ML19340E028
| ML19340E028 | |
| Person / Time | |
|---|---|
| Issue date: | 11/07/1980 |
| From: | Advisory Committee on Reactor Safeguards |
| To: | Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-1775, NUDOCS 8101060223 | |
| Download: ML19340E028 (9) | |
Text
.-
o DATE ISSUED:
NOV 7 1990
,;7L% ~ / 7'?6 n
MINUTES OF THE ACRS FLUID DYNAMICS SUBCOMMITTEE MEETING, LOS ANGELES, CALIFORNI A l
AUGUST 19-20, 1980 The ACRS Fluid Dynamics Subcommittee met to review recent difficulties with the hydraulic scram system of BWRs.
Discussions centered on events at Browns Ferry-3 which experienced a partial failure to sc. am, and Hatch-1 and Bruns-i wick which experienced damage to water level instrumentation on the scram instrument volume (SIV).
G. Young ( ACRS Staff) and D. Zukor (ACRS Fellow) presented a summary of their report to the, ACRS on the recent events.
The BWR control rod drive mechanism is operated with a hydraulically controlled double acting piston.
The scram system consists of an accumulator of pressurized water, a scram inlet line, the piston, and a scram outlet line for each control rod.
The inlet and outlet lines each have an air operated valve which when opened causes high pressure water so flow into the control rod drive, pushing the rod into the core, and expelling water from the back side of the piston through the outlet line into a scram discharge volume.
Each reactor is supplied with a scram discharge volume (generally two separate pipe headers, one for each half of the reactor) to contain the expelled water from the CRDs.
The scram discharge volumes (SDV) drain through (generally 2" diameter) pipes to the SIV and then to clean rad waste tanks within the secondary containment building.
The SDVs are provided w1.h vents (1" diameter lines) which generally also connect to the clean rad weste tank via the building drains.
The vent lines and drain lines each contain valves which normally open and close upon scram in order to contain the CRD discharge water and any seal leakage following scram.
B101060
~
Fluid Dyna:aics Aug 19-20, 1980 i
The drain and vent valves are reopened when the scram is reset and the control rod inlet and outlet valves close.
These valves are controlled by the plant air supply system and electrically operated solenoids which release air pressure to close the valves.
The air supply is not a dedicated system since its failure will lead to a successful scram.
(AE00 and C. Michelson have raised questions on the event sequence of a partial loss of air pressure that might open some scram valves prior to other leading to a water accumulation in the SDV.
IE has subsequently issued a bulletin to oparating -reactors requir-irg additional warning on loss of air pressure and a manual scram prior to limited scram valve opening due to air loss). The SDV is sized to accomraodate 3.3 gals / rod.
This accounts for.75 gal / rod from the water displaced by tne moving piston plus CRD leakage at a rate of 10gp:n for 10 seconds along with some additional allowances.
The SDV serves two functions during and after scraut.
First, it provides a collection point for the water discharged from the CRDs and second, it serves as a primary system boundary to contain CRD seal leakage following ".he scr4m.
In order to achieve the second function the vent and drain valves on the SDV must close to limit the loss of water through the CRD seals; however, this also exposes the scram systen to a possible failure mode due to water accumulation in the SDV.
The two Browns Ferry-3 SDVs are connected by a 2" drain line to a common scram f
discharge instrument volume (SIV).
One pipe run is about 20' with a 2' drop in elevation, the otner run is about 150' with a similar elevation drop. The Browns Ferry-3 partial scram failure appears to have been caused by water accumulation in the SDV with the longer pipe run to the SIV.
The SIV is connected to level instruments that are intended to warn of any water accumu-
+
__=
4 Fluid Dynamics Aug 19-20, 1980 lation in the scram discharge system and high levels are supposed to cause a reactor trip prior to water in the SDV.
Tests performed at Browns Ferry-3 following the scram f ailure showed that water drainage rates through the 150' 2" pipe were less than the drain rate through the SIV drain line, thus water could accumulate in the one SDV without being noted in the instrumentation in the SIV. The source of the water was not identified clearly, the leakage from CRDs was less than the drain rate in the 150' line. Browns Ferry-3 personnel indicated they believe there was a blockage in the drain line that was cleared during one of the scram attempts.
The design of' the scram system is dependent upon a number of different organi-zations.
GE sets the functional requirements of 3.3 gal / drive, the actual design of the SDV and piping system is left up to the utility, AE, and subcon-tractors, depending upon unique arrangements, under individual-contracts.
When various systems were reviewed by the NRC Staff they found that a number of different arrangements exist and that none of them appear to be exactly the same.
Several modifications in basic design have been made wit' the evalua-tion of the BWR design.
