Performance has eluded reso-to the sump-clogging issue have contributed lution despite significant efforts by industry and the greatly to the safety of U.S. nuclear power NRC. Although recent thought has been given to risk plants. Given the vastly enlarged advanced quantification (see for example Teolis et al., 2009) and strainers installed, compensatory measures early recognition of the need for risk evaluation was already taken, and the low probability of identified, (e.g. Darby et al., 2000), serious investiga-challenging pipe breaks, adequate defense-tion into risk quantification has not been undertaken in-depth is currently being maintained.

until now. Instead, resolution has followed a classical The operative words for Option 2 are in-deterministic approach. It is STPNOC's view, follow-novation and creativity. The staff should ing an initial 2011 PRA quantification, that a risk-fully explore the policy and technical impli-informed resolution path should be pursued in prefer-cations of all available alternatives for risk ence to a deterministic approach, thereby quantifying informing the path forward. These alterna-the safety margins and identifying any scenarios that tives include, but are not limited to, how pose significant risk in GSI-191.

50.-46a might impact this issue, and how the The primary issue here is that in a deterministic application of a no-transition-break-size ap-approach, examination of the uncertainty required to proach might work."

evaluate risk of plant operation is abandoned and an appeal to a hypothetical "worst case" scenario (as- It is worth reemphasizing that a risk-informed anal-sumed to encompass all uncertainty) is made. By as- ysis includes all scales of postulated accidents and suming a sufficient set of nonphysical processes along the full spectrum of possible plant responses. Ideally, with assumed equipment performance scenarios, ex- there should be no exclusive focus on "bounding" as-amination of the full range of possible scenarios is sumptions, no exclusive focus on "design basis" or avoided. Although this approach can be effective in "beyond design basis" assumptions, and no exclusive some cases, there are shortcomings associated with focus on "best estimate" assumptions. Every factor it (NEI, 2009) and, relative to GSI-191 closure, have in the accident analysis should be described as accu-become untenable. That is, the "worst case" is so rately as possible by a statistical distribution that is 1 of 15

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/Industry GSI-191 Closure Pilot Project consistent with available data, physical models, or ex- tion in order to ensure the Pilot Project met 2011 pert opinion, and so, factors with limited or no data schedule commitments. Section 2 provides a sum-necessarily have bigger uncertainties. Furthermore, mary of the major assumptions made in the initial the methods used to sample and propagate these un- quantification. Section 3 summarizes some of the sig-certainties should be unbiased. nificant insights obtained from the Pilot Project work In our initial quantification numerous conser- so far. Section 4 describes the several physical mod-vatisms are retained to maintain consistency with els that have been developed in 2011, others will be regulatory assumptions while developing a robust further refined and developed in the second project toolset and exploring parameter sensitivities. calendar year (2012). A new integration framework has been developed and is described as well in Sec-tion 4. A containment CAD model and toolset used 1.3 Pilot Project in the project are briefly described in Section 5. Sec-STPNOC is the plant working with the staff to de- tion 6 is a short description of the PRA modeling velop risk-informed closure strategies while preparing required to obtain the change in risk needed for Reg-a site-specific licensing submittal. Over the 2011 cal- ulatory Guide 1.174 decision-making. In Section 7 endar year, several public meetings were conducted the methodology developed in calendar year 2011 for to inform the NRC staff of the modeling approach uncertainty quantification is described. The status and to solicit feedback on the applicability and use of the Pilot Project LOCA frequency analysis is pre-of the approach for resolving GSI-191. The several sented Section 8. An overview of the Quality As-meetings included supporting material so that mem- surance Plan is provided in Section 9. The licensing bers of the public, especially other plants could be approach is designed around Regulatory Guide 1.174 informed as well: Rosenburg (2011), Thadani (2011), (Section 10). Conclusions are in Section 11.

Singal (2011c,a,b,d,e,f,g).

All materials provided have been posted by the NRC on their website for public access. 2 Major assumptions In design of the Pilot Project, STPNOC has been careful to make the proposed implementation at other Several assumptions had to be adopted in calendar plants as straightforward as possible. In particular, year 2011 to accomplish a quantification result within the physical models that typically would be folded the aggressive Pilot Project schedule. In some cases, into the event trees and fault trees of the PRA have the assumptions bias the resulting CDF and LERF been extracted into a flexible modeling tool and held higher, in some cases the assumptions bias the results outside the PRA. In the Pilot Project this part of lower. The Pilot Project has tried to adopt a policy the toolset is called CASA Grande (Letellier, 2011). regarding assumptions such that on balance, the CDF This way, the methodology is made more generic and and LERF results would be biased higher than with only simple changes should be required in the site- all assumptions relaxed.

specific PRA. The existing PRA LLOCA, MLOCA, The following list summarizes of the major assump-and SLOCA branches are maintained intact and only tions in calendar year 2011.

two relatively simple changes are required in the PRA ZOI: The zone of influence is assumed to be 17D for itself: the sump strainer demand failure likelihood is all fiberglass targets regardless of configuration replaced by a value that CASA Grande produces as or the presence of piping restraints. ZOI is not an output; and a top event must be added for long truncated by the presence of compartment walls.

term cooling relative to in-vessel effects.

