ML18153D184
| ML18153D184 | |
| Person / Time | |
|---|---|
| Site: | Surry, Trojan  |
| Issue date: | 03/16/1992 |
| From: | Muscara J NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Serpan C NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| Shared Package | |
| ML18153D185 | List: |
| References | |
| NUDOCS 9212040327 | |
| Download: ML18153D184 (31) | |
Text
-
UNITED STATES NUCLEAR REGULATORY COMMISSION WASHINGTON, D. C. 20555 MAR 1 6 1992 MEMORANDUM FOR:
Charles Z. Serpan, Jr., Chief Materials* Engineering Branch, RES,
t
\\
-* !ii.
""1:
FROM:
Joseph Muscara, Senior Metallurgical Engineer Materials Engineering Branch, Division of Engineering, RES
SUBJECT:
STEAM GENERATOR TUBE INSPECTION;** INTEGRITY AND PLUGGING ISSUES As you know, I have been the research program manager for NRC's Steam Generator Tube Integrity Program. This program started in FY 1977 and was completed 1n FY 1988 with the validation work on the Surry 2A steam generator removed from service. The validation work on Surry was conductQd with international cooperation. The total program cost approximately 18M dollars, 4K of which came from EPRI *and consortia from France, '.Italy and Japan.
Many results, analyses and insights were derived from this program.
Based on the results, several improvements have been incorporated in an update of ASHE Code Sec. XI Appendix IV on Eddy Current (EC) testing of steam generator tubes. Other improvements are -~ing incorporated in revisions to Regulatory Guides 1.83 and 1.121 *on steam ge.nerator tube inspection and
. plugging criteria. The results and insight~gained also lead to concerns with generic acceptance of industry proposals for alternate tube plugging criteria which would allow operation.of steam generators with known through-wall cracked (leaking) tubes.
My concerns with acceptance of this alternate plugging.criteria which would permit operation of steam generators with through-wall cracked tubes are on two grounds: Firstly, acceptance would imply violation of the long standing engineering principle of defense~in-depth used by the NRC and of the leak-tight reactor coolant pressure boundary (RCPB) concept in the spirit of 10CFRSO Appendix A - these concepts have served us well for many years.
Secondly, if we accept violation of the leak-tight integrity of the RCPB, then a strong engineering case must be made and actions taken to ensure maintenance of structural integrity during operation of power plants with leaking steam generator tubes. Based on NRC's independent research results, other evaluations and experience, I believe a case assuring structural integrity with today's capabilities for detecting and sizing flaws,.for estimating crack growth and progression during.. operation, and for evaluating crack growth and stability during operation through leak rate monitoring cannot be made.
The structural integrity of degraded tubes is governed by the flaw length and depth; for tubes with through-wall cracks, the important parameter is crack length. Under MSLB conditions the critical through-wall crack length for axial cracks is about one inch. Tubes with through-wall cracks of the order.
-of one inch and longer woul~ burst or develop large leak rates un~er MSLB conditions. The comonly used eddy current (EC) techniques have poor n fO~ fflVlW f
~ '{1
Charles z. Serpan, Jr.
2 MAR 1 6 1992 capability for detection and sizing of stress corrosion cracks.
The probability of detection (POD) for cracks of interest can be quite low and averages at about 0.5. After an inspection, as many significant cracks can remain undetected as were detected. Furthermore, the depth and length sizing is highly variable and inaccurate. Cracks of critical length and longer often are sized to be much shorter and would be classified as subcritical.
Even if cracks could be reliably detected and accurately sized, then knowledge of the crack growth rates and crack progression during operation is required to ensure that cracks remain below critical sizes, by a reasonable margin, during operathm :and :!~:-h:e~~-~f!SP~-.:1~-: Stress corrosion crack growth rates are known to be-*h*igh1y*.fartinle;-:even under laboratory test conditions where the same material is testid in-the same environment, under the same stresses, temperatu~ m1;~*:one _c1rcfer of magnitude variability in the crack growth rate 1s ea$tly prodltCed. Because conditions in operating steam generators are much-*more, complex, and considerable variations exist between pl ants and evetJ. w-itfi.tittn:**same generator with respect to operating environment, he~ts.~[itt-~ thermal hydraulics, crevice conditions, stresses etc ** :~
_,,... iability is expected in service. Further, because of the:__ **-
- g environment, many different types' of cracks have been exp~r.1~ * ;.. * ~_.
- - --j.*nd secondary side cracks, axial and circumfere~t-t-t.T:ir.a
- _,.!J.f':'....,~illiirf-enced at various locations: In.
tube sheet crevices;* at top of tube* sheet; in free span zones, within the tube support plate region, at U-bends*, etc. Intergranular attack (IGA),
intergranular corrosion cracking (IGCC) and crevice corrosion cracking have occurred. Int~rgr_-~~l*.t.s.f~--~l;l'ed wfth or wfthaut the Influence.
of stress. Se~-tt:'1:.f9nltt;lf:-
t£ijibf_~at ions of IGA and IGCC _have occurred. The 'ilifir.phaT~gy:::af;lam.*".
