ML20138P267

Probability of Disk Cracking Due to Stress Corrosion

ML20138P267
Person / Time
Site:	Comanche Peak
Issue date:	08/31/1984
From:	UTILITY POWER CORP.
To:
Shared Package
ML20138P244	List: ML20138P260 ML20138P267 TXX-4297, Forwards Engineering Repts ER-8401, Ultransonic Disk Insp & ER-8402, Probability of Disk Cracking Due to Stress Corrosion, for Advance Review Prior to 840919 Meeting TXX-4323, Informs That Engineering Repts ER-8401 & ER-8402,transmitted w/840911 Ltr,May Be Entered Into Docket Proprietary Documents
References
ER-8402, NUDOCS 8512260022
Download: ML20138P267 (18)

Text

.

l

, Utility PowerCorporation "J ENGINEERING REPORT ER-8402 PROBABIIJTY OF DISK CRACKING DUE TO STRESS CORROSION COMANCHE PEAK UNIT I August 1984 PROPRIETARY INFORMATION OF UTILITY POWER CORPORATION Not to be reproduced. copied or disseminated without the express prior written content of utmtv ro ., corporation.

P J 4

8512260322 841005 PDR ADOCK 05000445

^ '

PDR 4-PRINTED IN U S a

I o )

Probability of Disk Cracking Due to Stress Corrosion Introduction The probability of turbine missiles from our 1800 r/ min nuclear steam turbine-generators with 44-inch last stage blades is documented in the Engineering Report No.

ER-504 of October 1975 to be 2.1 x 10-7 per unit year for a 4-flow turbine.

This number actually defines the probability of the occurrence of a 2120% speed

~

~

or 120% overspeed event. '

Engineering Report No. ER-503 on turbine missile analysis describes our LP turbine design with the innercasings featuring crash rings at the circumference of the last stage blade rows. With this design, the threshold speed for producing an external LP turbine disk missile is:

Disk No. Threshold Speed for External Missiles

1. 1 2900 r/ min 161% of rated

1. 2 3220 r/ min 179% of rated

1. 3 2900 r/ min 161% of rated

1. 4 2960 r/ min 164% of rated

1. 5 3042 r/ min 169% of rated The following evaluation of disk cracking due to stress corrosion assumes speed operation below 120% speed and is, therefore, a probability of an LP turbine internal missile only.

Operating Speed .

Evaluating the probability of stress corrosion cracking for a speed 2120% of rated speed would lead to an insignificantly low probability because the probability of the accuracy of a 20% overspeed event being so small. Even an overspeed of 210%

presumes a failure of our highly redundant control system which could, however, occur 1

\.

a *

[

with a probability of about -

2 x 10-3 per unit year.

Since this again would lower th probability for a disk crack due to stress corrosion by about three magnitudes, we have, in the following report, conservatively evaluated the ,

probability of a disk failure at 110% rated speed. This will cover any small overspeed event which can occur during speed operation when the unit is not synchrozined and j which can also be the result of load rejections.

\

Crack Initiation y i,

Corrosion crack init a' tion has' occurred on LP turbine disks in U.S. nuclear power plants. To-date, 240 LP turbine disks of KWU design have been in operation in PWR and BWR powerI plants, accumulating approximately 1000 LP turbine disk service years. However, only three LP rotors were inspected and no crack initiation was found even after' 83,000 service hours. Since we do not expect stress corrosion cracking of the KWU-designed rotors which operate in KWU-designed power ' plants, 4

and because cracking has been, found in U.S. power plants, we have chosen as a probability base of disk crack initiation'U.S. nuclear plant experience as described

, below. m j .I z

y ,

Based on EPRI Report 2429 LD Vol.1 of June 1982, arid EPRI Steam Turbine Disk Integrity Seminar of, December 1 - 2,1983 (General Electric Nuclear SCC Experience),

the following probability of disk cracking can be established:

, c > , '

From 2429 LD Vol.1 Report Number of Number of Disks  % of Disks Probability of Inspected with with Cracks with 90%

Disks Indications Indicationr/ Confidence Level s

Disk #1 121 s 5 4.1 0.078 i Disk #2 121 17  ; 14.0 . 0.198 Disk #3 121 11 9.1 0.136 Disk #4 121 1 0.8 0.032 Disk #5 121 .

