ML20236D742
| ML20236D742 | |
| Person / Time | |
|---|---|
| Site: | Fermi  |
| Issue date: | 07/28/1987 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML20236D731 | List: |
| References | |
| NUDOCS 8707310047 | |
| Download: ML20236D742 (8) | |
Text
_ _ _ - - _ _ - _ _ _ _ _ _ _ _ _ _ _ -
- o g UNITED STATES 8 g NUCLEAR REGULATORY COMMISSION j g j WASHINGTON, D. C. 20666
,o SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION RELATING TO MAIN STEAM INSTRUMENT LINE LEAKS DETROIT EDISON COMPANY FERMI, UNIT 2 Docket No. 50-341
1.0 INTRODUCTION
On April 10-11, 1987, the licensee identified several steam leaks in the main steam'11ne. The leaks were caused by cracking in the welds joining I p
two small instrument pipes to the sain steam line and by a circumferential break through the tubes at the compression fitting. The licensee reduced j power to 25 percent on April 11, and isolated the leaks by passing steam directly to the condenser through the tu'rbine bypass lines. A flush con-nection on the east bypass valve blew off subsequently, however, requiring l the reactor to be shut down for repairs to the steam lines. The steam l 1eaks resulted in the release of some airborne radioactivity in the turbine l building, but no personnel contamination was reported.
The above leakage incident, which is a recurring problem, has been identi-fied by the licensee to be due to flow induced vibration of main steam
? lines resulting from throttling of valves. On January 2, 1987, the ,
L Ifcensee discovered three small steam leaks on two of the four main steam lines. The leaks were from broken spare ildrument lines located between the turbine control valves and the main turbine itself. The instrument I
line failures were at a dissimilar weld connecting the instrument tubing L
to a valve. The licensee plugged, capped, and welded the broken lines.
On January 5, the licensee identified a cracked weld on another instru-ment line/ valve connection. The licensee resolved the problem by adding i
a flex hose to the discharge end of the pipe. On January 6, a crack was l found at a dissimilar weld for a flushing connection, which was later l resolved by capping the connection. Then on January 29, a cracked weld was found on still another instrument line/ valve connection. The licensee agair plugged, capped, and welded the broken line.
As a result of the January failures, an extensive modification program was initiated and completed by the licensee for the instrument and tap connections on the turbine stop/ throttle valves, bypass valves, and associated loop piping system. Per the licensee's submittal of May 22,1987, a total of 112 tap connections were modified as follows:
- 1. 39 valves were removed and connections capped or plugged.
- 2. 65 stainless steel valves were replaced with lighter carbon steel valves and/or configurations shortened - 36 were leak detection connections.
- 3. 7 flushing connections were shortened and capped.
- 4. I flushing connection was plugged.
8707310047 B70728 1 PDR ADOCH 0500
_P -
fl 1
Some of these modifications have been found to be inadequate, judging from a subsequent series of failures which occurred in April 1987. 1
- 2. 0 EVALUATION On April 28-29, 1987, the staff conducted an on-site inspection of the
, affected instrument / tap piping systems that failed in April 1987. The
/ staff also conducted a review of the stress and fatigue life analyses of the systems as they were before the April 1987 failures and as they were modified after the April 1987 failures.
During the field inspection it was observed that the two cracked welds l found in the April 1987 incident, had been ground out and butt welded to j fill the weld up to the one and three quarter inch diameter of branch pieces.
The redesigned welds appeared to have very smooth profile and free of stress concentration. The failed compression fitting which occurred on l b< one of the two connection / tap lines with cracked weld, had been replaced i by a new fitting. In addition, a flex hose was installed to the outboard l end of the line. For the flush connection that suffered a circumferential break the licensee had installed a threaded plug and seal weld. All these field modifications appeared to be adequate from the point of view of l eliminating potential effects of flow induced vibration in causing future l main steam leakages.
,. Course of Failures g I supporting the above field modifications made after the April f 'ures, the licensee has provided a summary of the causes of
. sieures, as well as how they are addressing the question of main
-am pipe wall integrity, as detailed in the following:
- 1. Instrument Tap Failures
.t One of the weld failures that occurred in an instrument tap on April 10, 1987, was at a source valve connection where the weld had been redone in January 1987. There were several factors leading to this failure as follows:
(1) Design deficiency - Failure to recognize that if the source nipple were shortened sufficiently the toe of the inboard fillet weld would meet the tapered shoulder of the one and three quarter inch source nipple, causing an increase in stress intensification. The design was analyzed on the basis of a fillet weld to a three quarter inch pipe (one inch 0.D.) and analysis still showed that this would be sufficient if the fillet were away from the shoulder. The field workers were given a general direction
__________J
4 M
to make the nipples as short as possible. They were not specifically directed to fill the weld up to one and three quarter inch diameter, although it was suggested that this i be done if the nipples were short enough. '
(2) Communication problem - There was a communication pro-blem regarding the stress analyst knowing exactly what !
