ML20076F885

From kanterella
Jump to navigation Jump to search
Probabilistic Risk Assessment Evaluation of Tornado- Generated Missile Impact on Ccnpp EDG Engine Air Intake & Exhaust
ML20076F885
Person / Time
Site: Calvert Cliffs  Constellation icon.png
Issue date: 10/13/1994
From:
BALTIMORE GAS & ELECTRIC CO.
To:
Shared Package
ML20076F882 List:
References
NUDOCS 9410180228
Download: ML20076F885 (20)
v • d • e


Text

.-. - . . --

ATTACIIMENT (1) l J

J PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON TIIE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST i i

9410180229 941013 PDR ADOCK 05000317 P PDR

TisDLE OF CONTENTS

1.0 INTRODUCTION

.. .. . .. . . . . .. . ..I 1.1 Purpose.. . . . . . . . .. . . .1 1.2 Background.. . . .. .1 1.3 Limitations.. . . . . .1 1.4 Plant Model.. .

.1 2.0 ANALYSIS. . . . .2 2.1 Overview . . . . . . . .2 2.2 Tornado Strike Frequency . . . . . .2 2.3 Missile Strike Probability. . . . . . . .7 2.4 Component Damage Probability . . . . .13 2.5 Conditional Core Damage Probability. . .13 3.0 RESULTS. .14 3.1 Sensitive Assumptions. . . . .15 3.2 Conservative Assumptions.. . . .16

4.0 CONCLUSION

S . .

. .17

5.0 REFERENCES

.. . .18 I

)

4

ATTACIIMENT (1)

PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON TIIE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST j i

1.0 INTRODUCTION

1.1 PURPOSE To determine the dsk significance of tornado-induced missiles causing a failure of the Calvert Cliffs Nuclear Power Plant (CCNPP) Emergency Diesel Generators (EDGs).

1.2 BACKGROUND

The current design of CCNPP Units 1 and 2 does not provide for tornado missile protection of the EDG exhaust subsystem and air intakes. His risk assessment evaluates the Core Damage Frequency (CDF) due to a tornado damaging these key components coincident with a non-recoverable loss of offsite power (LOOP). As discussed in the cover letter, the acceptance criteria is based upon the probability of exposures in excess of 10 CFR Part 100 guidelines, rather than CDF; however, this evaluation conservatively assumes that CDF results in containment failure, and therefore, equates to exposures in excess of 10 CFR Part 100 guidelines. This assessment will aid in the evaluation of the adequacy of the current plant design with respect to tornado events.

Section 2 of this analysis presents the building blocks required to determine CDF, which is calculated in Section 3. Conclusions are presented in Section 4.

1.3 LIMITATIONS This analysis considers the effects of tornado missiles on the EDGs only. It does not determine the risk due to wind effects (tornadoes,1 urricanes, etc.) or the impact of tornadoes on other plant equipment.

1.4 PLANT MODEL The Probabilistic Risk Assessment (PRA) plant model used for this analysis is based on the model as of May 18,1994. This plant model is a refinement of the one documented in the CCNPP Individual Plant Examination Summary Report (Reference 1). The plant model used is based on Unit 1, but has been validated for Unit 2.

l 1

1 i i

1

ATTACIIMENT (1)

PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON TIIE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST 2.0 ANALYSIS 2.1 OVERVIEW In order to determine the risk significance of the postulated events, five key pieces ofinformation must be developed:

Tornado Strike Freauency (P,'): This is the frequency at which a tomado (as defined in Reference 2, Section 4.1) is expected to come in contact with the " site." Point strike and area strike frequencies are calculated in Section 2.2.

Tornado Missile Impact Parameter (y): This is the frequency ofimp tct of a missile on a structure or component, expressed as the frequency of impact per missile pcr unit target area per tornado (Reference 3, pg. G-35). This value is calculated in Section 2.3.' .

Missile Strike Probability (P,3): This is the probability of a tornado generated missile impacting a target, given a tornado strikes the site. P ms is based on the missile impact parameter (y), the target area and the number of missiles available. The value is calculated in Section 2.3.4.

