ML20216J100
| ML20216J100 | |
| Person / Time | |
|---|---|
| Issue date: | 03/31/1985 |
| From: | Chiramal M NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| To: | |
| Shared Package | |
| ML20216J082 | List: |
| References | |
| TASK-AE, TASK-E503 AEOD-E503, NUDOCS 9709170155 | |
| Download: ML20216J100 (20) | |
Text
_
AEDD/E503 l
l
,EllGIl4EERIl1G EVALVATIO!J REPORT PARTIALFAlldRES,0FC0!1Th0L ROD SYSTEliS TO SCRAli by 0FFICE FOR AllALYSIS AND EVALUATION OF OPERATIONAL DATA March 1985
'i
\\
Prepared by:
11atthew Chiramal
n
~
NOTE:
This document supports ongoing' AE0D and NRC activities and does not represent the positions or requirements of the responsible NRC program office, fd 9709170155 850311 PDR COMMS NRCC CORRESPONDENCE,PDR
TABLE OF CONTENTS, i
Page
~
~
SUlil4ARY f'
1.0 INTRODUCTIOil 1
3 2.0 DISCUSSION 3.0 FillDl!!GS At1D CONCLUSIONS 9
APPENDIX A Events Involving Control Rod System Failures in The Perfonnance of Reactor Trip Function 4
1 e
\\.
.w
-m 4
e 6
1-i l
i
SUMMARY
This report documents the identification and review of events at operating 4
nuclear plants involving partf al failures of the reactor control' rod sys' tem-to perfom its intended scram function.
The events under consideration are l
- those that have, occurred in U.S. reactors (with one exception) since the Salem ATWS events of February 1983. The scope of the report does not include individual component failures in the scram system, such as scram breakers, nor failures in the trip logic equipment.
To identify the events, AE00 has conducted a search of available operational experience data (Licensee Event Reports, Preliminary Notifications and i
10 CFR 50.72 Notifications) from February 1983 to December 1984.
Thirteen (13) events involving fai_ lures of control rods to perfom their trip function properly were identified. A brief description of these events with the names of the plants involved and the dates of the events are listed in Appendix A.
Of the 13 events, six occurred during an actual scram demand.
In all these six cases, the reactor was safely shut down by the proper functioning of the remaining operable control rods. The other seven events occurred during testing of the control rod systems. These events are of concern and significance because of the number of potential conynon-cause failure mechanisms that were identified and the potential generic implications associated with them.
Four of the 13 events are seen to have common-cause failure' implications and three of them (two at Boiling Water Reactor units and one at a Pressurized Water reactor unit) were detemined to have generic implications applicable to other light water reactors.
C The three potential common-cause failures that affect light ' water reactors were:
- 1) failure of scram pilot solenoid valves because of the presence of N-Loctite in the mechanism, 2) failure of scram pilot solenoid valves because of sticking disc holder subassemblies, and 3) looseness of rod assembly guide screws.
In all three cases, actions have been or are planned to be taken by the NRC and the industry to infom other,, operating nuclear plant licensees of the potential problem as follows:
- 1) IE Infomation Notice 84-53, "Infomation Concerning the Use of Loctite 242 and Other Anaerobic Adhesive /
Sealants" was issued on July 5, 1984; 2) General Electric issued a Service Infomation Letter (SIL)-on October 16, 1984 on T-ASCO solenoid valves; and
- 3) Westinghouse has advised the appropriate utilities, and the Office of Inspection and Enforcement intends to issue an Infomation Notice regarding
' the potential problem with the guide screws of Control Rod Drive Mechanism
' (CRDM) drive: rod assemblies.
I e-
-=
r m
o-y v.
y
-wp-.-
e y
<.g.g w
--a.
w -w
-n-m.-a.pponw
--sp w---m.-et-w y
1 The fourth event that was identified to have a common-cause " allure implication was the June 23, 1984 event at Fort St. Vrain.
ihe event involved the failure of six out of 37 rod pairs during a reactor trip.
Since the rod system at Fort St. Vrain is unique to that unit, this problem has no generic applicability.
Fort St. Vrain has been shut down since the date of the event, undergoing maintenance, repair and testing of the control rod system.
An additional problem with potential common-cause failure implication was identified in the reserve, shutdown system at Fort St. Vrain. The problem involved the failure of a shutdown hopper to discharge the designed amount of shutdown material during testing on November 5,1984.
The cause of this failure is under investigation. -This system, which 4
is an independent and redundant system, is also unique to Fort St. Vrain, and thus the problem does not have any generic implication. The licensee will obtain NRC approval of all corrective actions taken prior to reactor startup.
