ML18025B287
| ML18025B287 | |
| Person / Time | |
|---|---|
| Site: | Browns Ferry  |
| Issue date: | 12/31/1980 |
| From: | INSTITUTE OF NUCLEAR POWER OPERATIONS, NUCLEAR SAFETY ANALYSIS CENTER |
| To: | |
| Shared Package | |
| ML18025B288 | List: |
| References | |
| INPO-3, NSAC-20, NUDOCS 8101190500 | |
| Download: ML18025B287 (74) | |
Text
~
0
~
~ gy I
~-
I2VPQ 3 December 1980
~3lysis @II'ncomplete Coner e1 Rodl. csex am'il BIcwxxs I'erxy 3 Nuclear Safety Analysis Center Institute of Nuclear Power Operations
Analysis of Incomplete Control Rod Insertion at Browns Ferry 3 NSAC-20/INPO-3 December 1980 Prepared by Institute of Nuclear Power Operations Atlanta, Georgia and Nuclear Safety Analysis Center Palo Alto, California
OROERING INFORMATION Copies of this report may be ordered from Research Reports Center (RRC), 8ox 50490, Palo Alto, CA 94303, (415) 965.4081.
Keywords: arcwns Feny 3, Incomplete controt Rod Insenion. water Accumutation In sov. common Mode Fasure of Scram Systems NOTICE Mi'S repcrt we a preOared by the Nuoutar Safety Anaryak Gamer (NSACI OOerated by tne EieCtnC pCWer Researcn Institute. Inc. (EpRII and by tne tnsntute ot Nuciear power operatcns (INpo). Neimer NsAc. EpRI, INpo. members ct EpRI or INpo. other persons contnbunng to or asskbng in tne preoaratcn ot tne report. nor any person acting on the behalf ot any ot these parties (a> maxes any warranty or representation.
express or im plied. with respect to the accuracy. compieteness or usetuiness of the ir',formanon contained in this recon. or tnat tne use ot any information. aooaratus. method or process disciosed in tnk recon may not intnnge prNatety owned nghts: or (bi assumes any liaoiimes witn resoect to the use of. or for damages resulting trcm the use cf.
any information. apparatus. method or process dkctosed in tnk recen.
ABSTRACT The failure of 77 control rods fn the Browns Ferry 3 reactor to scram completely on June 28, 1980 has been analyzed.
It is concluded that the cause of the incident was that one half of the scram discharge volume, which should have been empty in order to receive the water from the control rod drives, actually was almost full of water.
Any one, or a combination, of several mechanisms can contribute to accumulation of water in the SOY under certain conditions.
It appears possible that there was an obstruction or a trap in the scram discharge volume exit line which impeded its drainage.
Condensation of steam from other parts of the clean rad waste drain system in the scram discharge volume may also have contributed to water accumula-tion.
i Oespite extensive investigation and analysis by TVA and others a unique cause has not been estab'lished.
Possible ways in which the above cited mechanisms could have been operative are pointed out, and suggested ways for guarding against them are given.
111
Contents Analysis of Incomplete Control Rod Insertion at Browns Ferry 3
l.
Introduction 2.
Executive Sugary 3.
Summary Oescrfption of Incident 4.
Findings and Conclusions 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 Principal Findings and Conclusions Cause of the Incomplete Scram Evidence for the "Obstruction Hypothesis" Evidence for the"Vent Hypothesis "
The "Stuck Valve Hypothesis" Effect of Scram on Fuel Operator
Response
Role of the Vent and Orain Valves Ancillary guestions Summary of Conclusions Appendices SOE Sequences of Events OES System Oescription and Operation CON Reactor
Response
and Consequences OPS Operator Actions and Procedures References Glossary
Analysis of Incomplete Control Rod Insertion at Browns Ferr 3
1.
Introduction This report presents the results of a study by NSAC and INPO of the June 28, 1980 incident at the Browns Ferry Unit 3 reactor in which 77 of the full complement of 185 control rods inserted only partially when a manual scram was attempted.
The purpose of this report is to present a description of the incident, and some possible causes.
In carrying out this analysis, NSAC and INPO performed a preliminary review at the site on July 3, 4, and 5.
Team briefings and meetings were held with General Electric and Tennessee Valley Authority, and individual consultations were held with GE, TVA, and several utilities.
NSAC and INPO also participated in several industry and NRC meetings on the incident.
NSAC/INPO contributors to this evaluation were:
Mr. R.
S. Baker Mr. 0. L. Gillispie Mr.
W. H. Layman Or.
M. C. Leverett Mr. J.
W. Power Mr. S. L. Rosen Utility representatives who also contributed were:
Mr. B. L. Clark, Philadelphia Electric Mr. B. B. Palagi, Commonwealth Edison Mr. John Stetz, Northeast Utilities
Event Ourfng a routine manual scram of the Browns Ferry 3 reactor, 77 control rods failed Wo insert completely.
~8ack round On June 28,
- 1980, a manual scram of the Browns Ferry 3 reactor* was attempted in conjunction with a planned shutdown for repair of a feedwater line in the turbine building.
Aside from the need for this repair, plant conditions were normal.
The shutdown procedure involved first lowering the reactor power level to 36'y reducing the recirculation flow and inserting a number of control rods to decrease the neutron chain reaction; and secondly pushing the manual scram buttons to insert all control rods completely to terminate the neutron chain reaction.
Complete control rod insertion is normally accomplished fn less than 3 seconds after both scram buttons are pushed.
In this incident, normal control rod insertion did not occur when the scram buttons were pushed.
Descri tion of Event Of 185 control rods, 10 were fully inserted prior to the manual scram.
77 rods failed to insert fully upon manual scram, wfth insertion ranging from posftion 02 (95% inserted) to position 46 (5%, inserted).
Observing this, the operator reset the scram; thfs procedure allows recharging of nitrogen-pressurized accumulators and draining of the scram discharge instrument volume.
Manual scram was repeated.
Insertfon progr essed
- somewhat, but 59 control rods remained only partially inserted.
After a third reset and manual
- scram, 47 remained partially inserted.
Recharging and draining of the scram discharge instrument volume was repeated and the scram instrumentation automatically initiated a fourth scram.
All rods were now fully inserted, placing the reactor fn normal shutdown condition.
This was accomplished within about 14 mfnutes of the first scram.
~unst ia a 1067 ias(e)nat.
General Electric toiiina water reactor, in commercial operation since March 1, 1977.
The following observations are significant:
The first scram decreased the chain reaction so that the smeared average fission power level was about 2% of full power.
The second scram terminated the chain reaction so that the only remaining heat generation was normal decay heat generation.
There were no indications that delay of full insertion until the fourth scram caused any damage to fuel, to the reactor, or to any other part of the plant.
Natural circulation core flow provided more than adequate overall core cooling.
General Electric calculations indicate that no fuel operating limits were exceeded.
Scope of Anal sis This is the only known instance of a large number of control rods failing to insert completely on demand.
Because of the crucial importance of being able to shut the reactor down quickly and reliably when needed, the reactor owner, the reactor designer, the
- NRC, and a joint NSAC-INPO team attempted to determine the causes of this failure to achieve full insertion on the first scram.
Efforts were also directed at identifying measures to prevent a repetition of this or similar failures in this and any other boiling water reactor.
The failure was of further interest to NSAC and INPO as a possible precursor to more serious events.
Findines After the first scram attempt, the operators observed that all but one of the control rods which failed to insert were on the east half of the reactor.>>
This behavior of the system led plant personnel and others to believe that the east scram discharge volume (SDV), a void into which reactor water is displaced when Tile scram hydraulic system is shown diagramnatically in Fig.
1.
The arrangement of the scram discharge volum is shown in Figs.
2 and 3.
high pressure water drives the east side control rods into the reactor, was for some reason already almost filled with water prior to the scram.
The SDV could not accoenodate the additional water discharged into it during a normal scram and therefore insertion was impeded.
This conclusion has since been re-examined and stands generally accepted.
Other possible causes of the incomplete scram have been analyzed and largely ruled out.
The origin of the water in the SDV and the cause of its accumulation have not been firmly established, but several possible explanations have emerged.
None individually can be confirmed, and a combination of them is also a possibility:
There may have been an obstruction in the 170'ong 2" pipe which connects the east scram discharge volume to the scram discharge instrument volume (SDIV). 'his line has only a very slight slope for
- drainage, and a slight obstruction would have been sufficient to back up water into the scram discharge volume prior to the scram.
It has been theorized that such an obstruction could have been disturbed and relocated by the violent hydraulic action in the SDV after the multiple scrams which took place.
It is not possible to state with assurance, though, that such an obstruction actually did exist.
It has been suggested that the obstruction could have been formed by the accumulation of solid crud.
Solid crud has been found in level switches on the scram discharge instrument volume.
