ML20116F302

Steam Generator Operational Assessment

ML20116F302
Person / Time
Site:	Arkansas Nuclear
Issue date:	07/31/1996
From:	ENTERGY OPERATIONS, INC.
To:
Shared Package
ML20116F285	List: 2CAN079604, Discusses Evaluation of Steam Generator Integrity for Cycle 12 Operation & Forwards Steam Generator Operational Assessment ML20116F302
References
NUDOCS 9608060308
Download: ML20116F302 (47)

Text

_.. -.

A

~

ENTERGY l

ARKANSAS NUCLEAR ONE UNIT TWO STEAM GENERATOR OPERATIONAL ASSESSMENT 4

July 1996 p$geoggggggggggg6e 4

PDR g

i Attachm:nt to 0CAN079606 Page i of 45 TABLE OF CONTENTS 1

EXECUTIVE

1.0 INTRODUCTION

2.0 ANO-2 STEAM GENERATOR DESCRIPTION i

3.0 EXPANSION TRANSITION CIRCUMFERENTIAL CRACKING i

3.1 Structural Requirements

' 3.2 Probabilistic Model 3.3 Accident Leakage Evaluation l

-4.0 FREESPAN/EGGCRATE TUBE SUPPORT PLATE AA1AL CRACKING 4.1 Structural Requirements l

4.2 Probabilistic Model l

4.3 Accident Leakage Evaluation 5.0 PROBABILISTIC DOSE MODEL i

6.0 -

PLANT OPERATIONAL EVALUATION

7.0 CONCLUSION

S l

l

8.0 REFERENCES

~

. 7 Attachm nt to OCAN079606 Page1of45' 4

EXECUTIVE

SUMMARY

i The purpose of this report is to provide the results of evaluations performed to assess the tubing i

integrity in the ANO-2 Steam Generators (SG) for the current cycle of operation. Previous o

assessments

,2) were performed for circumferential cracking at the expansion transition. This report builds on this previous work and addresses axial cracking at eggerate tube support plates i

and in the upper freespan region.

The evaluation performed for circumferential cracking built upon significant work previously completed. Improvements over previous assessments were as follows!

Use of the Plus Point coil to conduct eddy current examinations during 2R11 i

e Utilization ofhigher peak pressures for in-situ pressure tests of the larger flaws e

Use of a second Plus Point coil (high frequency) for improved flaw characterization -

e Use ofenhanced software that allows " profiling" of cracks e

Development of a leakage model to evaluate end of cycle (EOC) conditions j

Enhancement of probabilistic evaluation methods Development of a probabilistic dose assessment model

.)

e 1

This report also describes the ANO-2 SG Program and the success in identifying degradation in the SGs prior to challenging safe operation. Examples include the detection of upper bundle freespan degradation prior to its evolving into a safety significant mechanism and use of the best available equipment and technologies. Other program actions include preventive measures such as primary temperature reduction, installation and use of N 6 radiation monitors, comprehensive i

inspections, and a conservative operating philosophy. Strict controls are maintained on primary-to-secondary leakage limits, with procedures and rigorous operator training providing heightened awareness and guidance to assure safe shutdown in the event ofleakage is ensured.

Based on the results of the evaluations performed, the integrity of the ANO-2 steam generators is maintained, and ANO-2 can safely operate until the next refueling outage scheduled for April 1997.

I h

Attichment to OCAN079606 Page 2 of 45

1.0 INTRODUCTION

The purpose of this report is to describe the work performed to evaluate the integrity of the SG tubing during the current cycle of operation. Included in this assessment is degradation due to circumferential cracking at the expansion transition, axial cracking at eggerate tube support plates and in upper bundle freespan regions.

A detailed description of the ANO-2 operational history was provided in Reference 1, as updated by Reference 2. Currently, an equivalent of 9.2% of the tubes in the "A" SG and 6.8% in "B" are plugged. Included in this are 1029 tubes repaired with tubesheet sleeves.

2.0 ANO-2 STEAM GENERATOR DESCRIPTION l

The ANO-2 steam generators are of the U-tube design manufactured by Combustion Engineering (Model 2815). Each steam generator contains 8411 tubes constructed of high temperature mill annealed (HTMA) Inconel alloy 600 material with an outside diameter of 3/4 inches and a wall thickness of 0.048 inches. The tubes are explosively expanded the full depth of the tube sheet.

There are seven full eggerate tube support plates (TSP), two partial eggerate TSPs, and two l

partial drilled TSPs. The SG layout is shown in Figure 2-1. Commercial operation began March l

1980, and all volatile treatment (AVT) chemistry has been used since that time. Secondary side l

boric acid addition was initiated in 1983 to arrest denting at the partial drilled TSPs. The hot leg j

operating temperature was initially 607' F, but was reduced to ~600* F following the ninth refueling outage in the fall of 1992.

t i

1 i

i

Attachment to OCAN079606 Page 3 of 45 Figure 2.1 ANO-2 Steam Generators SEAM OUTLET NOZZLE d

STEAM DEFLECTOR

)

126 STEAM DRYERS OOOOOOO

- TUBE BUNDLE ACCESS WELL l

MANWAY

'I

--O I

SECONDARY l

8

~

[ SEPERATORS 0R

-- CAN DECK DRAIN (6 TOTAL) 166 STEAM DRANs f :3 UPPER LEVEL TAP p

s Leva - --.k e neo.7 ooes a

h er 1

w DEcu-o NORMAL LEVEL as arr w em.

g 9'

FEEDWATER FEEDRNG uMCowERED

/

5 NOZZLE e sus wt LEVEL STUBE d

/

,,,,h FEED RING J

,I or. west,use M RSLPPORT asmun : =

f r

It

• I3 # W

- LOWER NR LEVEL TAP l

BAT WING TUBE SUPPORT e 424 3 pots

-==l28 8=.-

g 294.2 NCHES ASOVE TUBE DEET lower SUPPORT PLATI te 10) lll DOWNCOMER evenu.

i f REGtON owam aan.

EGG CRATE 1UBE SUPPORT ASSEMBLIES

( ? FJ eypportsTotag N'

os wn. 2o NCHES AnovE TUst SHEri LOWER WR LEVEL TAP e 140 7 NCHES 22 8 NCHES AAOWE TVSE SHEET

'(~

q3 i

g -- BLOWOOWN NOZZLE 13 NCHES ABOVE TURE SHEET l

n i

]

TusE SNEET ri S NCHES aaNELSA s

l NOTLEo COto leo

_//

\\_

u UNIT 2 STEAM GENERATOR ARRANGEMENT

Attachment to OCAN079606 Page 4 of 45 3.0 EXPANSION TRANSITION CIRCUMFERENTIAL CRACKING Steam generator tubes with identified circumferential stress corrosion cracks are currently repaired on detection without regard to the size of the flaw. New flaws may develop over the operating period or flaws may become evident after they were inadvertently left in service due to non-destructive examination (NDE) detection limitations. Flaws that are present during operation could potentially degrade the required safety function of the steam generators. Continued operation of the SGs for the planned operating cycle is justified provided adequate structural integrity and acceptable leakage levels are maintained.

Circumferential cracking was first experienced at ANO-2 in March 1992. The cracking was discovered as a result of primary-to-secondary leakage from a tube in the "A" SG. The first inspection for circumferential cracking was performed during this forced outage. Since that time, five additional comprehensive examinations for circumferential cracking (during three refueling and two planned inspection outages) have been conducted. Because of their location and size, the circumferential cracks in the ANO-2 steam generators have been difficult to detect. The cracks are located at the tup of tubesheet, in the expansion transition region (ETR) of the tube, with sludge containing copper surrounding the tube. The damage consists ofintermittent cracks on multiple planes with ligaments between the cracks. These ligaments contribute significantly to the strength of the tube.

2R1I was the sixth inspection of the ETR with a rotating probe. The average size (both length and depth) of the cracks has generally decreased in successive outages, as shown in Table 3.1 and Figure 3.1. The increased number of circumferential cracks is particularly attributed to advances in eddy current technology. Based on studies performed by Entergy and pulled tubes from other utilities, the threshold of detection of the Plus Point coil for these type flaws is clearly lower than that of the pancake coil. This is also evident from reviewing the distribution of detected flaws, shown in Figure 3.2, which displays the fitted gamma distribution for each outage. Using the Plus Point coil results in the detection of smaller indications which would have previously gone undiscovered. The smaller average flaw size detected at 2R11 is most likely attributed to the increased sensitivity and lower threshold of detection of the Plus Point probe.

For the purpose of this operational assessment, adequate structural integrity of steam generator tubes is evaluated by both deterministic and probabilistic methods. The deterministic approach follows that outlined in Regulatory Guide (RG) 1.121. The probabilistic approach is based on an acceptable probability of a tube rupture under the conditions of a main steam line break (MSLB) and given the projected flaw distribution at the end of the proposed operating period. An acceptable cond;tional probability of tube rupture, as specified in Generic Letter 95-05, is considered to be lx10 2 Leakage integrity was also evaluated for the proposed operating interval. Operational leakage is governed by administrative shutdown limits (150 GPD and >60 GPD/hr) to ensure a defense-in-depth means for reducing the likelihood of tube rupture. Accident leakage is defined as the primary to secondary leakage which could potentially occur during the postulated MSLB given the projected distribution of flaws present at the end of the proposed operating period.

