i -v - w-X- - - -- I--- - - -

1 Don Lrsirkqhbrrv - first ctit at sIidp for Ashok Page 1 !- ---- Z' . -Z� ------ -

From:To:Date:Subject:

Gary Demoss /-Don Marksberry; Mark Kirk; Michael Cheok9/3/02 8:25AMfirst cut at slides for Ashok

For Dop and Mike, I have attached Mark Kirk's slides, which contain the pictures and graphics necessaryto really describe the event

Gary415-6225

( (10

f

[ Don Marksberrv - davis besse brief.wp~d Pane 1 0D o ..... r.. .vs.es........ n 1

Probabilistic Analysis Issues Associated theDavis Besse CRDM Nozzle Cracking

andRPV Head Wastage

Gary DeMoss, Don Marksberry, Mike CheokRES/DRAA/OERAB

PREDECISIONAL[NFORMATIONthZEedecasional Information --

- rr4 rrt�,------

V Don marKsIerry - davis besse briet.wpd Paae 2 4- .. .......I-- - . -~ ~ -Paae2-11

OVERVIEW

PERFORMANCE DEFICIENCY

* Licensee failed to properly implement a boric acid wastage program

* Licensee failed to properly implement an inspection program for thedetection of reactor coolant leakage

THREE POTENTIAL FAILURE MECHANISMS LEADING TO LOCA

* As-Found Cavity Rupture - Cladding exposed in as-found condition fails atnormal or transient conditions, pressure below SRV setpoints

* Enlarged Cavity Rupture - Cavity grows larger and fails at operating pressure

* CRDM Ejection - Circumferential crack grows and causes CRDM nozzle tofail and rod ejection

AS-FOUND CAVITY RUPTURE - MLOCAPre ca i ~oe aft rr _i-ai6h

[ Don Mrarksberrv - davis besse briefmwpd I ~Pane i fDon Mar - davs bessebri-wo .-. -. .

Initiating event analysis - based on the rupture probability distribution multipliedby the frequency of pressure transients between operating pressure (2165 psi)and SRV set point (2500 psi)

- expected values of rupture pressure are around 6000 psi.

* Probability of failure *1 x 10-3

Conditional Core Damage Probability given initiating event - SPAR Model andlicensee provided values are about 3 x 10-3 for MLOCA

ACDF- *3 x 106

Issues/uncertainties - Operating on the lower tail of the rupture probabilitydistribution; uncertainty in strength of cladding material

PredacseenMI0lTft& ton

nMarksberry - Davis bessebrief.wpdP

CRDM NOZZLE EJECTION- MLOCA

Initiating event - Have a probability of rupture vs. time curve from ANL. Assumednozzle 3 wetted for 8 years. Then assumed nozzle 2 wetted for 4 years andnozzle 5 wetted for 2 years. They added 64% and 4% respectively. Probability ofejection is 2.7 x 1 -2.

Conditional Core Damage Probability given initiating event - SPAR Model andlicensee provided values are about 3 x 1 0-3 for MLOCA

ACDF- *8 x 10-5

Issues/uncertainties - Crack propagation rates of wetted nozzles.

5gwxL l s rm-a ion

I - -- . -- -1 11[ Don Marksberrv - davis besse brief.wp~d Page 5 ~I Don Marksberrv - davis besse brief.wpd Paqe 5'.

ENLARGED CAVITY RUPTURE - LLOCA

Initiating event frequency - calculated from the probability that DB operatedlonger, thus had more time for cavity growth

Probability that DB was one of the five early B&W plants - 5/7=0.71

Probability that DB had a better availability record - 2/7=0.2

Probability that DB had 7 additional years - 5/7 *2/7 = 0.2

Probabilities based on B&W reactor availability factors:

Probability that DB operated for additional 1.44 years before Oconee 3cracking discovered - 0.57

Probability that DB operated for additional 2.0 years before Oconee 3cracking discovered - 0.14

Initiating event frequency range - 0.2 to 0.57 / yr.

TzadedsonalJnfomiatjon. - -A

I Don Marksberrv - davis besse brief.wpid 1 Pagie 6 gqDon arkserrydavsbesebrif~wp ------

Ie6

ENLARGED CAVITY RUPTURE - LLOCA (Continued)

Conditional Core Damage Probability given initiating event - SPAR Modelestimate is 2 x 1 -2, licensee provided value is about 1 x 1 -2 for LLOCA

ACDF- 2 x 10-3to 6.2 x 10-3

Issues/Uncertainties

Bounding corrosion rates

2"/yr - 4.5 to 7.5 years until rupture7"/yr. - 1 to 2 years until rupture

Likely ASP approach - At the time of discovery, the probability of LLOCA was notsignificantly greater than nominal.

