Text

Supporting Documents for Upcoming SNC NRC Public Meeting Regarding the Resolution of GS1-191
	ML15293A186
Person / Time
Site:	Vogtle
Issue date:	10/15/2015
From:	Pierce C; Southern Co, Southern Nuclear Operating Co
To:	; Document Control Desk, Office of Nuclear Reactor Regulation
Shared Package
ML15293A180	List: ML15293A186; NL-15-1904, Attachment F of Enclosure 2, Calculation No. ALION-CAL-SNC-7410-005, Revision 1, Head Loss Testing of a Prototypical Vogtle 1 and 2 Strainer Assembly; NL-15-1904, Vogtle, Units 1 and 2 - Supporting Documents for Upcoming SNC NRC Public Meeting Regarding the Resolution of GS1-191;
References
	NL-15-1904
	Download: ML15293A186 (17)
	v • d • e

Charles R. Pierce Regulatory Affairs Director Southern Nuclear Operating Company, Inc.

40 Inverness Center Parkway Post Office Box 1295 Birmingham, AL 35242 Tel 205.992.7872 Fax 205.992.7601 NUCLEAR A SOUTHERN COMPANY

.October 15, 2015 Docket Nos.: 50-424 50-425 NL-1 5-1904 U. S. Nuclear Regulatory Commission ATTN: Document Control Desk Washington, D. C. 20555-0001 Vogtle Electric Generating Plant, Units 1 & 2 Supporting Documents for Upcoming SNC NRC Public Meeting Regqardingq the Resolution of GS1-191 Ladies and Gentlemen:

A Nuclear Regulatory Commission (NRC) public meeting is being scheduled for later this year to discuss Vogtle Electric Generating Plant's planned resolution to GSI-1 91. In support of this upcoming public meeting, Southern Nuclear Operating Company (SNC) is submitting the attached reports to the NRC. While SNC is not requesting formal NRC review and approval of these reports, these reports help form a significant portion of the planned technical discussion for the meeting. To facilitate meaningful discussion, it is suggested that the NRC staff that plan on attending the meeting familiarize themselves with the content of these reports prior to the meeting. Because of the size of Enclosure 2, only the relevant portions to the upcoming meeting are provided. A full version of can be provided upon request.

This letter contains no NRC commitments. If you have any questions, please contact Ken McElroy at (205) 992-7369.

Respectul smitted, C. R. Pierce Regulatory Affairs Director CRP/RMJ

Enclosures:

1. Risk-Informed GS1-1 91 Uncertainty Quantification

2.

Head Loss Testing of a Prototypical Vogtie 1 and 2 Strainer Assembly

(

U. S. Nuclear Regulatory Commission NL-15-1904 Page 2 cc:

Southern Nuclear Operatin~q Company (w/o Enclosures)

Mr. S. E. Kuczynski, Chairman, President & CEO Mr. D. G. Bost, Executive Vice President & Chief Nuclear Officer Mr. D. R. Madison, Vice President - Fleet Operations Mr. M. D. Meier, Vice President - Regulatory Affairs Mr. B. K. Taber, Vice President - Vogtle 1 & 2 Mr. B. J. Adams, Vice President -- Engineering Mr. G.W. Gunn, Regulatory Affairs Manager - Vogtle 1 & 2 RType: CVC7000 U. S. Nuclear Regiulatory Commission Mr. L. D. Wert, Regional Administrator (Acting)

Mr. R. E. Martin, NRR Senior Project Manager - Vogtle 1 & 2 Mr. S. S. Koenick, NRR Senior Project Manager - Vogtle GS1-1 91 Mr. V. Cusumano, Chief, Safety Issue Resolution Branch Mr. L. M. Cain, Senior Resident Inspector - Vogtle 1 & 2

Vogtle Electric Generating Plant, Units 1 & 2 Supporting Documents for Planned November 5, 2015 SNC NRC Public Meeting Regarding the Resolution of GS1-191 Risk-Informed GS1-191 Uncertainty Quantification

OENERCON Excellence--Every project. Every day.

Risk-Informed GS1-191 Uncertainty Quantification This document discusses two methods that can be used to quantify the uncertainty associated with a risk-informed GS1-191 evaluation. The first method is a detailed statistical approach for sampling input parameter probability distributions and propagating the uncertainties. The second method is a simplified approach for selecting bounding input parameter values and calculating the uncertainty range using sensitivity analysis.