- However, no positive evidence of an obstruction in the 170'ong 2" line has been found.
A combination of trapping action in the east SDV vent line and in the east SDV-SDIV 2" connector pipe, together with condensation of steam in the east SDV, may have produced a partial vacuum.
The occurrence of such a partial vacuum has been observed in the SDIV of Browns Ferry I, which is very similar in piping configuration to Browns Ferry 3, and on at least two other BWRs.
There is no proof, however, that this was the cause of the incident.
Either of the foregoing may have been aggravated by momentary pressure or vacuum surges in the clean radwaste (CRW) drain system caused by large influxes of water from sources other than the SDIV.
Both the SDV vent lines and the SDIV drain line discharge into the CRW drain system.
This hydraulically complex system receives intermittent drainage from approximately 50 other sources.
Socle of that drainage is frequently hot water, which could increase pressure or suction surges.
Conclusions Since it has been impossible to determine whether any of these possibilities or combinations of them were responsible for the incident, corrective action recom-mendations should be comprehensive enough to guard against all of them.
It is concluded, therefore, that appropriate steps should be taken by all BWR plants to guard against:
An obstruction in the SOV-SOIV connection pipes.
A configuration of the SDV-SDIV connector or vent pipes capable of producing a trap or loop seal.
Such a trap or loop seal could possibly be the result of thermal expansion during hot conditions although it may not be present in a cold environment.
Interference by the CRM drain system with the operation of the SDV-SDIV system.
Failure of the SDY vent line valves to open.
Too slow drainage of the SDIY due to inadequate vent or drain capacity.
The review team also examined the procedures available to the operator for an event of this type, and the response of the operators to those procedures, par-ticularly with respect to the criteria for deciding whether to inject sodium pentaborate from the standby liquid control system.
It is concluded that these criteria should provide more explicit guidance in determining the need for sodium pentaborate injection.
The conclusions outlined above are intended to apply to all BWRs with hydraulic control rod drives, operating or fn construction.
It is recognized that the specific corrective measures which should be taken will vary from plant to plant because of design variations and differing procedures.
Water piping
~ Airline tel Open valve M Closed valve Electrical Outlet open Inlet closed vs,e solenoid l
N Check valve Control rod Drive housing Reactor vessel Atmosphere I Atr supply Atmosphere Scram pilot air header SDV vent From condensate storage tank Drive water pump Control rod I
drive and I
cooling water I I
I I
I Charging header Scram inlot valve Scram outlet valve H20 N2 Scram accumulator Typical for /
185 conlrol~
fod drives Level instrument SDlV SDV drain I
)
I Figure 1.
BWR Scram Hydraulic System (scrammed valve lineup)
1" vent 6" headers To CRW drain 21 ~
East Bank Scram discharge header 1" vent Instrument volume To CAW drain West Bank to CRW drain 2M 12" Qt ~
L 20'" headers Level sensors Figure 2.
Browns Ferry SDV Equipment Layout (isometric view)
SOV vent SDV vent Open SDV headers 2II 2
II SOV Scram discharge lines Solv
~4 level switches (scram) at 50 gal.
1 level switch (rod block) at 25 gal.
ltCUs 1 level switch (alarm) at 3 gal.
HCOs HCus SDV drain Other drains Floor elevalion 4" drain header Olher drains and vents Reactor building equipment drain sump Figure 3.
Sketch of Browns Ferry SDV Equipment Layout (elevalion)
3.
Sugar Description of Incident On June 28, 1980, it was necessary to shut, down the Browns Ferry 3 (BF-3) reactor.
This shutdown, which had been planned for a couple of days, was necessary in order to repai~
a feedwater line in the vicinity of one of the feedwater pumps.
The shutdown was taking place according to normal procedure.
Power was first reduced to 390 Mwe (365 of rating) by decreasing recirculation flow and inserting 10 control rods.
The operator then initiated a manual scram to complete the shut-down.
All control rod drives received a scram signal as reported by the operators and the shift technical advisor.
All 92 of the withdrawn west bank control rods inserted fully, except rod 30-23 which settled at position 02.* (It is not unusual for a rod to settle at position 02, t.ens 955 inserted, and it is of little reactivity significance since practically all the reactivity effect of the control rod fs associated with the first 95K, of its insertion.)
However, of the 88 east bank rods withdrawn, 76 failed to insert fully, coming to rest at various positions in the range 46 (inserted 6") to 02 (inserted l38" with only 6" remaining to complete insertion).
The average core fission power was reduced to less than 2" as reported by GE (Ref.
1) and as inferred from recorded LPRM (local power range monitor) readings after the first scram.
Upon observing that many rods remained only partially inserted, the operators re-set the scram after 271sseconds, resulting in opening the vent and drain valves in the scram discharge system so that water could drain from it, and at 364 seconds repeated the manual scram.
All of the partially inserted rods moved inward, but some remained still only partially inserted.
At 423 seconds the operators again re-set the scram, and at 476 seconds again repeated the manual scram.
- Again, those control rods not yet fully inserted moved inward, but some remained only partially inserted.
At 682 seconds the operators again reset the scram.
Upon removal of the scram discharge volume by-pass the reactor scranmed automaticaIly~
on high (50 gal.)
volume in the scram discharge instrument volume.
This resulted in full insertion of all the control rods which had not already fully inserted.
~antra rod travel is 144".
Full in is position 00.
Full out ls position 40.
manual and automatic scrams are functionally identical, so the fact that it was an automatic scram which finally inserted all rods is not significant.
Following the fourth scram, a fifth confirmatory scram was executed
- and, upon ascertaining that the reactor was shut down and being cooled in a normal
- manner, numerous tests and observations were undertaken to ascertain the cause of the failure to achieve ineediate and complete insertion of all control rods.
The tests and observations included the following:
(Ref.2)
Off gas and reactor coolant analyses Control Rod Drive Hydraulic Control Unit valve lineup Control Rod Drive Accumulator re-charge Exercising SDIV and SOV drain and vent valves Disassembly and inspection of vent and drain valves Vent system vacuum test SDIV level switch calibration Flushing SDIV level switch taps and examination for solid material Review of scram history at Browns Ferry (320 scrams successful except for a few instances of latching at position 02)
Control rod drive maintenance history review Review of other maintenance and modification work on BF-3 Verification of scram actuator de-energization time Measurement of control rod drive stall flows and friction Review of CRO temperature charts (high temperature could signal excessive water leakage through the drives)
The 170'ong 2" east connector pipe was cut in six places and rodded with a metallic tape to detect any blockages System flushing with water (after cutting the pipe)
SDV inspection with a horoscope (after cutting the pipe)
Measurement of elevatfons at various points on the east SDY connector pipe (after cutting and re-assembling the pipe)
Measurement of SDV slope Routing and configuration of afr lines traced out Unit 3 reactor building sump (reactor building equipment drain tank, RBEDT) inspected with an underwater TV camera and also raked for solid objects Drainage times for the east and west SDVs, with and without the vent valves being open Monitoring of the equipment (CRW) drain header for possible vacuum during the drainage time tests Individual scrams of all rods which failed to insert normally Manual and automatic scrams from power 10
These tests and observations failed to lead to a clear understanding of the cause of the incident.
The drain time tests showed that both the east and west SCVs would drain when the SDIV drain valve was
- opened, whether the SDV vent valves were open or not.
The time required to drain the entire system with the vent valves open was about 30 minutes.
With the vent valves closed the drainage time was estimated to be several
- hours, but drainage did occur.
(The postulated blockage would not have been present during these tests.)
With few exceptions these tests and observations indicated that the system was at the time of the tests capable of operating normally and as intended.
The principal exceptions were that the low level (3 gal.)
and rod block (25 gal.)
SDIV switches did not operate during the first calibration fill, and the SDIV west vent valve operator was found to have a
broken air tube and an air leak.
- However, recorded data and operator reports indicate that all of the SDIV level instrumentation worked during at least part of the incident.
The significance of the SDIV west vent valve air leaks has not been established but is not believed to be appreciable.
(On July 8 the reactor was shutdown briefly because of a bent stem on SDIY drain valve, but TVA states that this damage occurred after the incident of June 28.)
These tests established that the rate of drainage from the SDIV drain valve can exceed that from the east SDV bank.
- Hence, the fact that the SDIV instruments signal that little or no water is present there during drainage does not neces-sarilyy mean that the east SDY has drained completely.
- However, absent a trap or obstruction in the SDV connector pipe, it would not be expected that drainage of the east SDV would significantly lag drainage of the SDIY.
The reactor has operated normally since being restarted July g, with no indi-cations of damage to any equipment or the fuel.
(See Appendix CON.)
4.