Attachment to OCAN079606 Page 5 of 45 Acceptable leakage under these conditions is that which would not result in a dose exceeding the requirements of 10CFR100 and General Design Criterion 19.

Figure 3.3 shows the average amplitude, measured in volts, of the circumferential cracks detected each outage. In general, the average amplitude values have remained the same since the ninth refueling outage. The high average value in 2F92 is indicative of the " mature" cracks in service.

Since tubesheet inspections had not been previously performed, the cracks essentially grew to thc point of microcracks joining together to form larger cracks. These larger cracks produce higher amplitude signals, which makes them easier to detect. Table 3.1 provides the historical data and illustrates the changes in technology and the corresponding number ofindications-The average percent throughwall (% TW or average depth) is calculated based on the maximum depth and arc length. This average depth can also be viewed as the " crack area," which is indicative of the structural limit of the flaw. This has been demonstrated by numerous burst tests in the industry and as detailed in a report submitted to the NRC titled " Repair Limit for Circumferential Cracks in Steam Generator Tubing," dated August 28,1995 (2CAN089507)*

Calculation of average depth has typically yielded an overconservative picture of the flaw, and has led to false conclusions concerning the structural capability of certain cracks. An alternative approach, called percent degraded area (PDA) allows for " profiling" of the crack area (i.e., depth measurements made at =10 increments) which is a more accurate assessment of the actual crack dimensions and more comparable to metallurgical results. During the 2R11 examination, the largest crack was determined to be 73% degraded area. In terms of structural adequacy to meet RG 1.121, based on a limiting differential pressure of 4050 psig (3AP), a 79% degraded area would maintain the required safety margin. This conservative approach takes no credit for partial depth crack extent or ligaments which are expected to be present based on typical stress corrosion crack morphology.

Based on documented burst testing data, all 738 circumferential cracks in 2Rll would be expected to exhibit burst pressures significantly above that required by RG 1.121, with most being at or near that of a virgin tube. Further discussion on structural adequacy is presented in Section 3.1.

To further assess the structural significance of the flaws identified in 2Rll, in-situ pressure tests were conducted. Flaws were easily selected from eddy current testing (ECT) data based on PDA, length, maximum depth, and voltage. Based on these criteria, the three largest and the fifth largest ET region circumferential cracks were chosen for testing (fourth largest could not be f

reached from the same manipulator position). The pressure test device utilizes two expandable bladders approximately 3.5" apan, and dlows the chamber between the bladders to be pressurized i

up to 7000 psig. The device is designed to allow movement of the region between the bladders such that the axial load that would be applied to the U-bend of the tube during normal operation I

and postulated accident conditions is simulated for circumferential cracking. The results are summarized below:

i 1

~

Attachment to 0CAN079606 Page 6 of45 TUBE Max. Pressure (osin)

Leah oe 77-83

=6800 None 68-106

=6800 None 87-67

=6800 None 41-65

=6800 None Based on peak test pressure, all tubes met the requirement of 3AP (4050 psig). It has historically been relatively easy to distinguish the largest cracks at ANO-2. These larger cracks exhibit larger amplitudes and clear signals for the entire length of the crack, which is typically 360 During 2R11, the flaws'were generally smaller than had been experienced in previous outages. All of the 2R11 cracks are clearly bounded by the previous flaws and the in-situ pressure tests performed during those outages.

3.1 STRUCTURAL REQUIREMENTS The limiting RG 1.121 structural requirement relative to the significance of corrosion degradation in steam generator tubing is maintaining an end of cycle (EOC) burst pressure greater than three times the normal operating pressure differential across the tube wall. The required EOC burst pressure for ANO-2 is 4050 psig.

For circumferential cracking, the burst pressure is a function of the degraded cross-sectional area of the tube wall and the tensile properties of the tubing. Figure 3.4 shows a plot of burst pressure versus degraded area for pulled tubes and laboratory crack samples, demonstrating the correlation between the two. The lower bound curves (shown on the graph) are used for deterministic structural evaluations.

A review of ECT data was performed to evaluate crack growth. Determining growth rates for circumferential cracks is difficult since all detected flaws are repaired.

One must rely on indications not being detected in previous outages to obtain growth information. Re-analysis of previous outage data using the PDA approach shows flaw-like signals existing for some of the indications. From these indications, growth rates can be determined, as shown in Figure 3.5 for ANO-2. This was a conservative evaluation since the data set includes most of the larger flaws detected, and thus is representative of the largest growth rates. Only those indications which had good signal characteristics for both outages were chosen. Otherwise, assumptions concerning the beginning flaw size would have to be made, and growth rates would be predominantly determined by the assumed starting size. By using only those with two clear signals, the changes detected represent growth and/or ECT uncertainty. In an attempt to compensate for ECT uncertainty, all negative values were set to zero. This provides a more realistic evaluation of the flaw growth.

As a result, the average growth rate was determined to be 0.58% degraded area per effective full power month (EFPM), with the 95% upper bound being 2.47 % degraded area per EFPM.

Based on this information, a deterministic RG 1.121 analysis can be performed, as shown graphically in Figure 3.6, where the assumed beginning-of-cycle (BOC) PDA is 30%. A BOC PDA for 30% is justifiable for a 0.080" Pancake coil based on recent work performed

• Since

o Attachment to OCAN079606 Page 7 of 45 ANO-2 used a Plus Point coil, using 30% is conservative because the Plus Point provides a lower threshold of detection. Two crack growth rates, an average and a upper bound, are used to assess the expected conditions at EOC. The various limits (based on 95% lower bound tubing properties) are shown on the graph, with the lowest being a 95% lower bound from the burst curve (Figure 3.4). The 3AP limit from that curve is =71%. Even using the upper bound growth rate value, ANO-2 can safely operate for a full cycle.

3.2 PROBABILISTIC MODEL A probabilistic model has historically been used by ANOW to evaluate circumferential cracking damage and assess the operating intervals to ensure safe operation. The conditional probability of a tube bursting is determined by combining the probabilities of burst for specified extents of tube degradation when subjected to a differential pressure with the predicted population of degraded tubes for a specified operating interval.

An improved approach for calculating conditional probability of a steam generator tube rupture (SGTR) was developed in Reference 4. The improved approach takes advantage of additional burst test data, the ANO-2 burst correlation for circumferential cracks at the top of the tubesheet (TTS) and a model of the deviation of measured burst pressures from those predicted by the correlation. The method was applied to the predicted EOC 12 distribution of circumferential cracks to determine the EOC conditional burst probability Results from this analysis indicate that the conditional probability of one or more bursts is less than the 102 acceptance criterion of Generic Letter 95-05 and supports continued operation for the specified steam generator inspection interval.

The EOC population of degraded tubes is determined by several factors related to the predicted numbers of cracks at the end of the operating cycle, detection capability of the NDE inspection, defect sizing capability and the amount of predicted growth during the proposed inspection interval. Extent of tube wall degradation is measured in PDA.

The primary elements of the approach are:

EOC crack population e

The EOC crack population is described by two quantities: the total number of cracks of all sizes (detected + undetected), and the statistical distribution of crack sizes. A Weibull probability distribution for the incidence of cracks is used to relate the fraction of the total population of tubes with circumferential cracks to total operating time measured from an assumed " time zero." The number of tubes with cracks of each size at the end of a projected cycle is determined from the product of the total number of tubes, the predicted increase in the fraction of tubes with cracks and the probability of a crack of a given size.

4 Attachment to 0CAN079606 Page 8 of 45 For projections of the number of cracks in future inspections, probability of detection (POD) effects were incorporated to ensure consistency of data across inspections.

The basics of the method are that the observed distribution of crack sizes can be used to determine the "best-estimate" distributions by accounting for the effectiveness of NDE. A " weighted POD" can then be computed for the underlying crack distribution.

This " weighted POD" can be used to obtain best-estimates of the number of cracks before and after inspection and repair, After repair of cracked tubes the distribution of remaining undetected (mostly small) cracks can be computed. By using tmly best estimates it is expected that more accurate projections can be performed. When combined with the observed ANO depth and length distributions (2F92 through 2P95), these provide a weighted POD of 60% which is consistent with previous assumptions.

Burst correlation and model of" scatter" l

Burst testing of tube sections either removed from operating steam generators or produced in laboratory conditions (both cracks and notches) provides a correlation l

between the degraded area of the flaw and the pressure at which failure occurs.

Corrections for temperature and material properties were performed and the subsequent correlation developed. The difference between the " scatter" of the burst l

pressure test results used to define the correlation and the predicted average burst l

pressure obtained by applying the correlation to a flaw with a specific PDA is termed L

the residual variation of the data. Due to this observed variation of the burst data from l

the correlation line, a statistical model was used to represent the burst pressure j

residual variation of the data. This distribution is slightly skew to values below zero, while accommodating the long tail to high positive values which is characteristic of burst pressure test results. It also provides a means to accommodate occasional specimens which burst at or near virgin tube strength, even though the circumferential l

crack may have significant degraded area. Good agreement was indicated between the L

data and the model.

l.

Probability ofburst The ANO-2 burst correlation predicts the average value of burst pressure for a specific PDA. The probability of burst when subjecting a single tube with a circumferential crack of a given area to the maximum credible differential pressure is the conditional probability of burst for such a defect.

