ML20101G290

Nonproprietary Preliminary Data on Voltage/Burst/Leakage of 3/4 Inch Diameter Tubing for ODSCC at Tsps

ML20101G290
Person / Time
Site:	Farley
Issue date:	03/31/1992
From:	Wootten M WESTINGHOUSE ELECTRIC COMPANY, DIV OF CBS CORP.
To:
Shared Package
ML19311A852	List: ML20101G266 ML20101G282 ML20101G290
References
WCAP-13238, NUDOCS 9206250394
Download: ML20101G290 (46)

Text

2999lgQxgllWM nf.pMGgkudM%%w 5,

g a

g n

m%Lt+

n{Q l i Qg}

y gp;ns vuMdikMa M

}

- a;(:

q g_y r

,m

~

gy l@.

a g4 g

[-

$b

• 7 mp

+

h

,'; h.. gl %g y

"2

'y a

pg p

q e

3

+

u

, y$

g

~

+

1 a

qwg"fg%pg 44g y.

mp"apgagyweggg N #

P I$fM T QJ b 1

9(-.fl 4a p

1 9,.

g x

1

+1 g

c4 7

j w

.,4 a

s e

~,*j g

g.

(

~. :

ga' '

h.

L:

O N

i q

9

1

3 n

• f.N N;

A..

[

i r

h 3:

s$

,#^

gs l

4 y

{ "A g

i j @@ h A

- cy j q

.e f

f; nu &

a n

jgc

- :a y

' E is

[ MN y

j k

,' 3 d;

y cp

]O e

Af?

?

th

. !i s.

l '

q,

}(Q,,;*{,

ys g

- m.y?

h.

k se$

Ql N'k b

)

b q

l C g i kaa s$ k M M $ n

, s ner v

i s

"5

[

aw.

n e,

pggegNpaggs 4 gL 220$250394 oroo m 9.

afwa 4m r

2 81% N 5 % g m n@ M An a c~ d h l

. a, u e 1

von avoca ovoooaae nybf2'n2A k.

L_.

PD*

%iW

~

VESTNGHOUSE CLASS 3 l

WCAP 13238 SG-92-03-009 Preliminary Data on Voltage / Burst / Leakage of 3/4 Inch Diameter Tubing for ODSCC at TSPs March,1992

~

Approved by: ['

f M. J. Wootten, M nager team Generator echnology & Engineering c 1992 Westinghouse Electric Corporation All Rights Reserved

Table of Contents Section Ilue fim 1

INTRODUCTION 1

2 MODELBOILER SPECIMEN PREPARATION 2

3 NDE MEASUREMENTS 5

J 4

LEAK RATE AND BURSTTESTING 6

5 CORRELATIONS FOR 3'4 INCH TUBES 7

5.1 Burst Pressure Correlation 7

5.2 SLB Leak Rate Correlation 7

6 THEORETICAL SCALING OF 3'4 INCH RESULTS FOR COMPARISON WITH 7/8 INCH DATA 8

7 CORRELATIONS FOR THE COMBINED 7/8 INCH AND THEORETICALLY SCALED 3/4 INCH DATABASE 13 7.1 Burst Pressure Correlation 13 7.2 SLB Leak Rate Correlation 13 8

CORRELATIONS FOR THE COMBINED 7/8 INCH AND 3/4 INCH DATA BASED ON APPROXIMATELY EQUAL MEAN FIT 15 8.1 Burst Pressure Correlation 15 9

COMPAR' SON OF CORRELATIONS 16 9.1 Burst Pressure Correlations 16 9.2 SLB Leak Rate Correlations 16 10 PREUMINARY CONCLUSIONS 18 g

11 RetHENCES 20 i

._ _._=._.

c List of Tables Table lii!g Ea;te 1

Thermal and Hydraulic Specifications 21 2

Model Boiler Test Specimen Summary (3/4 inch Tubes) 22

.3 NDE Results for 3/4 inch Model Boiler Specimers 23 4

Burst Pressure and Leak Rate Data for 3/4 inch Specimens 24 5

Adjusted Data for 3/4 inch Diameter Specimens 25 6

Unear Regression Analysis Parameters for Burst Pressure and Leakage 26 4

ii l

I g

e..

7 g

-. -. ~.-i

++-m"--

-'e"-'fls ha'a--*

a

f List of Figures E!22I2 I!!!n Emle 1

Schematic of Vertically Mounted Single Tube Model Boiler 27 2

Schematic of Horizontally Mounted Single Tue Model Boiler 28 3

Burst Pressure vs. Bobbin Voltage for 3/4 Inch Tubes 29 4

Burst Pressure vs. Bobbin Voltage for 7/8 inch Tubes 30 5

SLB Leakage vs. Bobb;n Voltage for 3'4 inch Tubes 31 M

6 SLB Leakage vs. Bobbin Voltage for 7/8 inch Tubes 32 7

Burst Pressure vs. Unscaled Bobbin Voltage for 3/4 and 7/8 inch Tubes 33 8

SLB Leakage vs. Unscaled Bobbin Voltage for 3'4 and 7/8 inch Tubes 34 9

Bobbin Voltago vs. Crack Length 35 10 Bobbin Voltage vs. Crack Length Fit for Essentia!!y Single Through-wall Axial Cracks in 7/8 inch Diameter Tubes 36 11 Bobbin Voltage vs. Crack Length Fit for Essentially Single Through wall Axial Cracks in 3/4 inch Diameter Tubes 37 12 Normalized SLB Leak Rate vs. Bobbin Voltage Normalized to 7/8 inch 38 Diameter Tubing 13 Normalized Burst Pressure vr. Bobbin Voltage Normalized to 7/8 Inch 39 Diameter Tubing 14 Burst Pressure Correlation for the Theoretically Scaled, Combined Database 40 15 SLB Leakage Correlation for the Theoretically Scaled, Combined Database 41 16 Burst Pressure Correlation for the Empirically Scaled, Combined Database 42 iii

1.

