ML20059J277
| ML20059J277 | |
| Person / Time | |
|---|---|
| Site: | Mcguire, McGuire  |
| Issue date: | 01/18/1994 |
| From: | Lauer J, Schaffer R BABCOCK & WILCOX CO. |
| To: | |
| Shared Package | |
| ML20059J271 | List: |
| References | |
| 51-1228707-01, 51-1228707-1, NUDOCS 9401310422 | |
| Download: ML20059J277 (37) | |
Text
_ . _
NON-PROPRIETARY mn=eo ovam B&W NUCLEAR WTECHNOLOGIES ENGINEERING INFORMATION RECORD ;
Document Identifier 51- 1228707-01 l Title Evaluation of BWNT's Kinetic Sleevina Process Non-Pronrietarv !
PREPARED BY: REVIEWED BY:
Name t) t. A. k)}yRf Name kk $('N8Ef8$
Signature M Date b b Signature Date/~/h k Reviewer is[ Independent. YM
/ i Remarks:
1.0 INTRODUCTION
On August 22, 1993, McGuire-1 Nuclear Station shut down after -
detection of a 185 gpd primary to secondary leak in the "A" steam generator. The cause of a leak was traced to a stress corrosion crack in a tube which had BWNT kinetically welded tubesheet sleeve installed. The defect in the tube, R39-C72, was located adjacent '
to the freespan weld region of the sleeve to tube. joint.
The purpose of this document is to review the qualification and industry experience with BWNT's kinetic sleeve, discuss the root cause evaluation of the tube failure, and provide guidelines for evaluating in-service sleeves.
1 NOTE: This document is a non-proprietary version of BWNT 51-1228682-00. The location of deleted proprietary i information is indicated by brackets, and the reason (s) ,
for the deletion are indicated within the brackets by a code letter or letters from the following table:
i
- b. The information reveals data or material concerning BWNT research or development plans or programs of 3 present or potential competitive advantage to BWNT.
- c. The use of the information by a competitor would decrease his expenditures, in time or resources, in designing, producing or marketing a similar product.
- d. The information consists of test data or other similar data concerning a process, method or component, the application of which results in a competitive advantage to DWNT.
- e. The information reveals special aspects of a process, method, component or the like, the exclusive use of which results in a competitive advantage to BWNT.
~
9401310422 940126 PDR Page 1 of 37 P ADOCK 05000369 ppg
9 c:
NON-PROPRIETARY
- num EE10&W NUCLEAR ID WTEcH.VOLOGIES 2.0 EXECUTIVE
SUMMARY
The leak in tube 39-72 at McGuire-1 was caused by circumferential PWSCC in the tube at the freespan kinetic weld joint. Laboratory examination indicated that the defect in tube 39-72 is typical of that to be expected for a non-stress relieved freespan kinetic weld joint. However, review of site records, ECT inspection, and laboratory examination confirmed that the joint was stress relieved at approximately [ d ].
Destructive examination of tube 39-72 revealed properties which indicate a high susceptibility to PWSCC. Of particular note is an actual measured room temperature yield strength of 72.5 ksi, versus the stated material test report value of 64 ksi. A second sleeved tube from McGuire-1, tube 7-78, was also removed from the steam generator and examined. This tube was found to have a yield strength of 61 ksi. This tube was free of PWSCC. It had seen the identical service time as tube 39-72.
It was concluded that the stress relief temperature of approximately
[ d ] is insufficient to insure that PWSCC will not occur in highly susceptible tube material with properties like those of 39-72. Thus for this stress relief temperature, it is necessary to evaluate the PWSCC susceptibility of the tube material which has been sleeved. .
A method to do this was developed using corrosion test data which relate the susceptibility of Alloy 600 to yield strength and carbon level. This data was bench marked to the "real world" time to failure and properties of tube 39-72, resulting in an equation which will give a predicted time to . failure for any tube. Although the methodology is shown to be conservative, an additional safety factor of [d) is recommended when predicting time to failure using the equation. Examples are provided.
The consequences of sleeved tube failures were evaluated for both a single tube failure and multiple tube failure. The results, and experience, show that a single tube failure is well within typical plant analyzed parameters for primary to secondary leak rates. It was also shown that the likelihood of multiple tube failures is unlikely.
It should be noted that this evaluation ' is for sleeves already ;
installed. It is possible to perform a re-stress relief at higher temperatures to resolve the concern of PWSCC independent of tube properties. Also, future sleeve installations will [ b, c, e]
temperature to eliminate this concern.
1 i
UPL-78 51-1228707.01 )
Page 2 of 37 '
T NON-PROPRIETARY B W?tsCo"Enss"
3.0 BACKGROUND
3.1 Stress Corrosion Cracking in Alloy 600 Tubing Many recirculating steam generators (RSG) were constructed using low temperature mill annealed (LTMA) tubing. LTMA Alloy _600 tubing has been shown to be highly susceptible to pure water stress corrosion cracking (PWSCC) . Specific properties including high yield strength and small grain size with little or no intergranular carbide precipitation are major contributors to PWSCC susceptibility.
Plants constructed with tubing annealed at higher temperatures,
(>1800*F) are less susceptible to PWSCC than LTMA tubing. The improved resistance is most likely due to a decrease in yield strength, increase in grain size, and increase in intergranular carbide precipitation.
A [ e J thermal treatment of HTf:A Alloy 600 tubing further increases PWSCC resistance by precipitating even more intergranular carbides.
Plants with LTMA steam generator tubing first started having problems in Europe. By the end of the first fuel cycle, a problem had been identified with PWSCC of tubes in regions of high tensile stresses.
The two regions of attack were row 1 U-bends and in the roll transitions of tubing expanded into the tubesheet.
Shot peening of rolled tubesheet regions and stress relief of row 1 and 2 U-bends (UBSR) were implemented in steam generators with susceptible tubing in both Europe and the United States. However, it has been shown that small cracks, beneath the detection level of-NDE techniques, will continue to propagata after shot peening.. Thus BWNT began development of a new sleeve design to repair steam generator tubes experiencing PWSCC in roll transitions.