In addition to the concern regarding potential common-cause failure, the event at Susquehanna on October 6,1984, raised two other items of concern.
The first item is regarding the inadequacy of post-scram review practiced at the station which was identified by HRC Region I during a special inspection following the event.
The inspection found that during a scram of Susquehanna 1 on June 13,1984, one 2 X 2 control rod array exceeded the specified scram insertion time limit.
On June 25, 1984, the data obtained from this scram was used to demonstrate compliance with certain technical specification surveillance requirements.
The slow scram time was overlooked by the licensee's staff.
The significance of this was that two of the four control rods that failed.
to sc.am during the October 6,1984 event were in the 2 X 2 rod array that exceeded the scram insertion time during the June 13,1984 scram.
1 The second item of concern is the lac'k of attention given to*1essons learned from previous experience. On March 14, 1980, the NRC issued IE 4
Infomation Notice 80-11 entitled, " Generic Problems with ASCO Valves in Nuclear Applications Including Fire Protection Systems." This infomation notice described a potential deficiency of ASCO HP-1 solenoid valve.s regarding the effects of oil on ethylene _ propylene (EPR) elastomer materials which expands or swells when brought into tontact with oils, possibly causing
- valve failure. The notice also stated that Viton elastomer replacement kits were available from ASCO for NP-1 solenoid valves.
The information notice had an attachment which identified the potential incompatibility of solenoid valves with oil contamination in air systems.
The licensee'did investigate the problem of ASCO solenoid valves in 1981 and identified several valves at the Susquehanna station which had ethylene propylene (EPR) seals and which did fail due to oil contamination. However, the licensee's review did not identify the T-ASCO solenoid valves which employed polyurethane seals, other than EPR seals. These valves, containin were installed in all Unit 1 control rod drive (CRD) g polyurethane seals, assembites and in about one half nf the Unit 2 CRD assemblies.
a.-
i
- 111 -
i i
The renaining nine events are considered to be caused by random failures f
4 which do not appear to have common-cause failure or generic kmplications.
The corrective actions taken by the licensees in addressing these failures appear to be adequate.
However, one of these events, the October 25, 1984. event at Quad Cities 2, did raise a concern regarding the ade,quacy of l
post-trip review at that station because the plant operators were not asare, until 30 minutes after the event, that a control rod remained at the fully withdrawn position following the scram.
NRC Region 111 closely followed this event, and a Confirmatory Action Letter was issued and an Enforcement Conference was held to address the ' event and corrective actions.
The potential common-cause failure mechanisms were-identified promptly in most cases and the corrective actions regarding them were adequately. implemented.
However, the very existence of-such potential common-cause failure mechanisms in such an established and safety significant system as the control rod system remaine a scrious concern. Even though the Salem ATWS events and follow-up actions have made the licensees, vendors and NRC more responsive to events involving failures in the reactor trip systems, concerns identified after
~
the Salem events, such as inadequacies in post-trip review, post-maintenance i
testing, and identification of root cause of failure are still being noted.
When the actions discussed in the staff's Generic Letter 83-28 (Required Actions Based on Generic Implications of Salem ATWS Events) are established and fully implemented, these concerns should be alleviated.
Additionally, AE00 has established en in-depth periodic analysis'of unplanned reactor scrams as a major product of the Trends and Patterns Program.
A pilot study covering the first three months of 1984 was issued for staff comment in late November 1984.
The next effort, which includes a comparison with foreign experience, will cover all of 1984.
This report will provide o
an overview of the U.S. experience and the root causes of unplanned reactor i
J me
- 4.'
2 k
4 l
l i
._.u..___,._.
1-__-...,._
1.0 INTRODUCTION
1 During the semiannual briefing by the Office for Analysis and Evaluation of Operational Data ( AE00) to the Commissioners on November 20, 1984, Commt.ssioner Zech requested a listing _ and analysis of events that have occurred
- since-the Salem-1 ATWS events involving failures of control rods tb perforu' their scram function properly. This study was initiated in response to that request.
To identify the events of interest, AE00 conducted a search of operational experience data bases such as Licenseh Event Reports, Preliminary Notifications and 10 CFR 50.72 Notifications. As a result,13 events involving partial failures of control rods to perform their trip function properly were obtained. ; A brief description of these events with the name of the plants involved, the. type of reactor unit, and the date of the event are Itsted in
-Appendix A.