INTRODUCTION This report descr:bes the preliminary results from the testing of 0.75 inch Jecimens for tube burst strength and leak rate. This experimental work, which is continuing, was performed to support the development of tube repair limits based on bobbin voltage amp!ttudes of indications for outside diameter stress corrosion cracking (ODSCC) of steam generator tubes at support plate locations. Developmerit of tube repair limits for 0.875 inch diameter tubing has been discussed elsewhere, including Reference 1. Reference 1 also discussed in detail the tests performed for the 7/8 inch tubing. The same test procedures were used in the testing of 3'4 inch specimens described in this report. Therefore, no attempt is made here to repeat the more detailed test descriptions. The emphasis is placed on reporting the preliminary results of 3/4 inch test data.

An attempt is made to assess whether the 3/4 inch data can be scaled to merge with the 7/8 inch data using theoretical considerations as well as an empirical treatment. The rr.nerence 3'4 inch voltage normalization is based on scaling ronsiderations between tube diameters. Additional adjustments to the 3/4 inch data are necessary for comparisons with 7/8 inch data. The theoretical or first principle comparison is based on adjusting bobbin amplitudes to obtain approximately equal voltages for equal crack lengths combined with well established non-dimensional correin.tions of burst pressures. It is found that these first principle adjustments result in d;fferences in the mean fits to the data for the two tube sizes. An additional comparison of the 3/4 and 7/8 inch data is then made based on adjusting the 3/4 inch voltages to

[

obtain approximately the same mean fit as the 7/8 inch data.

The overall trends (burst strength vs. boboin voltage and leak rate vs. bobbin voltage) of the two sets of data (3/4 inch and 7/8 inch) are substantially similar. However, based on the limited work described herein, the 3/4 inch data may belong to a separate population than the 7/8 inch diameter data;i.e., different correlations may be applicable to the two data sets. Additional data and evaluation are required to finalize a position on separate or combined correlations for the 3/4 inch data.

1

l

2. MODEL BOILER SPECIMEN PREPARATION Cracked tube specimens were produced in the Forest Hills Single Tube Model Boiler test facility.

The facility consisted of thirteen pressure vessels in which a forced flow primary system transfers heat to a natural circulation secondary system. Appropriate test specimens were placed around a single heat transfer tube to simulate steam generator tube support plates. The tests were conducted in two boiler configurations, shown schematically in Figures 1 and 2. The majority of the tests were conducted in the vertically oriented boilers shown in Figure 1, in which four support plates were typically mounted on the tube. A few tests were conducted in horizontally mounted bollers, shown in Figure 2. Because there was no steam space in these boilers, seven support plates can be mounted on the heat transfer tube. Since capillary forces, rather than gravity forces, dictate the flow pattern in packed tube support plate crevices, the tube orientation should have little effect on the kinetics of the corrosion processes.

The thermal-hydraulic specifications utilized in the test are presented in Table 1. As indicated, the temperatures are representative of those found in PWR steam generators, and the heat flux is i

typical of that found on the hot leg side of the steam generator.

The cracks were produced in what is termed the reference cracking chemistry, consisting of either 600 ppb (1X) or 6 ppm (10X) sodium as sodium carbonate ir the makeup tank.

Typically a test was initiated with the 1X chemistry, and if a through wallleak was not identified aftc* 30 days of operation, the 10X chemistry was applied. The occurrence of primary to secondary leakage was determined by monitoring the boilers for lithium, which would ordinarily only be present in the primary system. Because of hideout in the crevices, the boiler sodium concentration was typically between 50 and 75% of the makeup tank concentration. Hyorazme and ammonia were also added to the makeup tanks for oxygen and pH control, respectively.

The tests utilized 3/4 inch (1.9 cm) O.D. mill annealed Alloy 600 tubing from heat NX7368.

The tubing was manufactured by the Plymouth Tubing Co. to Westinghouse specifications.

A summary of the test pieces which were subsequently leak and burst tested are presented in Table 2. Two groups of tests are listed; the EPRI test pieces were provided under the auspices of an EPRI program, while the Spanish test pieces were fabricated for a group of Spanish utilities Permission from the uti!ities has been obtained to use the results of these tests in other applications. The only difference between the two groups of tests is that the crevices were 2

l packed with ditlerent sludge formulations. The EPRI tests were performed with what is termed simulated plant sludge, which has been used in most previous rnodel boiler test programs, while the Spanish tests used a formulation more representative of that typically found in Spanish steam generators. As indicated in the fo"owing table, the only difference betwecn the two fL.~,ulations is that magnetite has replaced the meta!!ic topper content in the simulated p' ant sludge.

Webht _,,,_

Simulated Plant Spanish

.Conttituent

%te h

(

Magnetite 59.7 02.2 Copper 32.5 Cupric Oxide 4.5 4.5 Nickel Oxide 2.1 2.1 Chromic Oxide 1.2 1.2 As outlined in Table 2, three morns of packing the tube support plate crevices were utilized. In the fritted configuration, loose sludge was vibratorily packed into the crevice and then held in place with Alloy 600 porous frits placed over both ends of the crevice, in this configuration, cracks were typically produced near the interface between the sludge and the frits. In some cases, multiple cracks were produced at both ends of the crevice.