Following development and implementation of tubesheet sleeves for 3/4" OD tubing, BWNT developed a kinetic sleeve for generators with 7/8" OD tubing to address problems not only in the tubosheet, but also ODSCC at tube support plate (TSP) intersections. Subsequently, the increase in eddy current (ECT) indications at Westinghouse D-series tube support intersections, BWNT developed a 3/4 TSP sleeve.
3.2 BWNT Kinetic Sleeve Design and Qualification For steam generators with 3/4" OD tubing, two TS sleeve lengths [ o ]
and one TSP sleeve length [ e ] have been qualified to insure that any tube can be sleeved (Figure 3.1). The [ e ] inch sleeves (TS and TSP) can be installed into any tube in the steam generator.
[ e ] The tubesheet sleeves also come in two different designs: Kinetic freespan weld with rolled tubesheet joint and double kinetic weld (both freespan and tubesheet joints).
UPL-78 51-1228707.01 Page 3 of 37
NON-PROPRIETARY'
~
BW"!EE!Eisa With the exception of the upper TSP, BWNT's kinetic sleeves can be used to repair most defects found in recirculating steam generators.
Sleeving tube defects is preferred over plugging, because sleeving leaves the tube in service and thus has substantially less flow reduction and loss of heat transfer than plugging.
The structural analyses of the sleeves demonstrate that they meet the ASME Code Section III criteria for the steam generator design conditions. An analysis was also performed in accordance with NRC Regulatory Guide 1.121 to establish a plugging criteria for the sleeve.
The plugging limit was calculated to be a [ d ] through-wall defect.
To allow for possible defect growth between inspections, a plugging limit of [ e] of the original sleeve wall was established. The existing tube plugging limit of 40% remains unchanged for those areas not spanned by the sleeve.
Fatigue loadings for use in qualifying the sleeve joints were established and subsequently used in a series of leak and fatigue tests to qualify the kinetic expansion joint by experimental stress analysis per the ASME Code.
Corrosion tests were performed on both the ID and OD of sleeved tube specimens to evaluate susceptibility to SCC. The tubing used for the samples (Table 3.1) was prepared for [ b ] to be especially susceptible to PWSCC, [ d ] The samples were stress relieved at the minimum allowed stress relief temperature of [ e ] . This same minimum temperature was recommended by [ e ] . The corrosion tests indicated that after stress relief, the freespan kinetic weld should not be susceptible to PWSCC in the steam generator environment.
3.3 Sleeve Installation Process The steps for sleeve installation (Figures 3.2 and 3.3) are:
Tubesheet Sleeve Tube Suncort Plate Sleeve
[e] -
[ e]
The initial ECT examination used to detect tubing defects is also used to insure that no defects are present in the area where the kinetic welds will be made. The [ c, e ] is required to insure that a kinetic weld is formed between the sleeve and tube. A[e]. After
[ c, e ], the sleeve is inserted with the [ e ) . The TSP sleeve and the all-welded TS sleeve produce the [e ).
UPL-78 S1-122870).01 Page 4 of 37
, NON-PROPRIETARY 4 M CGWNUCLEAR V^
' b TECHNOLOGIES Following welding, a heater is inserted into the sleeve to stress relieve the freespan kinetic joints. The heater uses [ o ) on the heater to [ e ). The TSP sleeve has both kinetic jointa [ o ). A similar heater control system was used in BWNT's U-Bend stress relief process.
Following stress relief, an optional rolled lower joint is performed on the tubesheet sleeves. Rolling is performed by inserting the roll tool into the sleeve. Once in the sleeve, the roll is performed by
[ c, e ).
Following sleeve installation, ECT is used to inspect the sleeve.
A [ e ) is used to attain the required detection sensitivity of defects in all areas of the sleeve and in the expanded region of the parent tube. ECT is also used to [ e ).
3.4 BWNT Experience with Kinetic Sleeving BWNT first installed this type of kinetic sleeve in 1990. Through the end of 1993, BWNT has installed over [ e ] sleeves in 3/4" and 7/8" tubes, with both tubesheet and tube support plates in both tubing sizes. The following table breaks down the sleeves installed by tubing OD and sleeve type:
Tubing OD Sleeve' Type 3/4" 7/8" Tubesheet (Weld / Rolled Joints)
Tubesheet (Double Weld) [e ] [e]
Tube Support Plate In 1990 alone, over 1100 kinetic sleeves were installed in both the US and Europe. To date, there have been two tube failures associated with BWNT's kinetic sleeve, at Trojan in 1992 and McGuire-1 in 1993. >
3.5 Trojan Tube Failure In November of 1992, Trojan Nuclear Stiation shut down when a 200 gal / day leak in the "B" steam generator was discovered. The leak was isolated to tube 25-17 in the area of a kinetic sleeve installed at the first, tube support plate. ECT detected multiple indications in the parent tube at the lower weld area.
A review of log books found that the stress relief heater used on 25-17 was removed immediately after stress relieving this sleeve l because it was bent. Thus it was suspected that the heater may have i caught on the bottom of the sleeve and heated the arc a of tube beneath I the sleeve. [ c, d, e] I
[ c, d) l I
UPL-78 S1-1228707.01 !
Page 5 of 37 4
s
, NON-PROPRIETARY M111BGWNUCLEAR Rao W TECHNOLOGIES Thus the root cause of the defect in tube 25-17 was traced to a combination of operator error and equipment failure such that the sleeved tube joints were not stress relieved. The failure mode was attributed tc PWSCC in the parent tube.
All of the other sleeves installed at Trojan were confirmed to have been stress relieved.
Equipment and procedural changes to the BWNT sleeving process were implemented to prevent a reoccurrence of this problem.
3.6 McGuire-1 Tube Failure In August 1993, McGuire 1 Nuclear station shut down when a 185 gal / day primary to secondary leak was discovered in "A" steam generator.
The leak was isolated to tube 39-72 which had a kinetic tubesheet sleeve installed in the hot leg.