The conditional probability of burst is significantly influenced by the most extreme (largest PDA) flaws in the " tail" of the EOC crack population. This observation provides a strong basis for performing the flaw size " profiling" in 10 increments which more accurately assesses the size of these extremely influential data points.

4

=-

Attachment to -

OCAN079606 Page 9 of45 l

Maximum credible differential pressure between primary and secondary systems i

The maximum credible differential pressure is 2500 psi based on a primary pressure of j

2500 psi, and assuming the secondary side is at atmospheric pressure. This a very conservative assumption given the realistic nuclear steam supply system (NSSS) response to a MSLB.

The total number of circumferential cracks detected by Plus Point in 2Rll were 738. Adjusting for POD and projecting to EOC 12 raises this number to 935 for the limiting SG, which is ud for the analysis. Fragility curves were developed to estimate the tendency for a tube with a specific amount of degradation to burst when subjected to the differential pressure. There is a rapid increase in the conditional probability of burst for circumferential cracks with PDA values in excess of 75%. However, below 75%, there is a negligible probability of burst for an individual crack. The total conditional burst probability accounts for the contribution from all cracks which exist at EOC. Based on this data, the simulation model resulted in a probability of burst given a postulated MSLB at EOC (after 16 EFPM) of 6.9x10. This value is well within the acceptance 4

2 criterion of lx10 specified in Generic Letter 95-05, and thus full cycle operation is justifiable.

This value is combined with that obtained for other damage mechanisms, as described in Section 4.2, to evaluate the overall safe operating interval.

3.3 ACCIDENT LEAKAGE EVALUATION Operating experience has demonstrated that circumferential cracks, even those with large PDAs and/or arc lengths, do not leak significantly under normal operating conditions. This is most likely I

due to the presence ofligaments, which cannot be detected by NDE, and the tendency for cracks l

to form on multiple axial planes rather than on a single plane. Leakage under ncrmal operating conditions is limited by technical specification requirements and typically does not represent the -

basis for determining operating intervals. Potential leakage during a postulated MSLB scenario is the primary leakage consideration justifying operating lengths.

The leak rate from a circumferential crack is given by the same formula as for an axial crack

• Q = CKA 4Pejr ip Where:

Q = leak rate at operating temperature (GPM)

C = unit conversion constant l

K = flow discharge coefficient l

A = crack leakage area (in )

2 l

Per = effective differential pressure (psi) l p = fluid density (Ib/m')

l f

Attachment to 0CAN079606 Page 10 of 45 The flow discharge coeflicient, K, is an empirical adjustment to account for surface roughness and the tortuous flow path associated with stress corrosion cracks.

The effective differential pressure is the apparent pressure drop across the tube wall through the crack. This pressure drop is less than the full primary-to-secondary differential pressure because the primary water flashes to steam as the pressure decreases to the saturation pressure for the primary temperature. This flashing of the primary iluid effectively restricts the fluid flow to that which would occur if the secondary pressure were equivalent to the saturation pressure of the primary fluid at operating temperature.

A model for leakage through circumferential cracks has been developed using PDA.

The circumferential crack leakage area is modeled as a ftmetion of the PDA and throughwall arc length. Lateral support of the tube is considered in determining the crack leakage area (i.e., free bending is precluded). For the purposes of calculating an arc length, NDE indications greater than 90% are considered to be throughwall.

The model determines a leakage area for the specified crack size and steam generator conditions.

The leakage area is a function of the burst strength of the tube and the geometry of the flaw. The burst strength of the tube is calculated using analytic equations or an empirically developed correlation based on Combustion Engineering (CE) SG tube burst data. Given the leakage area, the primary to secondary flow is then calculated using the flow discharge coeflicient and the effective differential pressure across the tube wall.

The potential leakage under normal operation and potential accident conditions was calculated for the end of the current operating interval. The calculated values are based on the 2Rll flaw distribution, adjusted for potential growth of flaws remaining in service due to POD and the best estimate projected number of flaws for the limiting steam generator. Ninety five percent lower bound material properties adjusted to operating temperature are used. Due to the small number of cracks expected at EOC 12 and smaller size distribution, the contribution ofleakage from the cold leg is negligible when compared with the hot leg. Leakage under accident conditions was l

determined to be 7.44 gpm (95/95 upper bound) using a Monte Carlo simulation for the best estimate proportion throughwall. These numbers provide acceptable dose results, which are discussed in Section 5.

The calculated potential leakage remains in an acceptable range even using the conservative sizing assumptions. The majority of the leakage is contributed by a few j

tubes with larger flaws.

i l

l l

J U

i Attachment to OCAN079606 l

Page11of45 Table 3.1 l

Qutagg

1. Cracks Avg. Length. Deg.

Avg % TW Detection Probe and Coil 2F92 (3/92) 469 115 24.0 0.080 rotating pancake coil l

(RPC) l l

2R9 (9/92) 25 90 18.4 0.080 RPC 2P93 (5/93) 48 89 17.2 0.080 RPC l

2R10 (3/94) 170 82 16.1 0.115 RPC, low loss cables 2P95 (1/95) 283 85 17.0 0.I15 RPC, MIZ-30, low loss cables, terrain maps utilized l

2R11 (9/95) 702

• 69 11.6 Plus Point, MIZ-30, App. H settings, low loss cables, terrain maps utilized, high frequency Plus Point coil used for enhanced characterization Only includes the hot leg circumferential cracks. Total number including the cold leg is 738. Inclusion of cold leg cracks would reduce average % TW and average length values.

l J

4

4 Attachment to OCAN079606 Page 12 of 45 FIGURE 3.1 Circumferential Crack Data 800 120.00 700--

-- 100.00 l

600 --

3 500 --

- 80,00 4

8 eo b,

.j 400 --

-- 60.00 a

t

• 300 --

[

-- 40.00 4

200 --

[

l 100 --

N 1

-- 20.00 I

I 0

0,00 2F92 2R9 2P93 2R10 2P95 2Rll O u tage l

1. of Cire Cracks

-*- Avg Length

-+- A vg % T W l

1 FIGURE 3.2 Circumferential Crack Ganuna Distribution 2F92

_]

...... 2R9

_.._.2P93

~

_. --- 2R 10 3

- -- - 2 P95 8

s 2R11

~

f.sg#.?y

' o

..')q

_;11, 1

7!f I

i i

j 0

10 20 30 40 50 60 70 80 90 100 Crack Average Depth, %TW

Attachment to OCAN079606 Page 13 of 45 FIGURE 3.3 Circumferential Crack Average Peak Amplitude 6.00 4

\\

5.00 5 4.00 l

9 c

] 3.00

.nt a.

E R 2 00 1.00 l

l l

0.00 2F92 2R9 2P93 2R10 2P95 2Rll Outage FIGURE 3,4 '

US Steam Generator Tube Burst Data For Circumferential Cracks Circumferential Burst Data 12000 I

i j

i i

t i

f t

- {~-

--I---

-h-~

- A - - - k

- - I - - -V

-- b -- N -

11000 -

-_.-..._._!...-._._.l.._._...}..==================..=..-l

_-....j....._.-...

.d. -

l f

(

3g000 _

}

I

--~~-~--y}--

- - -~

--e---

e

$ 9000 -

-- [! ---- -f - -- f ---

8000 - ----- d-- - -d - -

-l--

i e

i 7000 -

^ -

-+

+

- + - -

-a

(

l 3e t

y 6000 -

f----+---

+ ~ - -- '

--}

8D I

l l

I l

ei l

7 5000 - --- '

D e'

---

i-~ ~ - -

5-

- + - -

4-

• i ec 6

[ 4000 - =

e e e a.t=

.a.o o

-a..o 4

y l

3000 -

-.7, NZ.,,,,...

---

f-

~j--

-+-

2000 -

-***=***********"*"a*'*'-'

d-

-- -~ -

t "l

-l-

~;' --j j-- ~

4 1000 -

- - - - + - ~ ~ f-1-

~+ --

[

i i

0 i

i 0

10 20 30 40 50 60 70 80 90 100 Percent Degraded Area (PDA)

l l

l-Attachment to 0CAN079606 Page 14 of 45 FIGURE 3.5 Circumferential Crack Growth Rate p

4.00 A

o 3.00

^

o 2.00 w

h o

g 1.00 g,

I

{

0.00

,m-o--m, "

d' e

-1.00 4

o g

-2.00 3

-3.00

-4.00 i

O 10 20 30 40 50 Initial Crack % Degraded Area j

o 2P93-2R 10 m 2R10-2P95 i

FIGURE 3.6 Circumferential Crack Deterministic Evaluation 100 yop l

90 --

Msta i

. 1.4 MStB i

80 --

30p f 70 --

95% ta 3

~~

Upper BotalGrowth e 50 --

2 m 49 __

b 30 -

Av& orath

=

s.

oN 20 --

ANG2 EOC ->

10 --

0 0

2 4

6 8

10 12 14 16 18 EFPM

i Attachment to OCAN07%06 Page 15 of45 4.0 FREESPAN/EGGCRATE TUBE SUPPORT PLATE AXIAL CRACKING l

Ennerate Crackinn Cracking at eggerates was first detected at ANO-2 in 1991. Based on the destructive examii2 tion l

results of tubes removed during 2F92*, the eggerate suppon indications are classified as axially oriented stress corrosion cracking.

