ML20211D252
| ML20211D252 | |
| Person / Time | |
|---|---|
| Site: | San Onofre, 07043057  |
| Issue date: | 09/30/1997 |
| From: | Begley C, Begley J, Woodman B APTECH ENGINEERING SERVICES |
| To: | |
| Shared Package | |
| ML20211D231 | List: |
| References | |
| AES-97043057-1, AES-97043057-1-1-R02, AES-97043057-1-1-R2, NUDOCS 9709290084 | |
| Download: ML20211D252 (53) | |
Text
. _ . .
_o ha e' ISAPPLIEDTECHNOLOGY AES 97043057-1-1 Revision 2 September,1997 A PROBABLLISTIC OPERATIONAL ASSESSMENT OF STEAM GENERATOR TUBE DEGRADATION AT SONGS
~ '
UNIT 2 FOR CYCLE 9 l
l Prepared by C. J. Begley B. W. Woodman J. A. Begley APTECH ENGINEERING SERVICES, INC.
PlTTSBURGH OFFICE Prepared for Southern California Edison Company San Onofre Nuclear Generating Station San Clemente, CA 92674 9'709290004 970925 PDR -ADOCK 05000361' G PDR 6
APTECH ENGINEERING SERVICES. INC.
200 FLEET STREET o SUITE 4040 o PITTS8UR3H a PA 15220 o (412) 920-6633 o FAX (412) 920 6644 HEADOUARTERS o SUNPNVALE, CA D (408) 745-7000 OFFICES o UPPEA MARGQRO MD r1(301) 599-23010 CUMMING, GA o (770) 7813756 o BETHLEHEM, PA 0 (610) 8G6-7347 HOUSTON. TX o (713) 558-3200 0 CHATTANOOGA, TN O (615) 499-3777 a GASTONIA NC o (704) 865-6318
1 P
4 i
i TABLE OF
, CONTENTS i t
.e , :
Section , - E,qgg !
t EXECUTIVE
SUMMARY
ll 4 i
1 INTRODUCTION 1 2 STRUCTURAL INTEGRITY AND LEAK RATE MODELS 4 i
i 3 ANALYSIS INPUT PARAMETERS 18 4 PROBABILISTIC MODEL 31 ;
5 STRUCTURAL MARGIN AND LEAKAGE EVALUATIONS 36 6
SUMMARY
AND' CONCLUSIONS 45 l REFERENbES 47
-I J &
t 1
i .-.
i
- Aptoch Engineering Services, Inc. . AES970430571 I l t
6
4 i
i EXECUTIVE
SUMMARY
1
~
A probabilistic operational assessment of steam generator tubing in SONGS Unit 2 vias conducted for operating cycle 9. Five modes of corrosion ;
degradation were considered:' ,
e Circumferential degradation at the top of the tubesheet
- Axial degradation at the top of the tubesheet e Axial freespan degradation ,
e Axial ODSCC/lGA at undented eggerate intarsections e Axial PWSCC at dented eggerate intarsect!ons .-
A Monte Carlo computer model was used to simulate the processes of crack initiation, crack' growth and detection via eddy current inspections over multiple cycles of operation. This allowed calculation of the conditional probability of tube burst at postulated steam line break conditions. The effect of tho five observed mo' des of corrosion degradation on the projected leakage for a postulated steam line break so,ldent also was determined.
The projected degraded tube conditions after 0.92 EFPY of operation in cycle 9 meets required NRC margins relativo to structural and leakage integrity.
The conditional probability of tube burst, given a postulated steam line break after 0.92 EFPY of operation is less than 0.01 for each of the five corrosion ,
mechanisms. The largest contributor is axial PWSCO at dented eggerate intersections, with a value of 0.0096. The figures of merit per the new draft Regulatory Guide on Tube Integrity are 0.01 for any single mechanism and a
- total of 0.05 for all rnochanisme combinod. Hence required structural Aptoch Engineering Services, Inc. AES970430571 '
il i
I l:. . - . - . - - - . . - . .- .-- - - - - . - . --
margins for 0.92 EFPY of operation in Cycle 9 are mot. The corresponding total projected 95% upper bound leak rate at 95% confidence at postulated steam line break conditions is loss than 0.033 gpm at 600* F.
The improved detection capabilities of the Plus Point probo compared to the bobbin probe leads to improved structural and leak rate margins for degradation at the top of the tubeshoot region relativo to degradation at eggerato intersections and frecspan locations. Hence, a mid-cycle inspection at the top of the tubeshoot is not required to maintain required margins.
Af ter 2.0 EFPY of operation, without a mid cyclo inspection at the top of the tuboshoot the total contribution to the conditional probability of tube burst during a postulated SLB from both axial and circumferential degradation in .
this region is loss than 0.004. The corresponding component of leakage during postulated accident conditions from those degradation mechanisms is loss than 0.39 gpm at 600* F.
Aptoch Engineering Services, Inc. AES970430571 lii
i
[
i l
Section 1- l lNTRODUCTION [
t A probabilistic operational assessment of steam generator tubing in SONGS-Unit 2 was conducted for operating Cycle 9. Five modes of corrosion degradation were considered:
e Circumferential degradetion at the top of the tubesheet; e Axial degradation at the top of the tubesheet; Axial freespan degradation; e--
, f
. Axial degradation at eggerate tube supports: undented tubes; e Axial degradation at eggerate tube supports: dented tubes.
i The onset of axial and circumferential corrosion degradation was observed in i
the SONGS 2 steam generator tubing after about 7.17 EFPY of operation.
I Eddy current inspections of the freespan and eggerate regions have been made using the bobbin probe. Eggerate axial degradation has been observed ;
on both the inside and outside tube diameters, the ID degradation in these regions being generally coincident with dented tubes. The presence of the oggerate tube supports, and to a greater degree, tube denting in the eggerate regions, tends to make crack detectico more difficult using the bobbin probe.
The simulation model employed in this work accounts for potential inspection difficulties. The projections made for the eggerate and freespan regions ;
reflect 0.92 EFPY cycle duration, in anticipation of a mid cycle inspection.
Circumferential and axial degradation at the top of the tubesheet has been
- observed using the RPC eddy current probe prior to Cycle 8, and the Plus Point probe thereafter. Degradation is present on both inside and outside P
Aptech Engineering Services, Inc. AES970430571 1
9
, - - - . - , , , .. ,.m4,r g -- . , , , ,- -
...<,_-__y_,- -er,.y,v.__,,.,.%,, , . . . , , , . , . , , . . . , , , , . , . _ . . . ,_.m ,.~..__.._.m.,,,m.., . . _, -mm
tubo diameters. Projections for the top-of tubosheet region are modo for a .
full 24 months of operation in order to assess the feasibility of foregoing the mid cycle inspection in this region.
Evaluation of the contribution of corrosion degradation to the conditional probability of tubo burst at postulated steam line break conditions and datormination of the upper bound leak rates expectoc during postulated accident condition form the main objectivos of the work described in this report. The now draft Regulatory Guido on Tubo Integrity' has established acceptablo values for the conditional probability of tube burst at SLB conditions as a measure of required structural margins. Accident induced leak rates are calculated for comparison with the site-specific acceptablo vatuo.
The basic calculatlanal technique employed is one of simulating the processos of crack initiation, crack growth and detection via oddy current inspection using Monto Carlo methods". *ino Monto Carlo simulation model follows those processes over multiple cycles of operation. This allows benchmarking of the moael by comparing calculated results for past inspections with actual observations. The simulation model tracks both detected and undetected populations of cracks and deals with actual crack sizes. When comparisons are modo betwoon calculated results and eddy current observations, an oddy current measuromont error is applied to convert predicted real crack sizes to predicted oddy current observations.'