Pradeqislonal Information

I , Uo Makbry-dvsbse bre .wp Paa 'I Uon MarKSDerry - Clavis besse briet.wpd Page 7 9

CONDITIONAL CORE DAMAGE PROBABILITIES

Soure .. LOCAICCDPs,~K

j SLOCA ^ L CA, sDB IPE not 6.87x1 0-3 1.08x1 0-2

provided

DB Recalculated for size N/A 2.91x1 0-3 N/Aof the exposed clad areaSPAR 2QA 6.46x1 04 Not AnalyzedSPAR 3i 3.47x104 2.54x10-3 2.1x10-2(MLOCA range is 2" to5")

SUMMARY OF RESULTS - VERY PRELIMINARY

Ewdectsional lnformation -. ~

flnnn Mqrkqhprrv - dnviq hp-,ce hripf-wnd --- - - - _- -- ----- __ , "Page 8 S

I1171=_'IZZ =I.,LI:7ZI711I; .1 -

Ruptre: Type LOCA Siz nti CCDP A CDFiEenty.

as-found medium <1 x10-3 2.91 x10-3 <2.9x104cavityenlarged large 0.2 to 0.57 1 .08x1 0.22.Ox1O-3

cavity to 6.2x10-3nozzle medium 2.7x1 0-2 2.91 x10-3 8x10-5e je c tio n_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Don Marksberry- 3 sep 02 thadani briefing.ppt.-..�rz........-..--'w ...

Paae 1 "A,-- 1- -1-- '- ---- - A... I -........................ ........ ...... . - P .I..-..-.-..........- v 5.5

MEB Inputs to NRR ProbabilisticAssessment of Davis Besse HeadDegradation

,4C 9 J RFQ,

* *

Mark Kirk, Bill Cullen, Wally Norris,Jim Davis, Nilesh ChokshiOffice of Nuclear Regulatory Research, Division ofEngineering Technology, Materials EngineeringBranch, United States Nuclear RegulatoryCommission

VG

I go|;v..OA:.Z

[ Don M~arksberry - 3 sep 02 thadani briefing.ppt ,Page .w21Don Mah - 3.A sep 6 2 .x o : t.h...n. .- b..p. .f.. ..... ._ .... .. .... 7

Overview of Presentation

* Photos / depictions of corrosion cavity atNozzle #3

* Contents of NRR Phase 3 RSD

* Stress analysisAs-found conditionPotential future states

v/ Corrosion rate

* CRDM circumferential cracking

VG

Don Marksberry - 3 sep 02 thadani briefing.ppt Paae 3;1S o h ... ov A _D ........._ .A. ._..........: a._AD oiO __ E.:__ __ _......................-.....-...-........

Nozzle #3Corrosion Cavity

Ift .-" am>".. , 0w.BS. AI.,^S=#*W S1- l Z ss c * .

I Don' Mark-s-b-errY-v -3 s-er)02 th-adani bfiefina.ppt - Pag 4,, .Don Makserv-3-DOtadn bieia-~ Pae41 4 *~ .. ~.... ~~~..." ..

Contents of NRR Phase 3 RSD

* The Phase 3 Risk Significance Determination that hasbeen drafted by Steve Long analyzes three accidentsequences that could contribute to risk

Vessel failure caused by rupture of the un-backedcladding

/ For the corrosion cavity as it existed when Davis Bessestopped operations in February 2002

Vessel failure caused by rupture of the un-backedcladding

/ For the corrosion cavity in a larger condition that mighthave existed .... ?????

Vessel failure caused by the development of a 3600circumferential crack, followed by ejection of thenozzle from the RPV head

VG

I Don ,Mprksbery - 3 sp 02 thadani bneiqtiq.pp1_vE<E~~~. x .. ..... e u s .

Page 5 1I Don Marksberry - 3 sep 02 thadani bnefing.ppt Pacie 5�

Philosophy Driving MEB Stress Analysis ofCladding Rupture

* Preliminary analyses (in 3/02 814/02) showed that theprobability of cladding rupturecaused by pressures up to theset-point pressure (2500 psi)is very low

* Most "uncertainties" in thisanalysis are knowable, atleast in principal

Cavity geometryCladding tensile propertiesCladding thickness & geometry

Because of the lowrupture proba ii lese"uncertainties ave beenbounded in our calculations(pending availability of betterInformation).

The one uncertainty that wehave retained in ourcalculation is theuncertainty in howaccurately our finiteelement model can predictcladding rupture

VG

[ Don Marksberry - 3 sep 02 thadani briefing.ppt z Page6 6

Basis of Finite Element Model

Cladding pTensile iRropertiesill(flower-bounded from avaW/abed data,)>

Er. 61 (A LA I f

40 -. ^Bi. t 5 " ~ t i 94 3.59z^ I B53::" =60 --Z.i i1 0 '

r, et-ti 7 .. W21,24 3 e

Dl iB 05 0 0 t5 - 0.2 |Bv:

Z'' ~ ~ ~ ru 3fu~ 0 ns it -) KM C r 3$W.,Z

Cladding� Thickness

(rn�n'rnum repottedby hcenseeX

2 ._ I I wI -. -

I __- __-__I __ I I _

- I I _ I _

.O le ___

_ T $ T T T _ _ dFI I I I .