The cracking can be single cracks or multiple cracks l

l interconnected in the tube within the eggerate support. The distribution of all axial eggerate cracks 20% TW and greater detected at ANO-2 is shown in Figure 4.1 for the "A" SG and Figure l

4.2 for the "B" SG.

I Since 1991, there have been 929 reportable eggerate indications in both SGs. Of those, 94%

l were one volt or less, 99.5% were less than two volts, and all but two were 2.4 volts or less.

Those two were a 5.7 volt indication discovered in 1992 in the "A" SG in a tube that had not been l

inspected for several inspections, and the 12.2 volt indication found in the "B" SG in 2R11. A L

plot of the 929 indications is shown in Figure 4.3.

In 2R11, the average size of all eggerate indications 20% TW and greater was 0.56 volts. This voltage reading is comparable to those for which alternate repair criteria were granted for l

l cracking at drilled tube support plates in Westinghouse designed plants. A comparison of burst pressures versus voltage from the Westinghouse data have been made and assessments of structural significance for amplitude values at ANO-2 have been performed. These comparisons show that eggerate cracks at ANO-2 have historically been structurally insignificant. In addition, burst testing of eggerate cracks of tubes removed in 1992 from ANO-2 resulted in burst pressures in excess of 8000 psig, which reasonably agrees with the Westinghouse plant data.

1 i

One eggerate flaw in the "B" SG was found by bobbin and appeared to be an exception based on its size. The flaw was sized by bobbin at 12.2 volts and 86% TW. This flaw was analyzed using the Plus Point coil and was sized to be'90% TW. Average voltages suggest this was an isolated

. case. All other flaws exhibited expected results. During the previous refueling outage,'this indication was sized at 1 volt and 22% TW. An in-situ test of this flaw was performed at 3 i

different pressures as shown below:

1 Leakage Step One

=1650 psig (AP)

=0.005 GPM l

Step Two

=2950 psig (Peak accident pressure)

=0.4 GPM Step Three

=1650 psig (AP)

=0.32 GPM t

Step Four

=4700 psig (3AP)

N/A

{

__7._.____

l i

Attachment to

)

OCAN079606 Page 16 of 45 The leakage values shown above are as-measured and have not been compensated for temperature. When temperature compensated, the leakage would decrease, thus indicating that the measured values are conservauve.

The tube was first pressurized to 1650 psig and 'a leakrate was observed that was comparable to that seen prior to shutdown. The pressure was then raised to 2950 psig where the leakrate increased to 0.40 GPM. A bladder was then placed over the; flaw and pressurized up to 4700 psig. A visual examination following the pressure test was performed which revealed that two parallel axi:1 cracks. Based on its peak pressure, this tube m,et the 3AP (4050 psig) guidance of RG 1.121.

Freesnan Crackine During the 2R11 bobbin coil inspection, several indications were identified as non<luantifiable indications (NQIs) and were subsequently==iaad using the:Plus Point probe. Two of these were located in rows 1 and 2 and were determined to be axial crack-like indications found in the free-span region below the U-bend tangent (not previously identified at ANO-2) and are bounded

.in elevation by the 4th through 7th eggerates on the hot leg side. Because of concems from the experience at Palo Verde about the ability of the bobbin coil to adequately detect free-span cracks, further investigation was warranted, and resulted in expanding the scope of the Plus Point examination to include 100% of the tubes in rows 1-3 from the stay cylinder out to line 167, with an additional 100 tube random inspection of rows 4-20 in the same area. Any tubes exhibiting free-span bobbin indications greater than 0% TW in the "A" SG were examined. These tubes (83) were distributed throughout the bundle at various elevations. Additionally, the bobbin data for rows 1-20 was re-evaluated with the objective ofidentifying any potential flaws. This re-evaluation yielded approximately 300 tubes with shallow indications on the low frequency channels. A total of 510 tubes were analyzed identifying 16 tuba with 18 indications (9 which were previously identified by bobbin, and 9 detected during the random inspection). A re-evaluation of the bobbin data for these 9 detected by the random inspection showed them to be low amplitude (0.08-0.65 volts) with a maximum TW value of 17 %. The largest of the 18 flaws was one of the original two identified as a NQI and measwed 1.9 volts and was approxinctely

,1.85 inches in length.

The tubes not originally flagged by bobbin are considered to be inconsequential indications with percent throughwall depths much less than the repair limit.

Overall, the bobbin examination technique was judged to be highly effective in the detection of free-span indications based on plant-specific studies at ANO-2 because all degraded tubes with detectable rotating probe indications also exhibited' detectable bobbin indications.

I

~

A full length in-situ pressure test of the tube containing the largest free-span indication wa*

performed. The tube was tested up to 4750 psi without burst or leakage. : Based on the peak pressure, this tube met the 3AP (4050 psig) guidance of RG 1.121. Because this was the largest indication, based on both bobbin and RPC/Plus Point, the pressure test bounded all of the other indications.

Attachment ts OCAN079606 Page 17 of 45 4.1 STRUCTURAL REQUIREMENTS A structural evaluation of axial cracking, both at eggerate and freespan regions, was performed.

As previously stated, the RG 1.121 structural requirement for ANO-2 is 4050 psig. Since the degradation mode ofinterest is axial outside diameter (OD) stress corrosion cracking (SCC), the burst pressure is a function of the degradation length, average depth and tensile properties of the tubing. Figure 4.5 shows a plot of calculated versus measured burst pressures for pulled tubes from Plant Am Calculated burst pressures are based on the Framatone equation".

d -

b p _ 0.58(oyp + outs)t i

R1 L + 2t where burst pressure P

=

o, yield strength

=

y tensile strength

=

tube wall thickness t

=

inner radius of tube R4

=

degradation length L

=

average degradation depth d

=

This equation gives conservative predictions of burst pressure when the average crack depth is based on observed crack depths over the total crack length. If the average crack depth is computed over the structurally significant crack length, then essentially no under-conservative calculated burst pressures are observed and there is no associated large penalty of over conservatism. If the crack depth versus length profile is known, then the structurally significant crack length ~is determined by ' repeated sampling of the total crack pro 5le to Snd the crack segment which dominates the burst pressure *.

Figure 4.6 shows that structurally significant crack lengths correlate with degradation length as indicated by a rotating pancake eddy current probe. The RPC crack length is either equal to or greater than the structurally significant crack length.

l If crr.ck depths are averaged over the structurally significant crack length, then tlm ratio of maximum crack depth to structurally signi6 cant crack depth is on the order of 4/x. That is, the structurally significant portion of a seemingly irregular crack profile can be idealized as a semi-ellipse. Figure 4.7 shows a plot of maximum crack depth versus average crack depth from pulled tube data of Plant A. This figure illustrates that the appropriate average crack depth for burst pressure calculations can be obtained from a measured or estimated maximum crack depth.

Figure 4.4 points to the success of this technique.

t

Attachment to OCAN079606 i

Page 18 of 45 With the above information and knowledge of lower bound tubing mechanical properties, a deterministic RG 1.121 analysis can be summarized in graphical form. Figure 4.8 plots maximum crack depth versus EFPM. The upper horizontal lines denote maximum depth RG 1.121 structural limits for various assumed crack lengths. Note that these limits are based on 95% lower bound tensile propenies. The length limit for a throughwall crack, fbr example, is 0.52 inches.

t Similarly, the figure identifies other combinations oflength and depth that would theoretically challenge RG limits. The assumed BOC maximum crack depth.is 50% of the tube wall thickness.

Two crack growth rates illustrate an average value and an upper bound value. The basis for those growth rates i described in Section 4.2.1. For the average crack growth rate, structural margins are easily met for a 16 EFPM run time. Note that the characieristics of flaws seen at ANO-2 are such that the population of flaws expected to be seen at EOd do not challenge the limits identified in Figure 4.8. The deterministic RG 1.121 analysis supports fbli cycle operation and is consistent with the past performance of ANO-2. In-situ pressure testing of ANO-2 flaws during 2R11 consistently support RG 1.121 compliance.

One measure of the past performance of ANO-2 is EOC bobbin coil signal amplitudes. Figures 4.1 and 4.2 summarize the distribution of bobbin coil voltages for 2R11 indications at eggcrate support locations. In terms of the bobbin voltage versus burst pressure correlation for 0.75 inch diameter tubing with ODSCC/intergranular attack (IGA), RG 1.121 structural margins begin to be challenged near six volts". All but one indication is far removed firem this level. In-situ pressure testing of the outlier indication demonstrated that EOC structural requirements were met.

The deterministic inputs to Figure 4.8 can be changed to selections from known distntutions of tensile properties and crack growth rates.

Crack lengths can be selected from measured distributions in similar plants. Finally, BOC crack depths can be input from a model which follows crack initiation, crack growth, and then periodic inspections with defined probabilities of detecting cracks Undetected cracks then remain in service and form the BOC population. A model of this type has been constructed'and is described in Section 4.2.

Freespan degradation under deposits observed between the fourth and seventh eggcrate supports is combined with ODSCC/ IGA at the support plates. Based on pulled tube examinations at Plant A, degradation morphology is. virtually indistinp h hle between freespan, batwing, and eggerate locations. Structural requirements are the same and crack length distributions need not be m

adjusted.