Actual degradation conditions in terms of number of cracks, real crack depths and longths can be calculated for any selected time porlod. Hence, the conditional probability of burst at postulated steam line break conditions Aptoch Engineering Services. Inc. AES97043057-1 2
h can be computed for the operating time of interest. Leak rate during such a ;
postulated accident can be calculated from the simulated numbers and sizes of cracks. .
In= the next section, a description of the methods of characterizing crack ,
shapes and critical dimensions for both axial and circumferential cracking is !
presented.' This is followed by explanations of burst pressure and leak rate ,
I
= calculations.; Next, input to the Monte Carlo simulation programs is defined, and the simulation steps are discussed. Finally, the preliminary conditional [
probability of bure,t and leak rate results specific to SONGS Unit 2 are
. presented. -
- L 0
Y
+
4 Aptoch En0 l neering Services, Inc. AES970430571
'3 i
- . . . _ . , . - ..--.2. _ ,.4.. _ . , _ . - . . ., . - . . . - _-_-_-._.._._._;.,..._.. _ _2. _ _2 2 m _ _._
- .-. - - - - - - . _ _ _ - . - - - - - - . - - - _ . - . _- .= -
i Section 2 GENERAL APPROACH AND UNIT HISTORY Burst strength and leak rate calculations for tubes exhibiting axial or circumferential corrosion degradation are based upon idealized crack profiles. ,
Both axial and circumferential degradation is modeled as planar cra: king.
The planar crack assumption is conservative for use in burst and leak rate calculations. The following paragraphs describe idealized morphologies for axial and- circumferential cracks and corresponding burst and leak rate equations.
2.1 Idealized Axial Crack Profiles From the perspective of tube burst strength and leak rate calculations, each axial corrosion adication is idealized as a single planar crack. This is ,
conservative in that the strengthening and leak limiting effects of ligaments between crack segments in physical crack arrays are neglected. In addition, the physical depth profile, which typically varies in a non uniform fashion over the length of the crack, is modeled as a simplified ideal profile for burst and leak calculations.
Figure 2.1 illustrates the idealized crack profiles used for burst and leak calculations, compared to tho' corresponding physical depth profile as measured during a pulled-tubo destructive examination. The idealized burst profile represents the portion of the physical profile that is structurally significant in computing burst pressure. The structurally significant dimensions are determined using the Structural Minimum Method", as l Aptoch Engineering Services, Inc. AES970430571 '
i 4
- \
l
- ~ _ . _ , _ - ~ . , . _ _
follows. The physical profile is discrollzod over its longth using a reasonable 1
number of segments, typically betwoon 20 and 50. For each contiguous portion of the crack (that is, for each potential structurally significant longth
'sogment), a corresponding dopth is. computed by equating the areas under i the physical and ideal profiles. Each longth and depth pair is then tested using the Framatomo burst equation' (described below) to find the dimensions that minimize the computed burst pressure. The length and depth that minimize the burst pressure represent the structurally significant dimensions, and henco defino the idealized burst profilo, it is essential to noto that historical measutomonts have shown the structurally significant length of a crack to be reasonably estimated by the portion of a physical crack longth detected by a rotating pancake coil oddy current probo8.
The idealized leak profile length is identical to the structurally significant length computed for the burst profile. The tont shaped leak profile is then datormined by equating the maximum depth penetration for both physical and ideal profiles, and by again balancing the areas under the respectivo profiles over the structural length. The profile form factor, F, is defined to be the ratio of the maximum depth, d . , to the structurally significant depth, d ,, , The distribution characteristics of this form factor are based on pulled tubo destruction examination data'. Soo Figuro 2.2.
Crack growth over timo is assumed to occur primarily in the depth direction.
The structural length for both burst and leak profiles is considered t6 be constant in time. For leakage calculations, the form factor is assumed to remain constant as the crack propagates through-wall, as shown in Figure 2.3. The profilo dimension d, is related to the structurally significant depth Aptoch Es gineenng Services. Inc. AES970430571 5
as d, = 2d,, d ... The length of the through wall segment, L.,,, is then defined by the geometry of the idealized profile to be:
f~!
L,, ',,b
,, (,e_ ,)
Lu,, #1 .
2.2 Axial Crack Burst Pressure Calculation Given the structurally significant longth and depth dimensions as computed above, the burst pressure for an axially degraded tubo is computed via the Fromatomo partial through wall burst equation:
~
0.58St
" '~ Ld L + 2t It '
R, where P is the estimated burst pressure, S the sum of the yield and ultimate tensile strength of the tubo material, I the tube thickness, R, the innor radius of the tubo, L the characteristic degradation longth, and d the characteristic degradation depth. The Framatomo equation, when used with the structurally significant dimensions (L,, and d,,), produces consistently conservative burst pressure estimates compared to measured burst data, as shown in Figure 2.4. It is an excellent lower bound to an extensive set of pulled tubo burst test data.
2.3 Axial Crack Leak Rate Calculation As described in Reference 9, Version 3.0 of the PICEP two phase flow algorithm was used to compute flow rates through cracks as a function of pressure differential (p), temperature (T), crack opening area (A), and total through wall crack longth (L). Friction effects and crack surfaco roughness were included in the model. Steam line break, room temperature, and normal operating condition leak rates calculated by PICEP were fitted to regression Aptech Engineerino Services. Inc. AES970430571 6
4 equations. The PICEP based leak rate regression equation for steam line break conditions is given as:
O = (a + b exp (c (NL)9*' + d (A/L). A p"', where a d are regression coefficients as determined oy an analysis of PICEP results. The leak rate O is expressed in terms of gallons per minute at room i temperature (70*F). To convert to gallons per minute at any other temperature, the calculated O is multiplied by the ratio of the specific volume , of water at temperature (T) to the specific volume of water at 70'F. The pressure, p, is in -units of psi, A is in inches8 and L (equivalently Lw as defined above) is in inches. The crack opening area is calculated using a , twice iterative plastic zone correction to adjust the linear elastic iution for plasticity effects. Further details of the PICEP regression equations and the crack opening area derivation can be found in References 10 and 11. A check of the validity of the leak rate equations is provided by a comparison of calculated leak rates versus measured leak rates listed in Reference 12. Measured leak rates at typical normal operating steam generator conditions are available for axial fatigue cracks in steam generator tubing and axial j stress corrosion cracks in steam generator tubing. Leak rates through stress corrosion cracks are less than those through fatigue cracks of the same length because of the more torturous cracking in stress corrosion samples. A good conservative leak rate calculation methodology is considered to be' one t l which is a closer match to leak rate results from fatigue cracks rather than -
- stress corrosion cracks._ Figure 2.5 shows that this criteria is met by the chosen methodology. Calculated leak rates, illustrated by the dotted lines, serve as a good bound to data from stress corrosion cracked samples of the l
Aptech Engineering Services, Inc. AES97043057-1 7 _ . ~ , . _ _ . . . , _ _ _ ._ _ . , _ _ _ . _ . . . _ -
same tubing dimensions. The calculated leak rates arc just below the measured data for fatigue cracked samples.