Findin s and Conclusions 4.1 Princi al Findin s and Conclusions The review team' principal findings and conclusions are that:
a) the failure to achieve complete insertion of 76 of the control rods in-the east half of the Browns Ferry Unit 3 reactor when it was manually scraneed on June 28 was due to water occupying a large part of the east bank of the scram discharge
- volume, and b) the source of this water and why it was present at this. particular time has not been positively determined
- and, due to lack of evi-
- dence, may never be determinable with certainty.
However, at least thr ee plausible explanations have been developed from the available
- evidence, namely:
1.
There may have been a solid obstruction, at some point between the lower end of the scram discharge volume (SDY) and the scram discharge instrument volume (SOIY), of sufficient size and strength to cause water to back up in the east SOV.
It has been calculated by GE (Ref.
- 1) that, if the SOV were 85-90%
full of water, the presence of this water plus the postulated obstruction in the 2" connector pipe would be sufficient to account for the failure of 76 control rod drives (CRDs) to complete their stroke.
In order to hold up the water in the SDV the postulated obstruction would have to be strong enough only to support' column of water of about 1/2'o 2 1/2'n
- height, depending on the location of the obstruction in the 2" line.
However, it would have had to be strong enough largely to survive the first scram.
The obstruction could subsequently have been dislodged and perhaps disintegrated by the build up of pressure in the SDV when the later scrams occurred.
The source or sources of the water in the SDV could, in this hypothesis, have been, at least in part, Outlet scram valve leakage Normal water discharge during and following the preceding (June 6) scram 12
The vent system (as liquid or as vapor which condensed in the SDV)
This hypothesis has been designated as the "obstruction hypothesis".
2.
Traps or loop seals may have existed in the system, particu>>
larly in the 2" connector pipe connecting the SDV to the SDIV, which, fn con)unction with a partial vacuum in the SDV, would be capable of preventing drainage of water from the SOV.
A particular place where, it is postulated, a trap could have existed is in the portion of the 2" line which forms an expan-sion loop.
This loop is in the steam tunnel.
No definite evidence of the presence of a trap in the 2" line is available since the line was cut apart (looking for an obstruction) before measurements to detect a trap were made.
In this scenario the partial vacuum could have been created by condensation of steam in the east SOV or its vent system.
The source of steam could be either the clean radwaste (CRM) sump or thermally hot discharges of CRW into the 8" piping to which the SDV vent line is also connected.
The source of the water in the SDV could have been the same as in the "obstruction hypothesis".
In particular, water may have been forced into or sucked into the scram discharge volume through the SDV vent line as a
result of discharges of water from other sources into the clean radwaste drain system, and retained in the SDV by either an obstruction in the 2" SOV-SOIV connector pipe or a trap or loop seal in that line, as in hypothesis 1.
This hypothesis has been designated as the "vent hypothesis".
3.
The SDV vent valve may have failed in the closed position at the time of the June 6 scram, trapping water in the SDV for-the intervening 3 week period.
Like the "vent hypothesis", this one probably requires that there be a trap or obstruction in the 170'ong 2" connector pipe since it has been demonstrated that the east SDV bank will drain (slowly) even if the vent 13
valve is closed.
This hypothesis has been called the "stuck valve hypothesis".
It is possible that a combination of these mechanisms were operative.
It is obvious that a mechanical obstruction in the 2" line could have essen-tially the same effect as the hypothesized trap if the obstruction werg big enough completely to seal off the pipe.
Even if the obstruction were not big enough to seal off the pipe it could still, in effect, reduce the diameter of the pipe, thereby decreasing the size of the dip or hump in the pipe neces-sary to create the hypothesi zed trap.
Thus, the behavior of the system is
'explainable by either a trap or a mechanical obstruction fn the 2" line.
Our principal conclusion regarding the cause of the presence of water in a large part of the SDV is thus that this water was present due to either an obstruction or to a trapping and partial vacuum mechanism, or to some combin-ation of these which could include failure of the vent valve to open after the June 6 scram.
These findings and conclusions raise certain questions:
Why is it believed that the incomplete scram was due to the presence of water in the SOV?
What is the evidence for the presence of an obstruction in the 2" connector pipe?
What is the evidence for the presence of a loop seal or trap in-the 2" connector pipe' What is the evidence for the existence of a partial vacuum in the east SOV'?
What is the evidence for the "stuck valve hypothesis"?
Is there any substantial evidence which is clearly inconsistent with the three hypothesis outlined above?
14
Additionally the following questions were addressed:
Ofd the progress of the incomplete scram subject the fuel to any unusual
- stress, and if so were any hazards created'ere the responses of the operating personnel appropriate for the situation, and were the procedures adequate?
These questions are discussed in the following subsections of this report.
4.2 Cause of the Incomplete Scram It has been concluded that it was the presence of a large amount of water in the SDV which caused incomplete insertion of 76 control rods by hindering the free discharge of water from the east CRD units.
The fault appears to have
~
been hydraulic in nature, not due to any electrical or pneumatic malfunction, and not due to any mechanical malfunction in a
CRO or a hydraulic control unit (HCU).
This conclusion has been reached partly by elimination of other causes and partly by direct evidence of the hydraulic nature of the pro-blem.
The evidence bearing on this conclusion is summarized in the following paragraphs.
(Ref; 2).
All the tests and observations cited here are those made by GE or TVA personnel.
Electrical Failure The electrical scram signal is applied to four discrete groups of control rods.
Some rods in all groups moved inward toward full insertion.
Thus the failure was not one of failure of the scram signal to reach one entire group of the rods.
Moreover, all 76 of the rods which dfd not insert completely moved at least one notch.
In tests made after the event the scram circuitry functioned normally and the electrical independence of each rod scram group was verified.
The operators report that they observed that all the blue lights on the reactor control panel lighted up.
This too indicates that all the individual rod scram valves
- opened, as required for a scram.
These observations and tests show that the problem was not an electrical one concerned with the scram circuitry.
The possibility that the reactor manual control system, which controls and programs electrical power to valves in the CRD hydraulic system for normal insertion or withdrawal of control rods might have inter,erred with the 15
scram, possibly by introducing a false withdrawal signal to multiple control
- rods, was considered.
This could not cause control rod withdrawal or arrest since the insert force in a scram is about three times as large as the
'ithdrawal force resulting from a manual control signal.
It is therefore concluded that the problem was not one connected with the reactor manual control system.
The. possibility that there could have been an incorrect line up of. the valves on the 76 individual hydraulic control units (HCUs) was considered, since closing of the isolation valves on the HCUs could prevent rod motion.
All valve lineups were reported to be found, in two independent post-event inspections by GE and TVA personnel, to be correct, so this was not the cause of the partial scram.
Lack of Accumulator Pressure The possibility that the nitrogen pressure in the scram accumulators might have been too low to get complete scram of all rods was investigated.
The control room indications of accumulator pressure during and after the inci-dent were normal.
Even had the accumulator pressure been low, scram would have occurred since the design of the drives is such that reactor pressure also scrams the rods.
Additionally, once charged, the accumulators are hydraulically isolated from each other and the chance of so many random failures occurring suddenly and simultaneously is very small.
Excessive CRD or Scram Valve Leaka e
Abnormal scram valve leakage could have supplied a significant quantity of water to the east SDV bank had it been present at the time of the incident.
The possibility that those drives which experienced only partial insertion had suddenly developed leakage of such size as to account for the incomplete inser tion was investigated.
Such leakage would have had to be very large and would have been signaled by recorded indications of wide spread high CRD temperatures.
Inspection of the pertinent recorder charts shows no such indications.
Post-event tests indicated that leakage thru closed scram valves from all the drives combined was about 2 gph.
Post-event individual stall flow tests indicated that seal conditions were normal in the 76 drives; leakage rates were in the 1.5 - 2.0 gpm range.
These results show that individual drive seal leakage was normal, and scram valve leakage was small at the time of. the post-event test.
16
pneumatic (Instrunent Air) S stem Failure The possibility of a pneumatic failure has been considered.
Complete and detailed information on the layout of this system is not yet available.
The available information is that each air-operated scram valve is fed from the instrument air system with no valves intervening between the closest central point of instrument air supply and the individual scram valves except for the individual maintenance valves on each HCU.
The design of the systsn is such that each scram valve is normally held closed by air pressure.
Loss of air
- pressure, whether intentional, by venting the scram pilot valve, or acci-
Additionally, back-up scram valves are located on the central instrument air supply to the 370 individual scram valves.
Open1ng, of these back-up scram valves removes air pressure from all the scram valves, allowing them to open if they have not already done so.
While the air supply to the east and west banks of HCUs necessarily has east and west branches, the available information discloses no way 1n which the air pressure in the west bank could be released without releasing the air pressure in the east bank also.
The possibility has been considered that some malfunction might have occurred on the east branch'f the pneumatic system only.