While bobbin probe indications are concentrated t the lower eggerate supports in the "B" SG, eggerate indications in the "A" SG are uniform as a function of elevation. This point is illustrated by the eggerate indication maps ofFigures 4.9 and 4'.10. The observation of freespan indications in the "A" SG, but not in the "B" SG, is consistent with numerous upper bundle eggerate indications in the"A" SG.

l

j I

Attachment to OCAN079606 j

Page 19 of 45 FIGURE 4.1 i

"A" SG Eggerate Amplitude Distribution 70 100 %

- 90%

-- 80%

50

-- 70%

g 8

3 40 --

-6%

.g

.Q

-- 50%

5 1

1 e

3

.5 30 --

- 40%

i U

20 -_

-30%

- 20%

10 -

0 l

l l

l l

lE:

0%

i 0

0.4 0.8 1.2 1.6 2

2.4 2.8 4

Amplitude, volts FIGURE 4.2 "B" SG Eggerate Crack Amplitude Distribution 70 100%

~

60 --

- - 80%

50 -

-- 70%

8

-- 60%

3 40 __

y

.h

-- 50% }

.5 30 --

- 40% $

=

y 20 --

One beyond 30 %

3 volts 10 --

12.2 -

-- 20%

-- 10%

j 0

l l

l l"l 0%

0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 Amplitude, volts

Attachment to OCAN079606 Page 20 of 45 FIGURE 4.3 Eggerate Indications >20% (1991-1995) 400 1PJ%

350 -

- 90%

300 -

- 70%

250 -

(

g

- 60%

200 50 %

b

• 150 2 more beyond 30*4 100 -

3 vo ts-->

20*6 (5.7,12.2)

)

50 -

10%

0 0%

0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 Amplitude, volts i

l FIGURE 4.4 l

Historical Comparison of Eggerate Indications I

250 1.00 1

-- 0.90 200 0.80

--0.70 o

150 0.60 8

-- 0.50 o h100 l

0.40 y<

--0.30 50 0.20 j

-- 0.10 0-1 1

- 0.00

}

2R9 2R10 2R11 1

j Outage i

i m "A" #Ind i

J "B"#Ind

-*- A" Avg Voks

-e "B" Avg Vnits

(

l

o i

I Attachment to OCAN079606 Page 21 of 45 Figure 4.5 Predicted Burst Pressure Versus Normalized Observed Burst Pressure, Plant A Pulled Tube Data 12000 10000 5

8000 g

5 i

o e

=

g

=

6000 8

$m i

Q o

g G

o 5

E 4000.

o

. DEPTH AVERAGED OVER WHOLE J

LENGTH i

2M

, STRUCTURALLY SIGNIFICANT DEFrH 0

0 2000 4000 (1000 8000 10000 12000 NORMALIZED OBSERVED BURST PRESSURE, PSI

Attichment to 0CAN079606 Page 22 of 45 Figure 4.6 RPC Crack Length Versus

~ Structurally Significant Crack Length 2.50 l

2.00 --

l i

e

]

(4y 1.50 --

O E

(

C 5a M

1.00 --

2

• +

t

=

l O

e 0.50 --

l i

+

4 d

0.00 0.00 0.50 1.00 1.50 2.00 2.50 STRUCTURALLY SIGNIFICANT CRACK LENGTH, INCHES

i 1

Attachment to OCAN079606 Page 23 of45 Figure 4.7 Maximum Crack Depth Versus Average Crack Depth 1

o 0.9 - -

a o

o 0.8 -

e o

a 0.7 - -

J<g E

o Ie g

0.6 --

,g 8

O a

o E o z

eo O

0.5 --

U a

4 5

o E

o i

g 0.4 --

no wC 232 E

0.3 --

<2 0.2 - -

o STRUCTURAL AVERAGE DEPTH

/

0.1 --

u WHOLE LENGTH AVERAGE DEPTH l

i 0

O 0.1 0.2 0.3 0.4 0.5 0.0 0.7 0.8 0.9 1

AVERAGE DEPTHS, F:MCTION THROUGH WALL I

Attachment to OCAN079606 Page 24 of 45 Figure 4.8 Summary ofDeterministic RG 1.121 Analysis LIMITING RG 1.121 THROUGH WALL CRACK LENGTH = 0.52 INCHES 100 1.0" CRACK LENGTH 1.5" CRACK LENGTH 90

/

2.0" CRACK LENGTH I

I

/

v

/

80 2.5" CRACK LENGTH J

d 3.0" CRACK LENGTH 3

70 2.2%/EFPM 8

8 E

60 H

0.5%/EFPM i

7 50 8

5 g

40 ASSUMED BOC O

MAXIMUM DEPTH ig 30 20 10 GYP + UUTS = 124.367 psi AT TEMPERATURE O

O 2

4 6

8 10 12 14

~~6 18 1

EFFECTIVE FULL POWER MONTHS

Attachmtnt to 0CAN079606 Page 25 of 45 Figure 4.9 l

Steam Generator A 2Rll Eggerate Indication Maps i

140 l

a OlH

• o l

• 02H 120 x

A 03H

$*A e04H o

100 x

o05H o

x i

o 9

o06H

%o l

a x

to m

o a07H

• o 80 xo o Q,

,P I*

%*a o 08H x

1. . * *U

^

x 09H 60 -

o, l

e a

a 8 A e

A a A*,

l 40 e

a g

gp 3

8 sm

.s m

A l

\\

a m

0 m

,tp a

,,4*

aesg 0

0 20 40 60 80 100 120 140 160 Line 50 l

45 40 35 E_

$ 30

-l l--

n h 25 -

h 20 EE 15 -

10 i

i ;"; i"i";"i ;";";

i 0

i ;;i I

z I

m Z

x N

v o

p o

o O

O kkO c

E E

E E

E E

E E

8 Eggerate Location

l,'

Attachment to 0CAN079606 Page 26 cf 45 Figure 4.10 Steam Generator B 2Rll Eggcrate Indication Maps 140 m O1H

+

120 -

'8 m03H

(

m e 04H 00 -

e 05H Og#

• a*+

o o06H l

80 -

8,a

+ m,

, m' A

A07H l

is a

60 -

• a a

==

l m

o, e

A eo a

S 40 -

a u ma t

a e

a m

m B* {

" "o "a

m a

m*

g %,,em %"

20 -

a a

a a a

no e,

o8 m

e 0

0 20 40 60 80 100 120 140 160 180 Line 100 90 - I 80 -

70 60 --

a

.E 50 -

o f

40 --

r l

30 20 _

_E 10 -I l lI"l lh O

;m

=; ;

ssss=ssgsaaaae o

o o

o 8

o o

o o

o m

m j

Eggerate Location e

i

1 Attachment to OCAN079606 Page 27 of 45 4.2 PROBABILISTIC MODEL The probabilistic run time model consists.of a Monte Carlo simulation of the processes of crack initiation, crack growth, eddy current inspection, and removal or repair of defective tubes which are detected. The input parameters to the model include tubing tensile properties, the ' distribution of degradation lengths, the probability of detection function for the bobbin coil, and the distribution of crack growth rates. These input parameters are described in the following paragraphs.

4.2.1 ModelInput Parameters Tubing Mechanical ProMrties The actual yield strength and ultimate : ensile strength values were available for the ANO-2 SG tubing. Room temperature test resuus were adjusted to account for the variation of flow strength with temperature. A normal distri5ution was fitted to this data and this provided the tube strength input to probabilistic calculaties.

Degr_adation Length Distribution e

The distribution of crack lengths judged to be the most appropriate for ANO-2, shown in Figure 4.11, is based on the second upper bundle RPC inspection at Plant A, where a 0.115 inch diameter pancake coil was used. This probe supplies the stmeturally significant crack length, as described earlier. About 25% of the crack lengths are greater than 1.0 inch. This provides a reasonable long crack effect, but not the grossly over-conservative shallow long crack contribution of the Plus Point probe.

Probability ofDetection Bobbin probe detection of eggerate cr'evice and freespan ODSCC/ IGA has been good. The tubing inner diameter surface condition and relative. lack of geometry changes undoubtedly contribute to this experience. The use of pilgered tubing in newer Combustion Engineering steam generators coupled with a smaller wall thickness (0.042" versus 0.048") has led to problems with the use of the bobbin probe in detecting upper bundle ODSCC/ IGA. Higher eddy cunent test frequencies are used in the thinner wall tubing in order to obtain an adequate phase angle spread. Variations in the geometry of the pilgered tubing give rise to noise signals whose relative emplitude is increased as a result of using higher test frequencies.

Hence, masking of bobbin probe signals attributable to ODSCC/ IGA in System 80 designs can lead to problems in bobbin probe detectability. At ANO-2, tubing geometry factors and higher eddy current test frequencies are not issues and adequate bobbin probe detectability is maintained. The few instances at ANO-2 where rotating probes have found freespan indications, but original bobbin probe evaluations were no detectable degradation (NDD), are limited to very shallow, low amplitude indications. Comparable detectability conditions exist within and outside of eggerate supports. Consideration of the number ofindications found

' Attachment to 0CAN079606 i

Page 28 of 45 within eggerate suppons leads to the conclusion that bobbin probe data does not seriously underestimate the number of significant freespan indications.

• Industry-wide data has been used in the past to compute the probability of detection versus maximum relative axial crack depth for RPC (Figure 4.12). Pulled tube data from Plant A leads to a comparable but somewhat less sensitive probability of detection (POD) curve.

h.

