ML20216D782
| ML20216D782 | |
| Person / Time | |
|---|---|
| Issue date: | 08/13/1997 |
| From: | Sullivan E NRC (Affiliation Not Assigned) |
| To: | Strosnider J NRC (Affiliation Not Assigned) |
| References | |
| NUDOCS 9709090418 | |
| Download: ML20216D782 (34) | |
Text
_ _ _ _ _ _ _ _ - _ _ _ - _ _ _ _ - _ - - _
.e p at:g$
t UNITED STATES p
NUCLEAR REGULATORY COMMISSION f
WABHINGToN. D.C. 30H6-0001
',%...., #g August 13, 1997 MEMORANDUM TO:
Jack R. Strosnider, Chief Materials and Chemical Engineering Branch Division of Engineering FROM:
,[
Edmund J. Sulliven, Jr., Section Chief til Materials and Chemical Engineering Branch Divi lon of Engineering 9M
$ 19
SUBJECT:
SUMMARY
OF JUL 2 EETING WITH COMBUSTION ENGNgERING.@,NERS GROUP TO DISCUSS STEAM GENERATOR TASK F6MCE' ACTIVITIES 1
On July 29, the NRC staff met with representatives of the Combustion Engineering Owners Group (CEOG) to discuss steam geneator task force (SGTF) activities. Attachment 1 lists the meeting participants.
CEOG formed the SGTF in February 1994 to address various steam generator (SG) issues.
The task force is comprised of Combustion Engineering and utility representatives. At the July 29 meeting, the CEOG SGTF presented an overview of current activities. These include development of an alternate repair criterie for axially oriented stress corrosion cracking (SCC), development of a neural network to evaluate and predict the location of freespan SCC indications, development of insitu pressure test leakage guidelines, and an assessment of SG internals degradation. Attachment 2 is a copy of the CEOG presentation view graphs, ce; Mr. Kevin Sweeney Mr. Gordon Bischoff SGTF Chairman CEOG Plant Manager 5801 South Wintersburg Road ABB Combustion Engineering Mail Station 7696 M.S. 96151932 Tonopah, AZ 85354 7529 2000 Day Hill Road Windsor, CT 06095 Mr. David Pilmer Mr. Charles B. Brinkman, Manager CEOG Chairman Washington Nuclear Operations Southern California Edison ABB Combustion Engineering k
Irvine Operations Center 12300 Twinbrook Parkway, Suite 330
,i Room 109C Rockville, MD 20852 I\\
23 Parker Street Irvine, CA 92718 7,
O f, d d /lg yMf7cA/
em,,,. u" D e
A t
Distribution / Concurrence:
see next page g
cp 7 IlllliIllIlllEllillOljllilll 9IO9090418 970813 PDP TOPRP ENVC-E C
PDR j
~ ~ ~ ' '
Akhmns D bljat 01 n,. n k r kr~r d
$pphC&/4r, __
__ _An:Rfor/IMLG_
_30MLEZl?f TaL Boytvy yt,-
Jntz.Jbe-/5MCE----- JCI'HK=51%
llACH_ZTto.W/DM A/t(K/dr/eM_ CA__
jol-Mir*- t 7 %
MicHsk kr
.__ Sot Csuk. D&_ _ - _ - _7/9'- 348:42 y[
KMiLSukwey_
An.84/si_h4E
- ___ Jn.... M 3
.s W 9 0!clE.5amep
..--._._.-. Eaaw. L b n..-_-
. N 7-E?-$8l?
d~kke _Th Yt.o n
- _-.fM/aC M4/uM 3_ -.
569'-739-4/M.S
. %klyr Cons -en.ce ry, fe%4r.>
9 \\ c - w_4 - t s_%
. bimMshtntm*w_ _._.__._ Ar tr.s.cwhtwesem3-_ _SM-M%ASM t% 4u Bakr A le,yp. -
401 Ja 2?L-9 D u l4L. h.
O w.s
. Ro.d%>+
UtU t.tLeb ko~Wo-OSsh
-.L sw4.hv.wcu._-
.. -..3 pn W\\e.c.$ean/__. _ ___. _ de}ugl DG \\eH3 -.. _-..-_.