For example, it might be speculated that water had been introduced into part of the instrument air system through a maintenance or modi'fication error, or that the coamon lnstrsaent air dryers had failed to dry the air adequately, resulting in water being accumulated more in the east than in the west branch of the HCU pneumatic system.
No evidence for such events has come to light.
The possibility has been considered that the instrument air system pressure might have drifted low enough to allow the scram valves to leak excessively; air pressure ts required to keep the valves closed.
Such leakage fran a large number of drives could have been large enough in the aggregate to cause the east SDV partially to fill with water, even if the 170'ong 2" connector pipe had not been obstructed.
However, there is no record of a low instru-ment air system pressure, and without more knowledge of the details of the instruaent air system it is not clear how a pr essure deficiency in that system could have affected only those scram valves on the east side of the reactor.
In any event a deficiency in air pressure would have interferred with scraneing only through premature filling of the SDV with water since it is loss of air pressure which causes the scram valves to open.
Thus, there
is no evidence that the fault was a pneumatic one.
- However, NSAC intends to investigate this possibility further in its longer range generic safety studies.
Presence of Mater in East SDV The possibility that the failure to insert all the, rods was due to the pre-sence of a large amount of water in the east SDV was suggested almost iaeediately by those who were on the scene or-were notified promptly.
The principal points of evidence were:
All the control rods which failed to insert completely discharged into the east SDV.
Those on the west SOY all, scr armed completely, except for the one rod mentioned fn section 3 of this report, which stopped at position 02.
The time to receipt of the 50 gallon level signal from the SDIY was about 18 seconds.
The fill time for the 10 preceding scrams on unit 3 fell between 42 seconds and 54 seconds.
It is'hus implied that a substantial volume of water was already present in the SDVs at the time of the scram although it cannot be determined whether this water was in the east or west SDV or in both.
The scram was re-set and repeated three times after the initial scram.
Each time the rods which were still not fully in moved toward and finally reached full insertion on the. last scram.
This.,
behavior is consistent with the physical picture that between each re-set and the next scram water was draining out of the east SDY at a slow rate, making roan for more water to enter the east SDV from the east CROs.
Subsequent measurements of water drainage rate from full SDYs indicated that the east bank drains in about 30 minutes.
with the vent valve open.
The west bank drains in about 10 min-utes.
The drain times between scrams were 93, 53 and 160 seconds.
Thus it would be expected that the volume voided during the drain times between scrams would be sufficient to accoaeodate the dis-charge from only a fraction of the still not inserted rods..
(The discharge per full scram is about I gallon per rod for drives which are new or newly maintained.
This is made up of 0.7 gallon dis-placement and 0.3 gallon leakage.
With use. this may be increased: by seal wear to 2.5 gal/rod, full out to full in.
The subsequent 18
testing showed that the 8F-3
- drive seals were in.good condition. )
It is evident that the time between scrams was insufficient to drain more than a small fraction of the SDV volume particularly if the SDY connector pipe were partially obstructed.
Since the first scram must have essentially filled both branches of the SDV it would not be expected. that the succeeding scrams would necessarily be complete if the postulated obstruction were still pr esent.
General Electric has calculated (Ref. I) that the observed behavior could be accounted for if (a) the east SDV already contained about 85-90% of its volume filled with water at the time of the first scram and if (b) there were a partial obstruction in the connector pipe.
GE bases these calculations on earlier tests on a single CRD and its appropriate discharge volume in a full scale test apparatus.
l 4.3 Evidence for the "Obstruction H
othesis" It has been theorized that the presence of water fn the east bank of the scram discharge volume was due to an obstruction in the 2" schedule 160 pipe which has an I.O. of about 1.69 inches and connects the SDV and the scram discharge instrument volume.
Attempts to confirm this theory have not yielded positive results either for or against the theory.
The unit 3 reactor building equipment drain sump, into which the SDIY dis-
- charges, was inspected.
It contained some sludge but no object was found which could block a 2" pipe.
The sump inspection did not include draining the sump completely; instead the bottom of the sump was raked, and inspected with an under-water TY camera.
The examination of the sump was thus incon-clusive and provided no support for the "obstruction hypothesis".
After the incident the east connector pipe was cut by TVA in six places and each section of pipe was rodded out with a metal tape.
No obstruction was found.
The east SOV was inspected by making cuts at each connection between the 6" SDV volumes and their respective 1" vent lines.
Fiber optics and horoscopes were used to inspect the SDY 6" volumes.
Small quantities of silt were observed but no larger objects were seen.
It is conceivable that the silt could have accumulated and perhaps solidified at one point in the 2" line (say, at the expansion loop in the steam tunnel) to a degree such that complete or partial blockage of the line occurred.
During post-incident
examinations some silt or crud was observed in lines leading to the SDIY level measuring instruments, partially clogging these lines.
It has been theorized that an obstruction could have formed in the 2" con-nector pipe as a result of flushing of silt from the SDY banks (this had been done only once) in order to reduce local radiation levels.
The flushing did little more than to push the local hot spots from the SDV 6" piping into the smaller 2" line according to TVA outage personnel.
The silt is theorized to have formed a soft plug, possibly in the expansion loop in the 2" connector line.
This loop is located fn the steam tunnel where the high temperature may have hardened the soft plug into a hard partial obstruction.
Presumably, if such an obstruction had been present, it could have been swept out by the high pressure differential resulting fran the opening of the scram discharge valves and dispersed so that little if any evidence of it would be expected to be left after five scrams in rapid succession.
In any event, none has been found.
During the post-event investigation by TVA the SDY headers were again flushed out with water.
Only a little silt was observed in the discharge.
It was observed that the flushing produced some movement of the'reviously existing radioactively hot spots in the headers toward their discharge ends, i.e., in the direction of the flushing flow.
Presumably the hot spots represented solid radioactive materials.
- However, these hot spots do not necessarily confirm the presence of materials capable of clogging the connector pipe.
Also, earlier scrams, such as that on June 6, gave no hint of an obstruc-tion.'hus, while the presence of a mechanical obstruction cannot be ruled
- out, no direct evidence for it has been found.
Some indirect evidence for an obstruction may exist in the following.
A mathematical model of the hydraulic aspects of the CRO-SDV-SDIY complex has been devised by General Electric as earlier noted (Ref. I).
Details of the model appear in Ref. 8.
This model reproduces correctly the behavior of a single CRD and its "share" of the SDV in the San Jose test facility.
Accordingly the model has been used by GE to predict the behavior of the entire Browns Ferry 3 CRO-SDV-SDIY system.
When this was done it was found that in order to make the model reproduce the behavior of the Browns Ferry 3
system as observed on June 28 it was necessary to assume that the east SDV was 85-90>> full of water and that the frictional loss coefficient of the 20
connector pipe was six times as large as would be expected from standard hydraulic correlations.
This could be taken as indirect evidence of an obstruction in that pipe.
However, there are uncertainties which make this conclusion questionable.
The two aspects of observed Browns Ferry 3 behavior which ft was desired to make the model match were a) the fact that 76 rods on the east side inserted only part1ally or, more precisely, that there was only 47% average insertion of those east side control rods not already fully inserted, and b) the report that the 50 gallon SDIV signal was rece1ved 18 seconds
( 19 seconds by some accounts) after the first scram.
If the con-nector pipe had had a normal loss coefficient the time to the 50 gallon signal would have been 8-10 seconds according to the model.
It is not clear that either the model or the data are suff1ciently reliable to justify concluding that there was a mechanical obstruction present.
The ability of the instrumentation used to measure the SDIY level to respond correctly to very rapid changes in level, such as must have occurred in this case is not known.
Also, the arrangement of the instrument piping is such as to make it difficult to interpret level indicat1ons during the period of rapid filling of the SDIY, particularly if there were some partial binding or blockage of them by crud, as was reported.
4.4 Evidence for the "Vent H
othesis" The principal features of the "vent hypothesis" are that:
The
- SDV, SDIY and connected piping, after each
- scram, could become f1lled with steam drawn in through the SDV vents from the complex system of drains into which other than SDY vents and the SDIY drain lines connect.
(Some parts of this drain system receive relatively hot water fram various sources in the reactor building.
This is why there is a cooler for the RBEDT into which all these dra1ns dis-charge and this is why steam can be seen rising frcm this tank when it is open to the reactor building atmosphere)
Another possible mechanism by which the SDV could become filled with steam is the
flashing of hot water, resulting from a scram, upon draining the system.
It is not known whether this occurred in this incident.
Some of the steam drawn into the SDV vent may condense on the rela-tively cool walls of the I" vent pipe, the condensate flowing by gravity to the SOV vent valve which ts located in a horizontal run of pipe at a level about 5'elow the high point of the. vent line and at that point forming a seal or trap.