.ll<<.t/,.. (including cellular IGA) combinations is such'* tlta'l~ as..~Jie** c_riek1 nf becomes deep or throughwa 11 the tube failure would be by burst wfttf'acct>nipanying large leak rates.
In the support plate region, cra~ks have occurred within the support plate region, have extended beyond tHe sup.,.,~ pl*te-and ~taf cracks have been accompanied by
- circumferential cracks. It 1s known from research work that environmentally assisted cracking once.;1nit1ated (by whatever mechanism) its growth 1s sustained. Cracks ihemselies can act like crevices and crack growth continues.
Even t.f cr,,c_k,~*.1nit1ate within the tube support plate region, their growth beyond tfle*su.,pport plate could be expected with time.
Many cracking modes ~ave bee~ recognized after the cracks developed, but their occurrence has na.t* been predicted or easily controlled. The mechanisms, causative factors and iynergisms are not well understood and therefore crack growth outside given zones and growth rates between operating periods cannot be reliably estimated:.
Assuming for a moilf.wit'tffff:.~if~dl!ij,.. trid/or through-wall cracks can be reliably detected and accurately sized, 'b) crack growth can be reliably estimated and c) changes in cracking ~echantsms, morphology and growthj outside of initial zones can be reliably predicted, then if these cracks are left in service, they must be monitored during operation to assure that they will not approach critical sizes and that thetr growth fs within predicted bounds.
To accomplish this monitoring, leak rate measurements and specifications are established. To assure structural integrity, the leak rate must relate to the
~
- e.
Charles Z. Serpan, Jr.
3 MAR 1 6 1992 through-wall crack length. Unfortunately, for the types of cracks of interest, correlations of leak rates to crack length and measured leak rate to predicted leak rates, show approximately two orders of magnitude variabilities. Further, in service, cracks tend to be tight, can become fouled with small particles and corrosion products, and may be surrounded and constricted by corrosion products and support structures. Therefore, through-wall cracked tubes in service tend to leak very.little whether the cracks are short or*long. Furthermore, approach of cracks to critical sizes cannot be determined since very small changes in leak rates are expected in service *
. These considerations indicate that critical cracks cannot be distinguished from subcritical ones based on the observed leak rates. *finally; when hundreds or even thousands of tubes may be leaking, how does one determine or distinguish those tubes with cracks that are or may be approaching critical size under MSLB?
How many are there? Small leakage does not necessarily mean short cracks.
In recent proposals from industry, the EC voltage has been used to relate an inspection parameter to tube integrity. This parameter does not uniquely measure the crack depth or length. For some crack morphologies of interest the voltage is not expected to relate to tube integrity for the following reasons: 1) the voltage saturates at a crack length of approximately 0.5 inches. Longer flaws will not produce a larger voltage; 2) the voltage produced is related to the crack tightness, if cracks are tight enough (and conductivity paths exist), the voltage response is low whether the cracks are long or short; 3) voltage produced is insensitive to critical crack morphologies - short, tight cracks axially aligned and separated by short ligaments produce low voltage indicative of the short segments; but from a structural point of view, an effectively long crack exists which is not characterized by the voltage. Plots of voltage vs. leak rate and of voltage vs. burst pressure have been presented. As expected, those plots show little correlation and show two to five orders of magnitude variability in the data.
Further, these plots lack data for long-tight cracks and for short cracks axially aligned with short ligament.s in between which could produce low voltage and also low burst pressure. Also a *voltage growth rate* is proposed as a measure of crack growth rate. Again, since the voltage does not relate uniquely to crack size, for the cracks of interest, all these voltage correlations have little meaning.
Based on the abov.Jt discussions, variabilities and uncertainties, I believe that the NRC position of the recent past that through-wall flaws and cracks are not acceptable is still prudent at this time.
In recent years NRR *staff have agreed that through-wall flaws in steam generator tubes are not acceptable and have adhered to the concept of maintaining leak-tightness of the RCPBo Further, NRR staff has agreed that if alternate plugging criteria is considered on a case specific basis that a strong engineering case must be made and actions taken to ensure structural integrity of tubes during operation. Enclosure (1) to this memorandum briefly discusses the issues related to steam generator tube inspection, integrity and plugging, and provides limited examples from the large body of data and.
- e.
Charles z. Serpan, Jr.
4 MAR 1 6 1992 results available to illustrate the particular point.
PLEASE NOTE THE PROPRIETARY NATURE OF FIGURE 15 OF ENCLOSURE Cl}.
A review of this memorandum and its Enclosure (1) would indicate a need to consider and resolve the key issues and questions posed in Enclosure (2).
If you have conments or wish to discuss this memorandum further, please call me on 492-3828.
~sc1!:r--
Senior Metallurgical Engineer Materials Engineering Branch Division of Engineering, RES
Enclosures:
Steam Generator Tube Inspection, Integrity and Plugging Issues Issues and Questions Related to Assurance of Maintaining..