(. Arld)-__ __ _ __ S#-858 :#9/1
$$00p./wh
.N AL
......._ % &cto__/fc,un.
AM)/pG/Edtcg. _.
3n-ffp 2?w jJff p_E/En cB gy. 415. g y_
-._.h Soscp A86 M.ddd_._..-
_-. 40J-7.:9 283.5
. - -.5%%m--
m ka ua.as5 se.12 do4n_M mle
.c PPP docs.59 -6BM d ef e n._id._L w i t _.
_.AEE_cC._
h
~L o h n _Ts g.o _ _-._
/MAlos.l_eacG_.
_360_215_97.c s
_Se/7f4-zos s t-4ts.c //o a
.____ 7h ms O. Kexms C.s.,en E,ey____ 141 -6 9y-93et.
t*tS
%y_L. h. pes ar.t JJthtLG. Th.GKaa.,
AB/MG'
_-}423)_752-A29
- :ss w Asesuas ansca Ha-sm-sm
/k 6._ & co w n
,Mmwww W
(*e r-m3_
g do no.eer m oc -
.Om. reds'nns' cum AA 6 cgggj (gyj_zy$._J$t,t 34ech4el. son ABB Mo_285_27.96 ggsw.
~
- e.4
_m. sin mani.m.*T'WW
Albuknua (D e
e.
CEOG Steam Generator Task Force Presentation to NRC Kevin Sweeney (APS) l CEOG SGTF Chairman 7/29/97 cec 0 sGTF / NRC Meeeng,7/29/97 WTRoDUCTioN I
l MEETING PURPOSE o Provide overview and update of CEOG SGTF Activities e Address Staff Questions on SG issues in CE Plants e Provide CEOG SGTF/PSAWG information on Severe Accidents in the ARC Context e General Discussion / Continuing Dialog CEoO SGTF / NRC Moosng. 7/21V97 NTRoDuCTON
Atheknua (9 e
CEOG Steam Generator Task Force Presentation to l
NRC l
l j
Kevin Sweeney (APS) l CEOG SGTF Chairman 7/29/97 CEoG SGTF i NRC Meeting,7/2947 MTRoDUCT1oM MEETING PURPOSE
.a+.
o Provide ovetview and update of CEOG SGTF Activities e Address Staff Questons on SG issues in CE Plants e PrT<ide CEOG SGTF/PSAWG Information on Severe Ai; dents in the ARC Context e General Discussion / Continuing Dialog CEoG SGTFiNRC Moseng,7/29/97 HTRooVCTON
.g
_ - - _ - - - _ _ _ _ _ -. _ - - - - _ = _ _ _ _ _ _ _ _ _ _ -.
e AGENDA LTf.M PRESENTER e Introduction and SGTF Overview Kevin Sweeney (APS) e Axial ARC Program Don Streinz (ABB) e NeuralNetwork Analysis Kevin Sweeney e Severe Accidents Dave Finnicum (ABB) e Summary Mike Short (SCE) e Discussion All CEoG SGTF i NRC Meetag. 749/97 WTRoDUCTioN Introduction 4
- P**
i e Membership j
e Mission and Objectives e Design and Degradation of CE Steam
. Generators e Task Force Activites CEoG SGTF i NRC Moseng,709/97 WTRoouCTioN
Task Force Membership e Kevin Sweeney (APS) e JohnTaylor(Entergy) e Gary Boyers (FPL) e Jay Cate (OPPD) e Tom Rexius (CPC) e Charles Eames (MYAPC) e Dan Meatheany (Entergy) e David Owens (NU) e Mike Short(SCE) e Anthony SacCavino (BGE) e David Ayres(ABB) l J
l CEoo s0TF t MC Wesens,7/2s47 erTRoouCTioN SGTF Mission and Objectives CEOG SGTF was formed in February 1994 The Mission of the CEOG SGTFis to:
o ProactNely and aggressively seek cost effectNo solutions to issues wh6ch chauenge steam generator safety and rehabikty e Reduce assooeted plant capnal and O&M costs.