~2.4 Circumferential Crack Idealized Morphologies ,
Tiio cararnotor chosen to define the severity of circumferential degradation is the percentage of the tubo crossectional area, which suffers corrosion degradation. Hence the term PDA or percent degraded area. As with axial degradation, a planar crack morphology is the idealized representation of circumferential degradation. For burst calculations it is practical to consider the worst care crack morphology for a given value of PDA. Here a single dominant crack is essumed and all of the degraded area is assigned to a single throughwall crack. This assumption is conservative but not unreasonable for burst calculations' . For leak rate calculations, always assuming this single throughwall crack geometry is grossly unreasonable, if this absolute worst caso morphology is always assumed, then cracks which do not change the burst pressure from its undegraded value would be assumed to leak at more than 0.5 gpm at postulated steam line break conditions.' Clearly a more practical approach to the conservativo estimation of leaking crack lengths and leak rates is needed. A reasonable yet conservative estimation of end of cycle circumferential leaking crack lengths must be based on observed crack profiles. A thorough study of circumferential crack profiles was conducted as part of the EPRl/ANO Circumferential Crack Program' dealing with circumferehtlal degradation at expansion transitions. These results are summarized as follows. The morphology of circumferential degradation shows a substantial variation but it is remarkably consistent irrespective of ID or OD initiation or expansion transitior; type. The general picture is one of multiple crack Aptoch Engineering Services, Inc. AES970430571 8
Initiation sites distributed around the tube circumference. The axial extent of this band of circumferentialinitiation sites ranges from 0 to 0.2 inches. This initiation morphology gives ri% to a latter morphology of deep crack segments against a background of relatlyoly shallow degradation. A doop crack segment is considered to be a region where the local depth is more than twico the background depth. On this basis, the number of deep crack segments per degraded tubo circumference was found to rango between 0 to 4 from pulled tube examinations. A roughly uniform depth profilo is obtained when the number of deep crack segments is either 0 or 4. Typically,1,2 or 3-deep crack segments are encountered as a degraded tube circumference is traversed. The probability of 1,2 or 3 deep crack segments is about the same: 0.32. The probability of 0 or 4 deep cracks is taken to be 0.02 based on pulled tube data. The circumferential extent of an individual deep crack segment varies from 40' to 360'. The distribution of individual deep crack segrnent longths can be estimated from pulled tube data and from field eddy current inspection results. This has been dono in the EPRl/ANO program. One check of the idealized circumferential crack morphology description is to predict the distribution of total arc longths of circumferential degradation detected by pancake eddy current inspections from the frequency of occurrence of deep crack segments and the selected distribution of individual deep crack segments. As shown in Reference 10, predictions and measurements are in very good agreement. The idealized circumferential degradation morpho logy, together with the probability of occurrence of the number of deep crack segments and the distribution o' deep crack segment lengths, provido for reasonable yet conservative projections of through wall leaking i Aptech Engineering Services, Inc. AES970430571 , 9 1
crack lengths needed for leak rate calculations. The leak rate calculations are discussed in section 2.6. l t 2.5 Circumferential Crack Burst Pressure Calculation Data in the literature and testing conducted as part of the EPRl/ANO r Circumferential Crack Program shows that the burst pressure of tubing with , circumferential degradation is bounded by the single planar, throughwall crack idealization. Further, in the region of interest near steam line break j pressure differentials, the burst mode is dominated by tensile overload of the ; net remaining section, in this region of extensive degradation, a lower bound representation of the- burst pressure is given by equating the average net ; section axial stress to the material flow strength. The burst pressure for a , tube with outside diameter circumferential deg,adation, in the tensile burst modo region, is then given as: f P, = ((R,' - R,')/R,8)(1 PDA)(S/2) where P,is the burst pressure, PDA is percent degraded area. S is the sum of yield and ultirnate strength at the temperature of interest, R, is the tube outer radius and R, is the tube' inner radius. For inside diameter circumferential cracking, the pressure on the crack face itself reduces the burst pressure and dictates a correction factor in the burst pressure equation: P, = P R,'/(R,' + R,2 R ')PD i A), where Pi is the burst pressure corrected for ID degradation.
.i
'Aptoch Engineering Services, Inc. AES970430571 10 i
s a .- -. --
)
2.6 Circumferential Crack Leak Rate Calculations , The PICEP based formula presented in section 2.3 can be used for either axial or circumferential cracks if the appropriate expression for crack opening .
' area is used. For circumferential crack.s, a formulation for crack opening area !
from the Ductile Fracture Handbook" was' used. A plastic zone correction to the crack length was applied. Calculated crack opening areas matched actual measurements made as part of the EPRl/ANO Circumferential Crack Program". Hence crack opening area calculations are well benchmarked. ; Since the basic conservative nature of the PlCEP based leak rate equation is demonstrated by--the comparison of measured and calculated leak rates presented in section 2.3, the lone remaining input for circumferential cracking ; is the projected end of cycle leaking crack lengths. This projection is ; developed from calculations of the end of cycle PDA values. The preceding description of circumferential crack morphology provides a picture of deep crack segments against a shallower background of corrosion degradation. Leakage will develop as these deep crack segments penetrate 1 the wall thickness. A conservative estimate of leaking crack lengths is provided by assuming that all of the degraded area is assigned to deep crack , segments in a sequence that produces the largest total leak rate, in most cases, a shallower background level of degradation exists but, in order to be conservative for leak rate calculations, all degradation is assigned to deep crack segments until all segments in a given tube are driven throughwall. A crack morphology simulator program has been written using the data of section 2.4. A PDA value for a tube is selected,- the number of deep crack segments is se , pled accor "ng to the observed frequency of occurrence and deep crack . 3 ment lengths are sampled from an appropriate Weibull
?
Aptech Engineering Services, Inc. AES970430571 11
distribution. The program then apportions the PDA to the deep crack segments to determine if wall penetration is possible. If wall penetration is possible, the program determines, with the given number and lengths of deep ; crack segments, the largest leak rate, which can be produced. I Aptech Engineering Services, Inc. AES970430571 12 i (-- - - .- - , . -.,.- - . - ,, .--.. , ,. ,, , . , . ._ _ _ , , , , _ _ _ _ _ _ _ _,. _ _ _ _
4 i O n A b. A vy Ym \ . 5e dst 8.
^
Lst o m. _ oo u o. o. o. in u 4. 4. crack length (a) m
, A m
.. b dmax
.a 5
E* t. e= x 4 Lst > y no or o4 os on to u $4 se crack length (b) Figure 2.1 Idealized Crack Profiles for (a) Burst and (b) Leakage. Aptoch Engineering Services, Inc. AES970430571 13 i
I Maximum Depth versus Structurely Signmcant Depth ; 100 ; g; ; 90 + 80 4 !
+4 *
, ++
++
+
+
- h
- 1. .,
e +
+
30 20 ' 10 0 0 10 20 30 40 50 60 70 80 90 100 Structural Depth, %TW Figure 2.2 Maximum Depth Versus Structurally Significant Depth, Pulled Tube Data. Aptech Engineering Services, Inc. AES970430571
n 9 3 7 I leak I 1 ~ d max r t 4 4 ds . I 51 N l l Jr V V l . _ . . . . . . . _ _ . . .. 1 i l r l l l l l l . l i l I l i i; Figure 2,3 Idealized Leakage Crack Profile after Through-Wall Penetration I i Aptech Engineering Services, Inc. AES970430571 15 I
,,_, --. _.~. .,,,,-.,..~....._,.,---m,- . . _ . . . _ _ . _ - _ _ _ _ _ . _ _ _ _ _ . . _ _ _ _ _ _ . _ . . _ . . _ _ _ _ _ _ _
l I i 14000 1 e ANO Unit 1 e ANO Unit 2 ga 4 APS PV Una 2 ,.