Structural Integrity of Through-Wall Cracked Steam Generator Tubes cc:
E. Beckjord T~ Speis L. Shao R. Bosnak M. Mayfield E. Hackett.
T *. Murley F. Miraglia W. T. Russell
- J. Richardson
- 8. D. Liaw J. Wiggins
- Johftscar
~
K. Wichman E. Murphy S. Shankman P. Shewmon
e Leak Rate/Crack Size Predictions and VarjabiJjtjes Assuming for a moment that a) through wall cracks can be reliably detected and their length accurately sized b) the amount of crack growthi any changes in cracking mechanism and morphology, and growth outside of original zones can be reliably predicted and c) particular cracks will not approach critical sizes during the next operating cycle and are left in service, then these cracks must be monitored during operation to assure that they will not approach critical size. To accomplish this monitoring, leak rate measurements and specifications are established. Unfortunatelyf:for the types of cracks of interest, correlations of leak rates to crack sizes and measured leak rates vs. predicted leak rates (from fluid flow and fracture mechanics models) show approximately two orders of magnitude variabilities, Figures 14 and 15.
PLEASE NOTE THE PROPRIETARY NATURE OF FIGURE 15.- The-vartabtlity ts due to several unknown or uncontrollable factors. The length of cracks varies from the inside surface to the outside surface, these lengths are not always known or easily measured tn service; the leak path for IGSCC ts variable and highly tortuous; cracks can be very tight and of variable tightness. Further, in service, cracks can become fouled with small particles and/or corrosion products and may be surrounded by support structures and corrosion products.
Under these conditions it ts difficult to relate leak rate to crack length (to assure tt ts below critical length). Furthermore, through-wall cracked tubes.*
in steam generators leak very little tnservtce whether the cracks are *short or long because of tightness, fouling, and constrtctton by corrosion products or support structures. So, critical cracks cannot be distinguished from subcrittcal ones based on observed leak rates. Furthermore,. the approach of cracks growing to critical sizes cannot be determined since very small changes in leak rates are expected in service. Finally, when hundreds ~r ev.en thousands of tubes may be leaking a very small amount, how does one distinguish those tubes that have cracks of, or approaching critical size under MSLB conditions? How many are there? Small leakage does not necessarily mean.short cracks.
A recent draft EPRI report attempts to show correlations between leak rate and EC probe voltage, Figure 16. These correlations are used in submittals to support alternate tube plugging criteria. The log - log plot of Figure 16 shows very little correlation of voltage with leak rate. Five orders of magnitude variability ts shown for leak rates at a given voltage and one to two orders of magnitude variability in voltage for a given leak rate. The correlation coefficient for this plot is reported as 0.73 which also indicates very poor correlation. This plot also lacks data for cracked SG tubes which produce low volt!9eS.
Burst Pressure vs. Degradation The NRC tube integri'ty results indicate that tubes with short flaws exhibit more strength than tubes with longer flaws of the same depth, also tubes with shallow flaws can withstand considerably more pressure. Figure 17 shows plots from an empirical equation derived from the data for EDM notches and validated by testing of stress corrosion cracks. It is not surprising that tubes removed from service have exhibited high burst pressures; this can be predicted for short through-wall flaws or for other reasonably deep flaws.
5
e Points of ;nterest are that longer (above 0.8*) through-wall flaws exhibit low burst pressures and even relatively short but deeper flaws also exhibit low burst pressures. Figure 17 shows for example that a tube with 0.4-inch long through-wall flaw can withstand about 3,500 PSI pressure but a 0.8-inch long through-wall flaw would withstand only about 1200 PSI pressure. A tube with a 0.5-inch long flaw at SO percent through-wall penetration would withstand about 6,500 PSI pressure at 80 percent through-wall, 4,000 PSI at 90 percent through-wall, 3,400 PSI and the 0.5-inch long flaw, 100 percent through-wall, would withstand approximately 2,500 PSI.
Because voltage does not directly relate to crack length and depth for crack morphologies of interest, a good correlation between voltage and burst pressure is not expected. For example, short tight cracks (deep or through-wall) would produce a low voltage and a high burst pressure, however, a long tight crack (deep or through-wall) could produce a low voltage but also a low burst pressure. Similarly, a series of short tight cracks, axially aligned, with short ligaments in between would produce low voltage and low burst pressure. Figure 18 shows a plot of voltage vs burst pressure used by industry in support of alternate tube plugging criteria. The 95 percent lower tolerance limit on the data is shown. Besides the large variability of a factor of two 1n burst pressure and one to twg orders of magnitude in the voltage, the plot lacks data for the tight long flaws or tight short flaws,*
axially aligned that could produce low voltages and low burst pressure.
6
e TABLE 1.
ODSCC Flaw Dimensions and Bobbin Voltages*
Specimen Maximum uu surface 118W
- n Renormalized Number Deoth. %
Lenath. In.