The Objectrves of the SGTF Forum for experience exeiange e
improve SG safety and rehabihty through joint programs e
Estabhsh CEOG gener6c poonions on techrucal and regulatory issues e
Csoo 30TF t MC Moseng,7/2s47 effmooVCTioN
Generic CE SG Design Features e Two or three loop design e 'Steyo(tube sheet e Tnengulartubepitcti e ' Square-bon (upper tube bundle o Betwings e vwooelsuppens I
- Low yised strength Alloy 600 tutung
~
e Full depth empended tubeaubesheet joint n Exploshjointexpension'explansion' s Minwneltubesheetcrevice a Gradualtransition reducedresidualstress CEOG SOTF iMC Moseng.7tM7 Drm000CT10N Tube Bundle Support Design e
MN Mi l
sm suppone suppone (SeeWine)
Rany unsee Carson Sese Pises era sporaes Wwe on gene Moest 87 Carton Steel Eggerees ans Ptees WWee On tone Moest 70 C ^. Sessi Eggerse Narrow OffBone Vendeted syeesm so ! Femsc SS E0gorea Nonow Getow Sene l
caoG sGTFiMC Woosng,7tM7 ermm
e i
SG Tube Damage Mechanisms 9-esams 1
f as a vv.u s...
1 090 E
\\__W ensemang esen emu.se E
'- e g
--w m -
W m --
<W 6 i
)
1 4
--T--
~'~#i ^
'" g
/'
d 1 6 5
\\
s i
g i p
g g
')
d 0
3 V
D' J
SUMMARY
c wsmi omseneswa t*ms
%,.. J f._
wes_. wes. _esq,,.eae.esars eemass eemess e essee
_eart
.u,
w
,emee, was enemas me ames eence we F's Dead y 84 a p95 eaat thatumse senes <sme e1 ese tenu er e
Ecumme w*
ts 4Wte ge e
a a
a a
espe tsum, te s tse m ette tot tWW a
a a
a a
B Qw Was t 94 2 W6 "M0 W
9 I
a a
a essen see 3 13J tes "EE8 m
9 a
a a
a etens tete 98 te estes 9
9 SLhnet 14 9 IW 14 W
9 5
a a
a a
e k Lamp t 64 W
M M
9 R
a y
a Asquas i 13 e W
M 3s8 m
E B
a a
a emp tpuPae l 14 9 W
90F00 W
9 R
I E
f 8
a a
Sep tbune B 40 WP 999 9 Gi 9
3 e
a e
a a
sensees t te W
90BW
- W G
I e
ste vesen t 91 GH WWs 59 9
m a
a a
a e
a w wee t 40 en m
tm 0
m a
a a
a a
e 8en vasen t 42 GH m
W 9
I I
a a
a e
voumpeut 92 St 4 Gas e
e vousesse een We tese ass 9
s q
4 f
. ~.
~
Overview of SGTF Activities o Examples of Compieted ActMties e In-Situ Pressure Test Guidelines
- Adopted in EPRI Guidelines l
s Circumferential Cracking Root Cause investigation 1
- Program Plan
- NDE Presentation e Response to GL 95-03: Circumferential Cracking of SG Tubes
- Assessment of applicability of EdF lasues CEOG SGtF iMC Meowng.7/2s47 WTRODuCTION Overview of SGTF Activities
. Exampi.s of Ongoing Activmes a CondmonMonttonng/OperationalAssesmentSeminar/ Workshop
- July 30 31,1997 Axial-Onented SCC Altemate Repair Criteria a Neural Network Analysis of Freespan ODSCC a In-Situ Pressure Test Leakage Guidelines (Revision 1)
- Canddate Selection Cnteria
- Leakage Response Guidance a SG intomals Degradation Assessment CEOG SGTF IMC Maseng,7/2617 HTRODuCTION
Steam Generator Internals i
e History i
e IN 96-09 & Supplement i issued in response to EdF Andinge