The rest of this report summarizes the GSI-191 Chemical precipitants: All chemical precipitants risk-informed closure process and methods that have (formed based on the available material in con-been developed over the past eight months. Several tainment) are formed and introduced in bulk at assumptions were adopted in the initial quantifica- 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> following the LOCA regardless of the 2 of 15

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/Industry GSI-191 Closure Pilot Project operation of containment spray. In-vessel pre- (to avoid over-estimating or under-estimating cipitants are assumed to be of the same form the LOCA frequency in the source term). That and amount used by Bajer (2011). The head is, in each category of LOCA, the location-loss due to precipitants on the strainer debris specific values for that category closely sum to bed doubles the head loss through the bed. All its PRA initiating event frequency. For the pur-chemicals are passed through the strainer if it is pose of in-vessel hydraulic effects on core flow not entirely covered with a 1/8 inch-thick uni- and head loss calculations, the LOCA frequency form debris bed. is split evenly between hot and cold leg locations.

Debris generation: 60% of the debris is fines, 40% Qualified coatings: At time zero (at the time of is large. 1% of large debris is eroded to fines with the LOCA), 33 lb of Epoxy and 553 lb of IOZ 100% spray exposure. coating particulate is assumed to arrive in bulk to the sump pool. This amount is present in all Transport of fine debris: 100% go to the sump break scenarios (regardless of break size).

pool, 4% are retained on the strainer bed (per strainer), 13% go to dead volumne (other sumps, Unqualified coatings: At 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />, 247 lb of elevator shaft, dead volumes inside the bioshield Alkyd, 843 lb of IOZ, 268 lb of enamel, and wall).

294 lb of epoxy coating particulate arrives in the Transport of large debris: 1% is eroded to fine sump pool.

debris, 99% retained on gratings. 3 Latent debris: 170 lb of particulate and 12.5 ft of Upstream effects: Upstream effects are assumed fiber arrive at the sump pool at the start of the to have insignificant contribution. LOCA (time 0).

Strainer configuration: The strainer meets design Strainer bypass: Strainer bypass is a function of at the start of the LOCA transient (open bypass face velocity and the number of operating trains paths' contribution from damage for example, is only (Zigler, 2011). All debris bypassed is de-assumed negligible). posited in the core and evenly spread over the fuel assemblies.

Scenario event timing: The scenario progression and associated event timing are assumed to follow nominal values in each LOCA category Strainer head loss: Head loss follows the behavior described by Zigler et al. (1995). The strainers (small, medium, large LOCA). This also applies do not collapse from overload due to differential to Operator actions with the exception of con-pressure.

tainment spray operation. In the case of contain-ment spray, the Operator is successful in stop-ping the single train (when three trains have started) as required but fails to terminate con- 3 Findings in 2011 tainment spray operation (the remaining two pump trains) after meeting termination criteria. In calendar year 2011, a primary objective was to Containment spray is never terminated when less obtain an initial PRA quantification of the risk as-than three trains have started. sociated with fibrous insulation in containment. The results of the initial quantification were included in LOCA frequency: The existing STPNOC PRA the project plan as a criteria for going forward or LOCA branch frequencies for small, medium, terminating the project and taking aggressive correc-and large LOCA are preserved in the calcula- tive action to correct any high risk scenarios revealed tion of location-specific frequency calculations in the risk assessment. A major finding of the 2011 3 of 15

Enclosure NOC-AE- 12002784 STP Nuclear Operating Company's NRC,/Industr'y GSI-191 Closure Pilot Project Illustration of 2011 "Initial Quantification" on RG 1.174 Regions

• ........... . . Region I

" No changes allowed Region II

. Small changes LL* Track cumulative impacts Region III

< -.Very small changes

. More flexibility with respect to baseline CDF

. Track cumulative impacts t-5  :... .*.". .:* W&.i::

1U Region II 10-6 Fibirou's aisuiation risk at Sip (conservative)'

III

.Region 4 CDF 10-6 10-5 10 Figure 1: Illustration of the core damage (CDF, ACDF) risk associated with as-designed fibrous insulation in the STPNOC containment on the Regulatory Guide 1.174 decision-making Region map.

quantification is the risk is "very small" in the lan- the changes are not very small. In this case, based guage of Regulatory Guide 1.174 as illustrated in Fig- on the "very small" risk evaluation, STPNOC man-ure 1. Regulatory Guide 1.174 is concerned with the agement has authorized the project to go forward as need to make changes guided by risk. communicated in the calendar year 2011 Pilot Project plan.