1 0.8 ' O.032 Disk #6 118 0 0 0.019 l

1

,2 '

. . _ . _ , _ _ _ _ . . _ _ _ . l_ , ,. _ L - _ _. . n.. . _ _ _ . . _ . . - . .

~

~ ~

From EPRI Seminar of December 1-2,1983 ~ ~ ~ ~ ~ ~ ~ "

, Number of Number of Disks  % of Disks Probability of Inspected with with Cracks with 90%

Disks Indications Indications Confidence Level i

0.2 in.

Disk #1 180 0 0 0.013 Disk #2 180 0 0 0.013

, Disk #3 180 3 1.7 0.037 Disk #4 180 17 9.4 0.133 Disk #5 180 25 13.9 0.200 Disk #6 180 1 0.6 0.022 Disk #7 180 0 0 0.013 Disk #8 146 0 0 0.016 These statistics from U.S. plants reveal the highest probability for corrosion cracking is found in Disks #2 and #3 and Disks #4 and 5 respectively. Since these disks operate under similar conditions to our disks #2 and 3, we have chosen 0.2 as a conservative probability from these actual experiences. Disks which operate above the Wilson line have not shown any indication of stress corrosion. It must, therefore, be concluded that the probability of cracks in earlier disks as listed above must have been arrived from units without reheaters. Our units are reheat turbines which only would operate for very few hours without reheaters. However, we assumed 30 hours3.472222e-4 days <br />0.00833 hours <br />4.960317e-5 weeks <br />1.1415e-5 months <br /> operation without MSR per year which would reduce the probability of cracks for our disks #1 to 1

0.2 x 30 Hours / Year 6000 Service Hours / Year = 0.001 -

All other disks, #4 - 6 and #6 - 8 respectively show probilities in the range from 0.032 to 0.013. Because our Disks #4 and 5 operate under similar conditions, we have chosen to apply a probability of 0.03 as basis for all of our Disks #4 and 5.

Corrosion crack initiation is influenced by:

Operating Steam Conditions i

Stress Level and Disk Material Purity of the Steam / Water Cycle Local Stagnation of Flow

- Crevices l

3

Operating steam conditions for all turbines in U.S. power plants are assumed to be the same. Stress levels and disk materials vary, but no significant differences have been found in regard to disk crack initiation. A major difference, however, seems to exist  ;

in the steam / water cycle cleanliness, but this fact has not been considered as part of this study.

Flow stagnation and crevices play a vital part in initiating corrosion cracks. These phenomena are design-related and do not exist with our disk-type rotor design. Our keyways are located where the metal temperature is higher than the temperature of the surrounding saturated steam which eliminates condensation (see Figure 1). The keyway areas are stress relieved by a large circumferential groove in the disk as shown in Figure 2. The groove is open over the entire circumference to the outside by a 1 mm (40 mil) gap between the disk and shaft which allows breathing and therefore, keeps the keyway from becoming a crevice trapping corrosion products.

Crack initiation testing under various conditions revealed that disk material with 145 Ksi yield strength is not susceptible to stress corrosion under controlled environmental conditions, even when oxygen is present. However, crack initiation must be expected when carbon dioxide or other impurities are nracant fenm air-in leakages, as described in the A'merican Power Conference Paper " Design, Operating and Inspection Considerations to Control Stress Corrosion of LP Turbine Disks".

Further test results shown in Figure 3 indicate that not only tests with carbon dioxide, but also with air, led to similar crack initiation. An even more important finding is that tests with oxygen did not initiate cracks after up to 30,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br />, but it took only approximately 2,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br /> until crack initiation when the refreshing of the test cycle .

with mixed bed ion filters was shut off and the low conductivity level was not any longer maintained. This test closely simulates conditions such as flow stagnation and crevices because all these circumstances drastically increase the conductivity. Such conditions do not exist with our keyway design. Therefore, it must be conluded that our advanced keyway design is a reason why stress corrosion cracking has not been found with our disk-type rotors.