The dual consorduted configuration consisted of two sludge regions, in which the oute region contained chromic oxide, while the inner region contained either simulated plant or - pansi, sludge.,'he regions had the following design, with the dimensions given in millines:

11.4 9

3.2 4 Y 4.1 &

a r Simulated Plant Sludgc 19,0

6. 4 Chromic 4

0xide

)

h38 J.

4 a

I l

The two region sludge configuration was specified in order to limit cracking to the sma!! inner f

a region, containing an oxidizing sludge. Chromic oxide is nonoxidizing, and previous testing had found that accelerated corrosion does not occur in its presence. The outer region provided thermat insulation for the inner region, so that the temperature in the inner region was i

suthciently high to produce accelerated corrosion. The two sludge regions were baked onto the tube using a mixture consisting of 5% sodium hydroxide 2.5% sodium sulfate, and 0.6.. sodium silicate. The support plates were then mounted on the tube over the sludge and held in place with

l ex1ernally mounted set screws. Since corrosion should be confined to the inner region, this configuration was intended to produce short, individual cracks.

The mechanically consolidatea sludge configuration was fabricated by mechanically compacting sludg9 Y ' "in a tube support plate simulant, drilling a hole in the sludge for the tube, and then sliding the tube through the hole until positioned properly. This configuration was used because relatively low voltage indications had been produced in previous tests having this configuration.

As indicated in Table 2, there was considerable variation in the time taken for a crack to be produced in a given test piece, in general, cracking was produced in shorter time spans with this heat of material than for the heats used in similar tests performed with 7/8 inch diameter tubes.

Cracks were typtca!!y produced most rapidly with the fritted configuration and most slowly with the dual consolidated configuration, although a few cracks were produced very quickly with the l

dual consolide,ed configuration.

i 4

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

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

l

3. NDE MEASUREMENTS P,0bbin coil tests were performed on the specimens prior to leak rate and burst testing. The bobbin probe with a 0.020 inch diameter was used and the resu!ts reported in this report is based on the d.fferential mix frequency of 550/130 kHz. Selecten of this probe and frequency mix is related to the 0.720 in;h probe and 400/100 kHz frequency used for the testing of 7/8 inch diameter tubing and the follow;ng considerations.

To a first approximation, bobbin probe voltages for different tubing sizes would be equal for scaled f!a v sizes if:

1.

The tubes are geometrically similar, that is, alllinear dimensions are scaled by the same factoi. This applies for the 3/4 inch and 7/8 inch d,ameter tubes.

2.

Respectrve odd / current dimensions are scaled by the same scale factor. This applies for the 0.720 inch and 0.020 inch probe diameters used in this report.

3.

The probe excitation frequency is inversely proportional to the square of the tubing e

thickness, This applies for the 550/130 kHz and 400/100 kHz frequencies used in this report.

I Thus the bobbin coil dimensions and test specifications were intended to yield similar responses for the 3/4 inch diameter tubing as in it'e 7/8 inch tubing. Bobbin vc!tages were normalized to 4.0 volts for the four,20% deep holes in the ASME calibration standard, in addition to bobbin exit testing, RPC (rotating pancake coil) tests were also performed on the laboratory specimens. The RPC voltages were obtained at 550 kHz based on normalization to 20 volts for a 0.G !..cii, throuvi wall EDM (electric discharge machining) slot.

The results of the NDE measurements for the 3/4 inch diameter model boiler specimens are summarized in Table 3.

5

i

4. LEAK RATE AND DURST TESTING The procedures applied for leak rate measurements and burst testing of 3/4 inch diameter tubing were the same as applied for 7/8 inch d;ameter tubing. Those procedures are described in prior reports dealing with 7/8 inch diameter tubing, Leak rate measurements were performed at normal operating and SLB pressure differentials. The results for the leak rate and burst 4

pressure tests for the 3'4 inch diameter model boiler specimens are given in Table 4.

l l

t i

Y C

),

6

'*+ptsqwe ymsw-g vr pMegPy-t-

y *-

'ry-

,9-gy-

+t D'W-T'e***'%T--

'"D-VW wu'WW w 'W h N""'irV F7 Mf" W-e*e*"m8-*'88-WWw*""+"*8^

^'N-'

-*^**-el'-

l 5.

BURST AND LEAK RATE CORRELATIONS FOR 3'4 INCH DATA 5.1 Burst Pressure Correlatioi The mod 61 boiler and pulled tube burst pressure data available for 3'4 inch diamter tubing is plotted in Figure 3 as a function of bobbin coilvoltage. The overall trend of the burst pressure vs bobbin coil data is quite simila.r to the comparable plot for the 7/8 inch diameter tubing (see Figure 4). It was judged that a linear equation on the semilog plot will adequately fit the available 3/4 inch data. Therefore a linear regression analysis between bc.st pressure and log of the bobbin voltage was performed. The resulting linear equation for the mean curve is:

[

}O where BP is the burst pressure in ksi and v is the bobbin voltage (550/130 kHz differential mix). The coefficient of correlation for this regression fit is 0.73 and the error of the BP estimate is 1.326. The linear iit is plotted in Figure 3 as a straight line.

5.2 SLB Leak Rate Correlation The leak rate data for the 3/4 inch diameter model boiler specimens under steam line break pressure d;fferential are plotted as a function of bobbin voltage in Figure 5. A similar plot for the 7/8 inch diameter tubing (model boiler and pulled tubes) data is shown in Figure G. It is observed that the overall trend with voltage is quite similar. Therefore a correlation similar to the 7/8 inch data was developed for the 3/4 inch leak rate results.