- CEOG completed formal evaluation of applicability a NEl formed Task Force to assess domestic impact
- ABB and CEOG utilities represented e CEOG suthonzed Task to further assess impact of intomational and domestic euponence o SG intemals Assessment as Part of NEl-led Industry initiative a incorporates recent SONGS 3 Eggerate degradation
- ActMttes tracked via long range program plan a
i CEDO SGTF INRC Meeeng,7/2s/97 NTRODuCTION Steam Generator internals
. _,p 5
e SG intomats Task Action Plan e Operability Assessment-September 1997
- Preliminary Andings e CE SG design tolerant of eggerate degradation e Aggressive degradation consWred unique e Noimmediatesafetyconcem e SusceptibilityAssessment
- Root Cause evaluation
> Generic Safety AnaWie
- Permits CEOG members to perform impact assessment
- Repair options e inspechen and Repair recommendsbons. February 1996 CEOQ SGTF / NRC Moseng, fr2617 WTRODuCTION
---.m
.,.4 w--
r w
- - - -,-~-.--
Tube Bundle Support inspection RECENT BUNDLE VISUAt. INSPECTION SG Model Comolated fjaDDAd Earty Units Yes Yes Model 67 Yes Yes Model 70 Yes Yes System 80 Yes Yes e All SG Model Types have been inspected e Most (60%) of CEOG Owner Units have been inspected e Except for One Unit. Inspections indicate no significant degradation CEOO SGTF i NRC h4seWg. 7/2s47 NTRODUCTION t
i 4
m'
- ^__
Alternate Repair Criteria l
for OD Axial Tube Degradation Don Streinz (ABB)
July 29,1997 CEoo SGTF / MC hemng 74617 00 Amel Degreesson ARC Alternate Repair Criteria for OD AxlalIGSCC
, =.
o Objective e Program Plan e Methodology e Laboratory and pulled tube data e Schedule e Industryinterfaces e Summary CEDO SGTF iMC Moseng 712697 OD Amal Degreasson ARC
Program ' Objective
~
e Objec"u?
a Develot.achnicalbasis for Altemate Repair Cnteria for OD Axial Intergicaular Stress Corrosion Cracking in CE designed and manufactured SGs e Need for art axial crack ARC
= Currently one of the more active degradation mechanism in CE SGs n Observed at tube supports,in sludge piles and in free span regions a TubepullresultsconfirinIGSCC a Plug-on-detectiot.cnterion may be overty conservative a Repairoptionslimited CEOG sGTF / NRC Meetmg 7/26W OD Amal Degreasson ARC Program Plan
)
e Develop axial crack evaluation methodology in three stages
- 1. Operational Assessment Methodology for ' plug on detection *
- 2. Repair Critens based on results of two EC coils having different sensitivibes
- 3. True plug-on-sizing ARC e Cooperate with EPRI and obtain EPRI ARC Committee review a Utilities assume active role e Support bad plant NRC acceptance CE00 SGTF / NRC Masong 7f2sW OD Amel Degreaanon ARC
[
Program Responsibility Matrix
--.:==
-,.m. =-
=_.
-e.:== 4 t= =
.e=
=
I
-.:=- -
--- = - - -.