4

In accordance with this finding, it must be conservatively concluded that the time until crack initiation with KWU-designed disks is at least 15 times longer:

30,000 Test Hours in Oxygen without Crack Initiation 2,000 Hours until Crack Initiation without Refreshment 15 This quotient for crack initiation time can directly be used for the reduction of the crack initiation probability in a given time:

Crack Initiation Probability of the KWU-designed LP turbine disks for Comanche Peak Units #1 and #2 are therefore:

LP Turbine Disks Crack Initiation Probability Disk #1 0.001 i 15 = 6.7 x 10-5 Disk #2 0.2 i 15 = 1.3 x 10-2 Disk #3 0.2 + 15 = 1.3 x 10-2 Disk #4 0.03 + 15 = 2.0 x 10-4 Disk #5 0.03 i 15 = 2.0 x 10-4 Crack Growth Crack growth in LP turbine disks can be defined as a function of the operating temperature of a disk and the yield strength of the disk material. The empirical equation for crack growth in LP turbine disks of nuclear units as given in the ASME paper 82-JPGC-PWR-31 seems to reflect the findings in nuclear plants and has, therefore, been used in this study:

(-4.968 - + 0.0278

• R pO.2 I The yield strength value has been taken from Engineering Report ER-8102 Nov. 81 Rev.1, which is the actual yield strength in Ksi for each disk measured at 20*C (68'F) ambient temperature. The disk metal temperature T used in the equation for crack growth is the maximum temperature in the keyways when operating at full load, arrived from a calculation of the isothermic lines in the disk / shaft system (see Figure i

1). The relationship of crack growth probability density over the crack depth of a corrosion crack in a disk i at the time t has been arrived from a normal distribution for the natural logarithm of the crack growth rate assuming the standard deviation of s=0.587. The distribution of f t(ao g) is calculated up to 3 s as the assumed maximum crack depth and is shown in Figure 4.

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

Critical Crack Depth Crack Initiation has been assumed to occur at the most critical locations with th smallest critical crack size which are the keyways. Critical crack depth is calculated from the calculated local stress at 10% overspeed and the fracture toughness arrived from the R p0.2 strength and the notch impact test results as measured from the individual disk material (see ER-8102). The remaining influence factors of the critical crack depth have been used as follows:

Variation of crack form from cracks of unlimited length to half circle cracks through a uniformly distributed random variable from 0.77 to 2.2.

Reduction effect of stress corrosion crack configuration on stress intensity through a normal distributed random variable with a mean value of/4=0.65 and a standard deviation of s = 0.175 which covers the presently available data on crack configuration, as shown in Figure 5.

The resulting density function for the critical crack depth g (ac g) has been cut off at the smallest possible critical crack depth which is the theoretical corrosion crack of infinitive length without branching (see Figure 6.)

The following Linear Elastic Fracture Mechanics (LEFM) equation is used:

2 a = K Q IC d .

1.21 x ,k/K .

c[ 2 ac = critical crack depth Q = crack form parameter with random variation from 0.77 to 2.2 K IC

= fracture toughness from tests for each disk (yield strength and notch impact tests) d

= calculated keyway stress for 10% overspeed kjg

= reduction of stress intensity due to branched stress corrosion cracking (mean valuef=0.65, standard deviation s=0.175) d/2 = radius of keyway in disk hub bore

Probability of Disk Rupture The probability of a disk I with an initiated crack to fall in the time period t is equal to the probability of a crack having grown to the critical crack size in the same time t .

Pg (t) = Probability [ao g (t) ? ac g)

The failure probability of a disk can be evaluated through the density functions f t ("o i) and g (ac g) with the convolution integral:

4c2 f f Pg (t) ='( f t(agg)

• g (ac i) *dso g' d acg J

0 0 This equation, with the data for crack growth and critical crack size for each specific disk allows the evaluation of the probability of disk rupture by numerical integration.

A distribution density from such evaluation is shown in Figum 7.