Linear regression analysis of the logarithms of the data results in the following mean leakage rate correlation where O is the SLB leak rate in l'hr and v is the bobbin voltage (550/130 kHz differential mir):

(

)

The coefficient of correlation for this regression fit is 0.64 and the error of the estimate is 1.165. The regression fit is also shown in Figure 5.

7

P 6.

THEORETICAL SCAtJNG OF 3'4 INCH RESULTS FOR COMPARISONS WITH 7/8 INCH DATA t

i As noted in Section 5, there is coristderable sim!!arity between the 3/4 inch and 7/8 inch tube data. Figure 7 shows the 3/4 inch and 7/6 inch burst pressure data on the same plot, without i

applying any scaling factors to either set, it is clear that, despite the similar trends, the data e

appear to come from two different populations. The same conclusion is derived from a review of the leak rate data (see Figure 8).

2 i

i It is desirable to scale one set of data to asses, whether or not the two data sets may be combined and the entire data base represented by a given correlation for tube burst and another for leak rate. By having a larger number of *cbservations," the uncertainty in the correlation will be reduced. However,in order to perform the scaling appropriately, a scaling f actor has to be i

developed. In addition to improving the fit of the correlation to the data, the scaling f actor should I

have a sound (theoretical) basis. An attempt at such scaling is made here. This work is

]

a considered quite preliminary and the results indicate that additional development is needed. it is

)

1 discussed here merely to present the current status of this task, rather than to derive any conclusions.

As discussed in Section 3, selection of the bobbin coil probe and test specifications were intehded to provide similar results for comparable cracks in 3/4 inch and 7/8 inch diameter tubes.

The principal difficulty in comparing the 3/4 inch and 7/8 inch results is developing the appropriate adjustments to the bobbin voltages, leak rates and burst p'essures to permit direct comparisons of the data for the two tubing sizes. The tubing sizes are geometrically similar in that the wall thicknesses (0.050 inch for 7/8 inch tubing,0.043 inch for 3/4 inch tubing) are proportional to the tube diameters. However, adjustments in addition to appropriate scaling are required to directly compare the two tubing sizes. The approach applied is as follows:

i o Eddy current probe diameters (0.720 inch,0.620 inch) are scaled by the tube diameters.

However conventional bobbin coil width and spacing are not scaled.

Eddy current inspection frequencies (400/100 (Hz,550/130 kHz) are appropriately o

scaled as inversely proportional to the square of the tubing thickness.

8

- _. _ _ _.. _ _ _ _. -.m.

1 A limitation on theoretical scaling of bobbin coil voltage responses results from the fact that, although probe diameters are scaled, tht conventional bobtin probes do not t.cale coil width aad coil spacing, in addition, ASME calibration standards for the 20% depth flaw in 3/4 inch and 7/8 inch tubing have the same 0.187 inch hole diameter.- Consequently, conventional eddy current practices would not be expected to yield exact scaling. To obtain the same voltage for equal crack lengths in 3/4 inch and 7/8 inch tubing, the 3/4 inch voltages would be expected to increase by the diameter ratio of 1.17 if voltage is assumed to be linear with crack length over the range of interest of ~0.2 to 0.7 inch hng cracks. An empirical factor of 1.25 is found for the model boiler specimens as discussed below.

i Using an empirically derived scaling factor for the bobb!n voltages of 0.875 inch and 0.750 inch diameter tubes having through wall cracks of the came length. voltage normalization procedures are described in the following paragraphs to permh the scaling of bobbin voltage leak rate and bobbin voltage burst pressure data between 0.730 inch and 0.875 inch diameter tubing.

The bobbin voltages of model boiler spechnens with ODSCC at simulated tube support plate intersections were determined using standard, well calibrated procedures as described in Section

3. Following leak rate tests at normal operating and steam line break conditions, burst tests were performed at room temperature. After burst testing, the fracture surfaces were examined at a magnification of 20X. The average lengths of the outer diameter strt.ss corrosion cracks were measured. Figure 9 shows a plot of bobbin voltage versus ax:al crack length. Most, but not all, of the tubes had through wall cracks. At a given crack length of the dominant crack in the tube, very high voltages are measured when multiple axial cracks are distributed around the circumference of the tube. Conversely, when cracking is only partially through the wall or when ligaments exist between coplanar axial cracks, the measured bobbin voltages form the low -

side of the scatterband at a given crack length.

i If only tubes with essentially single, through wall, axial cracks without ligaments are considered, then the relationship between bobbin voltage and crack length can be determined empirically as a function of tube size. Recall that apriori considerations indicate that bobbin voltage versus crack length correlations for 0.875 inch and 0.750 inch diameter tubes should be nearly_ the same allowing for a scaling factor of about 1.17. Figure 10 shows that the bobbin voltage (volts) of a 0.875 inch diameter tube is 31.9 times the crack length (inches). Figure 11 shows that for a 0.750_ inch diameter tube this slope is 25.5. Hence, at a given crack length, 9

.~

_.__m._______

the bobbin voltage of the 0.750 inch diameter tube will be less than that of a 0.875 inch diameter tube as given by:

V.875 31.9 L.875 L.875

- 1.252 V.750 25.5 L.750 L.750 Thus, if the bobbin voltage of a crack in a 0.750 inch diameter tube measures x, then the voltage of this same crack in a 0.875 inch tube will measure 1.252 times x. With this basis, bobbin voltage. leak rate and bobbin voltage-burst pressure correlations are scaled between tubing sizes as follows.