Component Failure Probability (Pp): Given a tornado, a component can fail as a result of being struck by a tornado-generated missi!c. The probability of this event occurring and causing a failure of the EDG is discussed in Section 2.4 Conditional Core Damage Probability (PCCD): Given a tornado and the failure of the EDG, the likelihood that core damage occurs is calculated. Section 2.5 summarizes the results of this analysis.

'Ihe following equation is used to determine the contribution to CDF due to tornado-induced failures of components. The strike frequency (point or area) which results in the higher CDF (which is conservative) is used.

For tornado missiles, CDFm = P,'

  • Pms
  • Py
  • PCCD 2.2 TORNADO STRIKE FREQUENCY The tornado strike frequency is the annual frequency of a tornado striking the site. The estimates for tornado strike frequency are based on data from:

2 l

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

ATTACIIMENT (1)

FROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON Tile CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST NUREG-4461 (Reference 4), which includes tornado histories from 1954 through 1983 for the 1* box containing Calvert Cliffs. A 1* box is an area enclosed by a rectangle with sides equal to 1 latitude and 1* longitade. Calvert Cliffs is at approximately 38 26' North and 76*27' West (Reference 5, Figure 2.2-5)

National Severe Storms Forecast Center (NSSFC), which includes tornado histories from 1950 through 1984 within 50 nautical miles (NM) of Prince Frederick, Maryland.

Strike frequency is calculated by determining the likelaaod that a tornado coincides with a target (in this analysis, the site). Strike frequency calculated using an Area Strike Model is dependent on the area of the target. A Point Strike Model is independent of the target area and can be used for relatively small targets (with respect to the area of the region in which the data is collected).

For the generation of tornado missiles, strike frequencies were calculated using both the Point and Area Strike Models. 'Ihis is required since the Tomado Missile impact Parameter (y) varies depending on which model is used for the strike frequency (see Section 2.3.1). The area strike calculations were developed with the assumption that a tomado will strike the site from a random direction and at a random orientation to the site. Although tornadoes may preferentially come fram certain directions at certain orientations, this infonnation is not available for the CCNPP vicinity. Non-random tornadoes (with respect to direction and orientation) do not afTect the results of the area strike frequencies, but can affect the missile-impact parameters (v - see Section 3.3.3) 2.2.1 Point Strike The Point Strike Frequency (P,) is calculated using the methodology described in NUREG-4461; data from -

both NUREG-4461 and NSSFC is used. A point strike frequency is estimated in this analysis since some of the y values used are normalized to point strike frequency. The equations for P, are:

P, = A /(A,

  • Ny) [ Reference 4, Eqn. (11)]

A = (Number of Tornadoes) * (Average Tornado Area) [ Reference 4, pg.16]

A r= Area ofinterest = 4774.3cos(Latitude at middle of box) [ Reference 4, Eqn. (14)]

OR i

l l

l 3

)

I

ATTACIIMENT Q)

PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON TIIE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXHAUST A r= x (radius of area under consideration)2 = xp2 Ny= Number of Years in the period of record [ Reference 4, pg.16]

Values are taken from Reference 4, pages E.1 and E.5 (1 box at 76 Longitude and 38* Latitude) and site-specific data attained from the NSSFC.

1* box 50 NM radius No. of tornadoes 15 51 Avg. tornado area 0.105 0.137 Ny 30 35 NUREG-4461 (1* box) Data A t= (Number of Tornadoes) * (Average Tornado Area)

= 15

  • 0.105 = 1.575 A r= 4774.3cos(Latitude at middle of box)

= 4774.3cos(38.5) = 3736.4 mi2 Ny = 30 Therefore, P

P, = 1.575/(3736.4

  • 30) = 1.41x10-5/yr l

P, must be corrected to account for large bodies of water in the region, since only tornadoes which occur .

over land are captured in the databases. To determine the percentage ofland in the 1* box, a trace of a map was made. A grid was made on the map and, based on grid squares, the estimated percentage ofland ,

is 62%

i P,'is the corrected Point Strike Frequency, accounting for the percentage ofland versus water.

P,' = P, * (Total Arca of the region / Land Area) (Reference 4, pg.17)  !