The probability P (t) of a disk failure within a turbine-generator with n disks that would occur in the time t, is with small value of P g(t) equal the sum of the failure probabilities of all disks. With a probability of the crack initiation of qi the following equation is applied: .

n P (t) 2, Pg (t) 'qg i=1

. 7

r

• Conclusion The probatiility of a postulated disk falling within an LP turbine of the Comanche Peak unit #1 within a time span of 50,000 service hours has been evaluated and listed

~

in Tables I and H. The total probability for the turbine-generator from both LP turbines amounts to:

E=NIf=SEE====I======

This result covers any overspeed event with f.110% speed, since the probability of a '

10% overspeed has been assumed to be 1. The probability of reaching an overspeed of more than 10% assumes a failure of our control system. The probability of such a control system failure is about 2

• 10-3 which reduces the total probability of disk cracks at higher than 10% overspeed to be an insignificantly small value.

4 i

A

l

)

l REVISION

/

0

__ b c

g Superheated Steam T = mn9'F I '

310 j j ,s iI Ii Ii '

226 C g 395 / '

U I, ,[k if 5 155 T = 500'F Saturated f Saturated Saturated Saturated Steam Steam If } j 355

! Steam .I Steam

. T = 316'F ll T = 225'F T = 192* F . T = 158'F i Ik 240 /158 287 , ii l

3b8 251

/ / 170 380 35st \ 230 l 163 332 1 3 19s t 4, 5

,@ L 80 2\b225\ 205 192 ! s /;

j 1 4 181 \ / 167 356 '  !

287 1 I 198 192'

\ S 260 290 k; \

s

\

175

[ f185 167 180:

i ii j190 i \. \ r u ti s

LP Turbine Rotor Half with Disk #1 through #5 is n!

d  ?

IE eg Ilt'!

yu il&

5 I'I sg

0,s':1

$:580

ill ISOTHERMAL LINES OF l'IllityI'tmer F OlSK TYPE NUCLEAR CorlM >riallott J LP ROTOR FIGURE 111E84.071

e Circumferential Stress Relief Groove ___ _

Radius:

R10 = 0.39 in.

' op R22.5 = 0.89 in. ,

REVISION

. AVA / ,

Diameter: #_ _ _ _ _ _ _ _ ;