Over the past year, the industry has been moving in the direction of using simplified methods for risk-informed GS1-191 evaluations [1, 2]. Implementing simplified methods is beneficial since it allows the ECCS strainer performance issue to be resolved more efficiently and reduces the time and effort required for NRC technical review. Therefore, this document focuses primarily on the simplified approach for uncertainty quantification.

1. Introduction Uncertainty quantification is a key requirement in Regulatory Guide (RG) 1.174 for a risk-informed evaluation [3]. As defined in RG 1.174 and explained in more detail in NUREG-1 855 [4] and two corresponding EPRI reports [5, 6], there are three types of epistemic uncertainty that should be addressed:

1. Parametric uncertainty

2. Model unciertainty

3. Completeness uncertainty Parametric uncertainty refers to the variability in input parameters that are used in the risk assessment. Due to the wide range of plant-specific post-LOCA conditions related to GS1-191 phenomena, this is a very important aspect for understanding the overall uncertainty.

Model uncertainty refers to the potential variability in an analytical model when there is no consensus approach. A consensus approach is a model that has been widely adopted or accepted by the NRC for the application for which it is being used [4]. For example, the use of a spherical zone of influence (ZOI) to model the debris quantity generated by a high energy break is a consensus model that has been widely adopted and accepted by the NRC [7, 8]. In general, plants implementing a simplified risk-informed approach are using standard models that have been widely accepted for deterministic evaluations (e.g., accepted insulation and qualified coatings ZOI sizes, the use of WCAP-16530 [9] to model chemical effects, and prototypical strainer module testing for head loss and penetration). By using these consensus approaches, the effort to address model uncertainty is minimized.

Completeness uncertainty refers to 1) the uncertainty associated with scenarios or phenomena that are excluded from the risk evaluation, and 2) the uncertainty associated with unknown phenomena. Although it is not practical to quantify the uncertainty associated with factors that are not explicitly modeled (e.g., secondary side breaks, or breaks downstream of the first isolation valve), their potential impact can be qualitatively assessed. Uncertainties associated with unknown phenomena, on the other hand, cannot even be qualitatively assessed. Uncertainties T. Sande T.

Sande ~~Rev.

E, July 8, 2015 Pg f1 Page 1 of 14

O ENERCON

Excellence--Every project. Every day.

associated with unknown phenomena are the reason that it is important to maintain defense-in-depth and safety margins.

This document primarily focuses on the method for quantifying parametric uncertainty, since this is the most important uncertainty for a simplified risk-informed GS1-191 evaluation that can be quantified. The uncertainty quantification will ultimately be used to define the confidence in the calculated mean values for the change in core damage frequency (ACDF) and change in large early release frequency (ALERF) due to GS1-191 effects.

2. Method for Parametric Uncertainty Quantification There are two methods that can be used for quantifying parametric uncertainty. The first method is a detailed statistical approach for sampling input parameter probability distributions, propagating the uncertainties, and producing probability distributions for the results. This is a rigorous and commonly used method for uncertainty quantification.

However, the preferred method for the simplified risk-informed approach is to calculate mean ACDF and ALERF values using a combination, of bounding and nominal/realistic input values, and define the uncertainty as the maximum change in ACDF and ALERF when the nominal input values are all changed to bounding values. Although this approach does not show the shape of the ACDF and 6LERF probability distributions, or the weight of the tails, it does provide a good indication of the overall uncertainty in the evaluation.

The simplified uncertainty quantification method fits very well with the overall simplified GS1-191 approach. At a high level, the evaluation methodology would include the following steps:

1. Select appropriate models for debris generation, transport, chemical effects, strainer head loss, etc. based on models that have been generally used and accepted for past GS1-191 evaluations.

2. Select input values (for calculating the mean ACDF and ALERF) where each input is either:

a. A bounding value similar to a value that would be used in a deterministic evaluation (e.g., a maximum latent debris quantity based on plant walkdowns)

b. A nominal value based on realistic plant design and operation (e.g., an initial RWST level that is above the low level alarm based on historical operating conditions).

3. Execute the integrated models with the specified inputs to determine which breaks lead to success and failure.

4. Use the conditional failure probability to calculate the mean ACDF and ALERF.

5. Evaluate a set of sensitivity cases where each nominal input value is changed to a bounding input value (minimum and/or maximum) to determine the minimum 1 and maximum ACDF and ALERF values.