The risk of ongoing operation is maintained in the site-specific PRA average risk model for CDF and LERF. The STPNOC PRA meets the ASME/ANS 3.1 Chemical effects PRA Standard as Capability Category II and has Based on early prototypical experiments performed successfully provided the technical basis for several by Dallman et al. (2006) and (much later) based on risk-informed applications at STPNOC, for example studies by Sande et al. (2011), STPNOC has main-RMTS (Yilmaz et al., 2009; EPRI, 2008). A change tained that in the initial quantification, chemical ef-to the plant may increase risk (CDF and/or LERF) fects could be minimized. STPNOC understood from and is expressed as ACDF and ALERF. Clearly, if a the start of the project that any assertions about change results in a large change in risk, it could affect chemical effects would require experimental verifi-the plant average risk. But if the ACDF is less than cation. However, based on the feedback from the 1.0X 10-6 and the ALERF is less than 1.0x 10', the NRC staff during the Pilot Project meetings, (Sin-change in risk is considered to be very small and the gal, 2011g), chemical effects were added to the initial cumulative risk to CDF and LERF do not need to be quantification. The approach was modified to include considered in decision-making. On the other hand, the results found by Baier (2011) in experiments in-the effect on plant average risk must be considered if tended for deterministic evaluations.

4 of 15

Enclosure NOC-AE- 12002784 STP Nuclear Operating Company's NRC/Industrgy GSI-191 Closure Pilot Project With the inclusion of Baier's chemical effects us- ing the system characteristic to steepen) due to fiber ing mass of fiber as the success criteria, the ACDF loading and chemical precipitant addition. In fact, increases from extremely low to roughly 5 x 10'. NRC staff has stated in Pilot Project review meetings The ALERF remains unchanged. Based on this re- (Singal, 2011e,f) this behavior has been observed dur-sult and based on earlier NRC staff Pilot Project ing experiments as chemical precipitants were added feedback, an important finding is that chemical ef- at constant flow and the head loss increased signif-fects on debris bed head loss needs to be better un- icantly (Baier, 2011). However, and again as illus-derstood. STPNOC has accelerated and expanded trated in the figure, this behavior might not be re-the risk-informed chemical effects experimental pro- alized in an actual scenario because in the unlikely gram in order to better understand chemical effects event of a LOCA that produces an in-vessel debris on head losses in ECCS strainer and fuel assembly bed, the required flow decreases dramatically (lin-debris beds. The chemical effects experiments will early with decay heat which is effectively an expo-also be designed to produce results that can be used nential function). Additionally, chemical precipitants to develop correlations for realistic physical models simply can't be created while the required flows are of chemical performance for use in CASA Grande. high (say within the first day to one week).

A significant result of the risk-informed analysis 3.2 Strainer and downstream perfor- that takes into account the evolution of the physi-cal processes is that: there are no ECCS failures due mance to strainer blockage (this needs to be examined for As mentioned earlier, the deterministic approach uses some infrequent potential sequences that, for exam-a hypothetical "worst case" scenario assumed to en- ple may cause mechanical collapse of the strainer);

compass all uncertainty assuming nonphysical mod- all hypothesized small LOCA events succeed; almost els and scenarios. Although it does create a bound- all medium LOCA events succeed; and most large ing case in some sense, the deterministic approach LOCA events succeed. All scenarios resulting in core has assumed chemicals as precipitants are produced damage were associated with in-vessel effects, not effectively at the start of the LOCA and during a hy- strainer blockage. That is, in a realistic setting that pothetical equipment alignment that produces max- retains significant conservative assumptions, chemi-imum ECCS flow rate through the strainer. None of cal effects turn out to be less important than previ-these conditions would ever exist simultaneously in ously thought. For example, a 15 grn/FA acceptance the unlikely event of a LOCA requiring recirculation criteria would cause failure for essentially all cases switchover with fibrous debris present. based on the latent debris alone. But in a realis-Debris beds in the core and on the strainers in- tic setting, it was demonstrated (again with conser-crease resistance to flow as the bed builds up thick- vatisms) that most hypothesized LOCA events would ness, but more importantly (based on experiment re- not cause core blockage.

sults for deterministic evaluations), the resistance in-creases more when chemical precipitants collect on 3.3 Destruction zone them also. Generally, head loss is higher order (nor-mally second order) with flow in a closed channel sys- A somewhat surprising result from the initial quan-tem. tification is the insensitivity to the "zone of influence" Zigler et al. (1995) utilized porous media correla- or ZOI for use in for example, ANSI (1988). A great tions where head loss, H, equals to H = aQ + bQ2 , deal of effort in experimentation and study by the in-where Q is the flow, and the coefficients a and b are dustry has been devoted to understanding if the ZOI determined by the fluid and debris bed conditions. could be reduced to much lower volumes of damage Referring to Figure 2, one can imagine how dra- than the widely accepted deterministic value defined matically the head loss would increase at constant by a sphere of 17D, where D is the break diameter.

flow rate as the coefficients a and b increase (caus- In fact Schneider et al. (2011) have been studying jet 5 of 15

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/Industry GSI-191 Closure Pilot Project Later, increasing debrisichemical Earlier, when debris/chemical luads ae small, loads steepens system characteristic:

and the cooling flow requirement is high, the however, the cooling flow requirement system characteristic is flatter: lower head losses decreases rapidly as well.

at higher flows.