ECT examination revealed a circumferential crack in the parent tube at the upper end of the freespan kinetic weld region. [ d ) which would indicate that the tube had been stress relieved after sleeve installation. The kinetic weld region of 39-72 and one additional sleeved tube (7-78) were removed for destructive examination.
i 1
UPL-78 '
51-1228707.01 Page 6 of 37 ,
i
, NON-PROPRIETARY BW?!sMa523 l 1
. TABLE 3.1 TUBING MATERIAL PROPERTIES CORROSION TEST SPECIMENS MATERIAL HEAT NUMBER 96834 96834 96834 LOT 2 LOT 5 LOT 6 NOMINAL SIZE (OD) 0.750" 0.875" 0.875" WALL TIIICKNESS (AVG) 0.043" 0.050" 0.050" YIELD STRENGTH (PSI)
ULTIMATE STRENGTH (PSI) [d]
% ELONG. IN 2" ASTM GRAIN SIZE FINI.L ANNEAL:
TEMPERATURE TIME (MIN)
CHEMISTRY (%)
NICKEL CHROMIUM IRON CARBON [d]
COBALT MANGANESE SULPHUR SILICON COPPER UPL-78 51-1228707.01 Page 7 of 37
. . . . = , . . _
' N0*J-PROPRIETARY
. B W 9tc %"sa s M i FIGURE 3.1 RSG SLEEVING C:PTIONS 3/4 RSG SLEEVE w -
. X 7 TSP J %,
x TSP >
- SLEEVE ,
o j ,
i c
t
\
ic m I
7 FDB
- J (OPTION) -
x m '__ * ?, #
,;,4:;'l
.. 2:*';.
I
- dTummaitli ,
i s Ce-JTS a CAJ rS SLEEVE SLEEVE ,
I l
TUBESHEET i i
I UPL-78 51-1228707.01-Page 8 of 37
, NON-PROPRIETARY MUHOGWNUCLEAR ,;
RmW W TECJINOLOGlQ FIGURE 3.2 RSG TUBESHEET SLEEVE INSTALLATION STEPS
[c ]
UPL-78 51-1228707.Ol>
Page 9 of 37
' NON-PROPRISTARY M W TECHNOLOGIES FIGURE 3.3 I RSG TSP SLEEVE INSTALLATION STEPS
[e]
b l
1 4
UPL-78 51-1228707.01 Page 10 of 37 ;
A NON-PROPRIETARY 1HB&W NUCLEAR W TECllNOLOGIES 4.O ROOT CAUSE EVALUATION Evaluation of the root cause for the failure of tube 39-72 at McGuire-1 was pursued by review of the sleeve installation records, in-situ non-destructive examinations, destructive examination of the pulled kinetic joints from tubes 39-72 and 7-78, and laboratory examination of additional kinetic joints.
4.1 Review of Sleeve Installation Records I
The first item investigated was to determine whether or not there was some anomaly during the sleeve installation. This was accomplished by reviewing the available documentation of the sleeve installation of 39-72 and other sleeves installed at the ssme time. The review included _ stress relief [ d ] used to check heaters for proper operation, kinetic expansion [ d, e ) and heater manufacturing records.
The record review revealed no anomalies that would help explain why tube 39-72 failed.
4.2 Non-Destructive Examinations Eddy current examination (ECT) was performed on all sleeved tubes to look for [ d, c ] by the stress relief process. The examination confirmed that all of the sleeves installed at McGuire-1 displayed.
A [ e ] ECT exam of 4 le sleeved tube revealed a circumferential defect in tube 39-72 at t a [ d ] weld. An ECT " geometric distortion" that was not a defect indication was called in the freespan weld region of tube 7-78. Both tubes had been sleeved in the same refueling outage. The [ e ] ECT showed no indications in any of the other sleeves.
Ultrasonic (UT) inspections of both 39-72 and 7-78 were performed to examine the kinetic weld bond. The UT examination revealed that the sleeves were kinetically welded with a sound bond in both tubes.
1 4.3 Laboratory (Destructive) Examinations The freespan kinetic joints of the sleeves installed into tubes 39-72 and 7-78 were removed for destructive examination in the laboratory. ;
The [ d ] of the kinetic joint with respect to the [d] material ,
properties of the parent tubing, and [ d ) of the kinetic joint were ~1 performed to gather information to aid in evaluation of the root cause of the tube failure. ,
)
Visual and destructive examination of the sleeves removed from 39-72 l and 7-78 revealed no abnormalities from the sleeve installation. j In fact, the joint [ d ] completely normal. Destructive examination J of the leaking tube (39-72) revealed that the tube failure was caused by stress corrosion cracking initiated from the primary side of the tubing, or PWSCC.
i UPL-78 51-1228707.01 Page 11 of 37
. NON-PROPRIETARY !
Et111R&W NUCLEAR li3 %AITrCHNOLOGIES i
The crack was circumferential in nature (Figure 4.1) and had [d]
at the end of the kinetic weld, identical to non-stress relieved specimens that were subjected to accelerated corrosion testing during the original qualification program. Destructive examination of tube 7-78 revealed no defects in the tube or sleeve.
Metallographic examination of tube 39-72 revealed small grains, ASTM 10.5, with no carbide decoration of the grain boundaries. The material test report for this particular heat of material indicated high carbon (0.04%) with a yield strength of 64 ksi. However, a room tensile test of a section of 39-72 revealed a yield strength of 72.5 ksi.
The carbon content of 39-72 was also found to be higher than reported during tube manufacture, 0.05% (Table 4.1).
Examination of tube 7-78 revealed slightly larger grains, ASTM 9.5-10, with a mix of intragranular and intergranular carbides. The material test report for this heat of material indicated a carbon content of 0.03% and a yield strength of 63 ksi. Testing of tube 7-78 showed a yield strength of 61 ksi and a carbon content of 0.035%. Both results agree fairly well with the material test report (Table 4.1) .
Industry evaluation of PWSCC has revealed that certain material factors increase the susceptibility of Alloy 600 to PWSCC. Two of the most important factors are yield strength and grain boundary carbides.