( ` ).I

:{ z . I

L

I - -F Don Marksberrv - 3 sep 02 thadani briefinampt Page 7 t,

Don Marksberry - 3 sep 02 thadani briefina.ppt Paae 7�

Finite Element Model

Reinforcement providedby J-groove weld was notmodeled. /

- 11 , -submodel

VG

, Don Marksberry - 3 sep 02 thad-an'iW br-ieffnagpptw t ~Paae 8Al... -.- . .. -... .1. � 1. � . � -- Im, A

Uncertainty in Rupture

2 , = O tAy 3B

,I iuI (kiQ

_ tNomal Distibbutlon I3 DMean .1.0975

r Std t v.O: 12 81t {

0 25I mean = 1 098= 25r med~an = I OM

a.0D

WeabullDtnbuaon /\t tocaboin = C 825 i

15r Scajle *2- 0 /308hpe r B2 30r

B6 OS t 1.2 lJ4Exp Burst PressuriPrediclet Bunt Pressure, az r,7'03,2.KiVG

t Dn Mrkserr - sp 02 thadai briefing. pt .

a rks b e r3 S ep O2th ad g .ppt Page 9

Results for "As Found' ConditionPressure at Numerical Estimated Median

.. Pressure at Numerical .Estimated MedianClad Thickness [in.] Instability [psi] Failure Pressure [psi]

Minimum Measured 0.24 6650 7355

Spec. Nominal 0.1825 5180 5732

Failure ProbabilitiesDistribution at Operating Pressure at Set Point Pressure

(2165 psi) (2500 psi)Log-Laplace 4.14E-07 2.15E-06

Beta 0 0Gamma 8.17E-19 1.50E-15Normal 8.44E-1 0 8.90E-09

Random Walk 0 0Inverse Gaussian 4.01 E-29 1.79E-22

Averages 6.91 E-08 3.60E-07

VG

[ Don Marksberry - 3 sep 02 thadani briefina.Dot ... Pq n i 10....-- in.L.1Z7 ~-

Analysis of Larger Cavities--[ 40 M a+IlC+ J+Em El

Ellipsoidal Growth Patterns

2O A6253 6 in

4' 7 16)

Circular Growth Patterns_3 'b.o#m-Q R'El00 Es3***) HDf)*4 Chakrabarty & Alexander, 1970

'Pumv+ituE

VG

[ Don Marksberyv _- 3 "sep 02 thadani briefing.pptD 's s ~mii ' i<E ~ ;v '-, i'^i ', i 1

Don Marksberry :3 sep02 thadanibriefinq.ppi Paoe11�1

Results for Larger Cavities10 .0 -

e 9.0

GA

. ~0

tM 7. 0EIM

.r 6 0CL

0.

i 4.0-

*8 3.0-

Z 20-D.a

e

2 0.0W&0.0

_ __ ___ ___ ____ _ __ _ ,____ , ___

CLad Thilness = 0 24 InchesSS 308 bounding stress-strain curve I . ;

05',;

x.

ound'

f * SBe-Stmilar Growth-- As Found

* E lipsoidal Growth-- Operaln Pressure-- *SMtPcirt

I.- - 51Thcry)- -- 95% Theory

-Chakraborty Thocry

*J Aukrabarty mid 1. M. Alexasicr, 71ydrusirtic Rulging ofCirmutar taphirngnsJoitroal of SSrmm Anwl,,v,, S13),(197UI 155-161

Enror bars are 90% corfidence intervalsbased on Log-Laplace stochastic model

a 50 100 160 203 250 300 35C

Footprint Aroa (In1)

, Don Marksberrv - 3 seo 02 thadani briefina.Dpt t Pagle 12 d!I 7 - _Ilm -11, :1

Observations from Analyses of Larger Cavities

-* Cavity shape has only a minor effect on the predictedburst pressure, especially at pressures near theset-point and operating pressures

> Don M\'arksberrv - 3 seD 02 thadani briefina.ppt Pagqe 131Do -akse-v s- 2-haai rif nal- ag -~

Completing this Work

* Reporting2 ORNL letter reports

/ 1 is done/ 1 is in review

Provide comments to NRR on DRAFT RSD dated 8-28-02

* Maywish toCompare exposed cladding geometry modeled to thatmeasured from dental moldCompare tensile properties used in analysis to thosemeasured from Davis Besse cladding

VG