1Note that the minimum ACDF and ALERF are generally not as important as the maximum ACDF and ALERF, and wouldn't necessarily have to be calculated. Also, the use of some bounding input values in Step 2.a biases the mean ACDF and ALERF toward the maximum.

T. Sande Rev. E, July 8, 2015 Page 2 of 14

O ENERCON Excellence--Every project. Every day.

3. Input Parameter Selection Determining whether to use a bounding or realistic value for each input parameter (for the purpose of calculating the mean ACDF and ALERF) is a plant-specific process. This process involves a consideration of the level of conservatism that can be tolerated, the confidence in the test or analysis used to determine the value, how the overall analysis will affect the plant design and licensing basis, and other factors. Some plants may choose to use a set of inputs that are mostly bounding, while other plants may choose to use realistic values for more of the input parameters to get a more accurate prediction of the post-LOCA conditions.

If a bounding value is selected for an input parameter, it is essentially equivalent to a consensus model where the uncertainty does not need to be quantified. This is consistent with the guidance in Draft Regulatory Guide (DG) 1322 (e.g., Paragraph C.10.d) [10]. For example, a plant may use a latent debris quantity that is 50% higher than what is measured in a plant walkdown in order to bound any uncertainty in the measurements and provide operating margin for potential changes in containment cleanliness. In this case, since the latent debris quantity exceeds the expected maximum value, it is not necessary to quantify the uncertainty. However, another plant may choose to perform their analysis using the actual results of the latent debris measurements without including additional margin. In this case, it is necessary to evaluate the uncertainty associated with the latent debris quantity based on uncertainties in the walkdown measurements and potential future changes based on the level of rigor in the containment cleanliness program.

Depending on the models that are used, the worst-case direction for some input parameters may not be intuitively obvious. For example, a minimum water temperature could be worse with respect to strainer head loss, but a maximum temperature could be worse with respect to degasification.

Similarly, a minimum pool volume could be worse with respect to NPSH margin, but either a minimum or maximum pool volume could be worse with respect to the quantity of chemical precipitates predicted using the WCAP-16530 methodology. For these types of parameters, the' best approach may be to select realistic mean conditions to calculate the mean ACDF and ALERF values. Sensitivity analyses with each possible combination of bounding values could then be used to determine the best and worst case scenarios that would provide the uncertainty bounds showing the minimum and maximum ACDF and ALERF values (e.g., for the two parameters described above, a 2x2 simulation matrix could be run to evaluate the combinations of minimax pool temperature and min/max pool volume).

Table 1 shows a summary of several important input parameters for a GS1-191 evaluation with an indication of which direction is more limiting in terms of strainer or core failures. This table illustrates the logic for determining the worst-case conditions. However, the worst case (or best case) set of input parameters is highly dependent on plant-specific configurations as well as the models that are implemented. Therefore, the bounding direction shown in this table would not be applicable for every plant.

T. Sande T.

Sande ~~Rev.

E, July 8, 2015 Pg f1 Page 3 of 14

OENERCON Excellence--Every project. Every day.

Table 1: Bounding direction for imp*ortant input parameters Parameter Bounding Direction Bounding Direction Comments for Strainer Failures for Core Failures Fiber Insulation Debris Maximum Maximum Quantity Qualified Coatings Debris Maximum N/A Core acceptance criteria only a function of fiber quantity Quantity Microporous Insulation Maximum N/A Debris Quantity Unqualified Coatings Maximum N/A Debris Quantity Latent Debris Quantity Maximum Maximum Miscellaneous Debris Maximum Minimum Miscellaneous debris blocks strainer area, which could be Quantity "beneficial" for reducing penetration Debris Transport Maximum Maximum Fractions Pool Volume/Level Minimum or Minimum or Affects NPSH margin, degasification, partial submergence, time-Maximum Maximum dependent transport, pH, chemical release, and chemical solubility Containment Pressure Minimum N/A Affects degasification and NPSH margin Pool Temperature Minimum or Minimum or Affects NPSH margin, degasification, pool volume/level, chemical Maximum Maximum release, chemical solubility, head loss ECCS Flow Rate Minimum or Minimum or Affects head loss, NPSH margin, time-dependent transport, Maximum Maximum penetration, core accumulation, pool volume, and degasification CS Flow Rate Minimum or Minimum Affects head loss, washdown transport, time-dependent transport, Maximum penetration, pool volume, and core accumulation ECCS/CS Switchover Minimum or Minimum or Affects pool volume, NPSH margin, and core accumulation Time Maximum Maximum Hot Leg Switchover Time N/A Maximum Secure CS Time Maximum Minimum Boil-off Flow Rate N/A Maximum Boron Concentration Minimum or N/A Affects pH Maximum Buffer Quantity Minimum or N/A Affects pH