There fs no direct evi-dence that this actually occurred at 8F-3, but there is no way to have detected it if it had occurred.
Therefore, its occurrence cannot be excluded.
The connector pipe between the SDV and the SDIY has an overal'1 fall of about I'0" from end to end.
(Ref. 6).
This pipe is schedule 160, having a nominal I.O. of 1.69".
If, at any point in the 170'un, there fs either a
low spot or high spot differing more than about 1.69" in elevation from the immediately adjacent piping, such a spot will constitute a trap, such that if suction is maintained in the piping above it, a water column can be stably supported in that piping even though the piping is open and free of any mechanical obstruction.
It has not been determined whether the low point (or high point) in the connector pipe and the trapping action of the vent valve necessary for this scenario were actually present.
The connector pipe contains an expansion
'loop about I3'ong fn a nominally horizontal plane.
This loop fs the most probable site for the hypothesized high point or low point to have existed.
The slope of the piping in this loop at the time of the incident is unknown since the piping was cut apart for internal inspection before any measure-ments were made.
The I.O. of the elbows in the, loop may be smaller than that of the pipe.
Hence a local high point of 1.69" or possibly less would have sufficed to form a trap.
It has been stated that pipe restraints are present on the loop, but the extent to which pipe movement is prevented is not known.
It may be significant that the loop is located in the steam tunnel, where ambient temperatures are usually high.
Thermal expansion could con-ceivably warp the piping out of its cold configuration sufficiently to produce the necessary
- trap, even if the cold configuration were not such as to produce a trap.
It is concluded that definite knowledge of the slopes of 22
parts of the loop at the time of the incident is lacking and probably can never be established.
Water can arrive in the SOV from CRD leakage as well as by condensation, and even the 2 gph leakage rate through the scram discharge valves which was measured for all the CRDs together after the incident would be sufffc'ient essentially to filI the system in several days ff none of it escaped via the SDIV drain valve.
If, as supposed in the vent hypothesis, it was sub-atmospheric pressure in the SDV which permitted the SOV to fill to about 85-90~ of its volume with
- water, and if the trap in the connector pipe was located.
as supposed, in the steam tunnel, the suction would have had to be sufficient to support a column of water of height not more than about 2'.
Hot water enters one or another part of the clean radwaste drain system frequently, i, not continuously.
Steam from this water would be available to the SOV vents since the SOV vents connect into the same CRW drain system as those hotter streams.
Visual observation of the RBEDT has shown visible currents of steam rising from it at times.
A complication in the analysis of the system vfs-a-vis the venting hypothesis is the fact that there were several vents fn the CRW drain system known to be open to the atmosphere.
Some of these vents were of I/2" size, but it is also reported by TVA that there is a 4" vent to the atmosphere at the reactor water clean up overflow.
These vents are avenues by which air may enter the CRW system, creating an air/steam mixture which, if the CRW drain system were at sub-atomspheric pressure as hypothesized, would then tend to be drawn into the SDV and SOIV, finally filling them with air to the point that inflow of the air/steam mixture would cease.
It would be thought that these vents would be sufficient to prevent the creation of any substantially sub-atmospheric pressure in the CRW system and hence in the SDV or SDIV.
- Yet, while conducting individual rod scram tests on the similar system at 8F-1, the 3 gallon alarm on the SOIV came in shortly after the second rod scrammed and the presence of water in the SDIV was verified.
This incident occurred in Hrowns Ferry I on July 19, 1980.
While conductino scram tests,
.the 3 gallon alarm on the SDIV came in shortly after the second rod was scrarrmed.
The SDIV drain valve and the SOV vent valves were both open as is normal for individual rod scram timing tests.
Ultrasonic testing 23
disclosed that about 7 gallons of water was standing in the SDIV with no water indicated in the SDVs.
.Subsequently, a plug was removed from the bottom of the 3 gallon SDIV float switch assembly.
The result was that air was drawn in to the open hole for several
- minutes, showing that a sub-atmospheric, pressure existed fn the SDIV.
Once the subatmospheric pressure was relieved, the water in the SOIV drained successfully.
Several more rods were subsequently scram-timed without repetition of the phenomenon.
~
NSAC and INPO have been informed (Ref. 3) that additional incidents like that of July 19 have since been observed at BF-I again.
It is reported also that the July 19 incident symptoms (receipt of the 3 gallon level signal) have since been reproduced by the expedient of restricting the entry of cold water 1
into the CRW drain system, admitting only relatively hot water.
It is reported that when this is done the 3 gallon SDIV level alarm can be received during normal operation, and that the open I/g vents on the 565'evel (above referred to) have been observed to emit steam.
The presence of a water column upstream from the SOIV drain valve has been detected ultrasonically under these conditions and apparently under more normal conditions also.
(Ref. 3).
No water was detected in the 6" SDV headers, and no standing water has been detected in any of the testing done on all U.S.
BWRs since the June 28 incident at Browns Ferry 3.
The BF-I observations show that it is possible for a subatmospheric pressure to occur in the SDIV (and presumably therefore in the SDV also) even though the vent and drain valves are open.
Precise identification of the locations of the traps which permit this occurrence is not possible without more detailed analysis of the system.
4.5 The "Stuck Valve H
othesis" It has been pointed out that, if the east vent valve had failed to open following the June 6 scram, and if there were a partial or complete obstruc-tion or a trap in the connector pipe, water from the June 6 scram could still have been in the east SDV on June 28.
After the June 28 incident the east vent valve was disassembled.
Nothing abnormal was found.
The operator has an indication in the control rocm of the status of this valve, derived from valve stem position (not solenoid energization).
It is considered quite unlikely that the operator would have fai led to notice the light signifying that the valve was closed for three
weeks.
'Thus it is considered unlikely that the "stuck valve hypothesis" is valid.
A blockage at any point in the vent line would have had the same effect as if the vent valve had been closed.
To investigate the question of whether the vent line was open, the plant personnel connected a small vacuum pump to the vent piping on the downstream (RBEOT) side of the vent valve.
The results of the test appeared to show that there was no blockage in the vent line.
However, the conditions of the test were such that those per-form1ng them are reluctant to attach much s1gnificance to them.
4.6 Effect of Scram on Fuel The fact that all the rods which failed to insert fully were in the same half of the reactor could be viewed as an out-of-sequence set of rod movements, thus raising the question of whether any unusual stress was put on the fuel.
General Electric has performed an analysis to determine whether the critical power ratio (CPR) of any of the fuel went below the 'licensed limit for the limit1ng bundle during the transient.
(The critical power ratio is defined as Bundle power at which transition boilino starts ctua ue bun e power initial power, flow and rod pos1tions were'he actual ones at the start of the first scram.
Pr1or to the scram the minimum CPR in the reactor was 2.21.
At the end of the first scram the calculated minimum CPR was 9.34.
(Ref.
7)
There are no indications against the assumption that the change in MCPR from 2.21 to 9.34 during the transient was monotonic.
The recirculation pumps continued at constant speed during the transient and the reactor pressure fell smoothly from 920 psig to 900 psig with no recorded pressure spike which might cause a power spike.
Core average f1ssion power at the end of the first scram was estimated, from local power range monitor (LPRM) readings, to be about 2>> of rated power.
Oecay power was also approximately 2
of rated power.
Coolant flow through the core was estimated to be about 21" of full rated flow.
25
Recorded LPRM readings were examined in order to ascertain, if possible, whether the rod pattern. resulting from the first scram had permitted the development of any local regions in wh1ch there was a suff1cient m1s-match of power and flow to-just1fy concern about potent1al fuel damage.
While this concern has not been resolved in th1s. specific case, further analysis is pl'armed to expl'ore'he" question" of'hether there are partial scram conditions which could carry the potential for fuel damage.
The LPRM readings fn the present case indicated that there may have been local regions where neutron power was in the range of 10-15K, of rated power.
While gross core coolant flow is known, the flow through those channels in -the locally high power regions is not known.
Recorded coolant'nd off gas radioactivity levels before and after the scram, as, displayed in Appendix CON,, show no significant increase in released activity since June 28.
[t is thus implied that the fuel received no appreciable short'erm damage.
4.7 0 erator Res onse Append1x OPS contains 1nformation on the pertinent procedures which were in effect at the time of the incident.
The procedures were not written to include the possibility that a large number (not just a few) of rods would fail to insert completely since they called for the operator to scram each un-inserted rod individually. It is possible that sequential 1ndividual rod insertion could: have been used in. the-June 28 event.
However, if the oper-ator (or shift engineer) regarded the situation which he saw as one which was not-cove'red by written procedure,,it was. reasonable for him to do what he did, i.e., attempt repeatedly to. get-a full scram of the rods en masse, rather than scraaming each one individually.
Under the circumst'ances prevailing at, the time,, the operators responded in an appropriate manner.