Experimentalleak rate measurements on 0.875 inch and 0.750 inch diameter steam generator tubes containing through wall, sharp fatigue cracks show that leak rate versus crack length curves are similar but not identical. A leak rate calculational model, benchmarked with -

experimental data, shows that leak rates for 0.750 inch tubing are lower than for 0.875 inch tubing at the same crack length by 10 to 25% in the range of crack sizes of interest, this effect can be accounted for by adjusting 0.750 inch leak rates by a factor of 0.9. To compare bobbin voltage versus leak rate correlations on the same basis, the bobbin voltages of the 0.750 inch tubes should be multiplied by a factor of 1.252 to match respective crack lengths. Figure 8 shows steam line break leak rates versus bobbin voltage without any normalization factors while Figure 12 shows that normalization to ths 0.875 inch diameter size brings the respective scatterbands into reasonable agreement.

l l

Theoretical and experimental studies show that, for ductile tubes with through wall, axial cracks, the normalized hoop stress at burst is a function of the 1ormalized crack length,lamda =

L/(R*t).5, where L is the crack length R is the mean tube radius and t is the wall thickness. At equal values of lamda, the hoop stress at burst divided oy the sum of the yield and ultimate strength will be the same for different sizes of tubing. For both 0.875 inch and 0.750 inch I

diameter tubes, bobbin voltage is essentially directly proportional to crack length over the range of interest. Hence equal values of lamda for the same burst pressure can be converted to equivalent values of bobbin voltage for the same burst pressure.

10

For equallamda,

.5 L.875 1

V.876 (R * :).875

= 1.17

=

.5 L,750 1.25 V.750 (R

• t).750 V.875

- 1.25

• 1.17

• V 750 - 1.46

• V.750 Thus, to convert the bobbin voltage of a 0.750 inch tube to an equivalent bobbin voltage of a 0.875 inch diameter tube with the same burst pressure, the conversion f actor is 1.46. Figure 7 illustrates a plot of burst pressure versus bobbin voltage for the two tubing sizes without any normalization facto.s. Figure 13 shows the burst pressure normalized to a yield and ultimate strength total of 147 ksi and the bobbin voltage noimalized to a 0.875 inch diameter tube using the 1.46 factor on the 0.750 inch data. The normalization approach can be used to combine data on different tubing sizes and perform appropriate statistical analyses on this larger combined database.

A comment on apparent outliers in burst and leak rate correlations is worthy of consideration. It is clear that single dominant cracks essentially determine leak and burst properties. Multiple cracking can lead to very high voltages compared to a single crack and thus lead to the non conservative edgss of the respective scatterbands of leak rates and burst pressures.

Ligaments beneath partial through wall cracks and ligaments between coplanar through wall cracks decrease bobbin voltages relative to the no ligament case. The presence of such ligaments is considered to be responsible for the conservative edges of leak rate and burst pressure scatterbands when ligament sizes are in a very narrow range. For example, any substantial ligtment will decrease the leak rate and increase the burst pressure while leading to a lower voltage. The net effect in this case leads to another point in the body of the scatterband. Very smallligaments will mechanically fracture due to normal operating pressure d;fferentials.

Conservative side outliers are proouced by ligaments in a narrow size range which decrease the voltage while not significantly increasing the burst pressure relative to the no ligament case.

For leakage, the voltage reducing ligaments must fracture at steam line break conditions but not at normal operating conditions to produce an outlier effect.

11

~.

~

l The above development provides a procedure based on equal voltates for equal crack lengths to conipare the 3/4 inch and 7/8 inch data. An alternate procedure for comparing the data based on adjusting the 3/4 inch data to make the mean fits to the two sets of data approximately equalis gNen in Section 8. Table 5 provides the 3/4 inch data including the adjusted volts. SLB leak rates and burst pressures. The Table 5 data are used below to perform statistical analyses of the

~

combined data base.

i 4

1

)

i l

l

)

l l

i i

12 i

.. _ = _ _ _ _ _ __ _ _..-._ ___. _ ____. _ _ _ __ _ _ _.. _ _ _

f 7.

CORRELATlONS FOR THE COMBINED 3'4 INCH AND 7/8 INCH DATABASE a

7.1 Burst Pressure Correlation The adjusted 3/4 inch voltages and burst pressures from Table 5 are combined with the 7/8 data (from Figure 4) in Figure 13. A linear regression fit to this entire data base results in the following correlation between burst pressure (BP in ksi) and bobbin voltage (v in volts):

1

[

)

The coefficient of correlation for this regression fit is 0.88 and the error of the BP estimate is 1.083. The linear fit is plotted in Figure 14 as a straight line.

The lower 95% confidence interval adjusted for lower tolerance limit strength properties (LTL)

- at operating temperature for the combiner; data leads to 4.55 volts at 3aP and 22.1 volts at SLB burst pressures These values are compared to 6.2 and 28.9 volts, respectively, for the 7/8 inch data only. Thus the 3/4 inch data tend to lower the burst pressure limits compared to the 7/8 6

inch data only, in part, this is due to the lower mean fit for the 3/4 inch data as discussed in Section 8.

The destructive examinations of the 3/4 inch tubes with the lower burst pressure values tend to indicate smallligaments that could reduce voltages but not contribute significantly to the burst pressure. Even though there are more 7/8 inch model boiler burst tests than obtained for 3/4 inch specimens, the comparable voltage burst pressure data points have not been found in the 7/8 inch data base.

7.2 SLB Leak Rate Correlation The adjusted 3/4 inch voltages and SLB leak rates from Table 5 are combined with the 7/8 inch j

data (from Figure 6) in Figure 12. The following correlation between SLB leak rate (O in l'hr) and bobbin voltage (v in volts) is developed for the combined database:

[

]S 13 c,_.1

__~;.,_..

.. ~.

Tne coefficient of correlation for this regression fit is 0.72 and the error of the estimato is 1.439. The regression fit is shown in Figure 15 as a straight line.