= 1.41x10-5 *(1.0/0.62) = 2.27x10-5/yr l l

J 4

1

)

i ATTACilMENT (1) l 1

PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON TIIE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST NSSFC (50 NM radius) Data A t = 51

  • 0.137 = 6.987 2

A, = rp (including NM to statute mile conversion of 1.151) = 10405 mi2 Ny = 35 Therefore, P, = 6.987/(10405

  • 35) = 1.92x10-5/yr To determine the percentage ofland within 50 NM of Prince Frederick, a trace of a map was made. A grid was made on the map and, based on grid squares, the estimated percentage ofland is 75%. Therefore, P,' = 1.92x10-5 *(1.0/0.75) = 2.56x10-5/yr Results Since the Point Strike Frequencies calculated using both sets of data are very similar, the conservative value (2.56x10-5/yr from the NSSFC data) will be used in this analysis.

2.2.2 Area Strike A method for determining Area Strike Frequency was developed. The frequency for a tornado striking the site, characterized by a given radius, is calculated by multiplying the probability of a tornado striking the site by the frequency of tornadoes in the sample area. Given the average tornado strike path and site dimensions, the probability of a tornado striking the site is determined using trigonometric and probabilistic relationships.

The data required to calculate the Area Strike Frequency is developed below.

Short Leneth (Tornado Width Ix1) and Lone teneth (Tornado Leneth fyl) i The NSSFC data gives a true average length and width. No length or v,idth data is provided for 1 boxes in NUREG-4461. In the NSSFC data, the product of length times width (0.101 sq. mi.) is less than the true average area provided (0.137 sq. mi.). In order to be conservative (since the larger the tomado, the higher the area strike frequency), new lengths and widths are calculated.

l 5

L ATTACIIMENT (1)

PROBABILISTIC RISK ASSESSMENT EVALUATION l OF TORNADO-GENERATED MISSILE lhiPACT ON TIIE CALVERT CLIFFS j NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST It is assumed that the average tornado has a length to width ratio equal to 37.5 (from NSSFC data, true average length = 1.95 mi. and true average width = 0.052 mi.). By soh'ing two equations, x and y can be determined:

y/x = 37.5 and x+y = true average area.

Therefore, using the average tornado areas from Section 2.2.1 as the true average area:

l' box: x = 0.053 mi. = 279.84 8. y = 1.99 mi. = 10507.2 A.  ;

50 NM eirele: x = 0.060 mi. = 316.8 n. y = 2.25 mi. = 11880.0 A.

Site Radius fr)  ;

Two values of the site radius (2000 feet and 1 mile) are chosen to agree with the y values described in Section 2.3.1.

Area of Sample i

. %c land area within the sample area is calculated from the data in Section 2. 2.1. For the 1 box, the area is equal to:

3736.4*0.62 = 2316.6 square miles For the 50 NM radius eirele, the area is:

10405'0.75 = 7803.75 square miles Freauency of Tornado Strike in Samnle Area nis value is equal to the number of tornadoes divided by the number of years in the sample. From the  !

data in Section 2.2.1:

1* box: 15/30 = 0.5 tornadoes / year 50 NM eirele: 51/35 = 1.46 tornadoes / year .

[

6

ATTACIIMENT (1)

PROBAUILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON Tile CALVERT CLIFT', ,

NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND ExlIAUST Results Based on the above values, four area strike frequencies are calculated, one for each radius (2000 feet and I mile) using both sets of data (l* box and 50 NM radius). The results are presented below (in tornado strikes per year).

Strike Area TornaAp_ Data 2000 ft 1 Mile 1* box 4.38x104 1.54x 10-3 50 NM radius 4.20x104 1.43x10-3 The highest value considering a 2000 foot radius strike area (i.e.,4.38x104 /yr) will be used in calculations which use area strike frequency. The one-mile radius strike frequencies are not used, except for sensitivity calculations. See Sections 2.3.1 and 3.3.1 for discussions of the strike areas.