Of the 13 events in Appendix A, six-occurred during actual scram demand.
These six events are listed in Table 1.-
Table 1 Failure of Control Reds to Scram Upon Demand No. of Control plant Name Date of Event Rods Affected Type of Failure Cause of failure r
- 1. Quad Cities 2 12/22/84 3
Not fully Not known.
inserted
- 2. Quad Cities 2 10/25/84 1
- Stuck rod Scram dischargt riser valve closed.
- 3. Dresden 3
- 10/20/84 1
Stuck rod Manual isolation valve failure.
. 4. Lacrosse 07/16/84
'l Stuck rod loose' roller nut assembly.
S. Fort St. Yrain 06/23/84
- '6 Stuck rods Under investigatio_n.-
- 6. Peach Bottom 3 11/17/83
-2 Slow scram time Scram solenoid
- valve failure due to foreign material.
.,n
.T
-**''"-Y W
"F
-*"-W
N
- H'"+N
'F""*****F
*"""'**r"'"
.. 1 J
4 i
The remaining seven events occurred during~ the performance of control rod i
tests, and are listed in Table 2.
I l
?
Table 2 j
s-i
' Failure of Control Rods to Scram Upon Test No. of Control Plant Name Date of Event Rods Affected Type of Failure Cause of Failure f
1.- lionticello 12/05/84 Several rods slow scram time Plugged screens in hydraulic control units.
- 2. Ko-Ri 5 11/19/84 1
Stuck rod Loose assembly guide screw.
- 3. Susquehanna 1 10/06/84 Several rods Sticking rods Sticking
-i and slow scram pilot scram time solenoid valve disc holder subassembly.
- 4. Trojan 08/18/84 1
Stuck rod Top' hat. pin mi salignment.
S. Surry 1 06/20/84 1
Stuck rod Obstruction byi foreign object.}
- 6. Browns Ferry 2 03/22/83 1
Stuck rod Manual isolation
- valve failure.
- 7. Browns Ferry 3
_03/12/83 1_
Stuck rod-Scram solenoid valve 0-ring
!?'
failure.
4
~
~
i i
- q L e
,-.-,~-,.-an.
.---.n-~.--
~..--,,,,.,,,,-.,.,.,~...n
....,-.i..._,.,..-,,,,,_-
- 2.0 DISCUSSION 1
The data in Table 1 can be put into perspective by comparing the six events of failures on _ actual scram demand to the total number of scrams exper.ienced by domestic nuclear plants. There were 499 unplanned reactor
~
trips in 1983 and almost an equal number in 1984.
Of these approximately 1000 scrams over this two year period, 280 were experienced by General Electric Boiling Water Reactor unitt, 16 by Lacrosse BWR (an Allis-Chalmers unit), six by fort St. Vrain (an HTGR unit), and the remainder (approximately 700) by Pressurized Water Reactor units.
(Lacrosse and Fort St. Vrain are one-of-a kind reactors with unique ro'd drive systems). Of the six events of rod failures that occurred during an actual scram demand, four occerred at GE BWR units, and one each at Lacrosse and Fort St. Vrain.
PWR units had no failures on actual scram demand.
In the staff's review of these events, the real concern has been the po-tential generic common-cause failure implication associated with them.
Of the 13 events, three were identified by the staff to have such potential generic implications.
Two of these events occurred at General Electric designed Boiling Water Reactor (BWR) units (Peach Bottom 3 and Susquehanna 1), and the third occurred at a foreign reactor, a Westinghouse designed Pressurized Water Reactor (PWR) unit (Ko-Ri 5). The three common-cause failures were: 1) failure of the scram solenoid valve because of Loctite,
- 2) failure of the scram solenoid valve due to sticking of the disc holder assembly, and 3) looseness of rod assembly guide screws.
The details of these three events and the generic actions taken to address the potential common-cause failure concerns are discussed below.
The event that occurred at Peach Bottom 3 on November 17, 1983, is described in LER 83-018.
As described in the LER, during a post-scram investigation e of scram insertion times, two control rods were found to have exceeded the 1
allowable time limit of 7.00 seconds. The cause was identified as the failure of a scram solenoid valve in the hydraulic control unit (HCU) of
(
both rods.
An examination by the vendor, General Electric, detemined the cause of failure as foreign material associated with maintenance activities (Loctite 242). Loctite 242 was used to secure the acorn nut on'a1.1 sqlenoid housings and the excess Loctite 242 had..apparently migrated when the solenoid was returned to service. The migrated Loctite eventually caused sticking of the solenoid plunger and failure of the scram pilot solenoid valve.