Voltage, Vohs Voltage, Volts B-63-08 26 1.41 u.~
1.16 B-46-02 31 1.06 1.00 3.61 F-10 37 0.25 1.62 5.85 F-15 38 0.25 2.38 8.59 B-S0-10 38 0.53 1.37 4.95 B-61-07 42 0.25 0.43 1.55 8-07 43 0.66 1.48 5.34 B-83-01 44 1.14 0.46 1.66
&62-08 42 1.43 2.04 7.36 B-61-0S 47 0.89 0.62 2.24 E-11-05 50 0.&4 1.31 4.73 E-07~7 58 0.45 3.44 12.42 8-62~2 61 0.50 1.71 6.17 B:,46-04 58 0.70 0.37 1.34
. B-55-04 59 0.91 1.92 6.93 B-63-06 59 1.11 1.82 6.57 B-69-07 76 0.81 2.22 8.01 E-11-0S 86 0.44 4.57 16.50 B-10-08 99 1.09 7.24 26.14
- Data from NRC Steam Generator Tube Integrity Program-Phase II 7
e Probablllty of Detection vs 90/90 L TL for Baseline and DAARR Teams 1.0...---------111-------------
C 0
+: 0.8 0
CD...
CD C
0.6
't-o
~ 0.4
.c cu 98 0.2
- a.
+ *
+
)(
0 0
8
)C
+
t II
- * ---~---**
D A o B A C D
,c E
- X
+ y 90l90LR 0.0 ~
......,_...__,...&-___ ~-----.-------------........ ----.-~
0 20 40 60 80 Metallography Wall Loss, 0/o Figure 1 - Results from NRC Steam Generator Group Project Reported in NUREG/CR-5117 and NUREG/CR-5185 100 co
C 0
- 0 8 u
CD....
Cl)
C 0
0.6.
~
- 0.4
-*-.a ca
..Q o 0.2 C.
Team V 0.0 -I--.L.....,~-------------.r-----..------.-----,,-----.------,--~
0 20 40 60 80 100 Metallo.grap~y Wall Loss, 0/o Figure 2 - Results from NRC Steam Generator Group Project Reported in NUREG/CR-5185
~
0 0
..J --cu 3:,,
Q)....
cu E *-....
w 0 w Estimated and Actual Wall Loss for Baseline and DAARR Teams 100--~-------~~~--.---~~~-----------...
80 60 40 20 D
0,.
- ,.,.A 0 88 0
- 20 40 60 80 0
i,.
Metallography Wall Loss, 0/o 1do F;gure 3 - Results from NRC Steam Generator Group Project Reported ;n NUREG/CR-5185 0....
Team V 100 1
- -e
~
.t
~*
0
~,
L,\\
t
.. 80 tn i:
tn
- i**
0
)C
..J
- r.
)C 60
'l i
>( lie X X
X (G
~
X X X
>C X
X
,c,c "C
40 X
i
)C Q)....
(U E
.,i, 20 X
- \\,:
V) w
~~~ ~~~.
,~._*,.
0 0
20 40 60 80 100 Metallography Wall Loss, O/o Figure 4 - Results from NRC Steam Generator Group Project Reported in NUREG/CR-5117 and NUREG/CR-5185
e C
0 *--
0 Q) -
Q)
C....
0
~ -*--*-
.c C'O
.c 0...
fl.
- IGSCC Round Robin
- POD Results 1.0,-.-------------------------------
0.8 0.6 0.4 0.2 0.0 +--+
Convtnllon11 *,~
~<~------ Alltrn111 --------3t MB MC MD ME MF MH Ml MJ MK ML MM MN MO MP
. Inspection Team Figure 5 - Results from NRC Steam Generator Group Project Reported in NUREG/CR-5117 o-4
- Bobbin 80 D
?fl.
D
.c -
f C.
60 Q)
,C
'C Q) -
D m
E 40 I
Cl)
D' w
I-w 20 0 -------0--0--a:h-a::mo----.a.-----------1 0
20 i
40
- 60 80
. Measured Crack Depth, 0A, Figure 6 - Results from NR*c Steam Generator Group Project Repo~ted in NUREG/CR-2336 and NUREG/CR-5117 100 M
o-4
C *-..
.c
. 0)
C G)
..J "C
G) -
cu E *--
u, w
1-w
~ ii
~;*, J, 1.,.
. \\J!:>U i IGSCC Round Robin Team MD
- Bobbin 2.5 'T'---------------------------------------------------------------~
2.o.;,...
1.5 1.0 0.5 DD D
D 0.0 <:o-----o-----,---(%),---a-racc--..---a-..-----,--.---1"'""'"" ___,__.---t 0.0 0.5 1.0 1.5 2.0 2.5
- Measur~d Crack Length, In.
Figure 7 -
Results from NRC Steam Generator Group Project Reported in NUREG/CR-2336 and NUREG/CR-5117
.c :;:
- 1
. a. ;
Q) -
~c
... "C *,:
Q) -
(,J E *--
Cl) '
w tu C
f\\
- .i
- , *l> *.* *1
...... c IGSCC Round, A*o*bln~:ieam MF
- Bobbin 1:
d:
100------------~_-.-~-, c--~~_..--~_---_-----------------------
I.
80 60 40 20 1,,.,1 I
L,!
. I"
~-.
1.1 I
a:.
-r.
I'
-~*-
a* *..r a**..
Q.* G I,',
. T'
~/"t S,..
( *.'
/..
..1:...
. -~ C--.