-
- DV Una 1 . .
12000 ,, , OCC Und 1 , 4 O ONS Und 2 0, A SONGS Unit 2_ O 10000 l
^
% 6* ,
d j8000 [ ,'* g . . Da f m o o
%O
- 4
^
o-O a f 6000 3 00 3 db a a. O
.a o" 4000
's 2000
- 0 0 2000 4000 6000 8000 10000 12000 14000 Measured Burst Pressure, pal Figure 2.4 Calculated Versus Measured Burst Pressure Using tho Structural Minimum Method and the Framatome Burst Equation.
Aptech Engineering Services, Inc. AES970430571 16
100 __ _ _; _ _ ._ . _ _ - . _ _ , ._ _ _ j ..y_ __ _
. _ _ - ; 3 ----,. ,
' I '[
10 - _ . i t- /- 4 . H '
, .) _JI-~
l' r f * ~F J.* i
- ..- - /A ._
-4 jg.
[
/l g .. L5"
/. .
- ti -
- .,y .
I
.__g; i t- - - - '
- r .,i n'
. .g.~,a 5- o
,st,,'L-
. f O a
N'4 O 0 01 "7',[I k +
,,^
,4 o
s,-
.e U Y'$ * -- !
f y,t l 0001 #-
-Q b W FATIGUE CRACKED DATA
" * ' * '0 875" OD DY 0.050" WALL 4 -p
+ ~ " 0.750" OD BY 0.043" WALL 0 CEGOSCC
- CEGB FATlGUE CRACKED DATA 00001 00 1
O.1 CRACK LENGTH,1NCHES . Figure 2,5 Calculated and Measured Leak Rates for Axial ' Cracks in Alloy 600 Tubing at Normal Operating Conditions, l Aptoch Engineering Services, Inc. AES970430571 17 i I
6 1 i Section 3 ANALYSIS INPUT PARAMETERS - A number of input parameters are 'needed for the Monte Carlo simulation model, range of mrmrial proporties is considered rather than a lower bound strength value. E nce the distribution of tensile properties of the : steam generator tubing is needed. The distribution of structurally significant l axial crack lengths is equated to the distribution of measured lengths as , found by the RPC eddy current probe. Thus a sampling distribution of axial crack lengths is needed. The simulation model conducts virtual inspections. This requires knowledge of the probability of detection of degradation as a function of degradation severity for the various eddy current probes that are , used. Since degradation growth is simulated, distributions of crack growth rates for both axial and circumferential degradation are required. 3.1 Tubing Mechanical Properties Figure 3.1 shows a histogram of tube strength for both steam generators at SONGS Unit 2. An adjustment has been made to correct for operating temperature. . A normal distribution was fitted to the data of Figure 3.1 for application in the simulation model. This distribution was truncated at the measured extremos of the tensile property database. 3.2 Degradation Length Distribution During the most recent eddy current inspection at SONGS Unit 2, crack length measurements were recorded for axial degradation at various locations in the steam generators. Experience has shown that length measurements made with the Plus Point probe tend to over estirrate the structurally significant portion 'of a_ crack; hence a best-fit length distribution based on ( Aptoch Engineering Services, Inc. AES970430571 18 I y,y,.-w-yv,a-,- .y.- -,- - ,,.o w - - n4e-.- y- v- - -- , , -- v
.-- ,, ,,m...y -,r,4-1,ym- $ --
e---
the Plus Point measured lengths adds a degree of conservatism to the i simulation. Figure 3.2 and 3.3 show the cumulative fraction of observations versus the measured axial crack length for the top of tubesheet and eggerato areas, respectively, compared to the be,st fit log-normal distributions used in the simulations. The freespan axial cracic length distribution is assumed to be similar to the distribution used for the oggerate region. 3.3 Detection Capabilities of Eddy Current Probes in the computer simulations, a probability of detection (POD) function is used to model the detection cap 6bility of an oddy current probe. Because the effectiveness of the eddy current probo dictates the percentage of cracks that are able to grow deep enough to threaten the structural integrity of the . steam generator, it is important to employ a POD function that accurately reflects actualinspection practices. Several types of oddy current probes have been used at SONGS Unit 2, depending on the type and location of degradation. Axial degradation at the top of the tubosheet has been detected with rotating probes: RPC for inspections prior to Cycle 8, and Plus Point thereafter. The probability of detection of axla! cracking versus average crack depth for RPC and Plus Point probes is shown in Figure 3.4. These curves were determined from pulled tubo data and are perhaps the best-founded probability of detection curves in the industry. The RPC curve is a logistic curve fit to pulled tube data". The Plus Point curve is a log-logistic fit to pulled tube data". The probability of detection function used to represent the RPC and Plus Point capability for circumferential degradation at the top of the tubesbret are shown in Figure 3.5. These curves were derived from data obtained as Aptech Engineering Services, Inc. AES970430571 19
. i I
l part of the EPRl/ANO joint study on circumferential cracking and depth based I analysis methods". t
' The freespan and eggerate regions yyare inspected using a bobbin probe. ;
~
Destructive examinations were performed on three pulled tubes from SONGS l Unit 2 at EOC 8. Burst and leak rate tests were conducted along with extensive metallographic sectioning. Crack depth versus length profiles were determined on burst test crack faces. Maximum crack depths were evaluated at the numerous locations of transverse metallographic sections. This information, when combined with the results of analyses of field eddy current data allowed the construction of a curve of probability of detection versus crack depth. ,. ; Analyst performance in detecting axial degradation at SONGS Unit 2 using bobbin probe eddy current data was evaluated with a supplemental performance demonstration test. Destructive examination data from the SONGS pulled tubes provided a direct measure of the presence and severity of axial degradation. The worst detection performance of five teams of analysts in a blind test using SONGS field eddy current data was used to construct a log logistic probability of detection curve. This curve is shown in Figure 3.6 on the far left of the graph. Historically, bobbin probe detection
- and sizing capability has been referenced to maximum degradation depths.