2.3 MISSILE STRIKE PROBABILITY There are several variables which must be determined to assess the Missile Strike Probability (Pms). They are:

y, the missile impact parameter. It is defined as the probability of impact / missile / unit target arca/ tornado strike frequency (Reference 3, pg. G-35)

A, the area of the target (s) in question (in f12)

N m, the number of candidate missiles The equation for determining the probability of a missile impacting a target, given a tornado (i.e.,

probability per toedo) is derived from unit analysis:

P,=ANmY m

2.3.1 Missile Impact Parameter (y)

Ideally, the Missile Strike Probability (Pms) should be calculated based on plant specific configuration using a computer simulation (e g , the TORMIS methodology, as used in the Seabrook study

[ Reference 12, pg.1-1]). However, several PRAs have derived y values based on models developed for other sites, due to the complexity of the analyses and the somewhat generic applicability of the data (i.e.,

based on the tornado experience in NRC Tornado Region I [per Reference 8, Figure 1, CCNPP is in 7

i l

ATTACilMENT fil l PROBAHILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON Tile CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXilAUST Tornado Region I]). Due to the unavailability of a plant specific reference, conservative values are chosen from these derived y values.

The PRAs researched are described in Section 2.3.1.1. The y-values chosen for the analysis of CCNPP are provided in Section 2.3.1.2.

2.3.1.1 y-Values Used in Other PRAs Oconee PRA (NSAC-60)

The Oconce study (Reference 9, pg. K-10) considers two values for the missile strike parameter, based on an area strike frequency. One value is derived from EPRI NP-0768 and 0769 (References 6 and 7) which is based on assuming all missiles within 2000 feet of the plant. The second is from a SAIC study and is based on assuming missiles from within one mile of the plant. The value derived from the EPRI study is higher than the one from SAIC, since it is more likely that a missile generated close to a target will strike it.

Ilowever, fewer missiles are included in the missile population of the EPRI study.

The mean displacement range of missiles considered in the EPRI study was less than 350 feet. In 250 simulations of each type of missile, no missiles traveled more than 2000 feet (Reference 7, pg. 2-22). The comparison of missile strike probabilities between the EPRI and SAIC studies in the Oconce PRA shows that missile strikes are dominated by close-in missiles. Ilowever, the SAIC study did find that some missiles were transported from as far away as 4000 feet (Reference 9, pg. K-11).

The values for y are:

EPRI Study (2000 ft. radius) 3.3x10-11/ missile /f12 of target area / tornado strike frequency SAIC Study (1 mile radius) 4.52x10-12/ missile /fl2 of target arca/ tornado strike frequency St. Lucie and Turkey Point Shutdown Decay IIcat Removal Studies (NUREG-4710 & 4762)

In NUREGs-4710 and -4762 (Reference 3, Appendix G, Section 4.1 and Reference 10, Appendix G, l Section 4.1), several mean values of y are derived from EPRI NP-0769 (Reference 7) and a TORMIS study of the Seabrook site (Reference 12). The values derived in NUREGs-4710 and -4762 are normalized to point strike frequency and will be designated yn.

One value is given for large structures (i.e., buildings, tanks, etc.) in NRC Reg!cns I and 11; the large structure Yn is derived from EPRI NP-0769 (see Reference 3, pg. G-35 and G-36). Since this y is normalized to point strike frequency, it is higher than the Oconee PRA value (3.3x10-II), because point strike probability is less than area strike probability.

8

1 ATTACIIMENT (1)

PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON TIIE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST Hree values of Vn are derived for small targets (approximately 100 to 1000 ft2). They are from the Seabrook study and are subdivided into high, medium and low exposure areas. These subcategories are based on judgment and consider where the targets are located relative to the concentration of missiles and the shielding effects of other buildings. Note that the site area considered for the Seabrook analysis missile origination is estimated to be 4.8 million ft2, which is equivalent to a circle with a radius of about 1200 feet. He mean values from NUREG-4710, Table 4-lb are:

"Iligh exposure" 2.42x10-9/ missile /ft2 of target area / tornado point strike frequency

" Medium exposure" 8.64x10-II/ missile /ft2 of target arca/ tornado point strike frequency

" Low exposure" 1.54x10-II/ missile /ft2of target area / tornado point strike frequency NUREG-4710, pg. G-37 uses weights of 0.1 for high,0.4 for medium and 0.5 for low exposures. This results in a Yn for small targets of:

Yn = 0.1*(2.42c-9) + 0.4*(8.64e-1l) + 0.5*(l.54e-1l) = 2.85x10-10 NUREG-4762, pg. G-40 uses weights of 0.1 for high,0.8 for medium and 0.1 for low exposures. This results in a Yn for small targets of:

Yn = 0.1*(2.42e-9) + 0.8*(8.64c-1l) + 0.1*(l.54e-1I) = 3.13x10-10 2.3.1.2 y-Values Used for CCNPP Analysis Area Strike Model For area strike frequency calculations at CCNPP,2000 feet is chosen based on the findings of the EPRI study, discussed above. Additionally, a significant portion of the area between 2000 feet and 1 mile from the center of the site is water (the Chesapeake Bay) and does not contain any appreciable missiles. See Section 3.3.1 for more discussion of the assumption to use 2000 feet.