...*. u
~ Goi Q...\\,a,, *:
- . ".. -... -~
- '...... }
.' '(".~ *-"I-L.:*
~
<i".
- .... Vl,
- J
~-*** C
- ..-.,.. c:. -. :.
µ
.1:::
- ~* ~*..:.,. :1 :::.
,L:
- ~
r.,
-~
- i i
t'.:
f
- 1.
~-**
a a
0 -o----"""T-~~-O,..C:KD-,----o.,.-~----.----,,---1
- O 20 (40 60 80 100 Me*a~ure~. Crack Depth 1 °/o Figure 8 - Results from NRC*Steam Generator Group Project Reported in NURE_G/CR-2336 and NUREG/CR-5117
IGSCC Round Robin Team MF
- Bobbin 2.5 ~---------------------------------------------------------.....
- 2.0 C
.c -
C) 1.5
'C G) -
ca E 1.0 U) w 1-w 0.5 a
a a
D D a 0.0 <t>-----()----~----.---..........a,......... --a,-,-----..-----.-----.------1 '
o.o 0.5 1.0 1.5 2.0 2.5 Measured Crack Length, In.
F;gure 9 - Results from NRC Steam Generator Group Project Reported in NUREG/CR-2336 and NUREG/CR-5117
e r.1 IGSCC Round Robin Team MH m RPC 100 aa
~
a a
80 a
'if!.
a
.c i:
- a.
G) 60 C
.. a; "C
G)
-~i.
cu E
40 (I)
UJ I
t-w
~ 1 20 oo-------a--o--c>r00-....--0r--~--------1 0
20 60 80 Measured. Crack Depth, 0/o Figure 10 - Results from NRC Steam Generator Group Project Reponted i~ NUREG/CR-2336 and NUREG/CR-5117 100
e
- J,I..
IGSCC Round Robin Team MH
- RPC 2.5.....-----------------------...... -------------------------------------...
.: r.,
C 2.0
-:.l.
Ii
.c -
C>
C a,
1.5 "C
D a, -
ca E
1.0 D
u, w
D t-w 0.5 D
C D
0.0 -o------o-----,-.----,..-..__-r0a,--,-.--o,-.-----..,...---........ ___....,.. ___ --t o.o 0.5 1.0 1.5 2.0 2.5
. Me*~sured Crack Length, In.
Figure 11 - Results from NRC Steam Generator Group Project Reported in NUREG/CR-2336 and NUREG/CR-5117
C, IGSCC Round Robin Team MO - Array Coll 100-r-------------~-------------------------..
..c...
C.
Q)
Q "C
Q)...
cu 80 60 E
- 40 Cl) w 1-w 20
~.;.:
~~
- 1, 1J; i:., '.
DD D
D D
- 1.
D.
0 '9-----r---o--a>ra:,oa-;--ar------,------1 0
20
-40 60 80 Measured Crack Depth, 0/o Figure 12 - Results from NRC Steam Generator Group Project Reported in NUREG/CR-2336 and NUREG/CR-5117 100
e IGSCC Round Robin Team MO - Array Coll 2.5....... ----------------------------------------------------------....
£.'
2.0 C *-..
a
.c -
C)
C 1.5 CD
..J "O
CD -
al a
E 1.0 Cl) w I-a w
a 0.5 a
a 0.0 o.o 0.5 1.0 1.5 2.0 Measured Crack Length, In.
Figure 13 - Results from NRC Steam Generator Group Project Reported in NUREG/CR-2336 and*NUREG/CR-5117 2.5 0
N
Comparison of French Leak Test Data with Model Predictions Normal Operating Con*dition C.e ca C).
10 1
~ 0.10 ca a:
-~
ca a,
...I 0.01 Predicted
-*----6 Ran_ge of Crack Length as Measured from ID to* OD 0.001 0
~........___......,. _____....___.... ___....__....__.. ____
--J 0.2 0.4 0.6 0.8 1.0 Crack Length. inch-Figure 14 - Data taken from various sources, see NUREG/CR-2336 and NUREG/CR-5117 21
- E
- a.
tD -
I!!
~
~
~
UJ
~
t-u
- ~
CJ
~
- a.
tE c tE f iE a tE-1 e
e WESTlNGHOUSE PROPRIETARY CLASS 2
. :*'.-,1'1
- .~ r -'*I I
- cuuc°i U
C SUUA1 C CC UCC1 fE-5 MEl~UIEJ ttk RAfe2 (GP~,,-*
Figure 15 - Reproduced from Westinghouse Report WCAP-13129, Revision 1 22 If 0 tE I fE 2
I.
e e
,.a 1.0 Figure 16 - Reproduced from EPRI Draft 3 Report. July 1991 "PWR Steam Generator Tube Repair - Technical Support Document for Outside Diameter Stress Corrosion Cracking at Tube Support Plates" 23
}.. *~
0 Q.
~
- _j w
~
~
- _j w
Q BURST PRESSURE PARAMETER CURVES 1
EDM Slot S eclmens 7/8 x 0.0~0 inch Tubing) 0.9 0.8 0.7 0.6
. 0.5 0.4 0.3 h/t=0.80 0.2 0.1 0
0 0.4 0.8 1.2 1.6 LENGTH Onches)
I e Figure 17 - Results from NRC Steam*Generator Tube Integrity Program -
Equations for Curves and Similar Plots Reported in NUREG/CR-0718 2
N
Enclosure (1)
STEAM GENERATOR TUBE INSPECTION, INTEGRITY AND PLUGGING ISSUES The following discussion addresses issues related to operation of steam generators with through-wall cracked (leaking) tubes.