.i
- ,.,=
l CEoG SGTF INRC Mosen0 frN67 00 Amet Degreasson ARC l
Program Approach e Develop means for Operational Assessment with axial OD degradation s Quanhtative assessment for number and size of defects a Benchmarked against relevant field experience e Physically based (mechanistic) model e Crack process components sunulated and vennable
- Initiate, growth and detection a Tube structural and leakage characteristicr. simulated and vennable
= Each component desenbod by appropriate probability detnbute funcDons e Multiple cycle model responsave to evolving condsbons e Monte Carlo simulation a Quantitative deterministic results with probabilistic devotion CEOG SGTF 1NRC Weseng frm:7 OD Axel Degrasemon ARC
Phase 1 Plug-on-Detection Method
~~
e Several statistical approaches considered (including GL 95-05) e GL 95-05 approach considered overty conservative for depth based Operational Assessment based on:
s Compenson of margins for Bobbin voltage and RPC voltage (or depth)/ length sWJng approaches a Constant POD as a function of degradation regardless of severity a Voltage measurement and depth sizing uncertainties e Due to the degradation type and the required NDE approach, CEOG selected methods differ from GL 95-05 m Demonstratedsuccess i
CEoQ sGTF iNRc Mesang 749/97 Oo Anel oe0reeston ARC i
Operational Assessment Methodology e Qualify probabilistic methodology a Rely heavily on empirical burst, leakage and NDE correlations
- Extensive laboratory data base being developed
- Significant pulled tube date exists
- Supplemental pulled tube data, as needed a Growth rate data from field experience
- Refined growth rates from in-service indications
> Monte Carlo methods to assess / predict entire SG condition CEoQ sGTF/ NRC Mosen0 749/97 oo Amet Deptoston ARC
Phases 2 & 3 e Keyfeaturesinclude:
> inferred defect ranking considering both voltage (or depth) and length a Probability of detection function versus degradation severity
> Probability distribution functions for initiation and growth a Structural characteristics defined by Framatome curve Ligament strengthening model
> Leak rates determined by PICEP model
- Inc!ades crack mo@hology and pop-through effects a Models benchmarked against empirical data and operation croa scTFime usenne rfme7 oo Ame oernesean Ane Program Elements e Improved NDE charactenzation of defects
> Select app:opriate NDE parameters e Develop extensive flaw behavior database a Crack initiation and growth rates e Extensive laboratory test program a NDE, leak and burst testing e Pulled tube database a Range of defect sizes andlocations e Develop statistical and computational models
> Evaluate structural and leakage integnty esoc sCTFIMC Meeene rf39T oo Aseloogreenson ARC
NDE Methods e Probe type a Bobbin, RPC and Plus-point a Evaluate UT for depth screening o Techniqueassessment a Phase angie, maximum voltage, peak to peak voltage a Consider two coil approach with drfferent sensitivities e Analysttraining
= Training and testing data sets from pulled tube database e Probabileyof Detection a Blind test of five or more Qualified data ar1I/sts cE00 SGTF INRC Weetng 7/2947 OD Amal Deysaston ARC NDE Sizing Methods o Evaluating relative succass of several methods s.115 pancake phase angle n.115 pancake voltage - vertical maximum n.115 pancake voltage-peak to peak a + point phase angle s + point voltage. vertical maximum a + pointvoltage peak to peak e Peak to peak voltage correlates reasonably well with
- ground truth
- n + point probe evaluated at 300 Khz s Calibration set points 100%,60% & 20% EDM notches a Expect less anatytical vanability than with some other parameters, i.e., phase angle cE0G SGTF F NRC MeetnD 7/2s47 00 Amal Depedemon ARC
l Peak to Peak Voltage Comparison with Destructive Examination CC.1 Pwies Tune to, R:44, L: 78; 200 kMt P.*, Ampt. V p.,
,=
-.e _ %
g%,,,%4,, g gw ico --
"I
- =.= rw
- 1..
e0 i U
SO.
l k 40
- p\\
/
- e
-f so L i
1.
,,,,,,..,,,,7,
,o 1
Lanom een.. Onenl CEoQ sGTT / NRO Moseng,/29ts, oo Amet oogracemon ARC Analysis Methods Structurat integnty Burst correlation for large range of crack lengths a
Burst correlaton for both 0.042* and 0.048' tube walls a
Empirical evaluation of uncertaintes a
e Leakage integnty a Leak rates based on PICEP Code
= Crack
- pop-through* model n Empincal evaluation of uncertaintes e Defect initiaton and growth rates Operaung exponenCe for 00 axial degradation to support e
initauonfunction a Growth rate database CEoG sCTF 1NRC Aaseeng 7/29/97 oo Ames oogrosecon ARC
