ML030850768
| ML030850768 | |
| Person / Time | |
|---|---|
| Site: | Davis Besse  |
| Issue date: | 05/30/2002 |
| From: | Hickling J Electric Power Research Institute |
| To: | Office of Nuclear Reactor Regulation |
| References | |
| -nr, FOIA/PA-2003-0018 | |
| Download: ML030850768 (26) | |
Text
File from Steve Long's computer dated 05/30/02 12:15pm named" NRC 5-30 CGRI.pdf" Crack growth rate for thick-section Alloy 600 material exposed to PWR 0 primary water John Hickling, EPRI for the MRP Alloy 600 Issue Task Group NRC 5/30/02.1 EI-2I MRP
"* Goal was to establish appropriate CGR guidelines for generic application to thick-section Alloy 600 base material under PWSCC conditions
"* MRP panel of international experts on SCC (includes ANL/NRC Research) was established August 2001 and has met several times to date
"* Extensive consideration was given to the likely OD environment in the annulus between a leaking CRDM nozzle and the RPV head (prior to Davis Besse incident)
"* Relevant arguments remain valid today as long as leak rates are low (typically < 1 liter/h or 0.004 gpm)
"* Plant experience has shown this to be the usual case NRC 5/30/02.2 r= 8I21 MRP
ratk"Gtowdh"`Rate*-Aý'
................... ,,,'.',P proac h I...
"* Relevant, worldwide CGR results were obtained and re evaluated so as to screen out inappropriate test data
"° Recommended MRP curve for CGR as a function of stress intensity factor (K) was derived taking into account the statistics of heat-to-heat variations and the strong effect of temperature
"° Curve was compared with existing field data and recommendations developed for its use in assessing axial/circumferential flaws
"° Screened CGR data for base material feeds directly into the probabilistic risk assessment being carried out by SIA NRC 5/30/02.3 ErIeFI MRP
Most likely environments
"*Hydrogenated superheated steam, if pressure drop within SCC crack
" Normal PWR water, if boiling transition well above the J-groove weld
"*Concentrated PWR primary water, if boiling occurs at the exit of SCC crack:
- situation has been considered in detail for the case usually observed to date, i.e. low leak rates (< 1l/h) and little or no wastage of LAS vessel head
- full evaluation has not been performed for Davis Besse type situation involving cavity formation and extensive wastage as a consequence of boric acid corrosion NRC 5/30/02.4 E'=21 1 MRP
"* Consideration of oxygen/hydrogen effects common to all three possible environments:
" Oxygenated crevice environment highly unlikely because:
- Back diffusion of oxygen is too low compared to counterflow of escaping steam (2 independent assessments based on molecular diffusion models were examined)
- Oxygen consumption by metal walls would further reduce concentration Presence of hydrogen from leaking water and diffusion through upper head results in a reducing environment Even if concentration of hydrogen was depleted by local boiling, coupling between LAS and Alloy 600 would keep electrochemical potential low Corrosion potential will be close to Ni/NiO equilibrium, resulting in PWSCC susceptibility similar to normal primary water NRC 5/30/02.5 EII21 MRP
- Possible environment #1: hydrogenated steam
- Numerous laboratory tests in hydrogenated steam (e.g. Economy et al., 1986- 1995) have shown that PWSCC rates are similar to those in normal PWR primary water at the same temperature
° Possible environment #2: PWR primary water within normal specifications
- Main focus of subsequent CGR data evaluation by Expert Panel NRC 5/30/02.6 E[F"I MRP
Possible environment #3: Concentrated PWR primary water. For low leak rates (< 1 I/h) as mostly observed to date:
- pHT between 4 and 9.4 based on MULTEQ calculations
- Actual pHT range expected to be narrower due to precipitation of complex lithium-iron borates
- A French experiment simulating a leak detected such borate compounds and estimated that pHT of the liquid phase was between 7-8
- A further French test involving slow concentration of a fixed volume of primary water gave a calculated pHT of around 4.5
- Cleaning practices followed during head assembly should minimize contamination by sulfates and chlorides and steam flushing will help to remove any residual impurities NRC 5/30/02.7 E1[21 MRP
1*
Possible environment #3: Concentrated PWR primary water (con.)
- Ohio State study shows no significant effect of pHT on PWSCC CGR between values of 5 and 8.5 at 330 C
- For pHT values between 7.5 and 9, CGR increases slightly, but acceleration factor only around 1.5 even for pHT = 9
- Expert Panel recommended that a factor of x2 on CGR should conservatively cover uncertainties in the exact composition of the annulus chemistry for 4 < pHT < 9
- More acid environments as a result of large leak rates and local cooling of head were NOT considered, but limited data (Berge et al., 1997) suggests that high chloride and oxygen levels are required for IGSCC of Alloy 600 to occur NRC 5/30/02.8 Er[r21 MRP 2
- Key technical issues on which screening was based:
- Material within specifications including condition/heat treatment
- Composition within material specifications
"* Mechanical strength'properties
"* ASTM specimen size criteria
"* Straightness criteria and crack front mapping
- Standard procedure for welds
"* Environment (Li, B, and H 2 concentrations; hydrogen control; temperature; ECP)
"* Loop configuration (e.g., once-through, refreshed, static with H, control) and flow rate
- Water chemistry confirmation (e.g., Cl, SOO)
"* Crack length confirmed by destructive examination
"* Transgranular fraction on fractograph
"* Fraction SCC along crack front
"* Changing conditions during a test?