Thus combining the 3/4 inch and 7/8 inch data tends to increase the SLB leakage at a given bobbin voltage compared to the 7/8 inch data alone. Similarly, the upper 95% confidence band is also increased.

I

]

14

~.... - -. -. - -..

.... -. ~. -... -

8.

CORRELATIONS FOR THE COMBINED 3'4 AND 7/8 INCH DATA I

EASED CiN APPROX 1L%TELY EQUAL MEAN F1T The theoretical approach described in Section 6 and the resulting correlations developed in Section 7 do not fully resolve the differences in the fit of the burst pressure and leak rate correlations with bobbin voltage, it may be that the difference cannot be fully resolved.

However, a more rudimentary alternate approach is applied in this section to investigate the feasibility of resolving these d;fferences.

The basic premise in this approach is that the voltages (from the 3/4 inch specimens) can be scaled using a multiplicative factor such that the combined data (both 3/4 inch and 7/8 inch) can then be represented by a single correlation with the scaled bobbin voltage as the independent parameter.

E 8.1 Burst Pressure Correlation Applying this approach to the burst pressure data, a f actor was developed on bobbin voltage to obtain nearly the same values of regression analysis coefficients as obtained for the 7/8 inch data only. The scaling on burst pressure is described in Section 6. The resulting factor of 2.4 was applied to the bobbin amplitudes of 3/4 inch diameter tubes. The combined database after application of the factor is plotted in Figure 16. A linear regression analysis is performed to result in the following correlation:

[

]S The coefficient of correlation for this regression fit is 0.88 and the error of the BP estimate is 1.061. The linear fit is plotted in Figure 16 as a straight line.

The lower 95% confidence interval (LTL) for the combined data leads to 5.4 volts at 3 AP and 28.0 volts at SLB burst pressure. These values are comoared to 6.2 and 28.9 volts, respectively, for the 7/8 inch data only, in this case, the lower confidence limits are lowe:ed for the combined data compared to 7/8 irich only near the 3 AP limit with negligible change at the SLB pressure differential.

15

9.

CCMPARISCW OF CORRELATICNS 9.1 Burst Pressure Correlations Correlations have been developed for the 3/4 inch and 7/8 inch diameter individually. The correlations express the burst pressure (BP, in ksi) as a function of log of the bobbin amphtude (v in volts). A linear fit was developed for the 3/4 inch data (Section 5) and for the 7/8 inch data (see Figures 3 and 4). The 3/4 inch data was scaltd using a theoretical approach and a correlation developed for the resulting combined (3/4 and 7/8 inch) database in Section 7.

Further,in Section 8. an empirical approach was used by applying an adjustment factor to the 3/4 inch data to yield an "approximately equal mean fit." The resulting combined, adjusted database yielded another resulting correlation between burst pressure and bobbin voltage.

Table 6 provides the linear regression analysis parameters for the leak rate and burst pressure correlations. It may be observed that the 3/4" data (unscated) evaluated individually has significantly different linear regression analysis parameters than the 7/8" data. When the 3/4" data is scaled in the manner described in Section 6 and combined with the 7/8" data, however, the linear regression analysis parameters are comparable to the parameters for the 7/8" data, alone. In fact the data spread (reflected in the " Error of BP") is also similar (1,01 vs.1.08).

Thus, the statistical analysis methodology used to develop the prediction intervals for the 7/8" data is confirmed to be consistent with the additional 3/4" scaled data.

An alternate adjustment of the bobbin voltages for the 3/4 inch specimens by a f actor 2.4 was -

made to compare the 3/4 inch and 7/8 inch data as described in Section 8. This adjustment method results in essentially the same mean fit parameters while slightly increasing the data spread (" Error of BP") from 1.01 for 7/8 inch to 1.06 for the combined data. At the lower 95% confidence interval, the voltage at 3.iP structurallimit decreases from 6.2 to 5.4 volts for the combined data while the voltage for the SLB pressure differential is negligibly different for the two cases.

9.2 Leak Rate Correlations in the case of the SLB leak rate, the empirical approach of obtaining an "approximately equal mean fit" has not yet been applied. The correlation from the combined database is available only 16 I

J

l l

i for the methodology app!ying theoretica: scaling as discussed in Section 6.

The 1;near regression analysis parameters for the leak rate correlations are provided in Table 6.

As for the burst pressure correlations, it may be noted that the 3/4* data (unsealed) evaluated individually has significantly different linear regression analysis parameters than the 7/8*

data. When the 3/4" data is scated and combined with the 7/8" data, however, the linear regression analysis parameters are quite comparable to the parameters for the 7/8" data, alone.

In fa01 the data spread (reflected in the " Error of 0") is almost identical. Thus, the statistical analysis methodology used to develop the prediction intervals for the 7/8* data is confirmed to be consistent with the additional 3/4" scaled data.

17

10.

PREUMINARYCONCLUS!ONS The 3/4 inch and 7/8 inch data show very similar trends for voltage versus burst pressure and SLB leakage correlations. However, voltage normalizations based on current eddy current cractice (Section 3) lead to significant differences in the correlation parameters. To provide comparisons of the two data sets, the 3'4 inch voltages were scaled on a theoretical basis (approximately equal voltages for equal crack lengths) and an empirical basis (approximately equal mean fits) as summarized below.

At the present time, there is insufficient 3/4 inch data to fully evaluate the feasibility of merging the 3/4 inch and 7/8 inch data into a combined data base. The theoretical scaling of the 3/4 inch data, as described in Sections 6 and 7, does not adequately merge the means of the 3/4 inch and 7/8 inch data sets. For this reason, the theoretical scaling methods applied do not appear to be an acceptable basis for combining the data for a voltage / burst correlation applicable to either or both tube sizes.