Therefore, for area strike models, the EPRI value derived in the Oconee PRA (3.3x10-II/ missile /ft2 or target arca/ tornado strike frequency) will be used.

Point Strike Model For point strike models, the mean values for Yn (for high, medium and low exposures) derived in NUREG-4710 and 4762 will be used. The weighting factors associated with high, medium and low exposures are chosen based on examination of the layout of the EDG rooftop components.

9

ATTACliMENT W PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON Tile CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST For the EDGs, the weighting of exposures chosen in NUREG-4710 (0.1,0.4 and 0.5) seems reasonable given the protection provided by the buildings, structures and the parapet wall. Most of the EDG components on the roof are shielded by the parapet wall, the Auxiliary Building, Containment and the Refueling Water Tank. Therefore, the Yn for the EDGs is 2.85x10-10, j i

2.3.2 Target Area (A)

The target area is the exposed area of the components of concern. Target areas are calculated for one EDG (all three EDGs have the same target area). The calculations are simple area calculations for an open ended cylinder, given a certain length and diameter. The area of the intake filter is a cylinder with one end l closed.

Two areas are shown, one for the total area and one for 75% of the area. 75% of the area is considered a reasonable estimate of the area actually exposed to a tornado missile. Since the exhaust piping and silencer generally run along the roof or a wall,25% of the area is assumed to be shielded from a missile strike. The reduction in area due to this shiciding was not accounted for in the intake filter and two portions of the exhaust piping, since they are not shielded as such.

l The results are summarized below. All EDGs have approx.mately the same dimensions. l l

Compgneat o _

Total Area 75% Area 2 l EDG intake filter: 158.4 R2 158.4 R (same as 100% area)

EDG exhaurt silencer: 247.3 ft2 185.5 R2 EDG cxhaust piping: 229.4 R2 188.6 ft2 TOTAL Area (per EDG) 635.0 A2 532.4 R2 2.3.3 Number of Candidate Missiles (Nns)

The number of candidate missiles (those missiles which could cause dauage to the components of interest) is typically determined by a detailed survey and walk down of the area surrounding the site. Although a detailed survey of the site has not been performed, an examination of the area revealed that the number of missiles present at CCNPP seems comparable to the number at other sites described below. The number of missiles chosen for this analysis is calculated in Section 2.3.3.2, based on the data in Section 2.3.3.1.

I 10 1

1

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ . _ _ . . _ _ _ _______J

ATTACIIMENT (O PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON Tile CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST 2.3.3.1 Data from other PRAs The Oconee PRA (Reference 9, Section K3.2) mentions three values:

6000 missiles (within 2000 feet) from the EPRI study (Reference 6, pg. 3-24) 75,000 missiles (withia 1 mile) used in the SAIC study (Reference 9, pg. K-11) 69,000 missiles (within 1 mile) used in the Oconee study (Reference 9, pp. K-8 & K-11)

NUREG-4710, (Reference 3, page G-37) uses a distribution of missiles as described below (no area was given):

Probability Weight Number of Missiles 0.2 5,000 '

O.6 25,000 0.2 60,000 Therefore, the weighted average = (0.2*5000)+(0.6*25000)+(0.2*60000) = 28,000 The preliminary values to be used in the Turkey Point IPEEE are 100,000 missiles within approximately one mile. An earlier study of Turkey Point by Sandia (Reference 10, pp. G-40) used a distribution of missiles (no area given):

Probability Weight Number of Missiles 0.2 10,000 0.6 40,000 0.2 70,000 Therefore, the weighted average = (0.2** 10000)+(0.6*40000)+(0.2*70000) = 40,000 The preliminary values to be used in the Fort Calhoun Station IPEEE are 30,000 missiles, which is based on the mean value used in the St. Lucie study.