In the present'ase it. would have taken substantially longer to carry out 76 individual rod scrams, than it actually took to carry out 4 en masse
- scrams, in spite of the fact-that indiv1dual rod-scrams are carried out with the drain and vent valves'pen, thus facilitating drainage of the SOV and SDIV.
The fact that it was necessary for. the operators to deal with a situation of this, type emphasizes:
the desirability of hav1ng procedures which have broader coverage of potential operating problems.
26
- The review team also discussed the matter of the appropriateness of not injecting sodium pentaborate (the "Standby Liquid Control" or SLC) in this incident.
It is clear that, because the reactor was brought to a safe shut-down condition without ~LC, SLC injection was not necessary in this case.
The procedure governing SLC injection places the responsibility for deciding to inject (or not to inject) on the shift engineer or his assistant.
In this
- case, using the information available, an acceptable decision was made.
So far as could be ascertained, no specific criteria or guidelines for making such decisions have been set down. If the condition of the reactor (e.g.,
power level) at the time of the scram had been different than it was, and if the scram had been one which was necessary for a reason of safety, the SLC decision might have had to be made under more adverse conditions than in this case.
New procedures governing this subject are now in effect at Browns Ferry based on NUREG-0460, but their basis is not clearly defined in NUREG-0460.
4.8 Role of the Vent and Drain Valves The SDV vent and the SDIV drain valves normally have only the function of retaining water resulting from a scram and, when the operator elects, of releasing that water to the reactor building drain tank.
However, starting at a few seconds after the scram itself, the vent and drain valves close and become part of the primary pressure boundary.
If they should fail to close they would create a small break LOCA with coolant discharge directly to the reactor building. If the coolant were highly contaminated, operator access to vital equipment could be impaired.
It is likewise necessary that the vent and drain valves not fail to open when required,
- since, in the event of a partial scram it is necessary reliably to open these valves in order to empty the SOV in preparation for a follow-on attempt to complete the scram.
4.9 Ancilla Ouestions In the course of the study of the HF-3 incident it has been necessary to look briefly at various aspects of the scram system not already addressed, such as the pneumatic
- system, the assignment of the individual rods to the two scram discharge
- headers, the reliability of equipment used for level indication in the SOIV, and the interaction between the turbine/condenser system and the reactor in the event of a partial scram.
So far as known these features of the plant played no adverse role in the event, but additional generic study appears to be warranted.
Consideration of these topics will be subject to later NSAC study and reporting.
4.10 Sugar of Conclusions To sunearize the foregoing discussion, it is concluded that The June 28 partial failure to scram at Browns Ferry 3 was caused by water occupying a large fraction of the volume of the east scram discharge volume.
The origin of this water, how it got into the east SDV bank and why it stayed there are and probably will remain obscure, although several plausible explanations have been advanced.
None of these explanations is contradicted by the available
- evidence, but also none of them has a complete factual supporting basis.
The principal explanations are:
There may have been an obstruction in the 170'ong 2"
connector pipe from the SDV to the SOIV.
A combination of trapping action in the vent line and SDV-SOIV connector pipe together with condensation of steam in the SOV may have occurred..
Any of the foregoing causes may have been aggravated by momen-tary pressure surges in the CRW drain system such as could be caused by large influxes of water from systems other than the SOV, particularly if that water were hot.
Incidental examination of various parts of the BWR scram system suggests further generic study of them by NSAC.
Re-examination of the procedures governing actions to be taken in the event of a partial failure to scram, and of the bases for deciding whether to inject sodium pentaborate, suggests further study by each utility.
28
4 r
A 1
t q
J I
't P
APPENOIX SOE SE 1JEHQ~KNTS BR WNS F
RRY JUNE 28, 1980 Plant Status Prior to the Start of the Event:
Browns Ferry Unit 3 commenced a routine shutdown to perform maintenance on a reactor feedwater pump discharge line.
The initial power reduction was performed at a rate of 8 mwe/min from 1082 mwe to 542 mwe by reducing recirculation flow.
With the recirculation pump operating at minimum speed, additional power reduction to 390 mwe was accomplished by normal insertion of selected control rods.
The plant conditions following this power reduction were as follows:
Power Level Core Flow Reactor Pressure Vessel Level Generator Output Steam Flow Feedwater Flow Control Rod Pattern 36% of Rated 39 x 10 lbs. per hr.
920 psi 35 inches (203 inches above the top of the active fuel) 390 HWE 4.9 x
10 lbs. per hr.
4.9 x 10 lbs. per hr.
157 control rods were fully withdrawn 10 rods were fully inserted 18 rods were at intermediate position (See Reference 1)
TTiIll&
Ol 31 16 Event A manual scram was initiated to complete Remarks and References Within several seconds after the scram, in accordance with
Time Event Remarks and References the shutdown.
This was a normal part of the Drowns Ferry shutdown procedure.
normal procedures, the scram discharge volume (SDV) bypass switch was placed in "bypass" position, the reactor mode switch was placed in "shutdown" position, and insertion of the source range monitors (SRMs) and intermediate range monitors (IRMs) were initiated.
Within several seconds (approximately 5 to 10) after this
- scram, operating personnel noted that:
The blue scram lights and the scram accumulator lights were lit as expected.
A number of rods on the east side of the reactor were indicated to be not fully inserted.
During this entire sequence of events, operating personnel reported that they observed that key thermal hydraulic.
parameters (pressure, level, temperatures and flows) were well within normal limits for ordinary scrams.
Mowever a number of control rods were not fully inserted.
Subsequently, a review of plant recorders and process computer printout showed:
A total of 77 rods did not fully insert.
Of these, seven rods settled at either position 02 or 04.
The extent of insertion varied from 5 percent of fully inserted to 95 percent of fully inserted for these partially inserted rods.
The rod pattern at this time is shown by Fig. SOE-l.
Many of the local power.range monitors (LPRMs) in the east side of the reactor and in the upper region (level CfD) indicated above zero values (See Fig. SOE-2).
The readings of the fully inserted IRMs were approximately mid-range on range five or six.
Time Event Remarks and References A review of LPRH readings indicated that fission power in one or more localized regions of, the core could have been in the range of 10-15$ of full rated power due to rods not being completely inserted in these regions.
Ouring and after this scram, no other unusual or unexplained variations in plant parameters were observed.
01 31 24 Low reactor water level trip occurred.
(This corresponds to Level 3 which is 180.5 inches above the top of the active fuel.)
Total indicated steam flow dropped sharply during this scram to approximately 10 to 15 percent of the pre-scram value.
Reactor pressure decreased from 920 to approximately 900 psi (this pressure remained approximately constant throughout the remainder of the event.)
Core flow decreased to 22 x 10 lbs. per hr.
This decrease was due to the loss of natural circulation driving head when reactor power was reduced.
Based on steam flow indication which may be inaccurate at low scale
- values, the heat generation within the reactor was close to the heat generation expected from decay heat alone.
The insertion of the rods had significantly reduced power level.
The water level decrease was due to void collapse following the scram.
From plant recorders it was later verified that minimum water level reached during this event remained above the level required for emergency core cooling system initiation.
(This corresponds to Level 2 which is 110 inches above the top of the active fuel.)
The level variation, including maximum and minimum values, was well within the expected range during a normal scram.
Time Event Remarks and References Ol 31 34 The scram discharge volume "high-high" level was reached.
Ol 31 40 The main turbine was tripped as part of normal procedure.
Ol 32 01 The low reactor water level trip was reset.
This occurred l8 seconds after the manual scram.
This" normally occurs approximately 40 seconds after a scram.
Subsequent investigation confirmed thab all "high-high" instrument volume level switches were properly calibrated and functioning.
During the interval between Ol 32 01 and Ol 35 43 normal water level was achieved and one feedwater
- pump, two condensate booster pumps and one condensate pump were then secured.
01 35 43 01 37 20 The scram was reset and the hydraulic control unit accumulators were recharged until the lights cleared.
A second scram was manually initiated for the purpose of inserting the remaining 77 rods.
Low pressure switch resets at approximately 950 psig.
Normal accumulator nitrogen pressure is 1150 psig.
Within several seconds after this scram, operating personnel observed that:
All blue scram lights and accumulator lights were lit and scram group pilot lights were out as expected.
Rod movement was indicated, but motion appeared slower than normal.
Some rods were indicated to be not fully inserted.
Subsequent review of plant recorders and printouts indicated that:
A total of 59 rods were still not fully inserted.
The rod pattern at this time is shown by Fig.
SQE-3.
Hajor plant parameters (steam flow, core flow, water level and reactor pressure) remained essentially unchanged throughout the scram.
Time Event Remarks and References Fluctuations of IRH sensor readings prevent determination of actual effects due to the scram.