Failure of all valves for the same cause could not be ruled out.
- Further, since Loctite 242 was utilized at other operating reactor facilities for similar applications, the staff detemined that this problem had generic safety implications and as a result issued IE Infomation Notice 84-53,
- Information Concerning the Use of Loctite 242 and Other Anaerobic Adhesive /
Sealants."
The vendor, General Electric, issued a Service Infomation Letter also addressing the problem.
The event at Susquehanna 1 occurred on October 6,1984.
The problem with the rods was discovered during scram time surveillance testing and
, involved the failure of four control rods to scram and several control rods with slow scram insertion times. The failures were detemined to be due to failures of the T-ASCO scram pilot solenoid valves to actuate and vent air from the scram valves.
Preliminary analysis perfomed by the vendor, General Electric, indicated that the failure mode was sticking of the polyu'rethane disc holder subassembly (DHS) to the exhaust port of 'the
~
solenoid valve probably due to the combination of temperature, time and oil and/or water contamination in the instrument air supply.
The licensee shut down both units to replace the polyurethane DHS with Viton-A which has better temperature resistant and hardness properties than the polyurethane (note - all scram valves on Unit 1 had. identical polyurethane disc holders -
about one-half of the scram valves in Unit 2 also had these holders). NRC Region I issued a Confimatory Action Letter on October 17, 1984, and held an Enforcement Conference on November 30, 1984 concerning T-ASCO Scram Pilot solenoid Yalves at Susquehanna 1 & 2.
T-ASCO solenoids are used in the newer BWR units only in the control rod system and are intended as an improvement on the dual ASCO Solenoid Yalves used in earlier BWR units.
Other BWR plants that use polyurethane DHS in T-ASCO solenoids are plants that have not loaded fuci and include Hope Creek, Nine Mile Point 2, Grand Gulf 2, Perry, Clinton and River Bend 1.
In Taiwan, Kousheng 1 & 2 have been in operation since 1981-82 with T-ASCO polyurethane seals.
The vendor, General Electric, issued a Service Infomation Letter (SIL) on October 16,1984, which (1) describes the problem, (2)
I recommends replacement of T-ASCO seals with repair kits containing Viton-A, and (3) suggests verification of this appitcation b,y all plants.
In addition to the concern regarding potential common-cause failure, 'the event at Susquehanna raised two other items of concern. The first item is regarding the inadecuacy of the post-scram review practiced at the station c.
which was identifiet by NRC Region I.during a special inspection that reviewed the event of October 6,1984.
The inspection found that during a scram of Susquehanna 1 on June 13,1984, one 2 X 2 control rod array exceeded s the specified scram insertion time limit. However, the slow scram time was overlooked by the licensee's staff, and the data from this scram was subsequently used to demonstrate compliance with certain technical.spepifi.
cation surveillance requirements. The significance of this was that two of the four control rods that failed to scrani:during the October 6,1984
. event were in the 2 X 2 rod array that exceeded the scram insertion time during the June 13,1984 scram.
The second item of concern 13 the lack of attention given to lessons learned from previous experience.
On March 14, 1980, the NRC issued IE Infomation Notice 80-11_ entitled, ' Generic Problems with ASCO Valves in Nuclear Applications Including Fire Protection Systems." This infomation notice described a potential deficiency of ASCO NP-1 solenoid valves regarding the effects of oil on ethylene propylene elastomer (EPR) materials which expands or swells when brought into contact with oils possibly causing valve failure.
The i
5-notice also stated that Viton elastomer replacement kits were available from ASCO for NP-1 solenoid valves.
Attached to the infonnation lotice was a letter from EG&G Idaho, providing the results of an LER revi ew of failure i
of solenoid _ valves.
The letter identifled the cause of manyIsolenoid valve failures to be apparent incompatibility of solenoid valve materials with foreign material, specifically oil, which can be present in the air supply t
sys tein.
The licensee did investigate the aroblem of ASCO solenoid valves in 1981 and identified several valves at tie Susquehanna station which had -
ethylene propylene seals and which did fail due to oil contamination.
However, the licensee review did not identify the T-ASCO solenoid valves which use polyruethane material as having this potential problem.
t The event at the foreign reactor occurred at Ko-Ri 5 in Korea during pre-critical testing.
It was dounnined that a rod assembly guide screw which guidos and aligns the breech components of the drive had fallen out and prevented rod movement.