The Seabrook study (Reference 12, Table IV-10) uses 66,796 missiles within approximately 1200 feet (area from Section 2.3.1.2).

2.3.3.2 Missile Population for CCNPP Based on data from the studies cited above, a representative value is calculated for CCNPP. It is believed that the EPRI value of 6000 missiles is too low, based on significantly higher values used at other plants and a generrJ survey of the site. Additionally, since Fort Calhoun used NUREG-4710 as a source, the value from Fort Calhoun is not considered, to avoid double counting the data sources. The NUREG-4762 data will not be used since there is a more recent estimate (100,000). Although some of the other studies 11

ATTACIIMENT (1)

PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON TIIE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXHAUST used a one mile radius from the plant site as the area for the missile population, those values will be considered, since they should be conservative. ,

An average of the data from NUREG 4710 and the Oconce, Seabrook and Turkey Point analyses was used ,

to calculate the value for CCNPP.

N m= (75,000 + 69,000 + 28,000 + 66,796 + 100,000)/5 = 67,760 (approximately)

His value may be high due to the inclusion of data from sites that used missile populations within one mile and the fact that approximately 20% of the area within.2000 feet of the center of the site is water.

However, it does not 'seem overly conservative considering the number of trees, laydown areas and.

temporary structures in the vicinity of the plant.

2.3.4 Calculation of Missile Strike Probability (Pms) <

Missile strike probabilities for individual component strikes are calculated based c,n the values for A, Nm and y developed above. Using the equation from Section 2.3, the missile strike probabilities for individual components (one using the point strike and one using the area strike model) are calculated below. Both >

point and area missile strike probabilities will be used in Section 3.2 with the point and area tornado strike frequencies for calculating CDF.

Pms=ANmV Where, Nm= 67,760 (Section 2.3.3.2)

Yn = 2.85x10-10 (Point strike -Section 2.3.1.2) y = 3.3x10-11 (Area strike - Section 2.3.1.2) .i A = 532.4 fl2 (Section 2.3.2.1)

Here fore, Pms(EDG) = 1.03x10-2/tornadopoint strike Pms(EDG) = 1.19x10-3/ tornado area strike

/

12

i ATTACIIMENT 01  :

PROBAlllLISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON Tile CALVERT CLIFFS )

NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST 2.4 COMPONENT DAMAGE PROBABILITY This section discusses the likelihood of component failure, given a tornado missile strike on that component.

The conservative assumption made in this analysis is that if a missile strikes any part of an EDG, it is assumed to fail. More likely, the failure probability is less than 10%. With respect to the wind speeds experienced 'n the area (i.e., no tornadoes greater than F-3 intensity or 206 mph (Reference 4, pp. C.35 and D.85 and NSSFC data) and the damage which must be incurred on the EDG components to actually fail the EDG (e.g., crush the exhaust piping enough to greatly restrict flow), a value of 0.1 is likely conservative. Ilowever, given that this value would be strictly based on engineering judgment and significant engineering judgment is already used in this analysis, a more conservative value of 1.0 is used.

2.5 CONDITIONAL CORE DAMAGE PROBABILITY (CCDP)

The CCDP was estimated based on sequence examination and probabilistic analysis of the most recent plant model results. The failure of one or more EDGs was analyzed, considering a non-recoverable LOOP.

A CCDP of approximately 0.02 is used based on loss of one or more EDGs. However, there is a possibility that the damage to the EDG exhaust or intake components, due to a tornado, is recoverable.

This recovery is not taken into account.

I 13

ATTACIIMENT (1)

PROIIABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON TIIE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST 3.0 RESULTS (CORE DAMAGE FREQUENCY)

The core damage contribution due to the vulnerability of the exposed EDG components during a tornado is evaluated. For core damage to occur, a tornado must strike the site and damage the EDGs. A tornado striking the site is assumed to cause an unrecoverable LOOP. This is similar to the assumption taken in NUREG-4710 (Reference 3, pp G-40 and G-49).