Laboratory Test Program e NDE, leak and burst testing, characterization of defect specimens e EDM notch specimens e Over100testspecimens a Simple and complex crack geometnes a Range of crack lengths (0.25 to 3.0 inches) e Range of crack depths (50 to 100% TW)
Multiple samples of each e Laboratory grown crack specimens n Approximately40 specimens Range of crack lengths (0.25 to 3.0 inches) a a Range of crack depths (20 to 100% TW)
CEoG sGTFiNRC Meenne 7/2s47 00 Amal Degreasson ARC Removed Tube Database Removed tubes (32 tubes, > 100 indications) e
= FC (1984) 1 tube Verticalsupport a PSL1 (1985) 3 tubes Studge pile & eggerate a ANO2 (1992) 2 tubes Eggerate a ANO2 1996) 2 tubes Eggaste o PV2 (1993) 6 tubes Free span e PV2 (1994) 12 tubes Free span a CC1 (1996) 3 tubes Free span (mult. indications) m SONGS 2 (1997) 3 tubes Free span & eggerate (mult. Indications) e Supplementalremovedtubes e Plants applying the ARC will remove one or more tubes to confirm similanty of morphology and burst, leak and NDE characteristics CEOG sGTF / NRC Moseng 7f2 47 OD Amal Degraceman ARC
Program Schedule o Phase 1: Plug-on-detection Fall 1997 e Phase 2 & 3:
a Possible two coilapproach Spring 1998 1
a Plug on sizing, ARC Fall 1998 First possible application Fall 1998 CroG sGTFiNRC Memmg 7/29/97 oo Ammi oogresanon ARC l
l Program Interface with {ndustry e EPRI Rule / ARC lssues Resolution Group
> Independent ' peer" review of technical basis a Consistency with other industry submittals e NRCstaff a Penodic review meetings a Communicateprogress
> Obtain regulatory perspective CEoo soTFiNRc u ong 7 trois 7 oo Amai oogroo.acn ARC
Summary
- Develop an Operational Assessment methodology for 00 axial defects a Quantitative assessment of number and size of defects a Physically based (mechanistic) model. Differs from GL 95-05
- Probability functions for crack process (initiation and growth)
- Probability funcbon for POD
- Structural strength base on Framatome curve considering ligament strengthening
- Leak rates determined by PICEP model a Multi cycle, Monte Carlo simulation a Benchmarked e Utility and EPRIinvolvement e Periodic NRC review CE00 SGTF i NRC Meegne 7/2417 Co AmalDegraceton ARC
I Neural Ware Networks nevin Sweeney(APS)
CEOG SGT7iNRC Moserg 7/2947 Neural Netware Objective e Evaluate and predict tube degradation g [TW
(
n e Pronde higher order conolations than was possible with ATHO alone o Additional tool to identify tube trJndie regions for detailed insp0ChonS csoG SGTF1HRC Mosen0 7/29/B7 Neural m
===.
Background===
e Freespan ODSCC phenomena first studied in 1993 m Indication location correlated with "dryout" deposit parameter
- ATHoS model s Studies indicate that multiple factors affect initiation and seventy of defects
- Geometry Operating History, Materials e Recent inspections indicate that T/H correlation not sufficient on a generic basis e SGTF recognized that additional study was required I
e Use of Artificialintelligence/ Neural Networks CCoC SGTFiNRC Meetmg 7/29/97 Neural Netwere Methodology e input Neural Network
- Engine" Analy0 cal T H/ Structural m_
V Output
=
%__ e_...
T-.
"j '"
y EoTData
~
eTube Condibon M
tions O
CEoG SGTF 1NRC Weseng: 7/29/97 Neurg Netware i
Methodology e NetworkTraining a Uses Mathematical Functions and Statistical Techniques to Relate the Outputs to the inputs.
- Examines Historicalinput/ Output Data and Searches for a Model that Produces Output Values that match the Actual (ECT) Data.
> ' Trained
- Models can be used to Predict Future Tube Degradation from Current inputs under the Assumption that the Historical Relationship Between Tube Defects and input c
l Parameters will Continue to Govem the Process.
CEoO sGTFi NRC Mosen0: 7/2s47 Neurel Netwere Phase 1 Network e inputs a As-designedthermal-hydraulicconditions a Five sets of tubeinspection data a Definitionof'cleantube*
e DesiredOutput a Tubes with SVI, sal, or MAI defects a Tubes without above indications (* clean *)
CEoG sGTF / NRc Moesn0: 7/2s47 Neural Neessere
Phase 1 Network
~
e 631 totaldata sets / OD n 247 sets vnth SVI, SAI or MAI Indications
= 364 sets with no indications e Each data set has 21 Selds a Tube location (row, line, elevation) m Thermal-hydraulic data (8 quantrties)
EFPD and known tube indications CEoG SGTF / NRC Moseng 7f2947 Neural Netware Phase 1 Network Model Evaluation
_i e Results a Linear Correlation between the Target Oumut and Neural Net Prediction, R = 0.B4.