As noted in Section 2, the structurally significant average depth is the
- parameter _ of interest for burst pressure prediction. Figure 2.2 shows' the -
relationship'of structurally significant depth to maximum axial crack depth. The typically ratio of maximum to-structural depth is 1.28. This factor was l used to, convert maximum depth to structural depth in construction of the probability of detection curves. j; Aptech Engineering Services, Inc. AES970430571 I r
-----,-v-- . _ - - . . -
Detected and undetected (by analysis of bobbin probe eddy current data) cracks were found in freespan and.eggerate crevice locations. Data from both locations was used to construct,the basic POD curve. However, the largest undetected cracks occurred in oggerate crevice regions. For this reason, a measure of conservatism was added to the simulation model for this region. Poorer detection properties were assumed for eggerate compared to freespan locations. The basic bobbin POD curve was shifted to larger depths by about 15% throughwall to treat degradation at undented oggerates. If denting is present a further lessening of detection sensitivity is expected. A further shift of the basic POD curve is warranted. A total shift of 30% throughwall was first applied as a best estimate. A 38% throughwall shift lends to acceptable structural margins at a mid cycle inspection. This is a generous allowance for denting effects on the detection of eggerate region corrosion detection. The curve to the far right in Figure 3.6 is a 38% throughwall(maximum depth) shift of the basic POD curve. 3.4 Degradation Growth Rates During the simulation process, crack growth rates are sampled from a distribution of crack growth rates. A single crack growth rate distribution was used for axial degradation regardless of location. A very extensive growth study was conducted by ANO based on reanalysis of historical bobbin data. This data was analyzed in a number of ways to develop crack growth rates estimates. The end result of the growth study is that a reasonable upper bound growth rate distribution is given by a log normal distribution with a standard deviation of 0.65 and a mean growth rate of 6.89% TW/EFPY. This growth Aptoch Engineering Services, Inc. AES970430571 21
i rate distribution bounds the results of several growth rate studies for plants of similar design". Based on the growth rate study results, the maximum , growth rate was limited to 35% TW/EFPY. The cumulative distribution of the axial crack growth rates used in al.I, axial crack simulation calculations is shown in Figure 3.7. Note that growth rates are based on change in average crack depth rather than maximum depth. The average depth growth rate is used in the simulation model. For any particular crack, maximum depths are calculated based on a known distribution of ratios of maximum depth to structurol (average) depth. For circumf antial cracking, a conservative growth rato distribution was used. In this case, growth rates are expressed as changes in the percent . degraded area or PDA. A crack growth rate study was conducted using data from similar plants wi ht explosive expansion transitions. Negative growth rates were set to zero and the resulting cumulative distribution of circumferential crack growth rates is shown in Figure 3.8. Aptoch Engineering Services, Inc. AES970430571 22
_ - .. . . - . - - - . - _ . _ . - - - . _ - - _ . . ~ . . - _ - - _ .- . a i 2000 1800 1600 1400 l j 1200 I 1000 , . E 800 600 400 I 110 115 120 125 130 135 140 145 Yield + Ultimate Strength (ksi) l l l l Figure 3.1 Histogram of Tube Strength Data for SONGS Unit 2, i Steam Generators 88 and 89. l l Aptech Engineering Services, Inc. AES970430571 l l
F 1 -- 0.8
.f0.6 o
2 g .- I j Measured Distribution .,
- s -
E , j 0.4 -Log-Normal Fit 0.2 } l 0 0 0.2 0.4 0.6 0.8 1 1.2 Axial Len0th, Inches Figure 3.2 Distribution of TTS Axial Crack Lengths. Aptech Engineering Services, Inc. AES97043057-1 24 l
i 0.8
.f0.6 LL -
e 3 3 3
/ Measured Distribution _,
E 0.4 -Log Normal Fit 0.2 0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Axial Length, Inches Figure 3.3 Distribution of Eggerate Axial Crack Lengths. Aptoch Engineering Services, Inc. AES97043057-1 25
,v 1.0 4
Plus Point . O. B - RPC
~
^
.f0.6 j .
15 5 3 0.4 n.
, 0.2 0.0 -
0 20 40 60 80 '100 Average Crack Depth, % Through Wall Figure 3.4 Plus Point and RPC Probability of Detection for Axial Degradation as a Function of Average Crack Depth. Aptech Engineering Services,Inc. AES970430571 26
l 1 l I 1 1,0 -- 0.8 Plus Point i RPC
~ ~
0.6
-3 E
3-k 0,4 n. 0.2 v A 0.0 0 20 40 60 80 100 Percent Degraded Area Figure 3.5 Probsbility of Detection Versus Percent Degraded Area for Circumferential. Cracking (TTS). Aptech Engineering Services, Inc. AES97043057-1 4 . ~
.= . . . - _ - - . . _ . . . . - . . . . - . . . . . .
;_;.; 2 - -
' ,a ,- ,
.. s
/
. /
/ .
l
.. ./
/ /
o8- : l l l l l . l l . l
- - l I l l l' .
l l I
- 1 OA -
,o ,
l 1 l . l
. l I \
l j freespan 1 0.2 l
- l / ' "eggerate, undented l l I
l
~ '-
- eggerate, dented
. ./
l /
*. i
- * .2 0 0.2 04 0.6 0.8 1 average depth (TW fraction) i 1
l ' Figure 3.6 Probability of Detection Functions for Bobbin Probe, y . Adjusted for Eggerate Region and Tube Denting. (. Aptoch Engineering Services, Inc. AES970430571 .l 28 1 1 7 l ; l e
1 0.9 -- O.8 --
@ 0.7 -
P O g U.6 -- u
-- W 0.5 - .
] 0.4 - -
- s g ._
8 o3;; 0.2 --- 0.1 -- 0 -1 . . . . 0 10 20 30 40 AXIAL CR ACK GROWTH RATE, %TW/EFPY Figure 3.7 Distribution of Axial Crack Growth Rates Used in Probabilistic Analysis. Aptech Ergineering Services, Inc. AES97043057-1
9 I i 1.0- f
/
1 0.8 b 56 e 0
- u. ..
o 3
' 0.4 0.2 i-0.0 0 5 10 15 20 25 30 Circ Crack Growth Rate, PDA/EFPY L
f' Figure 3.8 Distribution of Circumferential Crack Growth Rates Used in Probabilistic Analysis. l - Aptech Engineering Services, Inc. AES970430571 30 l
h
.Section 4 PROBABILISTIC MODEL The probabilistic run time model projec't$ the processes 'that have contributed to tube degradation over the history of a steam generator in order to assess
~ the structural condition'of the generator at a future inspection. Specifically, '
Monte Carlo simulation of the processes of crack initiation, crack growth, eddy current inspection, and removal or repair of degraded tubes provides information. necessary to estimate the probability of tube burst and the - magnitude of leakage at the next scheduled inspection, given a postulated
- steam line break event.
The state of degradation of the steam generator tubing is simulated by a i defect population that is defined by several parameters. These are: the size of the population at risk, the initiation function that describes crack inception, the distributions of the defect geometries, and the growth rate distribution that determines the change in cra::k depth over time. The population at risk, in combination with the initiation function, determines the total number of defects simulated in the analysis. The choice of population sizo- primarily influences the computational time and memory. requirements of the simulation. In cases where the choice of population at risk is not obvious from physical considerations, care must be taken to avoid an unreasonably low value that can prematurely exhaust the initiated defect
- population. For degradation near expansion transitions the obvious population at risk is the number of tubes in the bundle. For cracking at l
eggerate intersections, some multiple of the number of tubes in the bundle is appropriate. If the total population of degraded sites is small compared to L Aptech Engineering Services, Inc. AES970430571 31
the total number of sites at risk, then the choice of the number of sites at risk is not of concern other than perhaps creating unwarranted memory requirements. The initiation function for defects is based on a modified Weibull function, which requires a scale parameter and a slope parameter. The scale parameter reflects the length of time required to initiate a given percentage of all potential crack sites. This parameter may be on the order of several decades. The slope parameter is a measure of the rate of increase in initiated defects over time. The scale and slope parameters are adjusted iteratively until the number of indications produced by the simulation matches the actual number of flaws detected at recent plant inspections. . Having matched the numbar in indications observed at recent inspections, other key benchmarking items include: predicting the measured severity of l degradation, confirming notable in situ test results, and reproducing observed inspection transients. A probabilistic valysis of degradation within a steam generator includes many thousands of simulations that track the condition of the steam generator through several past inspection periods to develop benchmark statistics. The model then projects the degradation mechanism through the current operating cycle in order to predict the structural condition of the generator as a function of cycle duration. The present study considers all past inspections for which eddy current inspection results are available. ' Each mock operating cycle and inspection event within a single steam generator simulation consists of several steps that trace the initiation and development of individual cracks. For each potential crack site, a crack t Aptech Engineering Services, Inc. AES97043057-1 l 32 l L
l l w F Initiation time is drawn at random from -a cumulative initiation function. A-certain percentage of the' crack sites'will have initiated during or prior to the operating cycle of interest. For each initiated crack, a set of descriptive parameters is drawn at random from appropriate distributions to describe the crack in detail. These-parameters include the crack length, the crack form factor, and the strength properties of the tube in which the crack resides. The crack retains these particular features throughout its entire life. 'A growth rate is then sampled from the growth rate distribution. The growth rate is applied to the crack - depth over the interval of time between inspections. The growth is assumed to be linear in time. A new growth rate is sampled after each simulated . inspection and appliw over the ensuing operating cycle, which accounts for potential changes in local growth environments due to start-up transients. The average depth of the crack increases with time, and the maximum depth is correspondingly adjusted according to the crack form factor. Simulated ' inspections are performed according to the plant-specific inspection schedules. The crack depth at the end of a completed operating cycle, together with the POD curve, determine the probability that a particular crack will be detected during an inspection. A random number is
. drawn from a uniform distribution and compared to the POD. If the random draw is less than the POD, the crack is detected and removed from service.