500 400 4 Flow rate needed to cool the core (gpml 700 g 600 Sao V_610 9 11rr1 o 400 300 ~1' 0 261 gpm 200 100 46 gpm 0

4 f 12 16 20 24 708 712 716 720 Time (hr)

Figure 2: Illustration showing how decreasing decay heat removal flow requirement over time reduces cooling flow requirements that would result in significant reductions in head loss across debris beds. Experiments of in-core performance with chemical effects show significantly greater tolerance (higher mass loadings) for fully loaded debris beds at lower flow rates.

formation for use in the risk-informed methodology. ducing the ZOI to about 1/2 17D. While some reduc-The sense is that these destruction volumes are much tion was observed, this change did not significantly too large. For example, a 31 inch pipe break would affect the risk. This surprising result can be explained create a sphere of destruction about 44 feet in diam- by understanding the scenarios leading to core dam-eter. A commonly accepted practice is to limit the age. Because the STPNOC strainer design was mod-ZOI such that it can't extend beyond the concrete ified to install very large strainers compared to the walls of containment compartments. original strainer design, risk for loss of ECCS NPSH is effectively eliminated. However, the sump perfor-In this initial risk-informed quantification, the 17D mance gain comes at a price for core damage risk, destruction volume was assumed in all break loca- especially when deterministic-based chemical effects tions without restriction for compartment walls. A are used for success criteria.

sensitivity study was conducted to see the effect of re-6 of 15

Enclosure NOC-AE- 12002784 STP Nuclear Operating Company's NRC/Industry GSI-191 Closure Pilot Project The amount of fiber arriving on the fuel is almost ple, the strainer loading and the downstream (core entirely governed by the strainer face velocity at the FA) loading.

time of recirculation switchover and for a few min- The inputs to CASA Grande are a combination utes following recirculation switchover. Except for of conservative, yet realistic, treatments of accident the smallest breaks, the fiber load is governed by the phenomenology and decision criteria based on pre-number of ECCS trains running and the train config- cursors to possible system damage. This methodol-uration. That is, the core fiber load (which as men- ogy enables risk-informed insights without compro-tioned above is limiting) is (almost) independent of mising the traditional safety basis. In the initial risk break size and ZOI. quantification for calendar year 2011, the plant per-formance metrics of head loss across the recirculation strainers and fiber deposition per fuel assembly were 4 CASA Grande used to assess the risk of flow blockage leading to core damage during recirculation scenarios. Detailed de-The risk analysis examines the full spectrum of scription of the models and methods that constitute LOCA accident conditions ranging from predomi- CASA Grande can be found in Letellier (2011).

nant, but small accidents that are easily managed up to and including extreme, but unlikely accidents that challenge the design basis. The analysis framework, 5 Plant configuration CASA Grande, ultimately develops cumulative distri-bution functions (cdf) for exceedance of a particular The basis for the insulation source term and locations value used in success criteria (for example, fiber load- for all welds and plant components in containment ing per fuel assembly) for each of the standard PRA is the current STPNOC plant design drawings and initiating event LOCA categories, LLOCA, MLOCA, design configuration database. Although other for-and SLOCA, and for each of the possible equipment mats may be equally useful, the Pilot Project chose configurations as analyzed in the PRA. Once de- to exploit the availability of computer aided design veloped, the failure likelihood for each scenario can tools (CAD) as the method to capture and integrate be determined knowing the associated value as illus- the spatial information required to accurately calcu-trated in Figure 3. Note that if a value coming from a late the weld LOCA locations, debris quantity (het-bounding deterministic experiment is available, then erogeneously distributed insulation) and type of de-below that amount, success is assured. bris. Additionally, specialized CAD interface tools For example, if the amount of fiber is the crite- were developed that efficiently and reliably interro-ria value, and say 75 gm/FA is the amount of fiber gate the CAD model. The output of the tool set re-from an LLOCA, two available ECCS trains, and that duces the spatial information contained in the CAD value falls below the intersection of the distribution to a database accessible by CASA Grande such that on the abscissa, conceptually one would draw a verti- the debris source term can be calculated accurately cal line from that value to the distribution curve and for all locations, break orientations, and break sizes.