For tornado missiles, CDF will be calculated using point strike and area strike models; the choice of model affects the tornado strike frequency (P,') and the probability of missile strike (Pg3). The risk associated with tornado missiles will be based on the highest CDF from these calculations. Core Damage Frequency due to tornado missile impacts is given by:

CDFTM = P,'

  • Pms
  • Py
  • PccD The likelihood that multiple EDGs are struck by missiles is less than a single missile strike. Also, the  !

CCDP used in this analysis is the same regardless of the number of EDGs struck. Therefore, the CDF calculated here is for a single EDG failure, which is bounding.

]

P,' = Tornado Strike Frequency

= 2.56x10-5/yr (Point Strike - Section 2.2.1)

= 4.38x104 /yr (Area Strike - Section 2.2.2)

P ms = Probability of Missile Strike

= 1.03x10-2 (Point Strike - Section 2.3.4.1)

= 1.19x10-3 (Area Strike - Section 2.3.4.1)

Pp = 1.0 (Section 2.4.1)

Pcco =0.02 (Section 2.5.1)

Therefore, CDFng(Point Strike) = 5.27x10Nyr CDFn1(Area Strike) = 1.04x10Nyr 14

ATTACllMENT 0)

PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON Tile CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST l

3.1 SENSITIVE ASSUMPTIONS 3.1.1 Site Area (5280 or 2000 ft radius)

If the area of the site for tornado missile generation is larger (i.e., by using a one mile radius), the tornado strike frequency will increase by approximately a factor of 3.5 (Section 2.2.2). However, the y will decrease by approximately a factor of 7 (=3.3x10-II/4.52x10-12 from Section 2.3.1.1). It is expected that the number of available missiles will increase, but the value chosen for this analysis is already biased with data from studies using 1-mile radius areas. The assumed missile population will have to double to cause an increase in missile strike frequency. However, none of the studies reviewed, even those using 1-mile radii, used greater than 100,000 missiles. Therefore, it is not expected that CDF will increase enough to change the conclusions of this analysis, if CDF increases at all.

3.1.2 Missile Population If the actual missile population is higher than the value used, there will be a direct impact (i.e., CDF will increase by the ratio of missile populations) on the CDF due to tornado missiles. Ilowever, the value is believed to be conservative as discussed above.

3.1.3 y Value The value chosen for y has a direct impact on CDF. Ilowever, based on a review of the literature and the complexity of performing a site specific analysis, it is believed that the best values were used. The use of the weighting factors for high, medium and low exposures (Section 2.3.1.2) is also believed to be representative based on the orientation and location of components and random tornado orientations and directions. Even if the y value is assumed to be totally based on the high exposure (2.42x10-9), the CDF due to tornado missiles will still be below 104 3.1.4 Tornado Strike Frequency ne tornado experience used to determine strike frequencies is based on fairly recent data, specific to the locale of CCNPP. Using a higher tomado occurrence rate, such as EPRI NP-2005, Region C data (Reference 13), would result in a strike frequency of approximately two times the local strike frequency.

His would increase the CDF (calculated using the point strike model) by a factor of two, but not change the conclusions of this analysis.

3.1.5 Conditional Core Damage Probability CCDP is 0.02 for failure of all EDGs. His was conservatively estimated based on sequence examination and probabilistic analysis of the most recent plant model results.

15 l

1

r; t e ATTACilMENT (1)

PROBABILISTIC RISK ASSESSMENT EVALUATION OF TORNADO-GENERATED MISSILE IMPACT ON TILE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST 3.2 CONSERVATIVE ASSUMPTIONS The following assumptions are considered to be conservative and, based on further analysis, could be made more realistic. They are not listed in any specific order.

1. Assumed that a LOOP is not recoverable following a tomado. There is some probability that offsite power can be recovered before the batteries deplete and/or the EDGs are repaired. This has not been quantified, but the impact of power recovery will be to lower the CCDP and thus, the CDF.
2. Did not account for EDG recovery (e.g., repair, cutting away of damaged piping, etc.). Based on the damage incurred by the exposed EDG components, there may be a simple repair which has not been factored into this analysis.
3. EDG failure probability is conservative (assumed guaranteed failure). As discussed in Section 2.4.1, a value of 0.1 could probably be used and still be conservative.
4. y values are based on NRC Region I tornado experience, which has a much higher incidence of high F-Scale tomadoes. There have been no tornadoes F-4 or greater recorded in our area (including the 5 box). Calculations of y are based on simulations using trials with various I tornado intensities. Since NRC Region I encompasses a large area, including the midwest where there are very strong tomadoes, y values calculated are believed to be higher since stronger tornadoes produce more, larger and farther-traveling missiles.