the field configuration of instrument taps was and the l precise location of the strain gauges. He thought there was a three quarter inch nipple extending from the one and three quarter inch source nipple and that the strain gauges were on the smaller nipple. Based on that, the stresses were at an allowable level in the fillet attachment weld. Actually, the strain gauges were on the one and three quarter inch source nipple and the valve was closely attached without the weld being filled but to full 1-3/4" diameter. With the correct configuration , I b.e analysis showed the design was marginal because of higher stresses and that corrective action would probably be required, such as strengthening the weld. j (3) Potential metallurgical problem - A contributor to the failure, although probably not a major one, may have been !
resideal chrome left on the nipple when the stainless steel j weld was removed. There are four source connections in the i inlet pipes near the turbine casing. One of these failed I and this weld was ground out before analysis could be made. j U
However, the other three non-failed valves were also removed and the weld area analyzed. Some residual chrome was detected in the filler metal near the nipple and it was determined that this was from residual stainless steel from ,
the old weld that was removed. This would cause a harder 4 alloy and could increase the stress intensification.
(4) Off-design operation - Another contributor is the extended operation at reduced power. Ideally, the plant should be able to operate at any power level, but experience at other plants has shown that there are problems with instrument taps when the plants are at low power levels during startup, but once full load is achieved the steam lines quiet down and the problem goes away.
The corrective action for the inboard welds is to make the welds I full diameter on those nipples where the source valve is close to the necked down portion of the source nipple. Also, additional I strain gauges will be installed to monitor all of the instrument !
taps in question. Subsequent analysis of field data will reflect '
i actual field configuration precisely. Had this been done before, the fact that the design was marginal would have been detected. l l
9
Q 1
Other corrective actions are to increase the fillet size on the outboard side of the source valve, and to utilize flexible tubing for the eight active connections in order to reduce the stresses in the welds.
- 2. Flushing Connections A flushing connection attachment weld at the number 3 stop valve failed in January. This was a one and one half inch diameter capped pipe nipple approximately 14 inches long. The fillet weld attaching this pipe to the throttle valve body was found to be martensitic, which is hard and prone to cracking. This is a
- result of insufficient preheat and postweld heat treatment. The connection was removed and the hole plugged. There were seven other similar connections on the other three stop valves and two g bypass' valves. It was felt that bad welds in the other valves g would have started leaking and it was therefore decided to leave the nipples in but shorten them to five iaches and perform a dye penetrant examination of the welds, which revealed no surface flaws.
l On April 11, 1987, one of these five inch stubs in the bypass valve !
failed totally, breaking through the pipe wall at the toe of I the fillet weld. Examination of the crack determined that this was a fatigue failure that had been developing for some time.
Like the instrument taps this was a high cycle vibration fatigue failure.
The corrective actions are to cut the stubs and to install threaded plugs and seal weld.
- 3. Main Steam Pipe Wall Integrity .
The data obtained from strain gauges on the main steam line pipe walls in January and February was rereviewed. It was determined that strcss levels for the main steam pipe walls are well within allowable values. Where lugs were welded on the pipes for snubber clamps, there is some increase in stress intensification. Stress levels are still safely below allowables, but strain gauges are being added to the pipe walls in the area of the lugs to more precisely describe the stresses and provide added assurance that there is no problem here.
The staff concurs with the above licensee's evaluation and believes that based on the understanding of the causes of failure the modifi-cations that have been adopted by the licensee in resolving the related failures of the instrument tap / connection and flushing lines are adequate, pendirg a satisfactory analytical stress / fatigue life evalua-tion of the affected systems. This is presented in the following.
l l
l Analytical Evaluation I
The licensee has retained Hopper and Associates as their consultant in I performing the stress and fatigue life evaluation of the instrument i tap / connection and flushing lines. The objective of the Hopper study is to evaluate the vibration fatigue sensitivity of the welds for the )
l tap / valve configurations. The valve is a part of the steam line. '
It is welded to a branch piece on the inboard side whereas a 3/4 inch Schedule 80 pipe is welded to the outboard side.
The two welds are analyzed with respect to a strain amplitude of 30 generated by vibrations in the pipes, on both sides of the valve. This strain amplitude is an approximate value and was measured on the tap between the process pipe and the valve in previous service application.
Mathematical finite element models were developed for the tap / valve con-h figuration. A typical tap is a 1-3/4" outside diameter branch piece welded to the main steam line. As stated previously, a valve is welded to this branch piece with a 3/4" Schedule 80 pipe welded to the outboard side of the valve. ,
All parts are made of carbon steel. A butt weld connects the branch piece to the valve. The piping is attached to the valve by a fillet weld with a 1:2 slope.
First of all, the finite element program ANSYS was used to determine the y stresses due to an applied uniform axial load. Stress intensities were m computed and stress intensification factors were determined for the inboard and outboaru configurations. The results from the computer stress analysis were ratioed to calculate the actual alternating stress intensities as corresponding to the measured strain in the field. .
l l An internal pressure was also applied to the finite element model and !
maximum stresses were computed at the welds. These stresses were con- !
l servatively assumed to be mean stresses for the purpose of constructing a Haigh Diagram.