/ _

-.{- 3 0 208 = 0.79 in. b ----

~~~~~ '

b Shrink- P V

C Fit 20s kB d

• k.

y

, er, Detail A r... ...$

,[

/ ~1mm (a

" '- 5 Locking Gap

,. e.3  :./

Pins

iGi jiji L

$$Ni:

n e.$$5; iii::a ..

Shrink-Fit r-

'l -.

.i;iS%f.4 .. _ ..__

8

• p[i 9 .,$

8g 3

.(...

.!.5 i i . . . . T!.k 18 'hs . - i!!!D * .,. -

.:<g j.

?

W csw :.

of Vny! )

5 g ,. .. Shrink-Fit 69 }

I Free Regions Sect. ion B-B E5 a!! i5 /

s .. . ._

maa Disk #S i

SHRINK-FIT AND t'ailleyINmer F 1 KEYWAY CONFIGURATION O >fl ""I* 3 OF LP TURBINE DISK #5 FIGURE 211E84.072

i l -

N A n W G \ '

PROPRIETARY INFORMATION OF \"D g UTILITY POWER CORPORATION Not .duced. Co.ied

]

M * '

...........'..e,itto , dbe em.nated.

,e . o e

u, .., e..., c .

I i,

l Test Time (Hours) l 1 a a -

a i

V. h ?n f

i

.I M

-4 Oxygen Saturated

@ Oxygen Weak ggi

$>0 M H>

o m

e, a c*

+

n x -

ze*

AE wu, Z#g am UEd E o,. A _ 8{ Oxygen Weak without lon Filter Refreshment

If $ g !!L i m.o cu 5 m a.L.

i

m<Z i A .<

3

! A 15 m .g

-< r * *-

j a$

.=

Carbonated CO 2 l m e. O "O 3 E ,

S a '

x

• i d lE$g-e

! me  !.

l 3 Iy Aerated  ::r, e o s o '

C Y,, .7 3 l . m ,

m D """

) u 2 #

2 2" 3 m =1 4

cm -m N ,

o l N u

i 9

1 .

I

I Operating Temperature 235*F Disk Yield Strength 133 Ksi 1/in. Ilmm

/

0 0 '

2.5 b

C d

e 2.0 E O

.w h 1.5 0.06-O o

>. fto Ic / (a i) b>, 1.0 0.04 =

E 3

h.

a.

Io

=

0.5 0.02 =

eS E e g 55! -

5 11 18

== !n .a6 0 . , , ,

25 j g 5 0 l 10 l l20 l 30 l 40 mm h ggl  %  %  % 1 1% 1% inch eg e u 3r jil s } f, Crack Depth 15 3$$

COMANCHE PEAK UNIT #1 l'8IIIIYI't* CF FMl CRACK GROWTH OF DISK #2 ('d'fl M'fd83"8 8 TURBlNE SIDE OF LP ROTOR #1 FIGURE 411E84.074

REVISION

/

0 b theoretical calct4ations experiments C muttelo bent branch it miiture cracks d - (a) a M (c) (d) (e) t *

--n / _4 44-4 al ~A

- 1,0 -V O I

.?': k I>

\

E

$ 0,0 E

sA \

. <1 g

E Y vb. _ "' _ _

E_ ~~~1~~ I-~~ Mean Value e, o,e. G*

-a  % vs n

ll4 p = 0.6s

. a u* -

4 . .i jg -

py .g Standard u g o,4 $ ge; regis Deviation g ,

/A d4 S = 0.175 c u ., -

85 $ll D E 0,2 8C3 lu'A

,e , stress coerosion mgj 4 5* A g crack rnodels 60*A

+-

I 'm "aC'o E se vklein/Hodusa_k1978 SKelagawa1975 OSpeedel1971/1983 SM WU 1984 if

[

g Cleida 1978 <lfheocaris1972 Owdsonicherop6o1983 Axwut983 Clo1978 Wieh 1977 BClark el al1982 g .

8:; a !,I, e m Ils 19 O E E u$ %1!!

[0

<g... lId b2 68

$3 8N

.a:23 z4 22 INFLUENCE OF CRACK I'8IIIIFIi*Cf IJl 1 CONFIGURATION ON STRESS C4 trlx >tiilloit INTENSITY FACTOR FIGURE 5/1E84.075

Fracture Toughness 202 Ksl 6 Tangential Stress at 110% of Rated Speed 84 Kal REVISION

/

0 itin. Ilmm

b C

g,3 0 004-d e

.c U 0.075 0.00t g(aci) o 3

0.050 0.002=

1 r

2 31

!j a. 0.025 0.001m 8 %!

k I5 gg h a t .& 6

$b

-8 !i$

366 (

o. i i 5*$ 0 h l 20 l $0 l 40 l 50 mm kIg u 1%

1 1% 1% 2 inch

$(5 15 315