The original specifications allowing for 1.1 gal /
control red was changed after an early incident of leaking CRD seals at an operating plant.
Two older plants, Oyster Creek and Nine Mile Point, still use the 1.1 gal / rad specification.
A more recent change has been the use of a so-called " hockey stick" SDV and SIV arrangement where the two. volumes are essentially continuous so that long runs of small diameter piping are elimi-nated.
J. Ebersole raised a question as to whether the changes were done in response to a perceived deficiency in the older systems and if so, why the older systems were not also changed.
The NRC Staff indicated that they did not know the reason behind the system changes.
Fluid Dynamics Aug 19-20, 1980 s
The event sequence at BF-3 included a manual scram from 30% power.
The operator noted that all of the scram valves functioned properly but that the rod position indicators on the east side of the core indicated that a large number (76) of rods had not f ully inserted.
The control rods which did not scram ranged fro.n almost withdrawn (Step 46 of 48) to almost in (step 02).
i f
The reactor power level (neutron power) had dropped off scale low (less than 2%), thJs the reactor may or may not have been subcritical.
Six minutes following *,he initial scram the operator again initiated a scram. A number of additionai rods inserted at this time leaving 60 rods partially withdrawn.
Two minutes later a third scram was initiated which left 46 rods partially l
out.
Six minutes later the operator reset the bypass on the scran high level instrumentation which initiated an automatic scram. This resulted in complete insertion of the rods.
In subsequent tests and analyses GE and TVA have concluded that the east side SDV was 80%-90% full of water at the time of the initial scram.
They also correlated the amount of movement of the indi-vidual control rods to their previously determined scram insertion times; with the slowest rods moving the least distance.
In reviewing the design of the SDV and SIV the identified a number of poten-tial problems and possible causes for water accumulation in addition to the long run of pipe from the east side SDV header to the SIV.
The vents on the system connect into the same clean rad waste system drain lines a.s does the SIV drain.
The building drains discharge in the clean rad waste tank under the water surface and, thus, have the potential of not providing a positive vent path to the SDV exits.
In addition, the system is not safety grade beyond the single vent valve or the single drain valve on the SIV; however,
Fluid Dynamics Aug 19-20, 1980 failures beyond the boundaries of the safety grade portion of the system can 4
clearly interfere with the performance of the system safety function.
The Brunswick and Hatch events involved damage to the SIV level instrumenta-tion.
In each case float balls which activate switches if the water level is too high were found crushed and inoperable.
In the Brunswick case the plant was operating with the drain valve closed because of difficulties with the solenoid operator.
The intent was to allow periodic draining of water when the lowest level alarm was received; however, the reactor scrammed automati-cally on high water level without alarming or having a rod block at the intermediate level.
Upon investigation it was found that the two lower level floats were crushed.
Brunswick personnel attribute the damage to an earlier water ha imer event in the scram discharge system.
It is believed that the wa.er hammer, which tore pipe supports out of the wall, was caused by the
- r. rain valve closing too slowly.
This would cause closure against a water stream rather than against the initial surge of air from the SDV and SIV.
Similar damage was found with the Hatch floats.
In the case of Brunswick, failure of the scram floats in addition to the level floats could have led to a buildup of water in the SDV that would prevent scram.
l l
The AE00 Office investigated the Browns Ferry-3 event and made a number of l
findings and recommendations. These incuded:
^
1.
The partial scram failure was caused by a water accumulation in the east side SDV header.
2.
The SIV high level trip did not and does not provide protection against filling the SDV.
3.
A single failure such as a blocked vent or drain line can disable the instrument volume scram signal.
I
\\
Fluid Dyna:nics Aug 19-20, 1980 f
e 4.
A single failure of a vent or drain path could cause a partial loss of scram capability.
5.
The current layout results in an automatic safety function being dependent upon nonsafety related systems.
6.
The RPS logic does not allow a scram reset with certain automatic scram signals present.
7.
The failure of a single vent or drain valve during a scram can result in an unisolable release fran the primary system.
8.
The emergency procedures at Browns Ferry did not cover a partial or total scram failure event.
9.
The operability of the instrument volume high level scram should be independent of the venting and draining requirenents.
4
- 10. The vent system must be improved to increase the reliability of the drain system.
11.
Tne instrumentation should be both redundant and diverse.
- 12. The vent and drain valves should all have redundant isolation valves to eliminate the single failure causing an uncontained primary system leak.
- 13. The procedures need to be improved to provide operator guidance on use of the standby liquid control system.
Mr. Ebersole indicated tha!. consideration should be given to the possible need for a second scrcm, thus one must assure that vent and drain valves can open (with a single failure?) following the initial scram.