l I

5. Assumed that core damage will result in exposures in excess of 10 CFR Part 100 guidelines. This assumption is conservative in that it assumes the containment function is guaranteed to fail, which is not necessarily true. <

l l

l i

l 1

16

i ATTACl! MENT (I)

PROBABILISTIC RISK ASSESSMENT EVAL.UATION OF TORNADO-GENERATED MISSILE IMPACT ON Tile CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIR INTAKE AND EXIIAUST

4.0 CONCLUSION

S The risk to the public associated with failures of the EDGs resulting from tornado-induced missiles is very low.

The contribution to core damage to each unit due to a tornado missile strike on the EDGs is nearly 100 times less than 104 /yr, which would screen it from consideration using the criteria in NUREG-1407 (Reference 11). This is classified as having icw risk significance, using the Potential Vulnerability Underlying Cause Review Criteria in Table 3.4.2.1-1 of the Individual Plant Examination Summary Report (Reference 1). Based on the conservatisms in this analysis (Sc.: tion 3.2), this value may be lower.

For the purposes of this evaluation, CDF is assumed to be equal to expected rate of occurrence of potential exposures in excess of 10 CFR Part 100 guidelines; therefore, the calculated value (1.04x10-8/yr) is acceptable using the methodology proposed in the cover letter, and the contribution to CDF and risk to the public due to a tornado-induced failure of one or more EDGs is negligible, i

l I

17

'^

r; ATTACIIMENT (1)

PROBABILISTIC RISK ASSESSMENT EVALUATION l l

OF TORNADO-GENERATED MISSILE IMPACT ON TllE CALVERT CLIFFS NUCLEAR POWER PLANT EMERGENCY DIESEL GENERATOR ENGINE AIH INTAKE AND EXIIAUST 1

5.0 REFERENCES

l

1. RAN 92-008, Individual Plant Examination Summary Report, Rev 0.
2. C.Y. Kimura & R.J. Budnitz, NUREG/CR-5042, Evaluation of External liazards to Nuclear Power Plants in the United States
3. W.R. Cramond et al, NUREG/CR-4710, Shutdown Decay IIcat Removal Analysis of a Combustion Engineering 2-Loop Pressurized Water Reactor.
4. J.V. Ramsdell & G.L. Andrews, NUREG/CR-4461, Tornado Climatology of the Contiguous United States
5. Calvert Cliffs Nuclear Power Plant, Units I and 2, Updated Final Safety Analysis Report, Revision 15
6. L.A. Twisdale, et al, EPRI NP-0768, Tornado Missile Risk Analysis, Final Report, May 1978.
7. L.A. Twisdale, et al, EPRI NP-0769, Tornado Missile Risk Analysis, Appendixes, Final Report, May 1978.
8. T.A. Reinhold & B. Ellingwood, NUREG/CR-2944, Tornado Damage Risk Assessment -
9. EPRI ' Nuclear Safety Analysis Center and Duke Power Company, NSAC-60, Oconec PRA: A Probabilistic Risk Assessment of Oconce Unit 3, June 1984.
10. G.A. Sanders, et al, NUREG/CR-4762, Shutdown Decay IIcat Removal Analysis of a Westinghouse 3-Loop Pressurized Water Reactor.

I 1. J.T. Chen, et al, NUREG-1407, Procedural and Submittal Guidance for the Individual Plant Examination of External Events (IPEEE) for Severe Accident Vulnerabilities

12. L.A. Twisdale & W.L. Dunn, Applied Research Associates, Inc., Scabrook Nuclear Power Plant Tornado Missile Analysis, Final Report C569, Revision 1, March 1984
13. L. A. Twisdale & W.L. Dunn, EPRI NP-2005, Tornado Missile Simulation and Design Methodology, August 1981 18 l
Retrieved from "https://kanterella.com/w/index.php?title=ML20076F885&oldid=2767369"