- Active constant or cyclic loading versus constant displacement loading (e.g., wedge loading)
- Load during "cool down"
- Crack length versus time data
- SCC crack increment
- Precision on measurement of crack length increase NRC 5/30/02.9 E "21 MRP
- No attention was paid to numerous tests where no crack growth due to PWSCC was actually observed
- Result of data screening was elimination from further consideration of 203 CGR data points for one or more reasons (main reason individually documented in report)
- Consolidated database contains 158 data points for average CGR during each test (consistent with ASTM practice for measuring fatigue CGRs) plotted at a representative K value (ranged from 14.8 to 49.7 MPaNm)
- All were obtained in controlled primary water using fracture mechanics specimens under either constant load or constant displacement conditions Some tests under active load involved periodic unloading (considered to give a potential accelerating effect which is relatively small, at least for susceptible materials)
NRC 5/30/02.10 E.1 21 MRP
A eg 0 Domestic and Overseas material suppliers represented:
"* B&WTP, Huntington, INCO, Standard Steel
"* Creusot-Ondaine, Creusot-Imphy, Tecphy, Arbed, VDM, Schneider Creusot, Sandvik, Sumitomo Metal a 26 heats of material with at least 1 screened data point per heat (maximum # = 32 for B&WTP heat 91069)
- Multiple product forms
"* Thick walled tube
"* Forged bar
"* Rolled bar
- Forged plate
- Rolled plate
- Information on thermal processing history of material incomplete, so likely effects could not be systematically considered in a deterministic way NRC 5/30/02.11 E-,r2i MRP
- Multiple Labs
- Westinghouse, U. S.
- EdF, France
° CEA, France
- CIEMAT, Spain
- Studsvik, Sweden
° Test temperatures ranged from 290-363 'C (554-686 OF)
"° CGR through PWSCC of flaws in Alloy 600 is known to be highly temperature dependent, so
"° All CGR data points were adjusted to a common reference point (the most typical test temperature) of 325 0C (617 OF) using an activation energy of 130 kJ/mole (31 kcal/mole)
"* This represents a consensus value for Alloy 600 crack growth data NRC 5/30/02.12 ErV'i--l10MRP
"* Because of the known importance of material processing parameters on CGR, the initial evaluation was based on a heat-by-heat analysis of the screened database
"* Insufficient data points were available from any single heat over a wide range of K values to determine the form of CGR dependence on stress intensity factor
- Shape of curve to be fitted was adopted from the Scott equation, originally developed (1991) using inspection data for axial cracks in the roll transitions of SG tubes
- This much larger database of CGR measurements is considered to provide a more reliable indicator for the form of the CGR versus K dependence:
- da/dt = oL(K-9)P with Scott exponent J3 = 1.16 NRC 5/30/02.13 ERIE1 MRP
- Adoption of the Scott equation results in an apparent crack tip stress intensity factor threshold, Kth, of 9 MPa\Im (8.19 ksi*/in).
- However, no actual CGR data for CRDM nozzle materials is available at K values < approx. 15 MPalm.
- In contrast, use of the Scott exponent P3 = 1.16 may result in conservative estimations of CGR at high K values, since some test and field data appears to indicate the appearance of a plateau in the curve NRC 5/30/02.14 E'P2i MRP
"* For each of the 26 heats of material in the database, a mean power-law constant az was then calculated according to f(a, P) In(K, - K,,, )]}2 j{in(L,) - [in(a) ++/-6
"° Distribution describing CGR variability was taken as the log-normal fit to the ordered median ranking of the a*
values for the 26 heats using most likely estimator methodology NRC 5/30/02.15 ErE2I mRP
r IV, t o. n,q!OT.
U:.4 I;K! 41 -
N-3 On:'V.
7 4 V.. T r v.,..,e o
1.0 0.9 0.8 0.7 0.6 U,il 0.5 0.4 0.3 0.2 0.1 0.0 1E-13 IE-12 IE-1I 1E-10 Power-Law Constant a at 3251C (617'F)
NRC 5/30/02.16 ER21 MRP
° Recommended CGR curve is based on 75th percentile level of the distribution of CGR variability as a function of material heat and represents the mean of the upper half of the distribution
"° MRP curve lies approx. 20% above the Scott equation
"° Approach is consistent with ASME code considerations, where the goal is to make a best estimate of crack growth
° Addresses the concern that cracking detected in operating plants would tend to be in components fabricated from more susceptible Alloy 600 heats
° Likely that CRDM nozzles supplied by some material vendors may crack at a significantly lower rate than indicated by the MRP curve NRC 5/30/02.17 EI-I2I MRP
If r m 1.E-09 1 MRP Curve
~0~MRPCurve- -
-Modified Scott Curve
-S------- g o S 1-- o "Cb 0 o MRP Lab CGR Database (I58 points) 0 Cook2 #75 Length Increase 1.E-10 N Cook2 #175 Depth Increase 1 mm/yr 1.E-11 0 0
-~0-- 0--
-i-ii
- - -0 --
U S.... 0 b~o--_-_ All data adjusted to 325°C (617TF) using an activation energy of 130 kJ/mole (31.0 kcal/mole) 1.E-12 I I
0 10 20 30 40 50 60 70 80 Stress Intensity Factor, K (MPalm)
NRC 5/30/02.18 E1=21 MRP
0 P
0 so0,,.