Two general areas are discussed; 1) engineering design philosophy and the policy of defense-in-depth and 2) technical issues related to assurance of maintaining tube integrity of cracked steam generator tubes during reactor operation.
Engineering Design PbiJosophy and Defense-in-Depth
. General Design Criteria (GDC) of Appendix A to 10CFRSO require that the reactor coolant pressure boundary (RCPB) have an extremely low probability of abnormal leakage, of rapidly propagating failure and of gross rupture.
Further, the RCPB is to be designed to permit periodic inspection and testing to assess the structural and leak-tight integrity. Using materials that exhibit leak-before-break behavior, maintaining leak-tightness of the RCPB and conducting inservice inspection (ISi) to assess structural and leak-tight
. integrity are important elements of defense-in-depth for maintaining safety and are not meant to allow operation with a leaking RCPB.
The GDC indicate and the NRC staff has interpreted that through wall cracks in the RCPB are not acceptable. Several recent actions attest to this interpretation: 1) GDC 4
- on exclusion of dynamic effects from ruptured pipes does not apply to materials susceptible to degradation; 2) ASME and NRC rules for evaluation of-cracked stainless steel pipe do not allow operation with pipes containing through-wall cracks even though these pipes may exhibit leak-before-break.
Pipes with cracks deeper than 75 percent through-wall must be repaired; 3) NRC guidance for leak monitoring of RCPB allows for a small amount of unidentified leakage, however, if leakage is from a through wall crack, the component must be repaired; 4) NRR connents from review of a proposed revision to Regulatory Guide 1.121 required the guide to state that through-wall flaws of any type and identified cracks of any size are unacceptable. Since the steam generator tubes comprise over SO percent of the RCPB surface area and hundreds, even thousands of tubes could be leaking with an alternate tube plugging criteria 0 it is important to adhere to the policy of non-penetration of the RCPB.
TECHNICAL ISSUES If it is decided that it fs acceptable to operate a nuclear power plant with the-primary pressure boundary violated, i.e. with through-wall cracked steam generator tubes fJ>r the situation under discussion here, then a strong engineering case needs to be made and actions taken to assure maintenance of structural integrity. The important parameters relating to the structural integrity of steam g~nerator tubes are the crack length for through-wall cracked tubes and the crack length and depth for other cracks. Cracked tubes can exhibit no leakage, small leakage or large leakage and burst behavior under normal operating and accident conditions. For through-wall cracked tubes, with axial cracks, the crack-length at which large leakage or burst occurs (critical crack length) under MSLB condition 1s approximately one-1nch. Various combinations of crack lengths and depths for part-through-wall flaws can lead to burst under normal operating or MSLB conditions. Therefore, 1
e to ensure maintenance of structural integrity in cracked steam generator tubes, cracks must not exceed certain sizes during operation and tubes with cracks above these sizes must be removed from service. Cracks present in tubes during a given inspection must*not reach critical sizes during operation before the next ISi. This requires that cracks must be reliably detected and accurately sized, that the sizing errors are known and that crack growth rates
{both in depth and length) are known for the wide spectrum of conditions and mechanisms that occur in steam generators. Furtherm~re, if cracks are accepted they must be monitored during operation to ensure that their sizes do not approach critical sizes which would place the tubes at risk of large leak or rupture during a MSLB.
The information from the monitoring must relate directly to crack size.
No single factor mentioned above by itself can assure maintenance of structural integrity, these must be applied together. Reliable crack detection and sizing 1-s required along wi-th accurate estimates of crack growth rates and reliable leak rate/crack size correlations for monitoring crack evolution and stability during operation. Discussions related to these capabilities follow.
crack-Detection and Sizing uncertainty Some of the IIOSt extensive* research conducted to eva 1 uate flaw dt!tecti o,r
- probability, (as a funct*ion= of flaw *size,),- anct fla~-s1zing*acCUf'kY** was the inspection of the Sclrry,,9enentor-:*re11DYed fromi serv1ct.'":*ngures* J-4 show the flaw detection probability as* a-functiow of flaw size-lnd* flaw* sizing accuracy obtained from EC ISi teams. Plots for all the teams and -for the** best team are shown. These data are-for flaws found in the Surry generator i;e. wastage and combinatieas of wastage* and* pitting; These flaws: are'*considered* to be large volume flaws and easier to*detett and s1ze than* small*volume flaws such as cracks. It is expected that the performance for cracks would be even less reliable. To supplement the data froa Surry, a round robin was conducted on a IS-tube box containing laboratory produced stress corrosion cracks. Sixteen tubes contained cracks of various lengths and depths. The depth of four cracks ranged from 25 to 40 percent through-wall, the remaining 12 ranged from 40 percent to through wall. The lengths varied up to 1.5 inches long.