look up the split fraction value (the failure probabil- Although not credited in the initial quantifica-ity) on the ordinate. It is important to understand tion, compartment wall information is included in that thousands of samples for different scenarios pro- the database so that when wall truncation is imple-duce the complementary distribution and it is pos- mented in 2012, the ZOI will be properly shaped and sible that the same likelihood would come from dif- limited by their presence. A configuration database is ferent scenarios. For example, a small break with a required because CASA Grande samples all welds at lot of fiber nearby may result in the same value (say every location in each replication of the total calcula-fiber loading) as a larger break having less damage tion. For each LOCA category and at each location, opportunity. From CASA Grande, each scenario can 10 to 15 samples, random in size and direction, are be traced from where the break occurs to, for exam- taken for the source term calculation. Each sample 7 of 15

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/lndustry GSI-191 Closure Pilot Project If the threshold of concern 1.0 value increases, the exceedance At sufficiently high values ea, the likelihood goes to 0.0 valueiincreaseseorese(smaller the efcodnc likeihoo folow alng pcm lativfrato ailurev o fo xape 0

0 U.uI Performance measure (e.g., core fiber mass)

Figure 3: Illustration of a CASA Grande (complementary) cumulative distribution and the method used to develop input for the PRA from it.

always includes the largest possible break (double- the time of recirculation. Additionally, in each of ended guillotine, DEGB) at each location. A full these branches, there are the preexisting scenarios for spherical ZOI shape is used for the DEGB sample. the different combinations of ECCS equipment de-Otherwise, a full hemispherical ZOI shape is used. pending on support system and component successes The configuration database is designed to support au- or failures. The Pilot Project exploits this existing tomatic calculation of the debris source terms (thou- structure by replacing failure likelihoods based on the sands of calculations in each CASA Grande repeti- results of CASA Grande as described in Section 4 in-tion). stead of the simplistic demand failure likelihood.

As described in Section 1.3, the physical mod-6 PRA els have been extracted out of the PRA to enhance portability, simplify incorporation of different plant The typical PRA will have initiating events for information and characteristics, and facilitate adop-LLOCA, MLOCA, and SLOCA. It is interesting to tion of the methodology. Another inherent complex-understand that before GSI-191 became an issue, ity fundamental to typical PRA design is overcome by each of the LOCA initiating event branches already performing the uncertainty quantification in CASA contemplated a sump screen plugging event (most Grande. In the Pilot Project, time-dependent and likely as in the STPNOC PRA) represented as a multivariate uncertainty distributions (described in demand failure of the containment sump screens at Section 7) have been identified in the physical models 8 of 15

Enclosure NOC-AE- 12002784 STP Nuclear Operating Company's NRC/Industry GSI-191 Closure Pilot Project Distributions: Scenariol Scenario2 . . .

The recirculation failure event LLO LLii L

is actually a top that includes other failures such as valves failing to transfer. However, all the failure scenarios include MLOC L L~ the sump blockage failure.

Figure 4: High level illustration of the inputs and outputs overlaid on a schematic diagram of the PRA event tree structure. The event tree structure, with the exception of the long term cooling top event, already exists in the typical nuclear power plant PRA.

required to understand GSI-191. CASA Grande top event was required in the STPNOC PRA to ad-also provides the platform to propagate the time- dress the possibility of core flow blockage (long term dependent and multivariate uncertainties outside the cooling). This is the only structural change required PRA. In fact, this is a basic requirement of a risk- in the PRA and is considered to be a relatively minor informed analysis because current commercial PRA change.

methodology is not fully capable of propagating the A calculation can be performed (at any particu-required uncertainties.

lar plant) by summing the current frequencies as-Referring to Figure 4, the preexisting event tree sociated with each LOCA initiating event that go structure and LOCA initiating event frequencies are through recirculation to success. With each of these not changed by the Pilot Project methodology. The initiating event frequencies in hand, one can multiply distributions from CASA Grande described in Sec- each by its associated failure likelihood from CASA tion 4 are used in the recirculation switchover event Grande. The resulting frequency sum is the new suc-to substitute specific scenario likelihoods for the (in- cess frequency and the difference between the origi-variant) simplified demand recirculation likelihood (a nal frequency (the unchanged frequencies) sum from single basic event in the preexisting model). A new the new frequency (multiplied by the CASA Grande 9 of 15

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/Industry GSI-191 Closure Pilot Project A

Figure 5: Uncertainty quantification for complex computer models likelihoods) sum is a close approximation to the new faced by the Pilot Project during the 2011 were the ACDF due to having fibrous insulation in contain- choice of probability distributions for the input pa-ment. rameters of CASA Grande and how to sample from them.