The laboratory tests on 39-72 revealed extremely high yield strength and essentially no grain boundary carbides, indicating that this tube would be expected to be extremely susceptible to PWSCC.
Based on the laboratory examination results, tube 7-78 would also be classified as susceptible to PWSCC. However, tube 7-78 would be expected to be more resistant than tube 39-72. The yield strength is lower, grain size is larger, carbon content is lower, and there are some intergranular carbides. As discussed earlier, destructive examination found no defects in tube 7-78.
[ d ] measurements were taken of both the sleeve and tube in the kinetic weld region of both 39-72 and 7-78 (Figure 4.2). The [ d ]
of both tubes were similar, indicating that both tubes had received similar stress relief processes.
4.4 Additional Laboratory Examination of Kinetic Welds The kinetic welding process expands the sleeve and tube, thus causing material [ e ). Residual stresses occur with this [ e ] in the tube and sleeve. Subsequent stress relief reduces the residual stress and [ e J . There is not a direct correlation between material [ e ]
and residual stress, i.e. , significant stress relief can occur without a significant decrease in [ e ]. However, if the [ e ] of the tube in the kinetic weld region returns to that of the parent tube, then most if not all of the residual stresses in the tube have been removed.
UPL-78 51-1228707.01 Page 12 of 37
4
. NON-PROPRIETARY B&WNUCLEAR TECHNOLOGIES r
Af ter the initial laboratory results were known, sleeves were installed in various heats of tubing, stress relieved at different tenperatures, and examined for change in the [ d ] (Figure 4.2) . The stress relief -
was performed using heaters identical to those used during sleeve installation at McGuire-1 and the tube OD temperature was monitored by attaching thermocouples at the [ d, e ].
Comparison of the [ d ] measurements in Figures 4.2 and 4.3 along with mockup testing and stress relief records indicate-that tubes 39-72 and 7-78 were stress relieved at approximately [ d, e ].
Figure 4.3 shows that stress relief at higher temperatures completely removes the [ d ) induced by kinetic welding. Thus existing sleeves can be given an additional, [ c, d, e ) to insure that all residual '
stresses are removed. Removing all of the residual stresses will ensure that PWSCC will not occur in tubes that currently have kinetic sleeves. [ b, c, e ).
4.5 Conclusions of Root Cause Evaluation The defect in tube 39-72 was an intergranular circumferential crack at the root of the kinetic weld region of the Alloy 600 tubing, caused by PWSCC.
ECT detection [ d ), a review of installation records, and [ d ]
measurements indicate that tube 39-72 did receive stress relief.
Based on [ d ] and comparison to mockup tests, the stress relief temperature of 39-72 and 7-78 was approximately [ d ].
- The stress relief process on tube 39-72 did not reduce the residual stress sufficiently to prevent PWSCC.
The high yield strength and lack of intergranular carbides in tube 39-72 greatly increases its susceptibility to PWSCC.
The lower end of the stress relief process [ d, e ) is inadequate i for tubing that is substantially more susceptible to PWSCC than that used during corrosion tests in the original qualification ,
program. ,
UPL-78 51-1228707.01 Page 13-of 37
NON-PROPRIETARY ~ 5 lllB&WNUCLEAR
- w TECHNOLOGIES TABLE 4.1 CHEMISTRY AND MECHANICAL TEST RESULTS
[d)
(
l l
i i
i i
UPL-78 51-1228707.01 , l Page 14 of 37 j i
I
, NON-PROPRIETARY Mi1IBGWNUCLEAR Rac W TECHNOLOGIES FIGURE 4.1 LAD EXAM OF TUDE 39-72
[d]
UPL-78 51-1228707.01 Page 15 of 37
NON-PROPRIETARY 5 B&W NUCLEAR '
3 ECHNOLOGIES FIGURE 4.2
[ c, d] OF KINETIC WELDS REMOVED FROM MCGUIRE-1
[ c, d)
[ c, d]
.=..
o UPL-78 51-1228707.01 Page 16 of 37
4 NON-PROPiuETARY-D%111R&WNUCLEAR I3 W TECHNOLOGIES FIGURE 4.3
[ c, d] OF SAMPLE KINETIC WELDS
[ c, d, e)
. 1
- UPL-70 51-1228707.01 Page 17 of 37
, 1 l
EON-PROPRIETARY j BW"lcWHo"nia%
l 5.0 EVALUATION OF KINETIC SLEEVES IN-SERVICE As seen from the examination of tubes 39-72 and 7'-78, identical sleeve installations operating for identical service time had two very different results. One tube exhibited PWSCC and the other did not.
In order to evaluate the potential for PWSCC to occur in tubes with kinetic sleeves, it is necessary to evaluate the variables involved.
The kinetic sleeving process is consistent from tube to tube. The kinetic weld device (KWD) loading is controlled to tight tolerances during fabrication. Thus the resultant tube expansions and subsequent strain levels are consistent from tube to tube following welding.
In addition, the [ o ] controlled stress relief heaters provide consistency in the final tube temperature during stress relief.
The final weld joint is located in straight regions in the tube remoted from geometry effects (such as the tubesheet) and OD sludge pile effects. Thus, the installation environment is consistent. Therefore, the remaining variable associated with the sleeving process to evaluate-whether PWSCC will occur in the freespan joint region is the tube material properties.
An algorithm (Ref. 8.6) was developed by Dr. Jim Begley of Packard Engineering to estimate the lifespan of a kinetic sleeve based on the materia 1 properties of the tube. The derivation of this algorithm is summarized below.
5.1 Life Ranking Factor In the absence of significant variations in welding and heat treating parameters, the behavior of other tubes relative to the cracked tube depends on differences in susceptibility to PWSCC and response to stress relief heat' treatment of the Alloy 600 material. The only distinguishing features of the cracked tube joint is the very high yield strength (72.5 ksi) , high carbon level (0.05%), and lack of intergranular carbides of the parent Alloy 600 tube.