Analyses ofindustry-wide bobbin data led to a POD curve very similar to the Plant A RPC L

POD curve. This curve was shifted to larger relative crack depths to provide consistency i

wit' *he shallow freespan indications found at ANO-2 by RPC, but not called in the original be%

iata evaluation.

Degradation Growth Rates The development of crack growth rate data is a dominant element in a run time analysis.

Fonunately, over 360 pairs of bobbin probe phase angle data are available to construct a conservative, yet reasonable, crack growth rate distribution. Data from both SGs for the past two operating cycles has been combined. Individual crack growth rate data points are simply the difference in maximum depth estimates from bobbin probe phase angle data divided by the cycle length,16 EFPM.

Crack growth rate versus estimated BOC depth is plotted in Figure 4.13. Approximately half of the crack growth rates are negative due to measurement uncertainties. If only the positive values are considered, the average crack growth rate is 0.48% throughwall per EFPM. The maximum observed crack growth rate'is 2.3 %/EFPM. Given a difference in wall thickness, this average crack growth rate value is nearly identical with that derived for Plant A by a totally different procedure

• The observed maximum growth rates are consistent with previous statistical evaluations of expected extremes in growth rates relative to the average or median values.

Crack growth rates are shown in Figure 4.14 from laboratory tests in caustic concentrated environments, analysis ofPlant A data, and analysis of ANO-2 bobbin data. These~ rates span ODSCC and IGA morphologies.

Average and extreme rates correlate well.

Lower temperatures in ANO-2, compared to Plant A, indicate that lower growth rates might be expected.

The positive crack growth rates of Figure 4.13 were sampled directly in the probabilistic analysis. No distribution was fitted to the positive data points. Crack growth rates were allowed to vary from one cycle to the next. In the sampling process, negative growth rates were considered as zero. Thus in a given cycle, approximately half of the crack population was assigned zero growth.

+

i,

}

4 i,

' Attachment to OCAN079606 i

Page 29 of 45 4.2.2 Probabilistic ModelDescription l

In the probabilistic model, times to crack initiation are selected from a Weibull distribution.

The Weibull slope and scale factor are based on the past history of reponed indications as a a

function of operating time. Since reponed indications have grown sufficiently to be detected by an eddy current inspection, the actual point of crack initiation must be at some earlier i

time. A constant time shift provides an adequate estimate of the crack initiation time. The i

magnitude of the shift depends on the average crack growth rate.

j e

After a crack is initiated, it is considered to grow at a constant rate in the depth direction until the next inspection. As noted in the previous section, the assigned dependent crack growth rate is random by selection from the observed growth rates in Figure 4.12. Negative growth rates were taken as zero. If a crack remains undetected in the simulated inspection, a new growth rate is selected for the next cycle.

a j

Crack length and tensile propenies are assigned to a crack at initiation and are considered intrinsic propenies of the crack. The observed distributions of EOC structural significant s

crack lengths are reasonably constant. Even mid-cycle inspections have little effect on the length distribution. However, the number of cracks of a given length does depend on the l

cycle length.

The bobbin probe POD curve of Figure 4.12 is used to perform a simulated inspection. The process i.t straight forward. The crack depth at EOC is known from the crack initiation time, the total time available for growth, and the past selected crack. growth rates. A random a

number is selected between 0 and 1. If the random number is below the POD curve at the depth ofinterest, the crack is detected, otherwise, it is missed and remains in service.

- Upon detection, the length, depth, and tensile propenies associated with the crack are used to compute a burst pressure. This determines if a tube burst at MSLB pressure could occur or if RG 1.121 stmetural limits have been exceeded. Crack depths required for bursting at MSLB or 3AP are generally large enough to ensure detection.

Simulation of the overall process for the desired run time history is repeated 10,000 times to obtain reasonable estimates of the probability of a tube burst given a postulated MSLB at EOC and to develop a distribution function for the number of RG 1.121 structural limit exceedances.

One benchmark of the probabilistic model is a comparison of the predicted versus observed number of bobbin probe indications. Since the model is probabilistic, there is no single prediction of the number of bobbin indications at a given inspection.- However, some outcomes are more likely than others, and this is illustrated by the histograms of Figures 4.15 and 4.16. The actual number ofindications observed correlate with the most likely calculated number of indications projected for 2R9,10, and 11 inspections. Note the number of j

indications are per SG. Approximately 200 new indications per SG are expected at the end of cycle 12.

Attachment to 0CAN079606 Page 30 of 45 4.2.3 Structural Model Evaluation The parameters ofinterest in a probabilistic stmetural model evaluation are the conditional probability of tube burst given a postulated MSLB at EOC and the likelihood of multiple RG 1.121 structural limit exceedances. A reasonable figure of merit for conditional probability of a MSLB tube burst is provided by a historical value of 0.025. A value less than 0.01 provides a good benchmark of stmetural integrity., In terms of meeting RG 1.121 structural margins, a reasonable probability criterion is a goal of at least 90% probability that one or fewer tubes will be expected to violate structural limits at EOC.

The results of the Monte Carlo simulation model regarding the probability of a tube burst due to axial ODSCC/ IGA at eggerate support and upper bundle freespan locations given a postulated MSLB at EOC aner 16 EFPM is 0.0024. The probability of having one or less RG 1.121 structural limit violations was determined to be 0.94 after 16 EFPM.

Based on these results, a full 16 EFPM between bobbin inspections for ODSCC/ IGA at eggerate tube support plates and upper bundle freespan locations is fully supported.

l l

3 e

h l

1 Attachment to 0CAN079606 l

Page 31 of 45 Figure 4.11 Weibull Plot of RPC Crack Lengths in Plant A 99.9%

[ll 99.0 %

^

i 90 %

!I I

l l

[

l l

lll 80%

I I l ill ii l

l I

1 I

l l l Y

70 %

80%

l 1

'l l'

!L II d

Ii I

I I

I I

III 50 %

l l

ll l

l 4

l l l l'

.1-40%

30%

I I

/[

l l

l l

3 l

20%

Ay s

i 6

a 10%

/4 l

g o

5%

a A

/

t

/

~

/

A i

1%

0.5%

l lI 0.1%

.01 0.1 1.0 10.0 RPC (0.116* Col 4 CRACK LENGTH, INCHES l

l

Attachment to 0CAN079606 Page 32 of 45 Figure 4.12 Selected Probability ofDetection Curves for RPC and Bobbin Probes 1

0.9 RPCINDUSTRY 0.8 DATA 0.7 5

5 0.6 RPC PLANT A g

DATA Q

O 0.5 C

E E

BOBBIN PROBE 0.3' O.2 1

0.1 0

0 10 20 30 40 50 60 70 80 90 100 MAXIMUM RELATIVE DEPTH, % THROUGH WALL

\\

Attachment to 0CAN079606 Page 33 of 45 Figure 4.13 4

1 ANO-2 Crack Growth Rate From Bobbin Probe Versus BOC Crack Depth i

SG A + SG B EGGCRATE INDICATIONS 5

1 l

52R10 2R11 j

O2R9-2R10 4

{

3 i

j 3

3 1

E i

y d

o g2 o

o O

0 g

E O

OE O

g.E O.E,

E i

E E

E Be 0

5 E

E' ao E*

a""

aE 8

E

=

E E

5 E

8EE o o

i

,il o,',=o==s,sO:E uBOo s 5

0$0 hgEO h

_h o--

E 5

0 gjDER 55 0 l00 EjC OB Q 3 O

E00 O

O 0

E O

0 5lE E

flO OBE EO g

O E 0

g QlEg 5

O g

O lE O

O 0

E O

E E

O O

E O

O O

ME o

50 g

g

-1 O

O E

O 5

g O

O O

O O

O O

-2 0

5 10 15 20 25 30 35 40 INITIAL CRACK DEPTH, % THROUGH WAU.

Attachment to OCAN079606 Page 34 of 45 '

Figure 4.14 Illustration ofLaboratory Growth Rate Measurements in Concentrated Caustic Environments, Morphology Variations, and Growth Rates From Eddy Current Inspection Data 1

eIGA GROWTH RATE o SCC GROWTH RATE 10% to 40% CAUSTIC

~

ENVIRONMENTS l

n o

o.1 0

o MAXIMUM GROWTH RATE FROM g

JNO-2 BOBBlN, PROBE DATA a

S I

i g

?

AVERAGE GROWTH RATE

} o.o1 o

FROM BOBBIN DATA a.

I o.001 e

ROWTH RATE DISTRIBUTION FROM ANALYSIS OF PLANT A RPC DATA DATA SOURCE: NUREG/CR 5117 ~

0.0001 590'F

Attachment to OCAN079606 Page 35 of45 Figure 4.15 Comparison ofPredicted Versus Observed Bobbin Probe Indications C U M U L A T IV E TO OUTAOE R9.PER OENERATOR see i

8

=uuosa or eaacna i

oesERvt0 (297) l i

1

. /

/

w-i 8

8 8

m 8

A E

E E

I E

E E

.I 5

1 av=ese oF caaesapaeonctro I

(

0UTAOE R 10. P E R OENERATOR I

i

(

j I

t.,

l98.