A check of another Korean unit, Ko-Ri 6, identified a number of such breech guide screws to be ' finger-tight' only.
preliminary information indicates that in the United States, the followi_ng reactor units have the same type of control rod drive mechanism: -Catawba 1 and 2, McGuire 2, Watts Bar 1 and 2, and Seabrook 1 and 2.
The Itcensees of all these units have been infonned of the problem by the vendor and the NRC Regional Offices. The vendor has initiated corrective actions which are planned to be implemented at all these units. The NRC Regional Offices are actively following up the corrective actions at the above plants.
The Office of Inspection and Enforcement is planning to issue an Information Notice on the subject.
Another event that had common-cause' failure implication was the one that occurred at Fort St. Vrain on June 23, 1984.
However, this event. was considered not to have generic implications since it occurred at the only operating High Temperature Gas-Cooled Reactor (HTGR) unit in the U.S.
The event involved the failure of six out of 37 rod pairs to drop when a.
q reactor trip occurred.
Af ter verifying adequate shutdown margin, the shift i
supervisor, according to procedure, pulled fuses to the drive mechanisms but the rods failed to drop.
All six pairs were then inserted to the full-in
\\
position by running the drive motor, All rods were inserted within 20 minutes of the reactor trip.
The licensee had verified that cold shutdown margin was achieved and maintained before manual insertion of the six ' and affected rods. Additionally, the reserve shutdown system, independent redundant to the control rod system, was available. The licensee has committed to fully investigate the event, examine _all six affected control rod mechanisms and obtain_NRC approval prior to reactor startup. The reactor has been shut down since June 23, 1984, and the Office of Nuclear Reactor Regulation (NRR) and Region IV_ are activel'y involved in the resolution of the problem with
- the control rods at Fort St. Vrain..
An. additional-problem was identified at Fort St. Vrain during the investi-gation following the June 23 event. While the reactor was shut down for control rod drive inspection-and maintenance, two reserve shutdown hoppers (CR00A #26 and CR00A #21) were functionally tested in the hot service facility on November 5,1984.
During testing of CR00A #26 all of the reserve
. shutdown material (20 weight percent boron) was released from the hopper;
=
. however, hopper assembly CR00A #21 (40 weight percent boronf did not discharge the full 80 pounds of the material as designed, but only 40 ipounds. The material that failed to discharge has been collected for licensee analysis and for independent analysis by the NRC. The investigation by the. licensee to determine why some of the reserve shutdown material was retained.inside the hopper assembly is also being closely followed by HRR and Region 10 A review of the remaining nine events shows that the events were caused by apparently random independent failures which do not appear to have generic implications.
In seven of these events, the failure involved one stuck control rod which is an analyzed condition. Hence, these events are of low safety significance.
The seven events are as follows:
plant Date 1)
Quad Cities 2 October 25, 1984
- 2) Dresden 3 October 20, 1984
- 3) Troj an August 18, 1984 4)
Lacrosso July 16,1984
- 5) Surry 1 June 20,1984 6)
Browns Ferry 2 March 22, 1983 7)
Browns Ferry 3 March 12, 1983 The corrective actions taken by the licensees appear to be adequate in addressing the probicms that caused,these events..However, one of these events, the one that occurred at Quad Cities 2 on October 25, 1984, raised a concern regarding adequacy of post-trip review at the station. During this event, the plant operators were not aware until 30 minutes after reactor scram that one control rod remained at the fully withdrawn position. The event details and follow-up corrective actions were closely followed by n
NRC Region !!! and a Confinnatory Action Letter dated October 26, 1984 was issued.to the licensee to confirm the corrective actions being taken by the licensee in addressing the rod f611ure that occurred during the event.
An N
Enforcement Conference was held on November 5,1984 to discuss the event.
The event at Dresden 3 that occurred on October 20, 1984, and the Browns Ferry 2 event of March 22, 1983, both involved the failure of a manual valve in the scram discharge line of the rod'slydraulic control unit (HCU).
The valve involved was the same model in both cases, -- Hancock 950W; and the failure mode was also similar -- separation of the valve stem from the valve disc, blocking the scram discharge water path and causing failure to sc ram.
At Oresden 3, the licensee inspected 10% of the control rod HCVs to verify the integrity of similar valves on these units.
No other failures were found.
Dresden 3 has 177 control rods and corresponding HCUs.
Each HCU has seven such manual isolation valves, and these are only used when an
. IICU is isolated for maintenance. The liCUs at all operating Ceneral Electric BWRs typically use the same isolation valves, llence, there s a large population of these valves (approaching 30,000) at operating BWRs.