In terms of parameters which are known for many of the kinetic welded sleeves in service, susceptibility to PWSCC can be ranked using yield strength and carbon level. Since stress relief depends on both high temperature yield strength and time dependent deformation, it is logical that room temperature yield strength will also provide an estimate of relative response to stress relief heat treatment. Using data in the literature on stress corrosion of Alloy 600 in high temperature water and the high temperature stress strain properties-of Alloy 600 relative to room temperature yield strength, a ranking approach was developed which relates the expected service performance of kinetic welded sleeves relative to the sleeve joint in tube 39-72 which cracked after 1.19 effective full power years of service. This relative life factor depends on the yield strength and carbon level of the parent tube. The details of the life ranking factor approach are presented below.
UPL-78 51-1228707.01 Page 18 of 37
NON-PROPRIETARY ;
PmEEIDGWNUCUUtR IDWTrCHNOLOGlES Susceptibility of mill annealed Alloy 600 tubing to PWSCC has been shown to have a strong correlation with yield strength (Ref 8.1 and -
n.2). The time to cracking of reversed U-bend (RUB) test specimens. R in high temperature water has also been related to excess carbon-level.
The term excess carbon level refers to the carbon level minus the
^
amount expected to be in solution at the final mill annealing temperature. .
The yield strength effect is dominant but there is a significant effect j of excess carbon level on time to cracking. Historically, metallurgical structure has been related to PWSCC susceptibility, {
j but in the present instance, this approach is not practical. In many .l cases, tensile _proporties and chemical compositions of the tubes of inter'3st are available but details of the metallurgical structure- l are not recorded.
1 The selected ranking approach is based on three factors which are multiplied together to arrive at a life factor which relates the-expected time to cracking of a tube with a kinetic welded sleeve 1 installed. The life factor, L (in f EFPY) is composed of a _ strength factor, S g , a carbon factor, C,, and a constant of 1.19 which is the number of effective full power years when cracking was observed for ,
the upper sleeve joint in tube 39-72. Thus the life factor is relative to the known cracking time of tube 39-72.
[ c, d] .....Eq.1 The strength factor was developed as follows. The log of time'to cracking of RUB specimens in high temperature' water was plotted versus ,
yield strength to develop an expression relating time to cracking versus yield strength. This expression was corrected to reflect the ;
differences in stresses existing in a RUB. specimen versus a heat treated kinetic welded sleeve joint. This correction is facilitated i by the fact that only ratios of stresses.need be considered,.since l all results are referred back to a parent tube yield strength of 72.5 '
ksi. Hence, the strength factor for a joint with the same tube yield strength as 39-72 is 1.0. If the yield strength is less than 72.5 ,
ksi, PWSCC susceptibility and.the stresses leading to PWSCC will l decrease and the relative life will increase. l Figure 5.1 shows a plot-of time to cracking versus yield strength- '
for RUB specimens of 0.75 inch diameter Alloy _600 steam generator- l tubing tested in 680* F water (Ref. 8.1 and 8. 2) . A best fit.line from a log-linear relationship is shown. The equation for this line '
is given-by:
[ c, d ] .....Eq.2 1
'l i
UPL-78 51-1228707.01 !
Page 19 of 37 ,
1
NON-PROPRIETARY BW"lMo"a!Es where t, is time to crack in hours and o g is yield strength in ksi.
Similar results are obtained from other data in the literature. Figure 5.2 shows a plot of time to cracking versus yield strength for RUB specimens from 0.875 inch diameter Alloy 600 tubing. These specimens were tested in simulated primary water at 690* F water (Ref 8.3).
Figure 5.1 shows the effect of yield strength on time to cracking '
of 0.75 inch RUB specimens in high temperature water. This curve can be normalized relative to a yield strength r f 72.5 ksi by dividing all times to cracking by the value for a yield strength of 72.5 ksi, which is [ c, d ] Figure 5.3 shows the normalized time to cracking curve.
[ c, d] .....Eq.3 Sin < hhe activation energy for cracking is not a function of yield str.ngth, the normalized curve of Figure 5.3 is temperature independent. Changing the test temperature would change all RUB cracking times by the same factor. It should be noted that the data from 0.875 inch RUB specimens shows a stronger yield strength effect on time to cracking than the 0.75 inch RUB data and supports the reasonableness of the de'a used for the strength ranking factor.
A RUB specimen from 0.75 inch diameter tubing has an equivalent plastic-strain of about [ d] 8.4). Figures 5.1 and 5.3 refers to material with about [ d ](Ref.plastic strain subjected to stresses which are a substantial fraction of the work hardened flow stress. This curve can be adjusted to a constant stress level. Figure 5.4 shows a plot of various flow strengths as a function of the initial room temperature yield strength. This data is taken from Reference 8.4.
The stress in a 0.75 inch RUB specimen at a test temperature of 680' F is some fraction of the work hardened flow stress at operating temperature. A linear regression through the_ flow stress data at 680*F gives:
[ c, d ) .....Eq.4 It is reasonable to assume that the proportionality constant, a, _ is independent of yield strength. A power law with an exponent of is typically used to relate time to cracking,in high temperature water ..
to stress level (Ref. 8.5) . Hence the normalized RUB curve of Figure )
5.3 can be further normalized to the stress level experienced in a !
RUB with a material yield strength of 72.5 ksi. N, is simply l multiplied by the factor [ c, d ) . Note that the value of a is not i needed since only ratios are of interest.
l i
The final strength factor is obtained by considering the stress level in a heat treated sleeve joint of a given tube yield strength relative to that expected in a tube with a yield strength of 72.5 ksi. The maximum plastic strain level in a parent tube in a kinetic welded sleeve upper joint is about [ d ) . Upon fabrication, the stress in UPL-78 51-1228707.01 Page 20 of 37 i
e.
NON-PROPRIETARY BW?lsHo4%ai ;
the tube is some fraction, , of the [ d ] strain hardened strength. After heat treatment the stress level decreases.
flow l Upon heating to the stress relief temperature, the stress level immediately
~
drops to the yield' strength at temperature and then further decreases-as time dependent creep converts clastic strain to inelastic strain.