One of the inputs to CASA Grande is the LOCA 7 Uncertainty Quantification frequency table, Fleming et al. (2011). The proba-(UQ) bilities associated with LLOCA are extremely small and simple Monte Carlo sampling approach will al-The modeling and propagation of uncertainties for most never produce that observation. In order to the GSI-191 project involves several steps. Figure 5 have the LLOCA in all of the scenarios, nonuniform shows the collection of information flow, methodolo- Latin Hypercube Sampling was applied, Helton and gies and challenges for UQ of computer models that Davis (2003). This methodology generates observa-approximate a complex real system like the one being tions from the whole space (i.e. the LLOCA is al-considered in this pilot project. The main challenges ways included) and weights them by the correspond-10 of 15

Enclosure NOC-AE-12002784 STP Nuclear Operating Company 's NRC/Industry GSI- 191 Closure Pilot Project ing probabilities. The Pilot Project will expand upon 9 Quality Assurance this approach in 2012 to be able to generate corre-lated samples and estimate quantiles using the data A quality assurance plan has been developed in cal-generated by CASA Grande. endar year 2011. The plan includes regularly sched-Another major task during the 2012 will be the ver- uled (nominally weekly) technical review teleconfer-ification and validation of the methodology applied ences supplemented at critical product development to solve the GSI-191. The verification part will make steps with on-site review. The STPNOC PRA ana-sure that all the code written is "bug-free" and repre- lyst (Technical Team Lead) is responsible for review sents correctly the theory and methods implemented. and verification of the PRA inputs developed. How-The validation part is challenging since it will have ever, the STPNOC PRA analyst review is supple-to be shown that the approximation produced is very mented by independent critical peer review intended close to the real system. Popova and Galenko (2011) to help disclose any overlooked technical gaps that provide a detailed description and illustrative exam- would compromise results and also help ensure that ples of the UQ methodology applied to the initial the overall product is academically defensible even quantification. though it is developed for the industrial setting. In-dependent technical oversight was a part of the STP-NOC 2011 efforts and was used to further focus anal-ysis efforts.

8 LOCA frequency The overall quality assurance plan is illustrated in Figure 6 as a flow chart. Due to the diverse technol-In order to explore all possible scenarios, the risk- ogy required to be implemented in the GSI-191 scope, informed analysis requires knowledge of the likeli- the PRA inputs originate with products developed by hood of a pressure boundary failure for all possible experts in their respective field. The CASA Grande locations and possible sizes of failure. Per design, integrating framework uses the inputs to generate the significant rupture failures of Class 1 piping in do- two main inputs to the PRA, the sump demand fail-mestic light water reactors have not been observed. ure likelihood and the in-vessel cooling failure like-Obtaining the appropriate likelihoods where there is lihood (for each category of LOCA and all possible no evidence presents a challenging problem for the equipment configurations). These elements are doc-GSI-191 risk-informed closure investigator. umented by the vendor and the normal vendor doc-In the initial quantification, Fleming et al. (2011) ument review process is followed to assure they are performed a substantial study designed to build suitable for use as input to the PRA. The overall Pilot upon the established EPRI risk-informed In-Service Project quality assurance methodology is expected to Inspection program (EPRI, 1999). EPRI (1999) be similar to most utilities' processes for PRA activ-methodology was used as primary basis to develop ities although the details are most likely different.

the size and location-specific rupture frequencies for the quantification. In the Pilot Project study, Flem-ing et al. (2011) showed the total frequency in 10 Licensing the standard PRA methodology (SLOCA, MLOCA, and LLOCA) were preserved. Although the over- STPNOC will rigorously quantify the risk contribu-all methodology is considered to be sound based on tion to core damage and large early release of recir-peer review (Mosleh, 2011), and reasonableness of culation scenarios that are encompassed by GSI-191.

the values obtained, NRC review feedback in the The licensing strategy is based on looking at the dif-Pilot Project has resulted in further review of the ference in risk between a hypothetical containment approach. In 2012 an alternative to the LOCA fre- design that has no fibrous insulation and the existing quency methodology will be performed on a new basis STPNOC design. The expectation is that the hy-to fully address NRC concerns. pothetical containment design would have lower risk 11 of 15

Enclosure NOC-AE- 12002784 STP Nuclear Operating Company's NRC/lIndustry GSI-191 Closure Pilot Project Responsibility: Contracted service organization Process: Local quality program Input development Ga 30 I I I I I I

Responsibility:LANL CL CASA Framework I

Process: Local Quality Program Responsibility STP ContracrTechnical Coordinator, Project Technical Lead ProcessaOTP Technical Document Review Process Procedure OPGPO4-ZA-0 108 Approved/Distributed Vendor Document Control Process' Input to PRA Internal review supplemented and supported by outside Peer Revuew Inputs to PRA Verified/Reviewed Responsibility: ABS Consulting PRA Quantification/Output I

Process: IOCFR5OAppendix B Program Responsibility STP Contract Technical Coordinator, ProjectTechnical Lead Process' STP PRA AssessomentProcess PRA Application ProcedureOPcSP04-ZA-r)6O'rProbabiiisitc RiskAssessment Program' Responsibility. STPLicensing Engineer I

License Amendment Process: 5fTrLicenseAmendment Process Request Procedure OPGCp5-ZN-*00r'Changes to Licensing Basis Documents and Amend.ments to the Operating License-Figure 6: Illustration of the major elements of the STPNOC quality assurance plan for risk-informed closure of GSI-191.