- Ilowever, AE00 knows of only these two failures that have contributed to failure of a control rod to scram.
Thus, based on operating experience, AE00 considers these two failures as low probability failures and no further action is considered necessary at present.
One of the two remaining events occurred at lionticello on December 5,1984 where, during post-outage surveillance testing of the control rod system to verify scram times, several rods were found to have slow response times.
The plant had been shutdown since February 3,1984 for an extended outage which included replacement of major portions of the recirculation systen pi pi ng.
Investigation following the event revealed the cause to be plugging of screens in the hydraulic flow path of the control rod drive mechanism.
The primary coolant system was cleaned up to remove the fibers that had clogged the screens and all control rod drives were either modified or had new 10 mil inner screens installed in them.
The plant was returned to operation on January 17, 1985.
The last of these nine events occurred at Quad Cities 2 on December 22, 1984 and involved the failure of three rods to fully insert following a manual scram.
The rods inserted to the 02 position (one step or six inches short of full insertion) and had to be driven in the final step.
This problem is generally due to tho. tailure of the. control rod drive stop piston seals due to dirt particles. The failure of the seal results in water at reactor pressure being introduced between two sets of stori piston seal s. Upon scram, this pressure between the seals does not create a problem untti the rod moves past position 02 on the way to notch position 00, the fully inserted position. At notch position 02, the buffer holes in the stop piston are designed to vent the water above the drive piston assembly, thus, slowing the control rod at the end of its sc' ram stroke.
Due to the failure of the seals, the buffering will not occur properly N
and the rod will stop between 02 and 00 notch positions, and eventually settle into position 02.
This problem occurs occasionally at BWR units due to failure of the drive inner filter which allows dirt particles into
~
the stop piston seals, tionnal corrective action for this problem consists of replacement of seals and filters.
Regularly scheduled maintenance of the control rod drive system also tends to reduce the incidence of this problen.
This problem is of low safety consequence since analyses have shown that adequate shutdown margin is present even with all rods in position 02.
Failures of control rods and other failures in the reactor protection system that occur during unscheduled reactor shutdowns are part of the ongoing NRC and licensee activities in implementing the actions discussed in Generic letter 83-28, " Required Actions Based on Generic Implications
p of Salem ATWS Events." The actions required would include gc neric considera-tion of post-trip review, post-maintenance testing verificati on, technical specification changes, reactor trip system reliability including trip breaker problems.
Events such as the potentially significant failure that occurred at Sequoyah 2 on January 12,1985, where a component (transistor) in one trip logic _ circuit failed, resulting in the failure of one trip breaker to operate, would fall into Generic Letter 83-28 activitiet.'
Additionally, AE00 has established an in-depth periodic analysis of unplan-ned reactor scrams as a major product of the Trends and Patterns-Program.
A pilot study covering the first three months of 1984 was issued for staff comment in late November 1984. The-n6xt report will cover all of 1984, including a comparison with foreign experience. This report will provide an overview of the U.S. experience and address the_ root _causes of unplanned reactor scrams.
j i
nW 4
i-
9 4
. 3.0 FINDINGS AND C0tlCLUS10tl5 j
During the approximately two year period covered in this study, a total of 13 events were found where there was a partial failure of the control rods to properly perfonn their reactor scram function.
In six of these events, the failures occurred during an actual scram demand, and in the rema,ining seven events the failures were discovered during surveillance testing activities.
In all cases, the plant was tafely shut down by the proper functioning of the rmaining operable control rods.
The real concerns associated with sucit failures are the potential for commor-cause failure, e.g., an inability to insert sufficient control rods to assure reactor shutdown due to a singic type or cause of failure; and generic implications, e.g., other reactors may be susceptible to the same type of failures.
Even though the design and manufacture of control rod systems are certainly mature, it is of concern and significance that 13 ovents over a two year period involved four potential common-cause failure mechanisms, three of which have generic implications on similar reactors.
In addition to those potential common-cause failures of control-rod scram systems, another potential common-cause failure mode was identified at Fort St. Vrain when one reserve shutdown hopper failed to discharge all of the boron balls as designed.
This design is unique to fort St. Vrain; and thus, while the event is of concern becaupo of the potential common-cause failure mechanism, it does not have generic implications.
- For the three events that involved potential common-cause failure mechanisms which had generic implications, adequate corrective actions and actions to alert other reactor units have been taken, fort St. Vrain has remained shut down since the event and will startup only af ter all corrective actions have been completed and approved by the NRC. The remaining nine events were caused by random failures which do not appear to represent common-cause failure or generic implications.