The efficiency of the stress relief process is expected to bet a function of the initial . yield strength of the tube. Hence the remaining residual stress (aggs) in the tube is defined by a linear. -
regression through the [ d ] flow strength data in Figure 5.4: l
[ c, d) .....Eq.5 The value for the proportional constant (#) is not needed since only '
ratios are of interest. The final strength factor as a function of 1 yield strength is then given by: I
[ c, d ] .....Eq. 6 This factor includes both susceptibility to PWSCC and stress levels -i in heat treated joints as a function of parent tube yield. strength and is plotted in Figure 5.5 versus yield strength.
The carbon factor is not as dominant as the strength factor and was-developed in a much simpler manner. The time to cracking data of Norring el. al..(Ref. 8.1, 8.2) for RUB specimens testing in 680*F water was plotted versus excess carbon level and the linear fit was 4 normalized to a value of 1.0 for a total carbon. level of 0.05% (The- .
value of tubo 39-72). The carbon level expected to be in solution at the final mill annealing temperature for the tubes of interest was taken as 0.021%. Figure 5.6 shows a linear fit of time to cracking ;
to the RUB test data. The normalized carbon factor curve is shown in Figure 5.7. The equation for this line is given by-
[ c, d ] .....Eq. 7 where C, is the carbon factor and %C is the carbon level in percent. -
The factor is normalized to a value of 1.0 for a carbon level of 0.05%.
The magnitude of the carbon factor.effect is illustrated by the. fact that changing the carbon level from 0.05% to 0.02% increases the PWSCC susceptibility by about a factor of [ d ].
In summary, the total life factor, L,, as written above, [ c, d ] ,
can be computed from the yield strength and carbon level of the parent ;
tube. It establishes the expected service performance of the upper :
joint of kinetic welded sleeves. It is based on the actual time to.-
cracking of the only heat treated kinetic welded sleeve joint which- l has cracked in service (tube 39-72) , .the relative PWSCC susceptibility of parent tubes expressed in terms of yield strength and carbon level and consideration of the relative stress levels in RUB test specimens '
and heat treated sleeve joints.
[
UPL-78 51-1228707.01 Page 21 of 37 k ., , . -- - .__ . _ _ _ - - _ _ . =. -- --- A
i i NON-PROPRIETARY' l US1ff0&WNUCLEAR I3wTEcHNOLOGIES l
5.2 Variation in Properties and Cracking Times !
Since the computed life factor depends on the yield strength and carbon content, it is appropriate to consider variations between reported CMTR values and actual measured values, The only significant database available is a comparison of CMTR reported yield strengths with pulled tube test results for tubes from McGuire Units 1 and 2. Figure 5.8 shows a plot of the ratios of pulled tube yield strength to CMTR value yield strength. The average of this ratio is [ d ] . Except for an outlier considered to be in error, the maximum of this ratio is If all CMTR yield strength values are multiplied by [ d ], then the life factor approach would predict [ d ) incidents of cracked sleeve joints at McGuire Units 1 and 2 after 1.19 EFPY of service. Since only one cracked sleeve joint has been observed, this is clearly overly conservative. It is reasonable to consider the influence of increasing reported CMTR yield strength values by a factor of [ d ]. The following table summarizes the conservatism of the predicted tube failures due to kinetic sleeves versus yield strength values:-
Tube Yield Strength Predicted Actual Failures Failures Per CMTR [d] 1 CMTR * [d) [d]
CMTR * [d) [d]
Thus it is not considered necessary to adjust the yield strength value reported on the CMTR to conservatively estimate the lifespan of a kinetic sleeve.
Strength factors and carbon factors were developed from best fit lines.
It is not necessary to use bounding fits since the life factor is keyed to the known service performance of a cracked sleeve joint.
Other sleeve joints with equal life factors did not crack after equal or longer periods of service. On this basis, the life factor equation is considered to be keyed to an observed cracking probability of about
[ d ]. Having made this point and considering variations in properties and uncertainties in analysis, a minimum safpty factor of [ d ] should be applied when comparing life factor.to expected service life.
5.3 Application of Life Factor Given a tube's yield strength and carbon content, the algorithm '
described above can be used to estimate the life of a kinetic sleeve.
The total life factor, L f , can be simplified by combining equations 2, 3, 4, 5, and 6 to the following formula:
[ c, d ]
UPL-78 51-1228707.01 Page 22 of 37
NON-PROPRIETARY
$ $ frP N [ou g Where: L, = Life Factor in EFPY a = Tubing yield strength, kai
%E = Tubing carbon content, %
The following table illustrates how the calculated life factor varies as yield strength and carbon. content varies.
y_ield Strencith Carbon Content Predicted EFPY 60 0.03 60 0.04 62 0.03 [d]
62 0.04 65 0.04 Assuming that the sleeves in these tubes had been in service for 3 EFPY and that the steam generators were going to be replaced in an additional 3 EFPY, a total of [ d ] EFPY would be required to insure that there were no tube failures. Applying the minimum recommended safety factor of [ d ], any sleeve with a predicted life of less than
[ d ] EFPY would either be removed from service or [ c, d, e J . This would insure that no tube failures occurred due to the installation of kinetic sleeves.
For reference, tubes 39-72 and 7-78 at McGuire-1 have a predicted EFPY of [ d ] and [ d ] respectively. The laboratory values for yield strength and carbon content for these two tubes were used to calculate their life factor.
UPL-78 51-1228707.01 Page 23 of 37
NON-PROPRIETANY IHOGW NUCLEAR W TECHNOLDGIES FIGURE 5.1 TIME TO CRACKING VERSUS YIELD STRENGTH 0.75 INCH RUB SPECIMENS
[ c, d]
9 Time to cracking versus yield strength for 0.75 inch RUB specimens of-Alloy 600.
UPL-78 <
51-1228707.01 Page 24 of 37
4 "o"-"noraraTAar ;
BWficN##&yp; FIGURE 5.2 TIME TO CRACKING VERSUS YIELD STRENGTH 0.875 INCH RUBS, SIMULATED PRIMARY WATER 1
h
[ c, d) i Time to cracking versus yield strength for 0.875 inch RUB specimens of Alloy 600 tested in simulated primary water.