In the case of the burst pressure correlation, the 3/4 inch and 7/8 inch data were also merged by normalizing the 3/4 inch data to obtain approximately equal mean fits for both tube sizes.

This approach might be considered to be arbitrary as there is no theoretical basis to support this -

approach. This approach requires that the 3/4 inch voltages be increased by a factor of 2.4 compared to the factor of 1.46 for theoretical scaling. Pending availability of additional 3/4 ind data and evaluation of associated adjustment factors for the 3/4 inch data, the equal mean fit is judged to be more appropriate than the theoretical basis for comparing the combined data with the 7/8 inch data set.

The equal mean fit approach for the combined data reduces the voltage at a 3AP burst pressure of 4380 psi from 6.2 volts for the 7/8 inch data to 5.4 volts for the combined data while the voltage at SLB burst pressure is essentially unchanged (28.0 vs 28.9 volts). The influence of the combined data on tube repair limits requires a reduction of the voltage at 3AP for NDE uncertainty and crack growth a!!owances. These allowances are typically about 16% and 35%,

respectively (although values of 20% and 50% have been applied conservatively). Reducing the 3 AP voltage limit by a factor of 1.51 for the NDE and growth allowances would lead to a tube repair limit of about 3.6 volts for the combined data compared to about 4.1 volts for the 7/8 inch data only. Applying the conservative 70% allowance and using the burst correlation with 18 l

l

7/S inch data alone, the 3.tP voltage of 6.2 volts yields a tube repair limit ol 3 6 vo!ts as developed in Reference 1. Thus the conservatism of combining the 3'4 inch and 7/8 inch data has the same effect on tube repair limits as conservatively increating NDE and growth allowances from 51% to 70%

At the present time, no firm conclusion can be drawn as to whether the 3/4 inch and 7/8 inch data should be evaluated separately or combined. Further data collection and a more extensive review, such as by the EPRI ad hoc committee for tube repair criteria, is required to establish the basis for 3/4 inch data evaluation.

19

11 REFEP.ENOES 1.

• J. M. Farley Units 1 and 2 SG Tube Plugging Criteria for ODSCC at Tube Support Plates,*

WCAP-12871, Revision 2, Westinghouse Electric Corporation, Proprietary Class 2, Febn'ery 1992.

20

Table 1 THEGMAL AND HYDRAULIC SPECIFICATIONS Primary loop temperature 327'C (620'F)

Primary loop pressure 13.8 MPa (2000 psi)

Primary boiler inlet temperature 324'C 13'C (015'F 1 5'F)

Primary boller outlet temperature 313*C t 3'C (595'F 1 5'F)

Secondary T at at 6.1 MPa (900 psi) 278'C 13'C (532*F 5'F) s 2

Steam bleed 0.1 - 0.2 l' day 8 cm3 min (continuous)

Blowdown

/

4 16.28 x 10 kcal/m2 hr Nominal heat flux 2

(60,000 Blu/ft.hr) 21

Table 2 TEST SPECIMEN

SUMMARY

SPEC.

CREVICE DAYS IN

jfL, GROLP

- CCff1GURt3ON TEST 590 1 EPRI Frit 8

591 1 EPRI Frit 8

590-2 EPRI Frit 15 590-3 EPRI Frit 15 591 2 EPRI Frit 10 591-4 EPRI Frit 10 596 3 EPRI Dual Cons.

5 590 4 EPRI Frit 19 591 3 EPRI Frit 21 596 2 EPRI Dual Cons.

10 595 1-EPRI

Dual Cons.

30 598 1 EPRI Moch. Cons.

27 598 2 EPRI Mech. Cons.

27 598 3 EPRI Mech. Cons.

27

-594 1 EPRI Dual Cons.

85 604-2 EPRI Fril 7

600 1 SPNSH Dual Cons.

10 600 2 SPNSH Dual Cons.

14 600-3 SPNSH Dual Cons.

38 601 1 SPNSH Frit 12 601-2 SPNSH Frit 12 601 3-SPNSH Frit 17 601 4 SPNSH Frit 17 601 5 SPNSH Frit 17 601 6 SPNSH Frit 17 22

Table 3 NDE Results for 3!4 !nch Dlarneter Model Doller Speeltwns Model Boiler Eh0_Co !

RPC ha spe;..No.

h3

,1.D m hb fh t!Crans

~

1 590 1 2

590 2 3

590 3 4

500-4 5

591 1 6

591 2 7

591 3 0

591-4 9

594 1 10 595 1 11 596 2 12 596 3 13 598 1 14 598 2 15 598 3 16 600-1 17 600_

18 600 3 19 601 1 20 601 2 21 601-3 22 601 4 23 601 5 24 601 6 25 604 2 RPC data have not yet been anatyzed.

23 ei nni mmme

Ta' le 4 a

Burst Pressure and Leak Rate Data for 3/4 Inch Specimens Mocol Burtt Destructive Eram.

Boiler

_ Leak Rate flM Pressure Loacth finrM th, Esse No.

N. De AD SLB AP Imid fan.

Thaawau

~

1 590 1

~

2 590 2 3

590 3 I

4 590 4 5

591 1 6

591 2 7

591 3 8

591 4 9

594 1 10 595 1 11 596 2 12 596 3 13 598 1 14 598 2-15 598 3 16 600 1 17 600 2 18 600 3 19 601 1 20 601 2 21 601 3 22-

-601 4-23 601 5 24 601 6

- 25 604-2 L

24

f Table 5.