$Ih Crack Depth COMANCHE PEAK UNIT #1 L'IllllYl'tmer FM CRITICAL CRACK DEPTH OF i C a>rl $ >t u ilt>n k W J DISK #2 TURBINE SIDE OF LP ROTOR #1 FIGURE 6I1E84.076

, .----%,,~ ,.--.--,..~-----,-----y-,v3, - , - - - ,-----.-,-,.--,,,w,,,- ,-,--.--.--..-.e-- , , ,. - _ ,-- - ,-.- - , - - - - - - -

50000 Service Hours Probability P,(t)= 7.6 x 10-4 REVISION f illn. Ilmm 0 90 b

C 8

d e , 3.0-8 70

/ ,

a '

e / -

///

60 /

O N:/,'l/

// l/j' 6 s0 i

W,/, / f',/ n }, '

I o {'/7l ,/

5 ,),{'

':;; / '

/, ,. /

i y <0

$ i, ,';l' ',! N: ,h,//h

, , / ,

ha hi 3 .

7 /

'fY'

'l, a.

e

, ,o ..  ?

/ ;/;oi, ' 'f h.;6 '!l ,5,,,

/ b. / ~ ,'

20

,,, , , -: ,/ ///  ;/./m/ ./ , ..,.,j , .,/

-  ? //.

j% .., Q'/ // - ,/j ' j' . ,

g

,i;,i f,'[', /,',:<,'/,,

, . , , ' / 3;, ,,,/, 's,-

' ,-),':h' .'/,,l/D '//,,/ ,//'?,.,//,. /'s's,/'- -/ *ha 0 ' -

ob b; '- .,, ,, / . .

/

. / ' '/ / ,';./,,l< /

i t i n

(( '/6/ _'O _/ /_

//, '

/'J_'i____ '

l$ ! g 0 )) 2'o 3'o /0 5o mm g$

O u

lg 1[4 8's 1 . 1h 184 I

2 2's mch

!h;33 [ Crack Depth y t

• s}

b5 hi G Jes COMANCHE PEAK UNIT #1 I*8 I l CRACKING OF DISK #2 I "' l M "HI"" Q2I 8III TURBINE SIDE LP ROTOR #1 FIGURE 7/IE84.077 e

_E1Y1 b '

L O

-,, -.. n --

[d e

~ _'. Turbine Disk # and

_ TurbinelGen. Side ITS IGS 2TS 2GS 3TS 3GS 4TS 4GS STS SGS

' Tangential Stress at

'N 110% Speed Ksl 80 80 84 84 75 75 83 83 81 81 Fracture Tou hness Ksl in 203 206 202 212 210 208 214 215 181 176 Operating Temp.

'F 287 287 235 235 196 196 170 170 167 167 Yleid Strength .

Ksl 131 131 133 129 117 114 131 127 129 127 Probabilities 7.4x 6.9x 7.6x 2.0 plg 10-3 10-3 10-4 10-4 0 0 0 0 0 0 6.7x 6.7 x 1.3 x 1.3 x 1.3 x 1.3 x 2x 2x 2x 2x 3

NI 10-5 10-5 10-2 10-2 10-2 10-2 10-3 10-3 10-3 10-3 k p 4.9 x 4.6x 9.9x 2.6 x

,qi 0 0 0 0

{[

g 10-7 10 - T 10-6 10-6 0 0 5$

h n! -

$[ ')- Probability P(t) for Rotor #1:

58 i' d*

1.34 x 10-5 NIf,e)i

f$)1%

D ,: . *' .

8t o3.*'. .I f, COMANCHE PEAK UNIT #1 L~lllliv li m t r F PROBABILITY OF DISK ( '8 H lH ililllHit CRACKING OF LP ROTOR #1 - - - - -

TABLE li1E84.078 B

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

f 4

_ R(VISION

/

0

_b C -

id Turbine Disk # and e

TurbinelGen. Side 1TS IOS 2TS 2GS 3TS 3GS 4TS 4GS STS SGS Tangential Stress at 110% Speed Ksl 80 80 84 84 75 75 83 83 81 81 Fracture Tou hness Ksl in 207 224 207 224 209 213 197 201 187 183 Operating Temp.

- 'F 287 287 235 235 196 196 170 170 167 167 Yield Strength Ksl 134 129 134 132 III 112 121 129 132 133 Probabilities 8.2x 2.3x 6.0x 1.5x Pi (t) 0 0 0 0 0 0 10-3 10-3 10-4 10-4 6.7x 6.7x 1.3 x 1.3 x 1.3 x 1.3x 2x 2x 2x 2x pt NI 10-5 10-8 10-2 10-2 10-2 10-2 10-3 10-3 10-3 10-3 I S.4 x 1.5x 7.8 x 2.0x l

I i 0 0 0 0 h 10-7 10-1 10-8 10-6 0 0 8si t lin

• &5 Probability P(t) for Rotor #2:

k 1.04 x 10- S 2, m;6!

Probability P(t) for Rotor #1 and #2

{2;j23 ;j 2.38 x 10-5 is.oi 15 )$ $

COMANCHE PEAK UNIT #1 I MF N"' I C(HlNHiHlOH gQ PROBABILITY OF DISK CRACKING OF LP ROTOR #2 TABLE lil1E84.079

.