UPL-78 51-1228707.01 Page 25 of 37
ISON-PROPRIETARY '
95111DGWNUCLEAR 135AUTECHNoLOGIES FIGURE 5.3 TIME TO CRACKING RATIO VERSUS YIELD STRENGTH 0.75 INCH RUB DATA
[ c, d ]
E i
l Time to cracking ratio versus yield strength for 0.75 inch RUB specimens of Alloy 600.
l UPL-78 51-1228707.01 Page 26 of 37 I
I
1 NON-PROPRIETARY -
DHlDGWNUClEAR '
E=0 w WCHNOLOGIES T FIGURE 5.4 l FLOW STRENGTH VERSUS YIELD STRENGTH
[ c, d)
Flow strength of Alloy 600 tubing versus room temperature yield strength.
UPL-78 51-1228707.01 Page 27 of 37 l i
a
, NON-PROPRIETARY' IS B&WNUCLEAR 13WTECHNOLOGIES FIGURE 5.5 STRENGTH FACTOR VERSUS YIELD STRENGTH FROM 0.75 INCH RUB DATA
[ c, d ]
f t
Strength factor versus Alloy 600 parent tube yield strength for heat treated -
kinetic welded sleeve joints.
UPL-78 51-1228707.01 Page 28 of 37
NON-PROPRIETARY BWrig#pgj; FIGURE 5.6 TIME TO CRACKING VERSUS % EXCESS CARBON 0.75 INCH RUB SPECIMENS
[ c, d)
I Time to cracking versus % excess carbon for Alloy 600 RUB specimens.
l l
l l
i i
l UPL-78 51-1228707.01 l Page 29 of 37 l 1
i
, NON-PROPRIETARY 95111B&WN'JCLEAR I3 WTECHNOLDGIES FIGURE 5.7 CARBON FACTOR VERSUS % CARBON
[ c, d]
i i
Carbon factor versus % carbon.
1 UPL-78 51-1228707.01 Page 30 of 37
NON-PROPRIETARY 95HIB&WNUCLEAR 13 WTECHNOLOGIES '
FIGURE 5.8 PULLED /CMTR YIELD RATIO VERSUS CMTR YIELD STRENGTH
[ c, d]
Ratio of pulled tube yield streng'th to CMTR yield strength versus CMTR yield strength.
UPL-78 51-1228707.01 Page 31 of 37
NON-PROPRIETARY-
- 9"h1110&WNUCLEAR 33 WTECHNOLOG!ES 6.0 CONSEQUENCES OF TUDE FAILURES This section examines the effects of tube failures associated with kinetic sleeves. The effect of a single tube failure and the possibility of a multiple tube failure are discussed.
6.1 Effects of a Single Tube Failure As discussed in the Section 3.5 and 3.6, there have been two tube leaks associated with BWNT's kinetic sleeves. While the root cause of each tube failure was different, the effects of both are remarkably similar.
- 1) NDE of both tubes showed that the tube defect was a circumferential crack from.180
- to 270
- around the tube. Even at Trojan, where no stress relief was performed on the effected tube, the tubo did not completely sever.
- 2) Both plants detected primary to secondary leaks in the range of 200 gpd and immediately conducted a safe shutdown of the plant.
- 3) No damage occurred to tubes adjacent to the failed tube.
BWNT's sleeve is designed such that the sleeve [ c, e ] (Figure 4.1) soc,that even if the tube suffers a 360* circumferential sever,
[ e ].
In the event of a tube rupture, the end of the sleeve [ d, e ]
Calculations show that with a severed tube [ d, e ] following a main steam line break; [ d j at full power conditions. For comparison, the [ d ) at full power conditions for a design condition double-ended tube rupture.
6.2 Multiple Tube Failure There are two possible scenarios which could result in multiple tube !
failures: two unrelated tubes failing at the same time, or a severed I tube impacting against and rupturing adjacent tubes. The probability of these two scenarios are discussed below.
l 6.2.1 Simultaneous Rupture of Unrelated Tubes The tube failures at McGuire-1 and Trojan are not the first instances of circumferential cracking caused by PWSCC of steam i generator tubing in the nuclear industry. Circumferential cracks (ID and/or OD) have been seen world-wide since 1986, for example:
- Plants in Belgium: Doel and Tihange; Plants in France: Bugey, Tricastin, Dampierre, Fessenhelm 2, and St-Laurent; UPL-78 51-1228707.01 Page 32 of 37
NON-PROPRIETARY h#hggu c
Plants with WEXTEX expansions: North Anna 1, Fessenholm - 1, and Trojan;
- Additional plants in the US: Arkansas-2 and South Texas-1 This problem has been addressed by the varitun utilities using a combination of inspection and repair. Compared to the number of defects found in the industry, relatively few have occurred between regularly scheduled outages. It is also important to note that none of these plants has experienced a multiple tube rupture.
As discussed in the background section, BWNT has installed over
[ e ] tubesheet and tube support plate sleeves since 1990. The large majority of the sleeves were installed to repair PWSCC of the parent tubing. Thus over [ c ] stress relieved kinetic joints have been installed in tubing with PWSCC indications and es: posed to the primary environment of a nuclear steam Jenerator. Tubo 39-72 at McGuire-1 is the only failure produced by a properly installed kinetic cleeve.
Thus industry experience with circumferential cracking has shown that a program of inspection and repair adequately addresses the problem of circumferential cracking. The chances of two unrelated sleeves failing at the same time is considered to be very small.
6.2.2 Severed Tube Impacting Adjacent Tubes An unrestrained tube is considered to be any tube that does not have a tubo directly above it. This occurs for overy outermost (periphery) tube in each column and for those tubes next to stay rod locations (Figure 6.1). If an unrestrained tube were to rupture and completely sever, the sleeve end may not be sufficient to restrain the tube end. If this were to occur, then there is also a chance that the severed tube would impact against the adjacent tubes, causing wear, and could eventually cause additional tube ruptures.