than the existing design. Clearly an ideal outcome analysis to the Regulatory Guide 1.174, Region III 1

would be the difference in risk between the two de- limit of ACDF< 10-6 yr- and ALERF< 10-7 yr-1 signs to be roughly zero. will provide a basis for GSI-191 closure via exemp-tion to certain requirements in 10 CFR 50.46. The The difference between a comparison of the risk 12 of 15

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/Industry CSI-191 Closure Pilot Project terms ACDF and ALERF in the context of this work appreciated by the Pilot Project due to the aggressive mean the difference in CDF and LERF between a hy- schedule and complexity of the issues.

pothetically perfect containment building having no The Pilot Project is designed generically so that fibrous insulation and the existing plant design. The other plants can easily take advantage of it. The purpose of the initial quantification in calendar year design provides for simple integration of site-specific 2011 was to understand if the risk associated with models in a method that extracts complexities out fiber insulation in containment would exceed the Re- of the PRA and integrates them in a flexible model-gion III limits. ing tool, CASA Grande. A straightforward licensing If the final analysis (calendar year 2013) continues approach is proposed and generally accepted quality to support recirculation failure as a Risk Region-Ill assurance methods can be applied.

event, preventative measures (safety margin, defense Even with conservative margins retained, the Pilot in depth) will be identified to address contributing Project in the initial quantification has shown that factors that carry the largest potential impacts in the risk from fibrous insulation in containment at the analysis. Regardless of the quantitative findings, STPNOC is very small (in Region III) based on the STPNOC will use the risk analysis to prioritize spe- decision-making criteria of Regulatory Guide 1.174.

cific actions and determine the degree of remediation This helps assure both the NRC and STPNOC that that may be required. Thus, it is essential that all defense against ECCS failures can be easily identified parties fully understand the theory, implementation and reasonable measures can be undertaken to avoid methods and interpretation of a risk-informed deci- excessive risk in a risk-informed closure path.

sion process. STPNOC intends to continue to com-municate regularly and openly with the NRC staff and industry in calendar years 2012 and 2013, and References continually refine and enhance communication and communication tools as the Pilot Project evolves. ANSI (1988). ANSI/ANS-58.2-1988: Design Basis for Protection of Light Water Nuclear Power Plants Against the Effects of Postulated Pipe Rupture.

11 Conclusions Design Standard 58.2-1988, ANSI/ANS, Washing-ton, DC.

The calendar year 2011 initial quantification in the Pilot Project has demonstrated the viability of a risk- Baier, S. L. (2011, Septmeber). GSI-191 Fuel Assem-informed closure path to GSI-191. Even at the early bly Test Report for PWROG. WCAP 17057 Revi-stage, the usefulness of the risk-informed approach sion 1, Westinghouse PWROG, Pittsburgh, PA.

has been demonstrated in initial findings related to ZOI, chemical effects, time-dependency, and LOCA Dallman, J., B. Letellier, J. Garcia, J. Madrid, frequency. In a short amount of time, a comprehen- W. Roeschy, D. Chen, K. Howe, L. Archuleta, sive understanding of the GSI-191 important physical F. Sciacca, and B. P. Jain (2006, December).

phenomena has been accomplished and has been used Integrated Chemical Effects Test Project: Con-by the Pilot Project to develop a new, robust, risk- solidated Data Report. NUREG/CR 6914, Los informed framework for study and closure of GSI-191. Alamos National Laboratory, Los Alamos, NM.

The Pilot Project has worked closely and effec-tively with the NRC to incorporate feedback and in- Darby, J., D. V. Rao, and B. Letellier (2000). GSI-form the NRC staff of progress and the technology 191 STUDY: TECHNICAL APPROACH FOR developed. In some cases, NRC feedback has redi- RISK ASSESSMENT OF PWR SUMP-SCREEN rected the efforts of the Pilot Project and acceler- BLOCKAGE. Technical Letter Report LA-UR-ated the schedule (for example, chemical effects and 00-5186, Los Alamos National Laboratory, Los LOCA frequency). The interaction and feedback is Alamos, NM.

13 of 15

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/Ilndust?-y GSI-191 Closure Pilot Project EPRI (1999). Revised Risk-Informed In-Service In- White Paper ALION-REP-STPEGS-8221-02, Re-spection Procedure. TR 112657 Revision B-A, vision 0, Jointly, Alion Science and Technology and Electric Power Research Institute, Palo Alto, CA. Univiersity of New Mexico, Albuquerque, NM.

EPRI (2008). Risk-Managed Technical Specifications Schneider, E., J. Day, and W. Gurecky (2011, De-

- Lessons Learned from Initial Application at South cember). Simulation Modeling of Jet Formation Texas Project. TR 101672, Electric Power Re- Progress Report, August - December 2011. In-search Institute, Palo Alto, CA. ternal Report Revision 0, University of Texas at Austin, Austin, TX.