________________Maximum__________

pH Minimum or N/A Affects chemical release and chemical solubility Maximum__________

Head Loss Maximum N/A T. Sande T. Sande ~~Rev.

E, July 8, 2015 Pg f1 Page 4 of 14

O ENERCON Excellence--Every project. Every day.

Parameter Bounding Direction Bounding Direction Comments

__________________for Strainer Failures for Core Failures Structural Margin Minimum N/A NPSH Margin Minimum N/A Degasification Maximum N/A Void Fraction Limit Minimum N/A Penetration Minimum Maximum_____________________________

Core Fiber Limit N/A Minimum T. Sande T. Sande ~~Rev.

E, July 8, 2015 Pg f1 Page 5 of 14

OENERCON Excellence--Every project. Every day.

4. Example Calculation with RWST Injection Model To compare the rigorous and simplified uncertainty quantification methodology, a calculation was performed to determine the volume and timing for water injected from a refueling water storage tank (RWST). Each input parameter was specified using a probability distribution defining the mean, minimum, and maximum values 2. The mean values for the output parameters were then calculated by a) sampling the input parameter distributions and propagating the uncertainties, and b) a simple hand calculation using only the mean input parameter values. The minimum and maximum values for the output parameters were calculated in a similar manner with the two methods described above.

The input parameter values are shown in Table 2. For the first approach, these values were fit using beta distributions, and each distribution was sampled 1,000 times to calculate the mean output values. For this calculation, the output parameters are defined as a) the volume of water injected at RHR switchover, b) the total volume of water injected from the RWST, c) the time to RHR switchover, and d) the time to CS switchover.

Table 2: Simple RWST ireto model inpu prmtrvalues Parameter Minimum Mean Maximum initial RWST Volume (gal) 210,000 -

250,000 310,000 RHR Switchover Volume (gal) 55,000 60,000 65,000 CS Switchover Volume (gal) 12,000 -

21,000 25,000

-RHR Pump Flow Rate (gpm) 1,000 5,000 7,000

-CS Pump Flow Rate (gpm) 3,000 4,000 7,500 The simple hand calculation equations for calculating the mean injected volume and switchover time output values are shown below along with the mean input parameter values:

VRHR Injected = Vinitiai -VRHR Switchover = 250,O00ga/ - 60,O00ga/ = 190,O00gal VTotal Injected = Vinitial -

Vcs switchover = 250,O00gal - 21,O00gaI = 229,000gal VRHR Injected 190,O00gal 1.m tRHR Switchover--RR S

5,0pn+4,0pm 2.ri VTota1 Injected -- VRHR Injected CSSwitchover = tRHR Swttchover +-Qc 229,000gai - 190,O00ga/

= 21.1mai +

= 30.9ramn 4,000gpm Figure 1 through Figure 5 show the probability distributions with the sampled water volume or flow rate values for each input parameter. The mean value of the input parameter is also shown with a vertical line on each figure.

2 The inputs used for the example calculations are hypothetical. For a real evaluation, these values would be determined based on realistic conditions and constraints. For example, the minimum and maximum values for the initial RWST level might be defined based on the low level and high level alarm setpoints, and the mean value might be defined based on operating history.

T. Sande Rev. E, July 8, 2015 Page 6 of 14

O ENERCON Excellence--Every project. Every day.

Inital RWST Water Volume 1.0E-O5 5.0SE-OS O.OEO06 0

II 0

50,O0I0 100,O0c 150,00 200l,0 250.000 300,000 Water Volume (gal)

Figure 1: Sampled values for initial RWST water volume 350,000 RHR Pump Switch Over Volume 2,5E-04 2 OE-04

>. 1 5E-04 a

1OE-04 5,5,000

$6,O00

.7,000 S8,00 59,000 60,W00 61,WO0 62,00 63,00 64,0W0 6SIX00 Water Volume (gal)

Figure 2: Sampled values for RWST water volume at RHR pump switchover T. Sande T.