Concerns regarding post trip review, post-maintenance testing, and,1de,ntifi-cation of root cause of failure are evident in some of the 13 events reviewed.
tessons learned from past experiences have still not resulted in complete correction of the problems identified as illustrated by the fact that failure of valves, similar to that experienced by T-ASCO valves at Susquehanna Station in October 1984, was the subject of IE Infonnation Notice 80-11 issued in March 1980.
When the actions discussed in staff's Generic Letter 83-28 (Required Actions Based on Generic Implications of Salem ATVS Events) are established and fully implemented, these persisting concerns should be alleviated.
4
, APPENDIX - A l
Events Involving Control Rod Systern' Failures In The Performance Of Reactor Trip Function Plant Name and No.
(Type of Reactor)
Date of Event Brief Description of Event 1.
? ad Cities 2 Dec. 22,1984 Reactor manually tripped from
( BW.,
3% power as part of nonnal shutdown procedure. Three rods inserted to the 02 position (one step from full insertion) and eventually had to be driven in manually the final step.
2.
Mentire11o Dec. 5, 1984 The plant had been shut down/since (BWR)
February 3,1984 for an extended outage which included the replacement of major portions of the recirculation system piping.
On December 5,1984, during syrveillance testing of the Control Rod System to verify that the scram times were within the Technical Specification reqrirements (90% in-serted within 3.8 seconds), it has found that the scran time on several of the rods was excessive (nearly 10.,
seconds).
Investigation revealed that the cause was partial plugning of the s screens in the hydraulic flow path of the control rod drive mechanisms.
The source of the minute, par,ticles
, plugging the screens was the primary
- coolant system which, although cleaned and flushed afteF refilling following the recirculation system piping replace-r..ent, still had particles present.
The primary coolant system has been furtner cleaned up.
Fifty seven con-trol rod drive units have had-new 4
I
)
l a
i
- P1 ant'Name i
No..
(Type of Reactor).
Date of Event.
Brief Description df Event 10 mil screens installed and the other.tixty four units.have been modified with a different-screen system.
3.
KokRt 5
' flov.19,1964 While performing hot rod drops as 4
(PWR - Korean) part of pre-operational testing, a control-rod became stuck dui'ing downward stepping.
Investigation had determined that the control rod drive mechanism (CROM) heavy drive rod 4
assembly guide screw rotated out of position, and fell from the drive rod landing on top of the CROM latch assembly where it lodged and prevented drive motion.
A check of another Korean Unit, Ko-Ri 6, identified a number of guide screws to be "_ finger tight."
Preliminary infonnation indicates that Catawba 1 and 2, McGuire 2, Watts Bar 1 and 2, and SeabrMk.
I and 2 have the same type of CROMs.
McGuire 2 is the only operating a
reactor, and Catawba has been grant-i
+
ed its low power testing license.
+-
4.
Quad Cities 2 Oct. 25,1984 With the unit in llot-Standby and all (BWR)-
outboard Main Steam Isolation Valves closed, the reactor scrammed due to
,,,an increase in reactor pressure re-sulting from a procedure deficiency.
During the scram, one control rod was not inserted because its scram discharge riser valve-was misposi-tioned. The procedure is being
e u
13 -
=
Plant Name and jlo.
(TypeofReactorj,, Date of Event Brief Description-of Event
~
revised and necessary mehsurer have been taken to guard against a recurrence of this event.:
5..
Dresden 3-Oct 20,1984 Following a feedwater system 'tran--
(BWR) sient caused by a faulty master level controller, a low reactor water level signal scram occurred.
One control rod failed to insert.
It was determined that a manual valve downstream of the scram outlet-valve had failed with the valve disc disengaging from its stem.
6.
. Susquehanna 1 Oct. 6,1984 During the normal 120 day scram (BWR) time surveillance testing, four control rods failed to scram and several control rods showed hesi-tation. The four failures were detennined to be due t6 failure of the scram pilot solenoid valves to ar.tivate and vent air from the scram valve. General Electric, the vendor indicates that the failure -
was due to sticking of the polyure-\\
thane disc holder subassenbly (DHS) i to the exhaust port of the solenoid valve.- The licensee = decided to shut \\
down both units and change out the polyurethane DHS with Viton A material.
(PWR)
~
,, During preparation for start $p 7.
Troj an Aug. 18, 1984 '. following the completion of tha L
. annual refueling. outage, control rod L-3, Bank D,. stuck at 210.