Although the total number of flaws in this test is relatively small, some trends are evident from the results. Four organizations inspected these tubes using several techniques a) the standard field practice techniques that met code and regulatory guide requirements and b) alternate techniques to represent the organization's best efforts and techniques. Figures 5-13 show some typical results. The probability of detection for these flaws ranged from 0.2 to 0.75 and on the average was approximately 0.5 for either conventional or (Jternate techniques; the conventional technique used the bobbin coil while the alternate techniques included rotating pancake coil, array coil and an alternate bobbin coil design. Sizing ac~uracy was poor.
The through wall flaw was sometimes missed ;nd other times reported as a shallow flaw.
Of particular interest was the poor length sizing ability even with the alternate techniques, where flaws up to 1.5 inches long were missed or sized at 0.2 to 0.5 inches. Note that some of these cracks are of critical length or longer and the EC would classify them shorter than critical length.
The above discussion on flaw detection and sizing is based on techniques and parameters in co11111on use. Multifrequency procedures, using amplitude and 2
- e.
e phase angle for detection and sizing of flaws.
Recently a parameter has been emerging, the voltage (or amplitude}, as a measure of tube integrity. This parameter does not uniquely measure the length or depth of flaws, the critical parameters from a structural integrity point of view. Table l shows data from laboratory produced part-through-wall stress corrosion cracks. For.various crack morphologies of interest the voltage is not expected to relate to tube integrity for the following reasons:
- 1) For flaws of given width and depth a correlation exists of increasing voltage with length up to a flaw length of approximately 0.5 inch.
Longer flaws will not produce a larger voltage than this saturation level; this saturation of voltage for approximately 0.5 inch long flaws and longer is based on the coil design.
- 2) The voltage produced can be related to the tightness of the cracks; if the cracks are tight enough, and conductivity paths exist, low voltage response is expected whether the cracks are short or long. Of course from a structural point of view the larger flaws are more important and the voltage parameter would not distinguish between them.
- 3) The voltage produced is insensitive to critical crack morphologies. For example a number of short, tight cracks (deep or through wall} axially aligned with short ligaments between them would produce a small voltage indicative of the tight short segments of the cracking.
From a structural point of view such cracking would behave like a long crack i.e.
tubes would have low pressure holding capability; under pressure the ligaments would join to produce critical length cracks and high leak rates. The voltage response from such cracking would not predict the structural integrity.
Cracking Mechanisms and Growth Rate variabilities The discussion on crack detection and sizing reliability indicates that
- important cracks can be easily missed and those that are detected cannot be adequately sized.
Even if important flaws were adequately detected and sized, th~ crack growth rates, both in terms of depth and length, are required in oroer to estimate the crack sizes at the end of the operati_ng cycle, before the next inspection, to assure that accepted cracks remain below the critical size by a reasonable margin. Research results show that variabilities of one order of magnitude can be easily expected in crack initiation times and growth rates for environmentally assisted cracking even under test conditions where samples of the same material are exposed to the same environment, temperatures, stresses, etc. Much variation in the operating environment of steam generators exists for the power plants in the U.S. Conditions of water chemistries, temperatures and thermal hydraulics can differ from plant-to-plant; geometries, crevice conditions, heat of material, temperatures, water chemistries, stresses, etc. can vary even within the same steam generator.
As a consequence, m~ny different types of cracks have been experienced at.
different U.S. steam generators and even within the same steam generator.
Primary and secondary side cracking has been experienced. Cracks in tubes at various locations has occurred such as in t~e tube sheet crevice, at*top of tube sheet, in free span zones, within the tube support plate regions,* at U-bends etc. Fatigue cracks, intergranular corrosion cracks, intergranular attack and crevice corrosion cracks have been experienced.
Some of the intergranular cracking is associated with stress such as at dented regions, other intergranular cracking is not associated with any significant stress such as at crevices in undented regions. Several forms of intergranular
. attack and combinations of intergranular attack and cracking have occurred.
3
Uniform intergranular attack has been experienced which essentially produces a thinning from an integrity point-of-view. Other uniform intergranular attack progresses to a given depth, then is accompanied by cracking through additional depth in the tube. Still another form, cellular intergranular attack, is a network of axial intergranular cracking connected by circumferential cracking. At different plants and within the same generators, axial and circumferential cracks have been found.
In the tube support plate region, cracks have occurred in tubes within the thickness of the tube support plate but have also grown beyond the support plate region and axial cracks have also been accompanied by circumferential cracks. Although, at first, cracks may be noticed only within the tube support plate region, cracks may grow beyond the support plate region in time. Research studies have shown that even for materials that are diffi~ult to crack, once cracks are initiated their growth is sustained and the crack growth *rates are similar to those for materials that are more susceptible to crack initiation~. To varying degrees, the crack itself may act as a crevice arid growth is ~ust~ined. There.fore, cracks that initiated in the tube support plate crevice *region could grow b1yond the original crevice. After the different cracks are observed, the modes of *.cracking are recognized; however these cracking phenomena have not been predicted nor thefl" occurrence e.asily controll~~.*-, The.. various mechantsrn1 *. causative factors.~nd synel',gism.be~e*en i~rtat\\~.P~~ameters are not well uqdersto~d. __ J_y.__if th~. e~f,c~ of :~_om~ i!.. t~~: ia,p,or-t~t.-parameters.
such as ctiemistry wer'e *"-ett underst"o~, *their contt'AJ*.tR _cr.u.cia~ ).ocat1ons such as in crevices is difficul~_if.not impossibl~ to achieve. Concentration factors; of different species under *different conditions_, from the bulk to the crevice as high as 10* to 10* can be expected...