Sande ~~Rev.

E, July 8, 2015 Pg f1 Page 7 of 14

O ENERCON Excellence--Every project. Every day CS Pump Switchover Volume 20E-04 L8E-04 L.6E-04 1 4E-04 1*.2E-04 S1.0E-04 BOE-05 4,0E-05 4,0E-0S 00E*O0 1S

.3 0

5,000cA 1O, 00 15,(3XJ 20X000 2S%000 30,00 Water Volume (g~al)

Figure 3: Sampled values for RWST water volume at CS pump switchover RHR Pump Flow Rate 0*

4.0E-04 3 5E-04 3,0E -04 2.5E04 20 -04 1 SE 04 1 0E-04 5.0E-05 O.OE.+00 S,

1,000 2,0001 3,000l 4,000 5,000l 6,0cX0 7

Flow Rate (gpm)

Figure 4: Sampled values for RHR pump flow rate 0

,000 8,0(f T. Sande T.

Sande ~~Rev.

E, July 8, 2015 Pg f1 Page 8 of 14

O ENERCON Excellence--Every project. Every day CS Pump Flow Rate 7.0(-04 60E -04 rlk 5.0E-04 2*' 3.0 -0,4 0-.

2,0-04 O.Ot 400 0

1,000 2,000 3,0(X) 4,000 5,000 6,000 7,0(X) 8,000 Flow Rate (8pm)

Figure 5: Sampled values for CS pump flow rate The mean output values based on the two methods are shown in Table 3.

Table 3: Mean output parameter values from the simple RWST injection model Mean Output Mean Output Calculated Using Calculated Parameter Input Parameter Using Mean Difference Probability Input Parameter Distributions Values Volume of water injected at RHR 189,684 190,000 0.17%

switchover (gal)_________

Total volume of water injected (gal) 228,695 229,000 0.13%

Time to RHR switchover (minutes) 21.5 21.1

-1.86%

Time to CS switchover (minutes) 31.5 30.9

-1.90%

Although these two methods don't produce exactly the same result, the simplified method provides a reasonable approximation of the mean output parameter values.

Similarly, by plugging in bounding values for the various input parameters, the absolute minimum and maximum output values can be compared to the sampled minimum and maximum values.

These results are shown in Table 4 and Table 5. Note that the absolute minimum and maximum values are always bounding compared to the values calculated using the probability distributions due to the fact that a large number of samples or a stratified sampling scheme would be necessary to fully capture the low probability tails.

T. Sande T.

Sande ~~Rev.

E, July 8, 2015 Pg f1 Page 9 of 14

O ENERCON Excellence--Every project. Every day.

Table 4: Maximum output parameter values from the simple RWST injection model Maximum Output Maximum Output Calculated Using Calculated Using Difference Parameter Input Parameter Bounding Input Probability Parameter Values Distributions______

Volume of water injected at RHR 234,941 255,000 8.5%

switchover (gal)_________

Total volume of water injected (gal) 274,103 298,000 8.7%

Time to RHR switchover (minutes) 26.1 63.8 144%

Time to CS switchover (minutes) 36.4 78.9 117%

Table 5: Minimum output parameter values from the simple RWST injection model Minimum Output Minimum Output Calculated Using Calculated Using Difference Parameter Input Parameter Bounding Input Probability Parameter Values Distributions Volume of water injected at RHR 160,739 145,000

-9.8%

switchover (gal)

Total volume of water injected (gal) 198,626 185,000

-6.9%

Time to RHR switchover (minutes) 17.9 16.1

-10.0%

Time to CS switchover (minutes) 27.1 24.7

-8.9%

5. Example Calculation with Detailed GS1-191 Model A second, more detailed example calculation was performed using a GSI-1 91 model incorporating plant-specific inputs and phenomenological models (including debris generation, debris transport, chemical effects, strainer head loss, degasification, strainer structural margin, NPSH margin, pump gas void limits, fiber penetration, core fiber accumulation, and core fiber limits). In this example caloulation, 13 of the input parameters were defined using probability distributions. (All other inputs were defined using fixed values or time-dependent profiles.) The parameters used to define the distributions are shown in Table 6.