Undetected cracks are left in service and allowed to grow throughout' the following operating cycle, and the process is repeated at subsequent inspections. 4
; Aptech Engineering Services, Inc. AES97043057-1 33
( , l
. All cracks, whether detected or undetected, are examined at the end-of cycle
': inspections to assess the probability of tube burst and leakage under steam line break conditions. The algorithm. records a burst if the accident pressure a differential exceeds'the burst pressure ,for a particular, flawed tube. If the maximum' crack depth exceeds the tube thickness, the flaw'is considered to be leaking. A potentially high leak rate can result from a " pop-through"
' event, which occurs when the length of a particular defect is not sufficient to cause a full burst, but the average depth of the crack is such that tne crack breaks through-wall over its entire structural length.
When all initiated cracks have been inspected over the course of prescribed past and future operating cycles, a single Monte Carlo trial of the steam . generator is -complete. Many thousands of such trials are necessary to generate the distributions of tube burst and leakage rates required -in the structural margin assessments. The output from the simulation algorithm consists of a record of all tubes that have burst during the simulation, and all defects that have penetrated through wall and are assumed to be leaking. Other pertinent data such as the operating cycle during which_ the burst or leak event occurred, the tube material properties, flaw length, and form factor are also recorded. For a given operating cycle of interest, the number of burst events are tallied - and a 95% upper confidence bound for the probability of burst is comp'uted using an appropriate F-distribution, as in Reference 16. For example, if 10,000 simulations of the steam generator produce 1 or more bursts in 30 of the trials, the 95% confidence probability of burst is calculated to be PoB = 4 0.00407.
. Aptech Engineering Services, Inc. AES97043057 34
,m - ,_ _ , _ . . . ~ , . . . ..r -, . -., , ...e. , , , . , , - , .
e ,- w. .- . - - - _ r , #
A leak ' rate is assigned to each throughwall defect according to the methods presented in Section. 2. -The total leak- rate for each steam - generator -
- simulation is - then+ computed, - the. simulation leak rates are sorted in :
ascending order, and the 95/95 probabjljty/ confidence leak rate is determined as described in Reference 16. For example, for 10,000 steam generator-simulations, the 9537* highest computed leak rate represents the 95* percentile leak rate with 95% confidence, i i-1 i: f i Aptech Engineering Services, Inc. AES970430571 35
.. , , - . - . = . . . - . . - . . -..--.- -. ..- - .. - . . - . . - . . - , - . . .
Section 5 STRUCTURAL MARGIN AND LEAKAGE EVALUATION Monte Carlo simulation models were used to project the progress of corrosion degradation of steam generator tubing in SONGS Unit 2. Five degradation mechanisms were considered. The severity of corrosion degradation was projected for various operating times of interest dealing with both detected and undetected levels of cracking. Using this information, the conditional probability of tube burst at postulated steam line break conditions was calculated for each of the degradation mechanisms. Also, the contribution of corrosion related accident induced leakage was evaluated for - each operative corrosion mechanism. Table 5.1 lists the history of the number of eddy current indications for a composite worst case generator at SONGS Unit 2. Eddy current results for the two generators are about equal. The main difference is the number of freespan indications. A total of 43 have been observed in steam generator 89 compared to 7 in steam generator 88. There are relatively few indications at eggerate intersections. However, both axial ODSCC and PWSCC have been observed at eggerate intersections. The most numerous indications are at the top of the tubesheet where both axial and circumferential degradation has been found. Again both ID and , OD indications have been reported. Note, in Table 5.1, the eddy current probe types at different locations and the use of the Plus Point probe at the top of the tubesheet in the last -inspection. Aptech Engineering Services, Inc. AES970430571 36
4 a 5.1 ' Axial Degradation at the Top of the Tubesheet 2
- . Axial degradation near expension transitions at the top of the tubesheet was
- first detected at SONGS Unit 2 in th'e1ast inspection at EOC-8. In the most affected generator, 91 indications were found using the Plus Point eddy current . probe. About one-third of this total was ID degradation, located below the top of the tubesheet. The constraining effect of the tubesheet was ignored as a conservative measure. Crack lengths evaluated-from the response of the Plus Point probe substantially overstate the crack length relative to the structurally significant crack length. Even when conservatively
~ ~
equating the structurally significant crack length to the Plus Point crack length, the severity of the axial top of the tubesheet degradation is mild. .. Very few of the indications are long enough to challenge the SLB burst pressure with the bounding assumption of 100% throughwall cracking. , Probability calculations regarding burst and leak rate behavior reflect the implications of the measured crack length sampling distribution. Both ID and OD degradation was considered together using a conservative distribution of crack growth rates. After 0.92 EFPY of operation in Cycle 9 the conditional probability of tube burst from this degradation mechanism was calculated as 0.0005. The projected leak rate at postulated SLB conditions is negligible. If a mid cycle inspection is not performed for this region, the conditional probability of tube burst is 0.0033 after 2.0 EFPY of operation. The corresponding projected accident leak rate is 0.009 gpm at 600'F. Axial degradation at the- top of the tubesheet is not a limiting degradation mechanism for SONGS Unit 2. A mid-cycle inspection in this region is not required to maintain structural margins. Aptech Engineering Services, Inc. AES97043057-1 37
5.2 Circumferential Degradation at the Top of tne Tubesheet
.Circumferential degradation at expansion transitions. at' the top of the .
tubesheet has been observed at SO G5 Unit 2 at the last three inspections. Both ID and OD degradation has been observed. Use of the Plus Point probe 2 at EOC 8 led to an inspection transient which was included in the simulation
- model. The measure of severity for circumferential degradation is the ,
percent degraded area of the tube annular cross section. PDA values were obtained following an EPRI voltage normalization procedure. As in the case of the top of the tubesheet axial cracking, both ID and OD circumferential
~ ~
cracking was considered together using an appropriately conservative growth rate distribution. The conservative nature of this analysis is illustrated in . Figure 5.1 where actual PDA measurements at EOC-8 are compared with ! predicted PDA measurements. Actual PDA measurements are less severe than predicted PDA measured values. After 0.92 EFPY of operation in Cycle 9 the contribution of circumferentia! degradation at the top af the tubesheet to the conditional i probability.of tube burst is 0.0003. This increases to 0.0005 at 2.0 EPFY of operation into Cycle 9 if a mid cycle inspection is not performed. Projected leak rates behave in a .similar fashion. After 0.92 EFPY of operation, the 95/95 upper bound projected SLB leak rate is 0.028 gpm for circumferential degradation.' If -a mid cycle inspection is not performed, the projected i leakage from circumferential degradation becomes 0.38 gpm at 600 F after 2.0 EFPY_ of operation. Hence,- a mid cycle inspection of the top of the tubesheet region'is not required to maintain structural integrity and leak rate
. margins.