To sumarize 'the above dlscuss,fon,* several modes,of crack:iftg' hav*e been
- experienced in U.S~ steam generators*and several modes can.~e experienced in a given unit. The cracking modes can change with time and_ cracking that might have initiated in a given region, tube support plate for example, can extend beyond.that region. The mechanisms, interactions and causative factors are not -well understood or controllable and the cracking phenomena were not predicted a priori. Laboratory testing of the same heat of material under the same environment and loading conditions produce crack growth rates differing by an order of magnitude. Conditions in an operating plant are not so well known or controlled and even higher variability in growth rates can be expected. Thus, changes in mechanisms, growth of.cracks beyond given regions and crack growth rates cannot be reliably predicted. Therefore no assurance can be provided for cracks found during an inspection (even if accurately sized) that they will not reach critical size during the next operating cycleo Recent proposals'try to use a evoltage growth rate* obtained from consecutive eddy current ISis as a measure of crack growth rate. As discussed previously, for the cracks and crack morphologies of interest, there is no unique correlation between the voltage and the crack length or depth (parameters of interest to structural integrity) therefore the voltage growth rate cannot be used as a measure of crack growth rate.
4
e Enclosure (2)
ISSUES ANO QUESTIONS RELATED TO ASSURANCE OF MAINTAINING STRUCTURAL INTEGRITY OF THROUGH-WALL CRACKED STEAM GENERATOR TUBES Has the Commission decided that it is acceptable to operate steam generators, whose tubes comprise over SO percent of the reactor coolant pressure boundary (RCPB) surface area, with hundreds or even thousands of through-wall cracked tubes?
If it is decided that it is acceptable to operate SGs with through-wall cracked tubing, and therefore will have eliminated the leaktightness of the RCPB as a.very important element in defense-in-depth for maintaining safety, then a strong engineering case needs to be made to assure maintenance of structural integrity during operation. To maintain structural integrity,_ flaw length must remain below a critical length.
Key issues in assuring structural integrity are knowledge of:
a) the.through-wall flaws present, b) the crack length and accuracy of measurement,
. c) cracking mechanisms and crack growth rate and d) knowledge of crack size and progression from leak rate monitoring.
Questions and co11111ents related to these issues are as follows:
What is the probabi 11 ty of detection ( POD) for deep and through-wa 11 cracks as a* function of crack length? Past experience and results indicate low POD.
What is the accuracy of length and depth sizing?
If voltage is used as a measure of tube integrity, how is voltage related to length (and depth for deep flaws)?
Voltage saturates as a function of length below the critical length.
What is the voltage response for tight cracks - even if long?
What voltage response and variation is expected for effectively long cracks made up from a series of short cracks axially aligned with.small ligaments in between?
Tubes with short cracks, even if through-wall, will exhibit high burst pressures. However, tubes with deep part-through ~all flaws (85 percent and greater) and through-wall flaws above approximately 0.6-inch long will exhibit burst pressures below MSLB.
Again, from the voltage, what is the flaw size for these flaws?
1
12 11 10 g
(I)
,jf, 8
m 7
L
- J (I) 6
(/)
- 0)
L
- a.
5
-tJ
(/)
4 L.
- J CD 3
2 I
0
- 0. I 1.0 10.0 Bobb i n. Vo I toge.. Vo I ts 1PVL T78R.EtG Figure 18 - Submitted in Support of Alternate Tube Plugging Criteria 100.0 200.0 l
In the burst pressure vs. voltage correlation, were effectively long, tight cracks (expected of producing low voltage and low burst pressure) considered?
Regarding voltage vs. leak rate, considering five orders of magnitude scatter in the data and correlation factors of 0.7 to 0.8, is. it considered that a reasonable correlation exists?
How well do we understand the various mechanisms of cracking? What are the causative factors and synergisms? What assurance is there that cracking mechanisms will not change during operating cycles?
Why wouldn't existing cracks grow beyond the initial locations, i.e. outside of support plate for crevice corrosion cracking? What are the crack growth rates to be applied to estimate crack length (or depth) at the end of operating cycle?
What reliable correlations exist between crack length and measured or predicted leak rates?. How do leak rates measured inservice relate to crack length considering corrosion products, fouling, residual stresses, etc. which tend to restrict leakage? What changes in leak rates are expected and can be measured as cracks approach critical lengths?
In Monte Carlo evaluations to predict expected leak rates under normal operating, and accident conditions, how are non-detections of through-*
wall cracked tubes considered?
2