T. Sande T. Sande ~~Rev.

E, July 8, 2015Pae1of4 Page 10 of 14

O ENERCON Excellence--Every project. Every day.

Table 6: Detailed GS1-191 model input parameter values Parameter Minimum Mean Maximum Initial RWST Mass (Ibm) 5,711,564 5,898,959 6,025,231 RWST Mass at RHR Switchover (Ibm) 751,547 1,650,000 2,049,981 RWST Mass at CS Switchover (Ibm) 167,685 700,000 840,121 RWST Boron Concentration (ppm) 2,100 2,537 2,900 Break Size-Dependent RHR Pump Flow Rate8010%1%

Variability CS Pump Flow Rate (gpm) 2,597 2,700 2,900 Containment Spray Termination Time (minutes) 150 180 240 Break Size and Time-Dependent Pool

-5 0

+15 Temperature Variability (0F)______

Latent Fiber Quantity (Ibm) 0 9

65 Strainer Structural Margin (ft) 22 24.7 30 Pump Gas Void Fraction Limit 1.5%

2%

4%

Fiber Penetration Fraction 2%

5%

8%

Core Fiber Accumulation Limit for Cold Leg 6

7.5 15 Breaks_(g/FA)_____________

Each of the probability distributions were independently sampled (using simple Monte Carlo sampling), and the sampled inputs were used to evaluate the range of potential breaks to determine which breaks would pass or fail the long term core cooling acceptance criteria. A total of 235 iterations of sampled input parameters were run in the simulation. For each iteration, approximately 28,000 breaks (including a range of 1/2A inch partial breaks to double-ended guillotine breaks at each weld location) were evaluated.

The simulated failures were used to calculate conditional failure probabilities for small, medium, and large breaks, and these conditional failure probabilities were subsequently used to estimate ACDF. Also, similar to the simple RWST injection calculation, the mean ACDF output was estimated using the mean input parameter values, and the minimum and maximum ACDF was calculated using the best case and worst case bounding values. Figure 6 shows a histogram of the ACDF values calculated from the Monte Carlo simulation, and Table 7 shows a comparison between the mean, minimum, and maximum values determined using the Monte Carlo simulation and the simplified approach.

T. Sande T. Sande ~~Rev.

E, July 8, 2015Pae1of4 Page 11 of 14

O ENERCON Excellence--Every project. Every day.

A~CDF Histogram (10 bins, 235 samples) 60 so t~40 0.,

E 30 E

Z 20 10 0

m m-m m

m mm 1.64E-08 1.97E-08 2.30£-08 2.62E-08 2.95E-08 3.28E-08 More 3.28-09 6.56E-09 9.84E-09 1*31E*08 ACDF (yr-')

Figure 6: ACDF histogram based on simulation with sampled input parameters Table 7: Output parameter values from the detailed GS1-191 model Calculated Output Calculated Output Parameter Using Sampled Using Fixed Mean Difference Input Probability and Bounding Distributions Input Values Maximum ACDF 3.3E-08 3.6E-07 991%

Mean ACDF 1.6E-08 1.4E-08

-13%

Minimum ACDF 0E-00 0E-00 0%

As shown in this more detailed example, using the mean values for the input parameters provides a reasonable approximation of the mean ACDF. In this particular example, the bounding maximum ACDF value is an order of magnitude higher than the mean value. However, since the bounding maximum represents an extremely low probability scenario (i.e., it is based on worst case bounding input values using the simplified approach), and is still within Region Ill of RG 1.174 (very low risk), there is high confidence that the mean risk is very low as defined by Region Ill of RG 1.174 [3]. This is illustrated in Figure 7. The three points show the best-case, mean, and worst-case ACDF values calculated using the simplified approach, and the dashed line illustrates the ACDF probability distribution that would be calculated using the rigorous approach. A plot of ALERF would look very similar to the ACDF plot.

T. Sande T. Sande ~~Rev.

E, July 8, 2015Pae1of4 Page 12 of 14

O ENERCON Excellence--Every project. Every day.

L CO-)

S0-IeIO 10egion II o fSt-caset c.f~

C'DF Figure 7: Illustration of uncertainty quantification results using the bounding input values from the simplified approach overlaid on RG 1.174 risk figure

6. Conclusions Parametric uncertainty is the primary type of uncertainty that must be quantified for a simplified GS1-191 evaluation.