Adjusted 3/4 Inch Diameter Volts. Leakage and Burst Pressures for Comparisons With 7/8-foch Diameter Data Model Boiler leakage Burst Pressure SLB Leak Adjusted SLB Burst Adjusted

& Soacimen Bobbin Volts Adiusted Volts Miusted Vofts Rate ft/hri Leak Rate it'hri Pressure Burst Pressure (Note 1)

(Note 2)

(Note 3)

(No,e 4)

_g 2

590-2 3

590-3 4

500-4 5

591-1 6

591-2 7

591-3 8

591-4 9

594-1 10 595-1 11 596-2 12 596-3 13 598-1 14 598-2 15 598-3 16 600-1 l

17 600-2 18 600-3 19 601-1 20 601-2 21 601-3 1

~^

601-4 23 60*-5 24 Gb.-6 25 604-2 I

Notes:

1

1) 0.750* bobbin volts adjusted by a factor of 1.252 to obtain equrvatent 0.875* volts at equal crack lengths.

2) 0.750' bobbin votts adjusted by a factor of 1.46 to obtain ecurvaient 0 875* volts for equal burst pressure.

3) 0 750 leakago ratos adjusted by a factor of 0.9 to obta;n leak rates comparabia to 0.875* tubmg.

l

4) 0.750" burst pressures reduced by a factor of 1.11 to obtain burst pressure at flow strength of 0 875* tubing.

l 25 i

Table 6.

Linest Regression Analysis Parameters for 11urst Pressure and Leakage t

?

g m

M e

26

CONDENSER REFERENCE COLUMN p--

PRESSURE I

$ WITCH l I

I CONDENSATE l

l TANK l

FROM " A" LOOP bI I

I I

JJ U

TO

" LOOP l

~"

r--'

l' )

O i

l L __ _ _

J i

l

~

PRESSURE f__

l SWITCH VOID l

DETECTORk MAKEUP i

TANK j

\\

l S A M PLE l

L_-

I l

N.

O l

l l

CELL

_ _.J i

I

~

L_JI l

+

9 l

v i

w l

L _ SAMPLE

_3_____________,_____J h

U Figure 1. Schematic of Vertically Mounted Single Tube Model Boiler 27

i, W

CONDEN$tR REFERENCE il l

COLUMN E

PRES $URE t-- _ -

SwlTCH l

I l

a g.l_J U

i l

CoNotN$att l

TANK ORIGINAL l

MODEL g

6 BOILER 4

h I

PRE $$URE r - --

I Switch VOID otTECToR$

l F~

l w

Tm I

I N

l t__

I j

N s l

I l

STEAM CELL

.)

LINE j

g 1

L_Jl I

l l

l L-S_A MPLE

_ _ __ a 7-HORIZONTAL MODEL BOILER

/

I j

~

'l 4

y RECIRCULATION r --

Il l M

f A

LINE L___

J

$ A MPL1 DRAIN PRIM ARY l

SYSTEM Figure 2. Schematic of Honzontally Mounted Single Tube Model Boiler 28

Figure 3.

Durst F ressuto vs. Bobbin Voltage for 3/4 Inch Tubes

~

_g 29

Figure 4.

Burst Pressure vs. Bobbin Voltage for 7/8 inch Tubes 9

30 l

Figure 5, SLB Leakage vs. Bobbin Voltage for 3/4 inch Tubes Q

i W

m l

31 m

---

wei

, --., - -,we w

,--i,-

~ -,

+

I l

~s Figure 6.

SLB Leakage vs. Bobbin Voltage for 7/8 Inch Tubes

_9

~

i

~

i j

i 1

32

. -.. - ~. - -

Figure 7.

Burst Pressure vs. Unscaled Bobbin Voltage for 3/4 and TiB Inch Tubes

.g L

m 3/4 inch O 3/4 inch

• 7/8 inch Model Boller Pulled Tube Model Boiler Specimens

- Not Tested

!,pecimens to Burst

(

0 7/8 inch

• 7/8 inch O 7/8 inch Model Boiler Pulled Tubes Pulled Tube Specimens -

- Not Tested Not Tested to Burst to Burst 33 Chart Figure 7 3/9/92

Figure 8.

SLB Leakage vs. Unscaled Bobbin Voltage for 3/4 and 7/8 inch Tubes

_.9 t

34 Chart Figure 8 3/9/92

FIGute 9 Bobbin Voltage vs. Crack Length - 3/4" and T'8" Afodel Boller Specimens 0

I l

35 Chart Figure 9 3/6/92 l

. - ~

Figure 10. Bobbin Voltage vs. Crack Length, Essentially Single Through Wall Axial Cracks, 7/8" Dia. Tubes 9

36 Chart Figure 10 3/6/92

Figure 11. Bobbin Voltage vs. Crack Length, Essentially Single Through Wall Axla! Cracks, 3/4" Dia. Tubes 9

37 Chart Figure 11 3/6/92

Figure 12. Normalized SLB Leak Rate vs. Bobbin Voltage Normalized to 0.675" Diameter Tubing

_,g l

1 38 Chart Figure C-7 2/5/92

I Figure 13. Normalized Burst Pressure vs. Bobbin Voltage Normalized to 7/8 Inch Diameter Tubing

__, g

(

l l

I l..

l l-1 l

39 Chart Figure 13 3/9/92

i 4

Figure 14.

Burst Pressure Correlation for the Theoretically Scaled, Combined Database Q

M l

40

Figure 15.

SLB Leakage Correlation for the Theoretically Scaled, Combined Database

_9 I

l 41

]

Figure 16.

Burst Pressure Correlation for the Empirically Scaled, Combined Database

_9 42

-