The NRC Staff presented a summary of their involvement in and response to the recent events at Brunswick, Hatch, and Brcuns Ferry.
Subsequent to the Brunswick float f ailure IE issued Bul.letin 80-14 requiring examination of the SIV i nstrument ation.
Following the Browns Ferry event Bulletin 80-17 was In response to these bulletins a number of other problens are surfacing.
issued.
These include cracked floats at Hatch-2, a sticking vacuum breaker on the Dresden-3 SDV vent, the SDV at Browns Ferry-1 was also draining slowly, have the 10 second delay relay connected electrically, Millstone-1 did not
Fluid Dynamics Aug 19-20, 1980 at Duane Arnold the SDV drain valve was installed backward; so that pressure tended to unseat the valve, at Peach Bottom Units 2 & 3 the backup scram valve solenoids were connected to 125 V DC rather than 250 V DC and were thus inoperable, and Fitzpatrick found a loop seat in the SDV drain piping. All of these items are being corrected.
In response to questions the Staff indicated that testing requirements for many of these backup valves and equipment were not required in the Technical Specifications.
Most of the equipment appears to be field-installed and the Staff is concerned with whether the utilities actually understand the as-built conditions.
Examination of the present designs has revealed that they all are somewhat different.
NRC activities have included personnel from both NPR and IE Offices as well as an investigation of the events by the AE00 Office (C. Michelson).
IE has been responsible for reviewing the events and reviewing corrective actions.
They have also issued IE Bulletins 80-14 and 80-17 and are checking to see that all operating utilities have positive venting and draining of the SDV as well as working instrumentation.
NRR will be responsible for coordinating any needed long term system modifica-tions, Tech Spec changes, and new designs.
One of the near tem actions has been the installation of instruments on the SDVs (ultrasonic meters) to assure that there is no water accumulation on a continuous basis (although the readouts are local).
Mr. Dick Gridley, GE, reviewed their activities following the Browns Ferry-3 event and the functional design of the CRD system.
GE has been working closely with the TVA Staff and other utilities to determine the causes of the scram failure and to develop corrective measures.
GE has performed a number
s Aug 19-20, 1980 Fluid Dynamics t
a number of tests on a CR0 mockup at San Jose and has tentatively concluded that water accumulation in the 50V was the cause of the event.
They believe i
that the 50V was 80%-90% full.
GE issued several Service Information Letters to their customers during the week following Browns Ferry-3.
These recommen-and drain systems and a daily check of the 50V to l
ded review of the vent assure no water accumulation. GE also provided the NRC Staff with a number of j
analyses for ATWS with and without recirculation pump trip and for partial scram f ailures similar to Browns Ferry-3.
In response to questions GE indicated that the discharge of each control rod j
i If the scram valves begin to leak is monitored with temperature sensors.
reactor water will tend to leak through the CR0 seals and at about.13 gpm the drive temperature will get to 273 F.
Alarms will indicate if the tempera-0 I
ture exceeds 250 F.
Thus if there should be a gross seal f ailure and a 0
stuck open scram valve on one CR0 the operator should be aware of it and take l
measures to scram the reactor without excessive build up of water in the l
i SDV.
In response to ACRS and NRC questions GE performed a number of analyses if the consequences of a BWR partial and complete scram failure.
If the Browns i
Ferry-3 event had. occurred at full power the reactor would have dropped to I
With RPT and MSIV closure the reactor vessel would be alright; al-l about 20%.
though the containment would require cooling and the reactor would.have to be Without l
shut down promptly to prevent excessive pressure in the containment.
l RPT and with MSIV at power levels above about 30% service level C stresses in i
l the vessel would be exceed (for a total scram failure).
Fluid dynar.ics Aug 1-20, 1980 o
TVA personnel reviewed the Browns Ferry event and discussed the actions they took to detenaine the cause of the event.
They believe there was probably a blockage in the 2" dia.Teter pipe from the east side SDV to the SIV but they do not have conclusive evidence for the blockage. TVA is still studying a number of possible modifications to the scram syste.n to improve venting and drainage and to assure there is no water accumulation in the SDV.
In the meantime they have installed ultrasonic level detectors on the SUV with continuous local monitoring of the level and a periodic check (30 minutes) by the operating staff.
Control room monitoring will be connected in the future.
The meeting was adjourned at 12:00 noon on August 20, 1950.
Additional details are available in the meeting transcript on file in the NRC Public Document Room at 1717 H Street, N.W., Washington, D.C. or from Alderson Reporting. 400 Virginia Avenue, S.W., Washington, D.C.
A complete set of all slides used is on file in the ACRS Office with the record copy of the minutes.
9
, ~ ~