01 38 19 The scram was reset and the hydraulic control unit accumulators recharged until the lights cleared.
01 39 12 A third manual scram was initiated.
Operating personnel observed that:
All blue scram lights and.ccumulator lights were lit and scram group pilot lights were out.
Rod movement was indicated.
Sene rods were indicated to be not fully inserted.
Information obtained later from plant recorders showed that:
A total of 47 rods were still not fully inserted.
The rod pattern at this time is shown in Figure SOE-4.
LPRH readings all indicated "downscale" except 40-57C which was failed.
All major parameters discussed
- above, remained approximately constant throughout this scram.
01 42 37 01 45 17 The scram was reset for the third time and the hydraulic control unit accumulators were recharged until the lights cleared.
The operator placed the scram discharge volume (SOV) bypass switch in "normal" which resulted in a scram since the SDV was not actually fully drained.
This was the fourth scram in this sequence of events.
Operating personnel observed that:
Rod insertion rate appeared normal for a scram.
All blue scram lights and accumulator lights were lit and scram group pilot lights were out.
Time Event Remarks and References 01 46 30 01 46 43 01 57 34
.The operator initiated a manual scram (confirmatory).
The scram was reset.
The discharge volume high water level trips (50 gallon scram trips) cleared.
All rods were indicated to be fully inserted as shown in Figure SOE-5.
Plant parameters and conditions were as normally expected following a scram.
PRIOR TO SCAN Pl 59 55 48 51 48 48 47 48 12 48 43 48 48 48 48 39 38 48 48 48 35 48 48 48 48 31 48 48 0 48 27 48 48 48 48 23 38 48 48 48 19 48 48I 48 48 48 12 48 11 48 48 07 03 48 48 48 48 48 48 0 48 48 48 0 48 48 48 48 48 48 48 48 48 0 48 48 48 48 48 48 48 48 48 48 42 48 42 48 48 48 848 8
48 48 48 48 4S 4S 48 48 48 48 48 48 48 48 48 48I48 4S 48 8
48 48 4
48 48 48 48 48 48 48 48 48 0
48 48 48 0
48 48 48 48 48 48 48 0
48 48 48 0
48 48 48 48 48 4s 48 48 48 12 48 48 48 48 48 48 48 48 38 48 48 48 48 48 0 48 48 48 48 48 48 48 48 48 38 48l48 48 4S 48 12 48 48 48 02 06 10 14 18 22 26 30 34 38 4Z 46 50 54 58 AFTER SCRAM 81 59 55 51 36 47 42 40 36 46 20 4 36 42 38 2
36 42 10 24 0131 Blank Indicate Rod Ful 1 In 27 42 36 30 23 8
38 19 34 42 34 14 26 32 20110 43 12 40 44 30 39 24 40 35 30 4 38 18 31 ZZ 10 34 28 28 40 38 34 32 8 34 30 28 36 24 12 36 28 22 26 15 40 11 26 07 03 36 6 14 40 38 40 40 24 42 42 30 02 06 10 14 18 22 Z6 30 34 38 4Z 46 50 54 58 Figure SOE1.
Control Rod Positions SOE-7
57 2
0 0
0 2
=
0 0
0 0
0 0
0 0
2W 0
0 Wajled 49 2
4 0
2 0,,
0 0
0 0
0 0.
0 0
~ 0 0
0 41 3
2 0
0 3
6 2
~ 6 0
0 0
0 0
0 0
0 33 6
3 0
0 3
2 0
0, 0
0 0
0 25 9
)0 0
0 10 0
0 0
2 0
0 0
0 0
0 0
0 0
0 0
17
)0 6
0 0
10' 0
0 0
0 0
0 0
0 0
0 0
0
'0 0
09 0
C 8
A 4
2 0
0 0
0 0
'0 0
0 0
0 0
0 0
08 16 40 48 56 Figure SOE-2.
LPRM Readings Following Scram 41 SOE-8
PRIOR TO SCRAM 02 59 55 36 51 36 34 40 4
46 38 36 36 42 10 0131 47 43 12 39 24 35 30 31 22 27 42 23 8
19 34 15 11 07 03 42 20 40 44 40 4 38 10 34 36 30 14 38 32 42 34 20 40 36 26 6
24 30 28 18 8
28 2
26 28 12 10 28 38 14 40 40 2
28 42 40 32 34 36 22 40 38 30 24 30 02 06 lo 14 18 22 25 30 34 3a 42 46 50 54 Sa AFTER SCRAM '42 59 55 30 26 42 28 14.18 0137 51 47 30 30 39 8
35 14 31 8
27 26 22 23 19 10 26 12 22 38 14 30 30 4
14 12 20 2 12 30 20 24 28 34 12 4 30 26 26 14 22 20 12 16 2
16 8
6 4
15 18 ll 07 03 12 18 30 32 30 14 34 22 4
02 06 10 14 18 22 25 30 34 38 42 46 50 54 Sa Figure SOE-3.
Control Rod Positions SCE-9
PRIOR TO SCRAM 43 59 55 26 51 12 22 47 30 43 30 38 39 8
30 35 14 30 31 8
14 27 26 22 20 2
23 30 20 19 10 26 24 30 42 28 28 12 14 4
26 12 12 12 14 18 34 30 26 14 22 20 16 2
16 6
4 0137 15 11 07 03 1
18 12 28 32 30 2 34 30 14 22 4
02 06 10 14 18 22 26 30 34 38 42 46 50 54 58 AFTER SCRAM f3 59 55 20 Sj 16 47 26 40 22 24 30 0139 43 39 35 31 27 23 19 24 36 2
24 6
26 18 16 14 2
26 12 16 18 20 4
6 24 18 14 16 2
2 15 4
10 11 6
07 03 24
)0 28 30 26 8
10 02 06 10 14 18 22 26 30 34 38 42 46 SO Sc Sa Figure SOEA.
Control Rod Positions SOE-10
PR?OR TO SCRAM 84 59 55 51 47 26 24 20 40 16 22 24 30 0139 43 24 36 2
39 2
24 35 6
26 31 2
4 20 24 14 18 16 27 1816 14 2
4 23 26 12 l9 16 18 6
2 15 4
10 ll 6
07 03 30
<<0 28 26 10 02 06 10 14 18 22 26 30 34 38 42 46 50 54 58 AFTER SCRAM 44 IIII' s
59 55 Sl 47 43 39 35 31 27 23 l9 15 ll 07 03 LLROS 02 06 l0 l4 18 22 26 30 34 38 42 46 50 54 58 0145 Figure SO&5.
Control Rod Positions SOE-11
h h
J F
= 1
~
r
.""",.':.'."."::"". AFPENOD OES h
r
=h ~
(
r I
r
'r Z
h C
= h h
~
r r
~ =
A' h
r r
4 I
\\
h 4
~
~ ~
'3i'
APPENDIX OES SYSTEM DESCRIPTION CONTROL ROD DRIVE HYDRAULIC SYSTEM I.
SYSTEM FUNCTION A.
General - The control rod drive system makes changes in core reactivity by incrementally positioning neutron-absorbing control rods within the reactor core in response to manual control signals.
The system is also capable of quickly shutting down the reactor (scram) in emergency situ-
. ations by rapidly inserting all withdrawn control rods into the core in response to manual or automatic signals.
B.
The functions of the system are to:
1.
Move the control rods in either direction at controlled rates, latching the rod stationary when not requested to move.
2.
Charge HCU scram accumulators.
3.
Provide cooling water to drive mechanisms.
II.
MAJOR SYSTEM COMPONENTS (Refer to Fig. OES-I)
A.
Two 1001, capacity, 5 stage, centrifugal pumps.
B.
Two 10Ã capacity drive water (discharge) filters.
1.
Parallel installed, 50 micron cartridge type.
C.
Two 10(C capacity flow control valves.
l.
Air operated, balanced trim globe valves positioned by automatic flow control loop or local manual control stations.
2.
Normally controlled in automatic mode to give approximately 50 gpm system flow.
OES-I
0.
One drive water pressure" control valve.
1.
Motor operated valve positioned by operator to maintain approximately 250 psi differential between drive water head and reactor pressure.
E.
One cooling water Pressure control valve.
1.
Motor operated valve positioned by operator to maintain the cooling water header approximately 20 psi above reactor
'pressure, to provide cooling flow.
F.
Two sets of solenoid operated stablizing valves.
G.
Hydraulic control units (Fig.
OES-2) 1.
Provide control interface between hydraulic system and individual CROs.
2.
Seven hydraulic piping risers with manual shut-off valves at each HCU provided as follows:
a.
Insert riser - Line connecting HCU directional control manifold to representive CRO under piston area.
b.
Withdraw riser - Line connecting HCU directional control manifold to respective CRD over piston area.
c.