f
- Aptech Engineering Services, Inc. AES97043057-1 38 ,
i l 1 l
[ ' 5.3 Axial Degradation at Freespan Locations Axial ODSCC/lGA was detected at SONGS Unit 2 at freespan locations in the last inspection. This is not unexpect' edin. view of the performance of similar steam generators. This degradation was discovered by the bobbin probe af ter chemical cleaning of the unit. Plus Point inspection were performed in tubes with bobbin probe indications. The low signal amplitudes of Plus Point indications argued for mild severity of freespan axial degradation. This was confirmed by burst tests of pulled tubes. The burst strength of tube sections with axial freespan indications was in excess of 10,000 psi. As noted in Section 3, pulled ' tube destructive examination results and supplemental performance demonstration testing verified the excellent detection capabilities of the bobbin probe relative to freespan axial degradation. Hence a conditional probability of burst after 0.92 EFPY into Cycle 9 of 0.0003 is not surprising, nor is a negligible projected leak rate. 5.4 Axial Degradation at Undented Eggerate Intersections Both ODSCC and PWSCC at eggerate intersections were indicated by field eddy current results. This was confirmed by destructive examination of pulled tubes. Degradation initiating on the tubing ID was associated with the presence of detected denting with one exception. One very large PWSCC crack was located at an eggerate intersection where eddy current data did not indicate the presence of a dent. However, mild ovalization of a tuba can generate stresses in excess of the yield strength. Hence it is possible to create conditions needed for the onset of PWSCC without creating a dent which is detectable by eddy current testing. Therefore ODSCC/lGA in this analysis is asr.ociated with undented eggerate intersections and PWSCC is associated with dented eggerate intersections. Aptech Engineering Services, Inc. AES97043057-1 39
As noted in Section 3, the bobbin probe probability of detection of axial degradation at undented eggerate intersections was conservatively adjusted in recognition of the fact that the large.st undetected axial ODSCC/lGA was found at Ondented eggerate intersections rather than at freespan locations. In the probabilistic analysis, this shift of the POD curve was partially offset by the low number of indications at undented eggerates. The conditional probability of tube burst from this degradation mechanism is calculated to be 0.0008 after 0.92 EFPY of operation in Cycle 9. As in the case of ODSCC/ICA in the freespan, the calculated 95/95 upper bound leak at postulated accident' conditions is essentially zero. 5.5 Axial Degradation at Dented Eggerate intersections As noted previously, PWSCC, on mechanistic grounds, is associated with dented eggerate intersections, even if there is no detectable denting via eddy current inspection. The largest observed axial cracks were ID initiated. One such crack led to a measured burst pressure of 3515 psi. There was some demonstrable damage to the remaining ligaments beneath and around this crack during the tube pulling operation. Hence the undamaged burst pressure is in the vicinity of the 3AP burst pressure requirement per draft Regulatory Guide 1.121. However axial PWSCC at eggcrate intersections is the limiting form of degradation at SONGS Unit 2. Dented intersections detected via the bobbin probe, with a signal amplitude of 5 volts or greater, were inspected with the Plus Point probe. This l effectively limited the possibility of very large cracks being obscured by the ! dent signal. However, PWSCC is not limited to large detectable dents. 1 Hence the bobbin probe detectability must be relied upon in general. There is Aptech Engineering Se . ices, Inc. AES97043057-1 40
t- . no firm database to construct a probability of detection curve for PWSCC at dented eggerate intersections. A best estimate curve was constructed from results at SONGS, other similar plants and the key benchmark of leakage during both in situ pressure testing and. pulled tube pressure testing. This led to a-30%1throughwall shift of-the bobbin probe' POD curve determined for ODSCC/lGA. As a further conservatism, this curve was shifted a further 8% throughwall for a total shift of 38% throughwall leading to a conditional probability _of tube burst of 0.0096 after 0.92 EFPY of operation in Cycle 9. The key benchmark of a reasonable prediction of the probability of observing leakage at SLB pressures at EOC-8 is met with the best estimate POD curve shift of 30% TW7 Leakage at EOC 8, if postulated SLB pressure were applied, was predicted in about half of the .:mulation model runs. This . compares well with in situ and pulled tube press,'re testing. The bounding POD curve giving an acceptable mid cycle structural margin is a generous allowance for the problems of detecting PWSCC at eggerate locations relative to the demonstrated detection capability of the bobbin probe for ODSCC at freespan locations. Use of the best estimate POD curve and pressure test results indicate that a mid cycle inspection is warranted. 5.6 Summary of Structural Margin and Leak Rate Evaluations
- A summary of calculated conditional probabilities of tube burst and -upper
-bound accident induced leak rates is provided in Table 5.2 for the five operative corrosion degradation mechanisms at SONGS Unit 2. The limiting mode of degradation is PWSCC at dented eggcrate intersections. Present conservative results indicate that a mid cycle inspection in the vicinity of 11 months of operation into-.' Cycle 9 is warranted due to this corrosion
= degradation mode. A 100% _ bobbin hot leg inspection is reasonable.
.However, structural margin and leak rate evaluation resuits, shown at the Aptech Engineering Services. Inc. ES37043057-1 41
. . .- . . _. .- . . ~ . . ~ . ~ - - - . . . - . . . ... .. .- - -.. - . _ . - . - . - - - .- . - - . .
F
= bottom of Table 5.2, demonstrate that more than 24' months of operation -
without- a mid cycle inspection ~ at the top of ~the tubesheet.is a prudent-option. - The excellent degradation _ detection properties of the Plus Point ' . probe used in this region at the las_t . inspection le. ads to' a mild BOC - degradatidn condition and no significant challenge to required structural and leak _ rate margins for a full 24 month cycle of operationi i 4 3 e i 4 , Aptech Engineering Services, Inc. . AES97043057-1 42 - 5-P
.-. +,,~ ,y.-.. ,_m.y.,.. . - -,,--, s.n ,,,. ,, .,c,,,ne..,.yy.,-_,, , , , 7 y .,. , ,
u TABtE 5.1 5
- r SONGS UNIT 2 j{ COMPOSITE WORST GENERATOR-g EDDY CURRENT INDICATION HISTORY E
g DEGRADATION MECHANISM a iP EFPY TOP OF TUBESHEET TOP OF TUBESHEET FREESPAN UNDENTED DENTED CIRCUMFERENTIAL AXIAL AXIAL EGGCRATE EGGCRATE (ODSCC & PWSCC) (ODSCC & PWSCC) (ODSCC/ IGA) AXIAL AXIAL - (ODSCC/ IGA) .(PWSCC) - g (NUMBER OF INDICATIONS / EDDY CURRENT PROBE) - 7.16 10/RPC 1/RPC O/ BOBBIN 3/ BOBBIN ' ' 0/ BOBBIN 8.62 12/RPC O/RPC O/ BOBBIN 2/ BOBBIN 0/ BOBBIN 10.09 93/PLUS POINT 91/PLUS POINT 44/ BOBBIN 7/ BOBBIN 8/ BOBBIN PROJECTED' 11.01 82 101 63 2 7 8 g 12.09 227 250 - - - E o ' Average projection result 2
$ Without mid cycle inspection S
---- . . =- .- . - - - - . . _ - - - _ . .- __ - .. . . - . . _ - - . .