A Haigh Diagram was used in this evaluation as a tool in a simplified approach to fatigue life estimation ar.d plotting of S-N curves. The Haigh Diagram shows the combinations of alternating stress, mean stress, i and notch factors (or fatigue stress concentration) that correspond to {
the fatigue limit or to a life of 5 million cycles. Both the cases with !
and without flaw were considered in the calculation of the notch factors.
l From the Haigh Diagram one obtains the fatigue strength at the endurance 1
limit of an S-N curve. Another point of the S-N curve is obtained from the 1 l
l l
1 l
l
)
6 j I
knowledge or estimate of the stress ocrresponding to a very short life (i.e. , stress for one cycle). These two points when joined by a straight line on log S-log N coordinates constitute an S-N curve.
For the inboard tap / valve weld, Hopper and Associates' analysis indicates i that for a mean stress of 1.2 ksi, the fatigue strength at endurance limit i equals 10.1 ksi and 15.8 ksi, respectively, for welds with and without I flaw. For the outboard tap / valve weld, the corresponding fatigue strength at endurance limit equals 12.8 ksi and 20.2 ksi, respectively, based on a mean stress of 2.1 ksi.
Comparing to the actual alternating stresses, corresponding to the field measured data, of 1.6 ksi and 1.0 ksi for the inboard and out-board tap / valve welds, respectively, the above obtained fatigue strength at endurance limit indicates an order of magnitude safety margin.
In addition, the staff also reviewed the Hopper's analysis on the integrity
. of main steam pipe wall and flushing connections, as provided in the l
licensee's submittal of June 23, 1987, and found the analysis procedures and results to be acceptable.
The staff, therefore, concurs with the licensee's assertion that, based on the above analysis results of Hopper and Associates, the main steam instru- i ment line configurations as modified will be adequate in preventing recurrence of the previous failure modes.
., Final Modifications With the aids of the previous Hopper and Associates' analyses and the understanding of the causes of April 1987 failures, the licensee has initiated and completed the following 51 connection modifications;
- 1. 10 valves were removed and connections capped or plugged.
- 2. 4 nipples and caps removed and plugs installed.
- 3. 4 connections had valve rewelded with butt weld and flex hosed added.
- 4. 4 connections had valve removed, nipple low pressure tested, valve replaced with filled in weld and flex hose added.
- 5. 7 flushing connections plugged.
- 6. 4 leak detection connections reworked.
- 7. 18 drain connections reworked -
12 on stop and bypass valves, 4 on turbine casing, and 2 on high pressure steam inlet pipe.
U e staff reviewed the detail information for each of the above modifi-cations as contained in the licensee's submittal of May 22, 198/. It was revealed that with the exception of drain connections most of those m
l n
modifications were to improve the modifications made as a result of i January 1987 failures. The staff, therefore, concurs that the licensee's !
actions taken after April failures indeed reflected its correct under- l standing of the root cause of failures. The staff also believes that, based on the information available, the above modifications will be able l to prevent recurrence of the incidents. l It should be noted that in the previously discussed Hopper and Associates' finite element analyses, the flex hose was not considered in the mathema- ;
tical model of the instrument lines. Although its effect is believed to l produce favorable stress intensity, a final evaluation by the licensee appears to be necessary in order to confirm the acceptability of that portion of the modifications. To this end, a recent shake table testing sponsored by the licensee has subjected the instrument line with flex hose to a hypothetical operating load of one million cycles that generated a strain amplitude of 150pc. The test result reveals no incipient failure of tubing itself and the weldment, and thus serves to further confirm the adequacy of the modifications.
Nut and Bolt Torque Verification During the plant site inspection, concerns on the loosening of nuts and bolts in the main steam and bypass lines were discussed. The licensee has since agreed to tighten the nuts and bolts as required. In the letter of May 26, 1987, the licensee has informed the staff that the following tasks have been accomplished:
g 1. Pipe Attachments Inner pipe clamp bolts for clamps on the main steam and bypass piping between the 52" manifold and the turbine or condenser have been veri-fied to have the required installation torque and the nuts have been staked to prevent loosening.
- 2. Actuator Bolting Bolts on actuator covers, access doors, and other similar non-critical applications have been verified to have the required installation torque. Lockwashers were used whenever possible.
- 3. Turbine Decking and Grating Bolts Screws and bolts used for grating and decking on the third floor, near the turbine, have been tightened to an installation torque as determined by the plant Maintenance and Modification personnel. In addition, these components have been inspected for any evidence of vibration damage (i.e., worn cotter pins, missing nuts, dislodged swivel bearings, loose lock nuts, etc.) and repaired as appropriate.
h$
j 4
W 8-
[$ The staff believes that the above licensee's actions have' adequately resolved y the concerns on the potential loosening of nuts and bolts in the stated piping systems and areas which may otherwise occur during plant operation.
3.0 CONCLUSION
Based on the evaluation presented in the above, the staff concludes that the licensee has satisfactorily resolved the problems concerning main steam instrument line leaks at Fermi-2 which occurred in January 1987 and then recurred in April 1987. ;
)
e