1 l

l 8

NuMSER oP enacaa

=

oesenvs0 ti97) j

~

te, e

1. i

/-

/

8 8

8 I

E E

E I

0 E

E I

E E

E I

E 3

4U=85R oF eRacE8 9R E D Ic T E D

i 4

1 Att:chment to OCAN079606 Page 36 of 45 Figure 4.16 Comparison of Predicted Versus Observed Bobbin CoilIndications OUTAOE R 11. P E R SENERATOR ii.

l, I

l t

ise I

=u...o, CRACKS I

oseeavs0 i

493F) l l

/

l i

1 a

a e

g 5

g g

g E

E E

I E

E E

I 5

mWMeER 0F CAACES PRED6CTED 0UTAGE R 12. P E R GENERATOR I

as.

see see I

i 1

I se i

R a

e g

3 g

3 3

g E

3 E

R R

9 5

.......,e.........erse

Attachment to OCAN079606 Page 37 of 45 4.3 ACCIDENT LEAKAGE EVALUATION Using the Monte Carlo simulation model, crack lengths and depths at the EOC have been calculated. Based on these results, the probability of observing instances of throughwall cracks after 16 EFPM of operation have been calculated. If a MSLB is postulated after 16. EFPM of operation, the probability of no leakage is 61%. If the instances where leakage occurs, the dominant expectation is only 1 leaking crack. Given that leakage occurs, there is a less than 1%

l chance ofmore than 2 leaking cracks.

Since EOC throughwall crack lengths are one of the outputs of the Monte Carlo simulation model, calculation of the asscciated distribution of leak rates due to a postulated MSLB is straightforward. The Pipe Crnk Evaluation Program (PICEP)"') code was utilized for these calculations. If a throughwal' crack is observed at EOC, the leak rate of that crack at accident conditions is expected to be distributed according to the Weibull plot of Figure 4.16 t

Approximately 95% of these leak rates are less than 1 gpm. Using the data contained in Figure 4.16, a distribution free estimate of the 95/95 upper bound accident leak rate given the occurrence of I leaking crack is 1.22 gpm.

Figure 4.17 Weibull Plot ofLeak Rates Under Accident Conditions Associated With Individual Throughwall Cracks 909%

j I I IIlllll l I l1llll1 I I lllllll l l 1Illlll 1 I Ililli i

,, I n,u,

,,, uou. i,, uin, e o uno i muu i i ninn i i n ini, i i wnr i i ninn i i, o ni i istion e',"thol

'HU i i ilihn i t il otn i, o o,, i,, n o,

,,, o, o, o o.

,,,,oo I i I lllll1 I i i i l,Jl'.I I i 11llll l l 1 linil i i ililU

~

l I lllllll lIfl l l l lillll

! l lIlllll I i !lilh

' !Ilillll lJ/illlll l IIllll11

! Ililllll I i Illlli j

[.(

I I

N I

i ii I

k l

I l

l l

i mil i

I 1%

I I

I I

I i

05%

[

' ' i

'; i a

! 'l l

j ll I

I l

1l

!I '!

f l

l l

ll i ll l l i:l l

,,i I;

hi d

j a 0.0001 0.001 0.01 0.1 1.0 10.0 LEAK RATE GPM l

l Attachment to 4

OCAN079606 Page 38 of 45 5.0 PROBABILISTIC DOSE MODEL Probabilistic dose assessment is utilized as a way to demonstrate that there exists high confidence that the predicted dose to the thyroid and whole body due to iodine releases for postulated steam line break induced steam generator tube leakage will be less than the 10CFR100 and General Design Criterion (GDC) 19 limits. Doses at both the exclusion area boundary and in the plan.t control room are considered.

Standard Review Plan (SRP) assumptions are utilized with the, exception of the following:

1. use of statistical models ofiodine spiking (for co-incident release rate increase only) and the total leakage from the limiting steam generator as predicted under MSLB conditions

2. use of probabilistic criteria to establish compliance with doses specified by a small fraction of10CFR100 and GDC 19 allowables.

Otherwise, the probabilistic dose assessment applies conventional SRP assumptions such as conservative assumptions on the fraction of the RCS inventory released during the accident and conservative values for dose factors (dose conversion and dispersion). 'In addition the Technical Specification limits for Iodine of 1.0 Ci/gm equilibrium concentration and 60 pCi/gm for short term increases in Iodine concentration were utilized.

For completeness, comparison to deterministic SRP methods is presented.

Two cases are required to be evaluated for a MSLB dose evaluation. One is a pre-existing spike case where it is assumed that an event occurs such that the RCS Iodine concentration spikes up to the short term Iodine Technical Specification limits (60 Ci/gm) just prior to the MSLB, The other case is that the event itself causes a Iodine spike to occur due to the rapid cooling of the fuel and depressurization of the RCS. This is referred to as the co-incident spiking case.

The Iodine release rate is treated as a random variable in this analysis with values defined by a probability distribution.

The probabilistic analysis combines the predicted steam generator leakage distribution with this random variation in the co-incident Iodine spike release rate to determine dose consequences. These distributions are varied using a Monte Carlo simulation algorithm to provide a leakage estimate for the dose calculation which is combined with a spike release rate from a spiking probability distribution (for the co-incident spike case). Conservative doses are then determined by application of dose conversion factors and dispersion factors. In this way the probability of control room and exclusion area boundary doses determine the everall probability that the acceptance criteria are satisfied.

Summarized below is the key. input assumptions to the dose calculation. These inputs are consistent with those used in the ANO-2 design basis analyses with the exception of the ICRP 30 dose conversion factors.

~

6 Attachment to 0CAN079606 Page 39 of 45 ANO Unit 2 Dose Model Factors Thyroid Whole Body Control Room Exclusion Area Control Room EAB Boundary (EAB) x/Q (s/m')

5.6 x 10

6.5 x 10" 5.6 x 10

6.5 x 10d d

d Breathing Rate (BR) 3.47 x 10 3.47 x 10 (m'/s)

Dose Conversion 1.48 x 10' l.48 x 10' Factor (DCF)(rem /CI)

DCF (rem /CI)(ICRP 1.07 x 10' l.07 x 10' 30)

Iodine Protection 144.53 Factor (IPF)

Pre-existina Soike Given the predicted leakrate of 8.7 gpm the EAB and control room doses for the pre-existing spike are summarized below:

Postulated Accident Condition Dose at EOC 12 Pre-Existing Spike Thyroid Dose (REM)

Whole Body Dose (REM)

Control Room EAB Control Room EAB Dose Limit 30 300 5

25 Deterministic SRP Analysis 8.1 77.6 0.9 0.1 Deterministic SRP Analysis 5.9 56.1 0.9 0.1 (ICRP 30)

Probabilistic Cases Mean 95/95 Mean 9595 SRP Dose Model 3.7 8.4 35.0 80.6 ICRP 30 Dose Model 2.6 6.1 25.3 58.2 The doses in the probabilistic case are slightly higher than the deterministic case due to the slight contribution of the tail of the leakage probability distribution. The probability of exceeding the dose acceptance criteria for the pre-existing case for ANO Unit 2 at EOC 12, based on 10,000 Monte Carlo trials, is summarized in the table below. These results demonstrate that the dose criteria of 10CFR100 and GDC 19 are satisfied at the 95/95 level for the pre-existing spike case for the ANO Unit 2 steam generators at EOC 12.

• ~

Attachment to OCAN079606 Page 40 of 45 Probability Accident Leakage at EOC 12 Exceeds Dose Criteria ICRP 30 Dose Model

)

Case Probability of Exceeding Criteria Thyroid Dose ControlRoom 1.0 x 10

EAB 1.0 x 10

Whole Body Dose

~

Control Room 1.0 x 10

EAB

<1.0 x 10

It can be seen from the above tables that, using SRP assumptions, the predicted doses fall well below the pre-existing spike case acceptance criterion of 300 rem.

Co-incident Soike Given the predicted leakrate of 8.7 gpm the EAB and control room doses for the co-incident spike using the ANO Unit 2 leak rate distribution for EOC 12 are summarized below. ' The dose limits are provided in the first row of the table, In addition, a deterministic dose evaluation using SRP assumptions for Iodine spiking (500x Iodine release rate) is provided for comparison. It should be noted that the equivalent probabilistic calculated equivalent spiking factor is 290 versus the 500 utilized in the SRP.

1 Postulated Accident Condition Dose at EOC 12 - Co-Incident Spike l

Thyroid Dose (REM)

Whole Body Dose (REM Control Room EAB Control Room EAB Dose IJmit 30 30 5

2.5 Deterministic SRP Analysis 7.0 66.9 0.9 0.1 Deterministic SRP Analysis 5.0 48.4 0.9 0.1

__(ICRP 30)

Probabilistic Cases Mean 9595 Mean 95/95 SRP Dose Model 1.1 5.1 13.9 40.5 ICRP 30 Dose Model 1.3 3.7 10.1 29.2 The probability of exceeding the dose acceptance criteria for the co-incident case for ANO Unit 2 at EOC 12, based on 10,000 Monte Carlo trials, is summarized in the table below. These results demonstrate that the dose criteria of 10CFR100 and GDC 19 are satisfied at the 95/95 level for the co-incident spike case for the ANO Unit 2 steam generators at EOC 12.