The conditions required to achieve the multiple tube rupture scenario described above are listed below in more detail:
- 1) Unrestrained tube with a kinetic sleeve installed fails ,
circumferentially due to PWSCC. J
- 2) The tube completely severs.
- 3) The tube is free to move through every TSP locati? fuov+. I the sleeve joint and the tube disengages from the s.' wye end. ,
i
- 4) The severed tube impacts adjacent tubes until multiple tubes }
are ruptured. '
- 5) No operator action is taken to shutdown the plant at any point in the scenario.
1 UpL-78 51-1228707.01 Page 33 of 37 1
' " "-rnovarerAnr
~
BWricM#mg There are several required conditions listed above that make this type of multiple tube rupture unlikely:
- 1) Complete tube sever The tube failures at McGuire-1 and Trojan did not result in a complete tube sever, even though the tube at Trojan had received no stress relief.
For those plants with drilled TSP holes, there is a good chance that the tubes are tightly held or " locked" at the TSP locations. The amount of friction generated at each TSP will vary from plant to plant based on design and operating history. An indication of whether or not the tubes are free to move through the TSP's can be determined if attempts to remove long sections of tubing have been made for destructive examination.
- 3) No operator action is taken throughout the scent,rio.
The tube failures at McGuire-1 and Trojan resulted in primary to secondary leak rates of approximately 200 gpd or 0.14 gpm. At both plants, the operators took quick action and conducted a shutdown of the plant using standard procedures. During the scenario described above, the primary to secondary leak rate would be [ d ) after the tubo completely severs and [ d ] after the tube separates from the sleeve end. It is extremely unlikely that plant operators would take no action to shutdown the plant before the severed tube impacted and damaged adjacent tubes.
6.3 conclusions
- A single tube failure is an analyzed transient and does not constitute a safety concern.
- Existing plant procedures are adequate to deal with a tube leak caused by a kinetic sleeve. , Each plant should ensure that operators are familiar with the indications and procedures concerning a primary to secondary leak.
- As long as the failed tube is captured, the kinetic sleeve design substantially reduces the primary to secondary leakage.
- Based on the industry's experience with PWSCC and BWNT's experience wit'. kinetic sleeves, it is not credible to believe that a multiple sleeved tubes would independently rupture at the same time.
UPL-78 51-1228707.01 Page 34 of 37
.s NON-PROPRIETARY BEWNUCLEAR TrCilNOLOGIES
- Based on the two tube failures caused by BWNT's kinetic sleeve, it is not likely that a tube will fail and completely sever without any operator action.
The only credible scenario for a multiple tube rupture would be a complete sever in an unrestrained (or uncaptured) tube.
Even though the chance of a unrestrained tube causing a multiple tube rupture is considered small, unrestrained tubes with a kinetic sleeve installed need to be addressed on a plant by plant basis.
1 I
i UPL-78 51-1228707.01 Page 35 of 37 i
4
" . e NON-PROPRIETARY
.a lifB&WNUCLEAR KAR TECHNOLOGIES
,m n .. . ....i ..
.!. ne .- :::. . .'... FIGURE 6.1
...t ..
_ sew ....
p ..
....... . . ........ . .. ...................... g c
g ,.ar . ...................................
E
. ...................... ......... .... a
- c. er ... ... ... -
t . ..
.. U
_m u>
g c-7; c,
, er . . . .
3 g . . .. ,3
...... .... .. .. . . _s as . .
c
..a............................................
. . . ...... :o rs n,
ca
. ... .. ... ... .. . .. ... .. . .. ... ... .. ... ... .. ... ... .. . .. ... .. . .. ... ... ., .. .c: . . . . . . .
.. -s w,
... ....... ....... ..... .... ... .... ... .... ... .... ... .... ... .... ....... ....... ........... ....... ...... .. .. .. .. .. .. .. .. .. rn aa uw ...................................
......................... . . . . . . . . . . . . . . . . .o, . . . . . . . .
r-
. . . . .. .. .. .. .. ..p .. .. . . c
. . o n as
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. ~
),
- a
... . ... .. . .. .. . E. . . . . . . . . c3 .......
.> un ..............................
z
. ....... u,
- . ..... .... . .u un . ..
... -s o
t
. . . . ~
~ <
n,
- o
., -a n .
rr
. . . . . r xs m
rn en
... .... -s x.. ... . .
. o
, 70 c
as ... ...
c z,
. m s-
.. m m
+
nr m
g .. .
,. 3 .
g .
. B 3 8 .
... mws ...
UPL-78 51-1228707.01 Page 36 of 37
)
NON-PROPRIETARY l 0 nGW NUCLEAR TECHNOLOGIES l
7.0 CONCLUSION
S The leak in tube 39-72 at McGuire-1 was caused by an intergranular, circumferential crack due to PWSCC. The crack occurred in the tube ,
at the upper end of the freespan kinetic weld.
The tubc/ sleeve joint created during the kinetic sleeve installation was stress relieved at approximately [ d, e ].
The combination of high yield strength, high carbon content, and no intergranular carbide ' precipitation make tube 39-72 extremely susceptible to PWSCC.
The lower end of the stress relief process [ d, e] is inadequate for tubing that is substantially more susceptible to PWSCC than that used during corrosion tests in the original qualification process.
An algorithm has been developed that conservatively estimates the service life of kinetic sleeves. This algorithm provides the basis of a method to determine whether or not corrective action is required for kinetic sleeves that are currently in-service.
A single tube failure is bounded by plant analyses and is not a safety concern. Existing plant procedures are adequate to deal with a tube leak caused by a kinetic sleeve.
It is extremely unlikely that kinetic sleeves could cause a multiple tube failure.
8.0 REFERENCES
8.1 [c]
8.2 [ c ].
8.3 [c]
8.4 [ c]
8.5 [ c]
8.6 [ c]
i UPL-78 51-1228707.01 Page 37 of.37