TABLE 5.2
SUMMARY
_OF STR.U_CTURAL MARG,1N,-
. AND PROJECTED SLB LEAK RATES.
PROJECTED FOR 11.01 EFPY (Mid Cycle) Degradation Mechanism Conditional Probability of 95/95 Leak Rate at. Burst at Postulated SLB Postulated SLB (95% Confidence Level) (GPM at 600'F) Axial ODSCC at Undented Eggerate. . 0.0008 0 Intersections . Axial PWSCC at Dented ' Eggerate Intersections 0.0096 0.0044 Freespan Axial ODSCC 0.0003 0 Circumferential ODSCC/PWSCC at 0.0003 0.028 Expansion Transitions Axial ODSCC at Expansion Transitions 0.0005 0 PROJECTED FOR 12.09 EFPY WITHOUT A MID CYCLE INSPECTIOfJ l Degradation Mechanism Conditional Probability of 95/95 Leak Rate at L Burst at Postulated SLB Postulated SLB-
- -(95% Confidence Level) (GPM at 600*F)
Circumferential. ODSCC/PWSCC at 0.0005 0.38 Expansion Transitions Axial ODSCC at Expansion Transitions 0.0033 0.009 Stech Engineering Services, Inc. AES97043057-1 44
, , ,. - , __ ~ , .~_ - - - . _ _ . _ _ _ _ . _ _ _ _ _ _ _ _
18 16 - 14 5 SG-88 and SG-89 combined 12 - - O simulation results 10 -
~
e f
.E .
V S 8
= - ..
6
=
4 2 n - _ 0 5 15 25 35 45 55 65 75 85 95 PDA Figure 5.1- Comparison of Actual Measurements and Predicted Measurements for Circumferential Degradation PDA Values at the Top of Tubesheet, EOC 8. Aptech Engineering Services, Inc. AES970430571 45
- Section 6
SUMMARY
AN CONCLUSIONS Monte Carlo simulation models were used to project the progress of corrosion degradation of steam generator tubing in SONGS Unit 2. Corrosion degradation was conservatively represented as planar cracking. The processes of crack initiation, crack growth and detection of cracking by eddy current inspections were simulated for multiple cycles of operation. Thus the severity of corrosion degradation was projected for operating cycles and times of interest. Both detected. and undetected crack populations are included, Burst and leak rate calculations are based on the total crack' - population. The simulation model is benchmarked by comparing simulation results with actual eddy current inspection results, notable in situ test results, and pulled tube test data. Good benchmarking results were obtained for SONGS Unit 2. >^ . Five modes of corrosion degradation were evaluated:
- Circumferential degradation at the top of the tubesheet e Axial degradation at the top of the tubesheet
- Axial freespan degradation e Axial ODSCC/lGA at undented eggerate intersections e Axial PWSCC at dented eggerate intersections ,
Projected levels of corrosion degradation severity allowed calculations of the l conditional probability of tube burst and an upper bound accident induced i leak rate. After 0.92 EFPY of operation in Cycle 9, the conditional L probability of tube burst, given a postulated steam line break event, is less Aptoch Engineering Services, Inc. AES97043057-1 46
- 14 than 0.01 for each of'the five corrosion mechanisms combined. The largest
' contributor to the condition; probability of tube burst is axial PWSCC at dented eggerate. intersections. This value of 0.0096 is far larger than those of the other degradation mechanisms; ,, ,
- An acceptable structured margin per the new draft Regulatory Guide on Tube Integrity is a value of 0.01 or less for the conditional probability of burst for any one mechanism and 0.05 or less for. all mechanisms combined. These limits are met for 0.92 EFPY of operation. Calculations for longer time periods show that a bobbin probe eddy current inspection is appropriate prior-to full cycle operation but that a mid cycle inspection at the top of the tubesheet is not required for a total operating period in excess of 2.0 EFPY. ,
f Aptech Engincoring Services, Inc. AES970430571
o REFERENCES
- 1. Draft NRC Regulatory Guide X'.XX,- Steam Generator Tube Integrity, 1996 Revision.
- 2. Sweeney, K., "Palo Verde Nuclear Generating Station Unit 2 Steam Generator Evaluation", Arizona Public Service Co. Report, August 1995.
- 3. Begley, J.A., " Analysis of ODSCC/lGA at Tubesheet and Tube Support Locations at St. Lucie Unit 1," APTECH Report AES 96052749 1-1, dated October 1996.
- 4. Begley, J.A., Woodman, B.W., "An Analysis of ODSCC/lGA at Eggerate Support Locations at Arkansas Nuclear One (ANO) Unit 2, APTECH Report AES 95102556-1-1, dated March 1996. ,
- 5. "Palo Verde Nuclear Generating Station Unit 3 Cycle 6 Steam Generator Evaluation," Arizona Public Service Co., Submitted to the NRC, July 1996.
- 6. Woodman, B.W., Begley, J.A., "Palo Verde Unit 2 Run Time Analysis Regarding the Impact of Upper Bundle corrosion Degradation During Cycle 7, AES 96072812-1-1 Rev.1, dated December 1996.
- 7. Cochet, B., " Assessment of the Integrity of Steam Generator Tubes -
Burst Test Results - Validation of Rupture Criteria (FRAMATOME DATA)," Palo Alto, CA, Electric Power Research Institute, NP-6865-L, Vol.1, June,1991.
- 8. Begley, C.J., and Begley, J.A., "A Generic Probabilistic Analysis of the Effect of Axial Freespan Corrosion on the Structural Performance and Leak Rate Behavior of OTSG Tubing", APTECH Report, AES96102886-1Q 1, June 1997,
- 9. "PICEP: Pipe Crack Evaluation Program (Revision 1)", Electric Power Research Institute, December,1987, NP-3596-SR, Revision 1.
- 10. " Depth Based Structure 2 Analysis Methods for 'SG Circumferential Indications", EPRI report, in preparation.
Aptech Engineering Services, Inc. AES97043057-1 48 I l
e e
- 11. Begley,~ J. A., " Leak Rate Calculations for Axial Cracks in Steam Generator' Tubing," APTECH Calculation AES-C 2797-2, dated April,-
1997. ,
'12. "PWR Steam Generator Tube Repair Limits Technical Support Document for Outside Diameter Stress Corrosi6r1 Cracking at Tube Support Plates",
EPRI Report, TR-00407 Rev.1, August,1993.
- 13. Zahoor, A., " Ductile Fracture Handbook", EPRI Report, NP.6301-D, RP 1757-69, June,1989.
- 14. "Palo Verde Nuclear Generating Station Unit 2 Cycle 7 Steam Generator Evaluation," Arizona Public Service Co., Submitted to the NRC, January 1997.
- 15. Woodman, B.W., " Development of a Probability of Detection Function for Calvert Cliffs Unit 1," APTECH Cal:ulation AES-C-2797-1, dated April .
1997.
- 16. "SLB Leak Rate and Tube Burst Probability Analysis Methods for ODSCC at TSP Intersections," Westinghouse Non-Proprietary Class 3 Report, WCAP-14277 Revision 1, December,1996.
l 1 Aptech Engineering Services, Inc. AES97043057-1 49
-. . . -. -- __ _____ _ __ _ - _ _ __-_ -}}