> Total Data Points:
631 Number of Known ECTindications:
247 False Negatives:
11/ 247 or 4.5%
a False Positives:
31/ 384 or 8.1%
(g J,M CEoG SGTFiNRC W 7/2s47 Neural Netnere
Phase 2 Network e inputs a SG #1 therma hydraulic conditions a Two sets of tube inspect!on data (SG #1 and SG #2)
= Definition of'cisan tube
- e DesiredOutput a Tubes with SVI, sal, or mal indications a Tubes without above indications (* clean *)
CEOG SGTF / NRC Meetmg: 749/97 Neural Netware Phase 2 Network 4
e 672 totaldata sets a 397 sets with indications a 275 sets with no indications e Each data set has 21 fields a Tube location (row, line, elevation)
= Thermal-hydraulicdata(8quantrties)
= EFPD and known tube indications CEOG SGTFiNRC Meeeng: 70s17 Neural Neenese
__x
Phase 2 Network Model Evaluation
r e Resutts VerticalTube Region SG#1 Linear Correlation between the Target Output and Neural l
Net Prediction, R = 0.94.
> Total Data Points:
358 Number of Known ECT Indications:
83 False Negatives:
4 / 83 or 4.8%
False Positives:
0 /275 or 0%
CEoO SGT7 iNRC Mosen0: 72s47 Neural Netware Phase 2 Network Model Evaluation e Results. U-Bend Region SG#1 Linear Correlation between the Target Output and Neural Net Prediction, R = 0.99.
Total Data Points:
379 a Number of Known ECT indications:
104 False Negatives:
1/104 or 1%
> False Positives:
0 / 275 or 0%
CEoG SGTF i NRC Meeeng: 749/97 Neural Netware
I Phase 2 Network Performance Evaluation e Network was Trained using SG #1 Data.
e " Trained" Network Captures Essential Relationship Between SG #1 input Conditions and Resultant Tube Degradation.
e Network Performance was Tested with the Similar but " Unseen" Data from SG #2 I
CE00 SGTF iNRC Moseng: 7/2s/97 lessi Netware Phase 2 Network Performance Evaluation e Steam Generator #2 Tube Defect Prediction with the
" Trained" Network using SG #1 Data a VenicalTube Reaion
- Number of Known ECT indications:
328
- False Negatives:
4 s U-Bend Reaion
- Number of Known ECT Indications:
180
- False Negatives:
1 CEOG SGTFiNRC Moseng: 7/2s/97 Neural Noguerg
Summary and Conclusions Work to date indicates Neural Networks provide promising technology for predicting tube locations susceptible to defects and developing eddy currentinspections plans
)
CEOG SGTF / NRC Moseng: 7/29/97 Neural Netware Summary and Conclusions e Ongoingwork a Better definition of a " clean
- tube a inclusionof additionalinputparameters a Additionalnetworktraining a Parametricstudies a Network verification and validsbon usN plant data and conventional statistical methods CEOO SGTF / NRC Meegg 712947 Neurel Netware
'W ADDRESSING TI-SGTR RISK IN SUPPORT OF AN ARC Dave Finnicum (ABB) l CEOG SGTFINRC Meeerg 7/29/97 Severs AcadentI ARC Purpose e Show Where CE Plants Differ from W Plants WRT TI-SGTR and Thus Should Be Treated as Separate Group e Outline ABB Approach Proposed to CEOG to Address TI-SGTR in Defining ARC.