E.
steps (almost fully withdrawal position) during cold drop tests.
4 4
4 7
1 b
t
-m,.
m m
+:,_
^
4 Plant:Nsme and No.
(Type of Reactor)
Date' of Event Brief Description of Event u.
The reactor was fully loaded and at cold shutdown at this time.
- 8.
. Lacrosse Jul. 16, 1984 During a reactor shutdown, control (BWR) rod #29 would not insert from its fully-withdrawn position electrically.
It did not insert hydraulically, either, in response to a manual scram signal.
The malfunction was found to be in the upper control rod drive (UCRD) mechanism. One of the three roller i
assemblies in the roller nut assembly was found to be loosely assembled, which had allowed the bottom' ball bearings to fall out of the' roller assembly. A ball had lodged against one of.the lead screw threads, causing the rod to jam. All three roller assemblies were missing a catch pid. The three roller assemblies were replaced and catch. pins installed.
The UCRD was reinstalled and successfully scram tested. Three other UCRDs were inspected. Their roller assemblies were in good condition.
D 9.
Fort St. Vrain Jun. 23,1984 On -June 22, 1984, the reactor-(HTGR) was being shut down in a controlled 4
manner due to a problem of-high moisture in the helium coolant.
On June 23,19d4, the re' acto'r tripped-*-
4-
- '>o. a high pressure signal resulting from a combination of increased helium' inventory and progranming down of the high pressure. trip N
P 7
4
1' l
s Plant Name and I
No.
(Type of Reactor)
Date of Event Brief Description of Event point as reactor power was: reduced When the reactor trip occurred, six of the 37 control rod pairs failet to drop.-
The shift supervisor, after verifying adequate shutdown margin and in accor-dance with procedure, pulled fuses to the drive mechanisms, but the rod pairs failed to drop.
lie then replaced the fuses and successfully lowered all six failed rod pairs to their full-in position by running the drive motors.
All rods were inserted within 20 minutes of the reactor trip.
10.
Surry 1 Jun. 20,1984 With the reactor at 29% power, a quadrant (PWR) power tilt of greater than 2% existed for greater than 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> because control rod B-6 was stuck at step position 56. During a subsequent refueling outage, the c.ause of the stuck rod was found to be due'to a foreign object (a hold down sprin,g) obstrutting rod motion.
11.
Peach Bottom 3 Nov. 17, 1983 A post-scram investigation of scram,
(BWR) insertion times identified that
'i control rods 34-35.and 34-27 exceeded the allowable limit of 7.0 seconds.
Reactor shutdown was in progress and no additional control rod drive problems were identified. The rate of reactor shutdown was not noticeably affected by-the excessive scram time. These control ro'ds did scram because of the proper operation of the back up scram solenoid valves. Cause was failure of a scram solenoid valve ( ASCO HVA-405) in both hydraulic control units.
Both solenoids in both HCOs were replaced. GE examina-tion has determined the cause of failure as foreign material associated with maintenance activities (Loctite 242).
1
-.=
a c.-
9 6-Plant Name and l.
No.
(Type of Reactor),
Date of Event ~
Brief Description-of Event 12.-
Browns Ferry 2 Mar. 22, 1983 With Unit 2 at 38.5% power. for refueling (BWR)-
tests, CR010-39 failed to' scram whife perfonning tests. The scram signal was initiated from the auxiliary instrument room. All other CRDs were operable.
The redundant system (Standby Liquid Control) was available and operable.
Technical Specification 3.3. A.2.F-allows plant operation with an inoperable control rod. Yalve 85-617 (Hancock 950W) was found to have the valve disc separated from the valve seat,. blocking the scram discharge water path.and causing failure to scram.
The CRD was inserted to '00' with drive pressure and the valve was repaired and success-fully tested. This appears to be a random event and, as such, no action to prevent recurrence is required.
13.
Browns Ferry 3 Mar.12, 198)
During nonnal operation while (BWR) perfonning scram timing surveillance CRD-38-31 failed to scram upon initiation of a scrani signal. The CRD was inserted with normal drive pressure to '00' position and tagged out for maintenance.
Technical Specification 3.3. A.2.F s
pennits operation with inoperable 1
control rods. The scram solenoid
\\
valve was inspected and the 0-ring on the inlet air side was found out of position which apparently. caused the m
failure. The valve-was rebuilt and the,,.
" CRD successuffly tested.
This is considered a random failure and, as such,_ no action to orevent recurrence is required, t
W 9
m y
m'
-y s