Attachment to OCAN079606 i

Page 41 of 45 Probability Accident Leakage at EOC 12 Exceeds Dose Criteria Co-Incident Iodine Spike Case -ICRP 30 Dose Model Case Probability of Exceeding Criteria Thyroid Dose 4

Control Room 2.0 x 10d EAB 4.3 x 10-2 Whole Body Dose Control Room

<l.0 x 10" 3

EAB

<l.0 x 10d From the above results, it can be seen that for the co-incident spike case using the probabilistic model and ICRP 30 dose conversion factor, the resulting control room and offsite doses are less than acceptance criteria. It should be noted that in both of the above cases, pre-existing spike and co-incident spike, the 8.7 gpm leakrate is assumed to remain constant for the entire event.

Conclusion Using the predicted leakrate for EOC 12, it is shown that the 10CFR100 and GDC 19 acceptance criteria for the pre-existing spike case using standard SRP assumptions is satisfied. For the co-incident spike case, the probabilistic model is utilized to demonstrate that with a more realistic lodine spiking prediction and the ICRP 30 dose conversion factor, the 10CFR100 and GDC 19 dose criteria are satisfied at the 95/95 confidence level.

Thus the dose criteria of 10CFR100 and GDC 19 are satisfied at the 95/95 confidence level for both the ICRP 30 model pre-existing and co-incident spike cases for the ANO Unit 2 steam gsnerators for the entire planned operating cycle 12.

e 1

- - -.. ~ -- -

'I Attachment to l

OCAN079606 Page 42 of 45 l

6.0 -

PLANT OPERATIONAL EVALUATION' l

An operational leak rate limit is established to provide reasonable assurance that flaws either missed.during inspection or growing more rapidly than expected will not render the tube l

vulnerable to tube rupture in the event of a MSLB. The ANO-2 technical specification limit is 0.5 GPM per SG, but an administrative procedurallimit of 0.1 GPM (144 GPD) exists to provide for added margin ay.iw burst. In addition, rate of change limits exist to ensure rapidly propagating cracks or damagt

-Q be addressed at the earliest possible stages.

j Upon any control room alarm indicating primary to secondary leakage, the Operations and Chemistry Departments enter abnormal operating procedures. If the leak rate is 20.1 GPM, a i

plant shutdown is procedurally required. In addition, a plant shutdown is procedurally required if the leak rate is projected to be 20.1 GPM in one hour. Stable leak rates of >0.01 GPM procedurally require management awareness for continued plant operations.

The Operations Department trends the steam, condenser off-gas, and steam generator sample systems in determining the indication of a steam generator tube leak. Steam lines are monitored via radiation monitors and nitrogen sixteen (N-16) gamma detectors, which provide the chemists and operators with the capability of quantifying leakage. Procedures are utilized when the monitors or trend recorders for the aforementioned systems exhibit increasing trends. The Operations Department enters these procedures to place the plant in a stable condition and to mitigate the consequences of a steam generator tube leak..

Extensive training for operators is performed and places emphasis on changes in plant parameters and develops an aggressive strategy to identify early signs of a steam generator tube rupture and mitigate the event. During a NRC Region IV inspection in June 1994 u inspectors noted that 0,

operators "... exhibited excellent capability and used diverse methods for detection of priman-to-secondary leakage..." In addition, an ANO-2 re-qualification scenario involving a SGTR event was witnessed by an inspector, who observed that the communication was " excellent."

The Chemistry Department routinely samples both the primary and seconday water systems, as.

well as condenser off-gas, to identify primay to secondary leakage and trends the sample results to identify possible priman-to-secondary leakage occurrences.

i l

l i

l

.{

i 4

~

~

Attachment to 0CAN079606 l

Page 43 of 45

7.0 CONCLUSION

S Entergy Operations has performed an extensive investigation into the circumferential cracking occurring at ANO-2.

The investigation includes comprehensive inspections, application of i

appropriate safety factors, use of statistically valid (95/95) material properties, NDE data, growth rate, and tube burst test data. First of a kind probabilistic evaluations were performed to assess the safe operation for previous and current intervals. ANO has performed extensive studies into circumferential cracking, and has sponsored several initiatives which have helped the industry.

These include pulled tubes, the generation of 54 laboratory cracks for NDE and stmetural studies, i

and a first of a kind probabilistic safety assessment of circumferential cracking.

From a structural standpoint, circumferential cracks on the order of 50% degraded area and smaller exhibited burst pressures equal to or near that of unflawed tubes.

This has been demonstrated in both U.S. and w;rldwide programs for burst testing of SG tubes with circumferential flaws.

These programs exhibited the irieren: strength in stress corrosion circumferential cracks. This st ength is generally attributed to the discontinuous nature of circumferential cracks, where W are ligaments of sound material which generally are not detectable 6 conventional ECT analysis techniques. These ligaments between crack segments add significantly to the drength of the cracked section of tubing. In-situ pressure testing of the most severely degraded tubes in CE plants have demonstrated the tubes would meet RG 1.121 requirements with added margin.

For NDE, the use of the Plus Point coil significantly enhances the ability to detect and accurately l

size the cracks. The new software has also allowed for a re-evaluation of flaw sizes by providing i

more accuracy. Growth rates of circumferential cracks have been shown through this enhanced analysis to be lower than that reported based on earlier analysis. The technological advances in NDE have resulted in the ability to detect smaller flaws, many of which would have not been l

detected previously. These small flaws constitute little, if any, challenge to the structural integrity of tubing, and have been shown to be acceptable for remaining in service as a result of the i

previous higher detection thresholds.*

Additional areas provide added margin above that required by RG 1.121. Throughout the assessments, planar depth is assumed, thus not accounting for the ligaments previous!y mentioned.

l In additiori, lower bound (95%) values were used for material properties, NDE uncertainty, and the strpetural limit, as well as an upper bound (95%) value for growth rue. A deterministic evaluation was performed which shows full cycle operation is justifiable.

l For freespan and eggerate axial cracking,100% bobbin examinations have been performed since 1992. These examinations have identified indications prior to any challenges to RG 1.121 structural limits. This was verified through the performance ofin-situ pressure tests, where even the largest flaws withstood a pressure of three times the normal operating pressure differential.

Using data on growth rates from several 100% examinations at ANO-2, a deterministic evaluation shows full cycle operation is justified.

_. - - ~ _ - -. -

. = _ -.

Attachment to 0CAN079606 Page 44 of 45 Probabilistic evaluations of the various damage mechanisms show acceptable results for full cycle operation. The conditional probability of burst was determined to be 2.40x10 for axial cracking i

4 and 6.90x10 for circumferent: ' cracking. Combining these yields results within the acceptable value of lx10 2 Other mechanisms at ANO-2, such as wear at diagonal and vertical straps, are l

considered negligible contributors to the probability of burst evaluation.

p The combined 95/95 accident leakage value was 8.7 gpm. The dose consequences associated l

with this number was evaluated with a probabilistic model. The results demonstrate that the dose j

criteria of 10CFR100 are satisfied at the 95/95 level for both the pre-existing and co-existing spike cases.

j Entergy has utilized a comprehensive integrated approach to managing steam generator tube integrity at ANO-2. Comprehensive inspections performed at selected intervals provided the necessary information to ensure safe operation. This approach supports full cycle operation in l

Cycle 12. ANO has an excellent primary-to-secondary leakage detection program with excellent opernv communication and training programs to ensure early detection and a rapid response should leakage occur.

Entergy is committed to operating ANO-2 safely, and believes the. inspections and analy is performed in 2R11 ensures safe operation of the unit for the entire planned operating cycle.

1 i

i l

1 I

i l

,o Attachment to 0CAN079606 Page 45 of 45

8.0 REFERENCES

1) 2CAN029505, "2P95-1 Steam Generator Inspection Results and Circumferential Cracking Evaluation", February 17,1995.

2) 2CAN089507, " Repair Limit for Circumferential Cracks in Steam Generator Tubing," August 28,1995.

3)

"End of Cycle Projections for Circumferential Cracking," presented by B.

Woodman, J. Begley, B. Keating at the EPRI ARC Circumferential Cracking Committee Meeting, Chicago, IL, May 15-16, 1996.

4)

" Probability of Burst Model for ANO Unit 2 TTS Circumferential Cracks," Tetra Engineering Group Repon TR-96-005, May 1,1996.

5)

NP-6864-L-Rev. 2 "PWR Steam Generator Tube Repair Limits: Technical Suppon Document for Expansion Zone PWSCC in Roll Transitions," August 1993.

, 6)

"hformation Update Concerning the ANO-2 Steam Generators," August 26, 1992.

l' 7)

AES 95102556-1, Rev.1, "An Analysis of ODSCC/ IGA at Eggerate Support Locations at Arkansas Nuclear One, Unit 2," April 1996, Prepared by Aptech Engineering Services.

8)

NP-6865L, Vol 1 Cochet, B., " Assessment of the Integrity of Steam Generator Tabes - Burst Test Results - Validation of Rupture Criteria (FRAMATONE

. DATA)," Electric Power Research Institute, June 1991.

9) i NP-7480L " Steam Generator Tubing Outside Diameter Stress Corrosion Cracking i

at Tube Suppon Plates - Database for Alternate Repair Criteria, Volume 2: 3/4 Inch Diameter Tubing," Electric Power Research Institute, October 1993.

10)

NP-3596-SR "PICEP: Pipe Crack Evaluation Program (Revision 1)," Electric Power Research Institute, December 1987.

11)

OCNA079416, "NRC Inspection Report 50-313/94-17; 50-368/94-17," July 13, 1994.

L