e Get NRC Staff Feedback on Proposed Approach CE00 SGTFi NRC Meeurg 7/29/97 Severe Acmeerdi ARC 1
Risk of Thermally Induced SGTR (TI-SGTR)
Per Draft NUREG-1570 e SGTRs resulting from adverse T-H conditions during core damage scenarios with possible bypass, a LERF from CD sequences -2E 06 for reference plant e Dominant sequences. characterized by a High pressure RCS/ Low Seconcary Pressure a induced SGTR nsk ted to RCP seal failure CEoG SGTF / NRC Mesen0: 7/N97 severe Acceent / ARC Significant GOG Plant Differences e Conclusions from NUREG-1570-
> lssue resolved for B&WOG
= Concems over RCP Seal Failure contnbuted to conclusion that genene position could not be drawn for non-B&W PWRs a RCP Seal leakage pnmanly W issue e RCP seal design significantly different than W design
- Seal LOCAs of neghgible procability e CEOG should have a separate identity, like B&WOG CEoG SCTFiNRC Meeeno. 7/ms7 severe Acceenti ARC 2
ABB Proposed Approach e Approach includes nsk assessment and supporting deterministic analyses e Demonstrate probability of achieving a High RCS pressure / low SG pressure CD State < or = 1E 6/yr e Common LOOP evaluation approach a realistic Loop frequences, ' convolution based
- offsne power recovery models. post-core uncovery power recovery Consistent treatment of battery management strateges a
e include impact of AMG strategies on recovery actions e Deterministic analyses Establish realistic core uncovery times.
a SG stress analysis to confirm feasibility of feeding a cry SG e Common position on failure of SG safety valves to resent CEoG SGTF f NRC Moseng: 7/29/97 severe Acceent/ ARC l
Advantages of CEOG Group PSA Approach e PSAWG has had success dealing with PSA issues on a group basis for TS Applications. Groups are useful since:
> Focuses revow on nsk significant issues for the application a Allows agreed upon consistent interpretation of PSA success entena e Account for plant-specific features while providing structure for common evaluation of assumptions and methods e Common review promotes stability in regulatory approach e Allows generic conclusion for TI-SGTR for CE plants.
CEoG sCTFiNRC Moonng 7/2947 severe Acceenti ARC 3
Summary e Risk and PSA regulatory guides have entered the ARC arena via the need to assess TI-SGTRs.
e CEOG plants should be more resistant to TI-SGTRs due to low probability of RCP seal failure on SBO.
e Best estimate PSA modeling if SBO (and related high pressure events) considenng delayed power recovenes. realistic battery management strategies, AMG strategies, etc. can improve the CDF for the most cha!!anging SBO events, i
e Joint analyses with cross compensons/ peer review would simplify NRC review and focus CEOG resources.
e CEOG based technicalinteraction with NRC essential to l
supplying an acceptable product.
CEoG SGTP / NRC Weseng 7/2697 Sowwg Acacenti ARC 4
Summary / Wrap-up Mike Short (SCE)
CEOQ SGTF i NRC Weseng: 7/2:47 summaryiWree up Meeting Review e Provided Update of CEOG SGTF ACtmtes e SGTF: Most Active CEOG Committee
- CEOG # 1 Strategcissues Pnonty
- Most CEOG Commeed Funchng a Tasks are Responsive to industry issues
- CircumferentalCraddng Esenence
- IN 9649 (SG treemals Degradaten) e Some Tasks Lead NRC Concems
- NeuralNetware Ans@is CEOG SCTFiNRC Mosen0: 7/887 summaryiWrap.up o
o Meeting Review w>
e Addressed Staff Questions on SG lssues in CE Plants SGTF Mesion: Pro +ctively and aggressively seek cost effemve solutions a
to issues wtuch chauenge steam generator safety and rehabit;ty SGTF is anecipanng issues and pn>ectnery seeking soluuons a
- Neweswei ere w
- Ameisc: Amc
- sG nummen osynesson e Provided information on Severe Accidents in the ARC Context a The CE design can t>e differentiated in key areas of the severe accident considerations related to ARC
> Comrnon CEOG approach can lead to acceptable CDF CEOG sGTF iNRC Meeeng:7f2s/97 summaryiWrap up Meeting Review e Seek Continuing Constructive Dialog between CEOG SGTF and NRC Staff e CEOG and SGTF Appreciates the Opportunity to Meet and Share Information
)
CEOG sGT7 INRC Moserg 7tJs/97 ggme,y p %
l
_