Fleming, K. N., B. 0. Lydell, and D. Chrun (2011, July). Development of LOCA Initiating Event Singal, B. K. (2011a, .June). FORTHCOMING CON-Frequencies for South Texas Project GSI-191. FERENCE CALL WITH STP NUCLEAR OP-Technical report, KnF Consulting Services, LLC, ERATING COMPANY (TAC NOS. ME5358 and Spokance, WA. ME5359). Memorandum.

Helton, J. and F. Davis (2003). Latin hypercube sam- Singal, B. K. (2011b, July). FORTHCOMING CON-pling and the propagation of uncertainty in analy- FERENCE CALL WITH STP NUCLEAR OP-ses of complex systems. Reliability engineering & ERATING COMPANY (TAC NOS. ME5358 and systems safety 81, 23-69. ME5359). Memorandum.

Letellier, B. (2011). Risk-Informed Resolution of Singal, B. K. (2011c, May). FORTHCOMING GSI-191 at South Texas Project. Technical Report MEETING WITH STP NUCLEAR OPERATING Revision 0, South Texas Project, Wadsworth, TX. COMPANY (TAC NOS. ME5358 and ME5359).

Mosleh, A. (2011, October). Technical Review of STP Memorandum.

LOCA Frequency Estimation Methodology. Letter Singal, B. K. (2011d, August). FORTHCOMING Report Revision 0, University of Maryland, College MEETING WITH STP NUCLEAR OPERATING Park, MA.

COMPANY (TAC NOS. ME5358 and ME5359).

NEI (2009). ECCS Recircultation Performance Fol- Memorandum.

lowing Postulated LOCA Event: GSI-191 Ex-Singal, B. K. (2011e, October). FORTHCOMING pected Behavior. White Paper.

PUBLIC MEETING VIA CONFERENCE CALL Popova, E. and A. Galenko (2011, Deecember). Un- WITH STP NUCLEAR OPERATING COM-certainty Quantification (UQ) Methods, Strate- PANY (TAC NOS. ME5358 and ME5359). Mem-gies, and Illustrative Examples Used for Resolv- orandum.

ing the GSI-191 Problem at South Texas Project.

Technical Report Revision 0, The University of Singal, B. K. (2011f, September). FORTHCOMING Texas at Austin, Austin, TX. PUBLIC MEETING VIA CONFERENCE CALL WITH STP NUCLEAR OPERATING COM-Rosenburg, S. (2011, January). PUBLIC MEETING PANY (TAC NOS. ME5358 and ME5359). Mem-WITH THE NUCLEAR ENERGY INSTITUTE orandum.

ON STATUS AND PATH FORWARD TO RE-SOLVE GSI-191. Memorandum. Singal, B. K. (2011g, November). FORTHCOMING PUBLIC MEETING VIA CONFERENCE CALL Sande, T., K. Howe, and J. Leavitt (2011, October). WITH STP NUCLEAR OPERATING COM-Expected Impact of Chemical Effects on GSI-191 PANY (TAC NOS. ME5358 and ME5359). Mem-Risk-Informed Evaluation for South Texas Project. orandum.

14 of 15

Enclosure NOC-AE-12002784 STP Nuclear Operating Company's NRC/Industry GSI-191 Closure Pilot Project Teolis, D., R. Lutz, and H. Detar (2009). PRA Mod-eling of Debris-Induced Failure of Long Term Cool-ing via Recirculation Sumps. WCAP 16882, West-inghouse Electric Company, LLC, Pittsburgh, PA.

Thadani, M. (2011, February). FORTHCOMING MEETING WITH STP NUCLEAR OPERATING COMPANY (TAC NOS. ME5358 and ME5359).

Memorandum.

Vietti-Cook, A. L. (2010, December). STAFF RE-QUIREMENTS - SECY-10-0113 - CLOSURE OPTIONS FOR GENERIC SAFETY ISSUE-191, ASSESSMENT OF DEBRIS ACCUMULATION ON PRESSURIZED WATER REACTOR SUMP PERFORMANCE. Letter from Annette L. Vietti-Cook to R. W. Borchardt.

Yilmaz, F., E. Kee, and D. Richards (2009, July).

STP Risk Managed Technical Specification Soft-ware Design and Implementation. In Proceedings of the 17th International Conference on Nuclear Engineering, Number 17-75043 in ICONE. Interna-tional Conference on Nuclear Engineering: ASME.

Zigler, G. (2011, January). Quantification and Char-acterization of By-pass Fiber Debris. Internal Re-port Revision 2, Alion Science and Technology, Al-buquerque, NM.

Zigler, G., J. Brideau, D. V. Rao, C. Shaffer, F. Souto, and W. Thomas (1995, October). Para-metric Study of the Potential for BWR ECCS Strainer Blockage Due to LOCA Generated Debris.

NUREG/CR 6224, Science and Engineering Asso-ciates, Inc., Albuquerque, NM.

15 of 15