!P RS 9
a ' 0 e.
- Large uncertainties exist in reported values of CGRs from operating plants due to:
- uncertainties in ultrasonic measurements of crack size at two or more different times
- uncertainties in the estimates of K, which depend on estimates of residual stress
- uncertainties in the actual operating temperatures of CRDM nozzles in different plants and in different countries.
NRC 5/30/02.19 E. 21 MRP
,.e ... EtM P aris
....... of MRP., atabase
-40 d GI h Vai a an d a t`ý
" Limited US data (from D.C. Cook nozzle #75) lies well below the MRP curve
"*Most extensive measurements of CGR in operating plants are from France.
"*The data have been extrapolated by the MRP from the reported operating temperatures in the plants to a standard value of 325 0 C for comparison purposes
"° Values are compared with the results of predicted CGRs calculated by using:
- the reported K values for the French field data
- random sampling from upper half of the MRP distribution for CGRs
- the K-dependence of the Scott equation NRC 5/30/02.20 Er=.-i" MRP
ý.Co Pans n o-..- jjl F,:da a"b ýIs 1.E-09 C
C 0 O EDF Field CGRs extrapolated 0 to 325'C 13 o Sample of 50-100 Percentile of 1.E-IO MRP Distribution 1 mm/yr U
1.E-1I L The MRP points were created by sampling the upper half of the MRP distribution for a and then using Eq 2 Q. to calculate the CGR for each EDF All data adjusted to 325 0C (617 0 F) field stress intensity factor K value using an activation energy of assuming Kh = 9 MPa'Im and/I = 1.16 130 ki/mole (31.0 kcal/mole)
! l 1.E-12 II II I I I I I 0 10 20 30 40 50 60 70 80 Stress Intensity Factor, K (MPaxlm)
NRC 5/30/02.21 ER2 MRP
CID f aM 1;P!7i PP. d a a as.
la le ava i. . ..
a a:
1.0 0.9 0.8 0.7 I-,,
0.6 0.5 0
Qp, 0.4 0.3 0.2 0.1 0.01E IE-11 IE-10 1E-09 Crack Growth Rate (m/s) Normalized to 3251C (6171F)
NRC 5/30/02.22 EPI'**2i MRP
,o m. a nSo 'O a0 S.. a a,,
"° Agreement with French field data is quite reasonable considering the uncertainties involved
"° Supports the choice of the 75th percentile curve from the MRP distribution as representative of the rates expected for axial crack growth in CRDM nozzles
"*In no case did the actual measured CGR in the through wall direction exceed 4 mm/yr (0.16 in/yr) for data from French plants of fundamentally Westinghouse design
"° This figure was adopted in France, independent of nominal upper head temperature, to justify continued operation with axial cracks up to 11 mm (0.43 inches) deep for a one-year fuel cycle NRC 5/30/02.23 .r--,r l MRP rk
- Z v.
zh'-.
PPnca i t',.Y
- The MRP recommended curve is intended for disposition of detected PWSCC flaws in thick walled Alloy 600 components exposed to normal PWR primary water
- Thus it is directly applicable to axial ID flaws detected in CRDM nozzle base material
- Its use at low crack-tip stress intensity factors
(< approx. 15 MPa/m) involves assumptions not currently substantiated by actual CGR data for CRDM nozzle materials NRC 5/30/02.24 E"=P21 MRP
A.... S S - S 17 S". at M R S P 12.0 II
__ __ CGR Adjusted to 318.3°C (605'F) 10.0 Using Q = 130 kJ/mole (31.0 kcal/mole)
Assume K = .1 or*sqrt(ira ) where a = 50 ksi S8.0
- 6.0 U 12.0 millimeters is the 75% through wall acceptance limit for ID axial PWSCC cracks in CRDM nozzles
" 4.0 2.0 0.0 0 12 24 36 48 Operating Time (months)
NRC 5/30/02.25 ER2I MRP
° For evaluation of (hypothetical) OD cracking above the J-groove weld, the MRP recommends that CGR values from the curve be multiplied by 2x to allow for uncertainty in the exact composition of the external chemical environment
- A subgroup of the Expert Panel have revisted the relevant arguments in the light of the Davis Besse experience and found that they remain correct as long as leak rates are low (typically < 1 liter/h or 0.004 gpm)
° Plant experience has shown this to be the usual case
° Analysis would no longer be valid, however, if leak rates were sufficiently high to result in a large, local decrease in temperature and appreciable corrosion of low-alloy steel NRC 5/30/02.26 Er=ri MRP