Using a combination of mean and bounding input values, the mean ACDF/ALERF as well as the ACDF/ALERF uncertainty range can be estimated. This is a simplified approach for uncertainty quantification that reduces the overall effort required to evaluate uncertainties.

Although the simplified approach sacrifices some accuracy in the calculated mean ACDF and ALERF values, the use of consensus models (which are generally conservative), and a mixture of bounding input parameters, skews the results in a conservative direction (i.e., higher ACDF and ALERF).

This document primarily focuses on the calculation of ACDF and the associated uncertainty.

However, it is important to note that CDE, LERF, and ALERF are also important parameters in determining whether the overall GS1-191 risk meets the RG 1.174 acceptance guidelines.

Using a simplified approach is beneficial for both the NRC and industry to work toward a more rapid and resource-efficient resolution of GS1-191.

T. Sande T. Sande ~~Rev.

E, July 8, 2015Pae1of4 Page 13 of 14

O ENERCON Excellence--Every project. Every day.

7. References

[1]

NEI Presentation to the NRC (ML14204A233), GS1-191 (CL 2004-02) Issue Resolution Strategy, A Simplified Risk-Informed Approach, July 22, 2014.

[2] STP Presentation to the NRC (ML15034A114), STP Risk-Informed Approach to GS1-191, February 4, 2015.

[3] Regulatory Guide 1.174, An Approach for Using Probabilistic Risk Assessment in Risk-Informed Decisions on Plant-Specific Changes to the Licensing Basis, Revision 2: May 2011.

[4] NUREG-1855, Guidance on the Treatment of Uncertainties Associated with PRAs in Risk-Informed Decisionmaking, Revision 1 DRAFT: March 2013.

[5] EPRI Report 1016737, Treatment of Parameter and Model Uncertainty for Probabilistic Risk Assessment, Final Report: December 2008.

[6] EPRI Report 1026511, Practical Guidance on the Use of Probabilistic Risk Assessment in Risk-Informed Applications with a Focus on the Treatment of Uncertainty, Technical Update: December 2012.

[7]

NEI 04-07 Volume 1, Pressurized Water Reactor Sump Performance Evaluation Methodology, Revision 0: December 2004.

[8]

NEI 04-07 Volume 2, Safety Evaluation by the Office of Nucelar Reactor Regulation Related to NRC Generic Letter 2004-02, Revision 0: December 2004.

[9] WCAP-16530-NP-A, Evaluation of Post-Accident Chemical Effects in Containment Sump Fluids to Support GS1-191, March 2008.

[10] Draft Regulatory Guide DG-1322, Risk-Informed Approach for Addressing the Effects of Debris on Post-Accident Long-Term Core Cooling, DRAFT: April 2015.

T. Sande T. Sande ~~Rev.

E, July 8, 2015Pae1of4 Page 14 of 14

Charles R. Pierce Regulatory Affairs Director Southern Nuclear Operating Company, Inc.

40 Inverness Center Parkway Post Office Box 1295 Birmingham, AL 35242 Tel 205.992.7872 Fax 205.992.7601 NUCLEAR A SOUTHERN COMPANY

.October 15, 2015 Docket Nos.: 50-424 50-425 NL-1 5-1904 U. S. Nuclear Regulatory Commission ATTN: Document Control Desk Washington, D. C. 20555-0001 Vogtle Electric Generating Plant, Units 1 & 2 Supporting Documents for Upcoming SNC NRC Public Meeting Regqardingq the Resolution of GS1-191 Ladies and Gentlemen:

A Nuclear Regulatory Commission (NRC) public meeting is being scheduled for later this year to discuss Vogtle Electric Generating Plant's planned resolution to GSI-1 91. In support of this upcoming public meeting, Southern Nuclear Operating Company (SNC) is submitting the attached reports to the NRC. While SNC is not requesting formal NRC review and approval of these reports, these reports help form a significant portion of the planned technical discussion for the meeting. To facilitate meaningful discussion, it is suggested that the NRC staff that plan on attending the meeting familiarize themselves with the content of these reports prior to the meeting. Because of the size of Enclosure 2, only the relevant portions to the upcoming meeting are provided. A full version of can be provided upon request.

This letter contains no NRC commitments. If you have any questions, please contact Ken McElroy at (205) 992-7369.

Respectul smitted, C. R. Pierce Regulatory Affairs Director CRP/RMJ

Enclosures:

1. Risk-Informed GS1-1 91 Uncertainty Quantification

2.

Head Loss Testing of a Prototypical Vogtie 1 and 2 Strainer Assembly

(

U. S. Nuclear Regulatory Commission NL-15-1904 Page 2 cc:

Southern Nuclear Operatin~q Company (w/o Enclosures)

Mr. S. E. Kuczynski, Chairman, President & CEO Mr. D. G. Bost, Executive Vice President & Chief Nuclear Officer Mr. D. R. Madison, Vice President - Fleet Operations Mr. M. D. Meier, Vice President - Regulatory Affairs Mr. B. K. Taber, Vice President - Vogtle 1 & 2 Mr. B. J. Adams, Vice President -- Engineering Mr. G.W. Gunn, Regulatory Affairs Manager - Vogtle 1 & 2 RType: CVC7000 U. S. Nuclear Regiulatory Commission Mr. L. D. Wert, Regional Administrator (Acting)

Mr. R. E. Martin, NRR Senior Project Manager - Vogtle 1 & 2 Mr. S. S. Koenick, NRR Senior Project Manager - Vogtle GS1-1 91 Mr. V. Cusumano, Chief, Safety Issue Resolution Branch Mr. L. M. Cain, Senior Resident Inspector - Vogtle 1 & 2

Vogtle Electric Generating Plant, Units 1 & 2 Supporting Documents for Planned November 5, 2015 SNC NRC Public Meeting Regarding the Resolution of GS1-191 Risk-Informed GS1-191 Uncertainty Quantification

OENERCON Excellence--Every project. Every day.

Risk-Informed GS1-191 Uncertainty Quantification This document discusses two methods that can be used to quantify the uncertainty associated with a risk-informed GS1-191 evaluation. The first method is a detailed statistical approach for sampling input parameter probability distributions and propagating the uncertainties. The second method is a simplified approach for selecting bounding input parameter values and calculating the uncertainty range using sensitivity analysis.

Over the past year, the industry has been moving in the direction of using simplified methods for risk-informed GS1-191 evaluations [1, 2]. Implementing simplified methods is beneficial since it allows the ECCS strainer performance issue to be resolved more efficiently and reduces the time and effort required for NRC technical review. Therefore, this document focuses primarily on the simplified approach for uncertainty quantification.

1. Introduction Uncertainty quantification is a key requirement in Regulatory Guide (RG) 1.174 for a risk-informed evaluation [3]. As defined in RG 1.174 and explained in more detail in NUREG-1 855 [4] and two corresponding EPRI reports [5, 6], there are three types of epistemic uncertainty that should be addressed:

1. Parametric uncertainty

2. Model unciertainty

3. Completeness uncertainty Parametric uncertainty refers to the variability in input parameters that are used in the risk assessment. Due to the wide range of plant-specific post-LOCA conditions related to GS1-191 phenomena, this is a very important aspect for understanding the overall uncertainty.

Model uncertainty refers to the potential variability in an analytical model when there is no consensus approach. A consensus approach is a model that has been widely adopted or accepted by the NRC for the application for which it is being used [4]. For example, the use of a spherical zone of influence (ZOI) to model the debris quantity generated by a high energy break is a consensus model that has been widely adopted and accepted by the NRC [7, 8]. In general, plants implementing a simplified risk-informed approach are using standard models that have been widely accepted for deterministic evaluations (e.g., accepted insulation and qualified coatings ZOI sizes, the use of WCAP-16530 [9] to model chemical effects, and prototypical strainer module testing for head loss and penetration). By using these consensus approaches, the effort to address model uncertainty is minimized.

Completeness uncertainty refers to 1) the uncertainty associated with scenarios or phenomena that are excluded from the risk evaluation, and 2) the uncertainty associated with unknown phenomena. Although it is not practical to quantify the uncertainty associated with factors that are not explicitly modeled (e.g., secondary side breaks, or breaks downstream of the first isolation valve), their potential impact can be qualitatively assessed. Uncertainties associated with unknown phenomena, on the other hand, cannot even be qualitatively assessed. Uncertainties T. Sande T.

Sande ~~Rev.

E, July 8, 2015 Pg f1 Page 1 of 14

O ENERCON

Excellence--Every project. Every day.