Coolin water riser - Line connecting hydraulic system cooling header to CRO insert line.
d.
Exhaust riser - Line connecting HCU directional control manifold and hydraulic system return header.
e.
Drive water riser - Line connecting system drive water header to HCU directional control manifold.
Provides hydraulic pressure during insert, unlatch and withdraw modes of CRO operation.
OES-2
f.
Charoino water riser - Line connects hydraulic system charging header to water side of scram accumulator.
Maintains scram accumulator at pump discharge pressure.
g.
Scram dfscharce riser - Line connects the CRD over piston area to the Scram Discharge Volume via the withdraw riser and outlet scram valve.
3.
Directional Control Manifold consists of a 4-way block manifold to port CRO hydraulic flows between the CRO mechanism via the insert and withdraw risers and the hydraulic system via the drive water supply and exhaust riser during normal rod movement.
4.
Scram Section consists of the following HCU mounted components to allow rapid insertion of the control rod in response to signals from the RPS.
a.
Scram Valves - Two normally closed, air operated globe valves.
Held closed by instrument air via two solenoid operated scram pilot valves, each powered from one of the two RPS trip channels and connected physically such that both solenoids must'e-energi ze to vent air from and to open the scram valves.
(1)
Inlet Scram Valve, when open, connects the water side of the scram accumulator directly via the insert header to the CRO under piston area.
(2)
Outlet Scram Valve, when open, connects the CRO over piston area, via the withdraw riser, to the scram discharge volume.
b.
Scram Accumulator consists of piston type, 5 gal capacity, accumulator connec.ed to a precharged l<2 cylinder at the bottom and the insert riser vfa the inlet scram valve at the top.
c.
S stem Res onse Ourin Scram - When reactor scram is initiated by the reactor protection system (RPS) the inlet and outlet scram valves open to admit the pressure in the scram accumulator to the area below the drive piston and to vent the ar ea above the piston to the scram dischai ge headers.
The scram discharge headers are maintained at atmospheric pressure during normal operation. 'he large differential pressures applied to the drive piston area produce a.,large upward force on the index tube and control rod, giving the rod a high initial acceleration and providing a large margin of force to overcome possible friction or binding in. the drive line.
The characteristics of the CRD hydraulic system are such that the CRD index tube rapidly accelerates to a scram velocity of approximately 5'/sec.
As the index tube closes off ports in the cylinder wall (i.e., buffer holes in the piston tube),
the increasing resistance to water flow reduces the speed of the index tube.
The number, size, and spacing of the buffer holes in the CRD piston tube which are progressively closed are chosen to provide a
gradual deceleration of index tube movement.
Each CRD puts approximately 2.5 gallons of water into the SDV for a maximum scram stroke; the water side of the scram accumulator holds approximately 1.5 times this volume.
There is adequate capacity in the hydraulic system accumulator supplying each CRD to complete a scram stroke in the required time at low reactor pressure.
At high reactor pressures,
- however, the accumulator provides
,the initial surge of water, but the normal accumulator discharge and line losses quickly reduce the pressure at the CRD to a level equal to reactor pressure.
This causes the ball check valve (built into the CRD flange directly below the under-piston water port) to shift its position and admit reactor water under the drive piston, thus admitting water from the accumulator and from the reactor vessel.
Reactor pressure, therefore, supplies the force required to complete the scram stroke at higher reactor OES-4
pressures, while the accumulator-alone supplies the force for low-pressure scrams.
III.
SCRAM DISCHARGE VOLUME (Fig.
OES-3)
A.
General
- Provided to receive and contain the water exhausted from all CRDs during scram.
During a scram the SDV becomes part of the primary system pressure boundary.
Consists of header piping which connects to the HCU scram discharge risers.
The east and west sections of this piping drain to a co+non instrument volume.
The east and west headers are equipped with vent valves.
Since a minimum SDV capacity is required to allow the water displaced during scram to accumulate fn the
- SDV, a
series of level switches is provided on the instrument volume to warn of diminishing capacity:
1 scram discharge volume not drained (3 gallons) 1 rod block (25 gallons) 4 SDV Hi Hi scram (50 gallons)
B.
SOV Vent and Drain - Air to-open, normally open valves which maintain SDV drained and vented to atmospheric pressure during normal oper-ation.
Valves are held open by instrument air from scram pilot air header via a 3-way double solenoid dump valve.
Each solenoid fs normally energized from one of the two reactor protection system trip channels.
'Ahen both solenoids are de-energized (scram) the valve exhausts the air header downstream allowing the vents and drain to close by spring pressure.
Hoth the SOV vent lines and the SDIV drain line discharge into the clean radwaste drain system.
This is a complex arrangement of piping which takes drainage from about 50 points in the reactor building and collects it in the reactor building equipment drain tank, (Ref. 4) (Ref. 5),
a lined rectangular concrete sump of about 6340 gallon capacity bottomed at elevation 509'n the reactor building.
Except when being inspected the top of the tank is closed with a concrete plug.
The RSEDT is equipped with pumps for removal. of the water and for passing-the water through a cooler when necessary.
Several potential DES-5
sources of hot water drainage into this tank have been identified including:
Reactor Mater Clean up System Leakage Fuel Pool Cooling System Leakage Steam Tunnel Piping or Valve Drains Residual Heat Removal Heat Exchanger and Piping The CRM drain system collects drainage from equipment at several elevations in the reactor building as follows:
(Ref. 4) 66p<6u 621'65'he SOV-SOIV system fs located on elevation 565'.
Measurements of the normal rate of CRM flow rate into the RBEDT are reported by TVA to have been made, with the result being about 4-5 gpm.
These masurements have not been documented to NSAC.
The CRM drain system piping ranges in size from a few 1/2" vents to 8" iron pipe size downcomers.
AII discharges from that part of the system into which the SDIV drains combine in a single 8" downcomer.
The design drawings for the system specify that it is to be air tight no funnel drains permitted, but some of the 1/2" vents have apparently been open for a long time.
There is also a 4" vent at the RWCU precoat overflow line.
OES-6
REACTOR VESSEL CRO
~ HCU BOUNOARY ORIVE WATER (WITHORAWI SCRAM VALVE ORIVE WATER IINSERTI I
I SCRAM VALVE
/
i AS iRSA~
SYRSII~
HCU'S
~ T I
ISO 7 R
HCJ'S MOLINQWATER
'klASUPPLY
-IS1 OIRECiIONAL CONTROL'O LE NO I 0 V*LVE E(HAUSS 'PYATER I Mo I CRO PUMP
~ AO I STABIUZINQ VAI.VES SCRAM OISCHARQE VOLUME NOTE: MANYVALVESANO INSTRUMENTS L4OT SHOWN ORAIN IVORMALPI.OW PATH l
AIR SUPPI..
REAC i OR PRoi c
< ICN SYSTEM Figure OES-1.
Typical CRD Hydraulic Control System (piping layout)
OES-7
ISOLATION VALVK SCRAM VALVEPILOT AIR ISOLATION VALVE EXHAUSTWATER RISKR ISOLATION VALVE INSKRT RISKR ISOLATION VALVE COOLING WATER RISER O
4 CHARGING WATER RISER ISOLATIONVALVE WITHDRAWALRISER ISOLATION VAIVK ORIVE WATER RISER SCRAM PILOT VAI.VKASSEMBLY JUNCTION BOX ISOLATION VALVE SCRAM OISCHARGE RISKR OUTLET SCRAM VALVE INLETSCRAM VALVE OIRECTIONALCONTROL VALVK IINSERTI MANIFOLO OIRECTIONALCONTROL VALVE IWITHORAWANO SETTLKI WIRING TROUGH ASSEMBLY UNIT INTERCONNECTING CABLE 0IRECTIONALCONTROL VALVE (WITHDRAW)
OIRECTIONALCONTROL VALVK IINSERTI SHUTOFF VALVE WATER ACCUMULATOR DRAIN SCRAMWATER ACCUMULATOR SCRAM ACCUMULATORN2 CYLINOER ACCUMULATORN2 PRESSURE INOICATOR CARTRIOGK VALVK ACCUMULATORN2 CHARGING ACCUMULATOR
'NSTRUMENTATION >
ASSEMBLY Figure OES-2.
Control Rod Orlve Hydraulic Control Unit (component assembly)
DES-8
1" vent 6" headers To CAW drain East Bank Scram discharge header 1" vent Instrument volume To CRW drain West Bank to CRW drain 12" Ql ~
L~20'evel sensors 6" headers Figure DES-3.
Browns Ferry SDV Equipment Layout (isometric view)
-4
' 'h,.'E=
'i>>S ~
C' E
~
~
. -"..'...j 'i"., ',.'.'"ppEMI"Ã EON'G~(UQKK;.OF FHE EV'N(T: TO TH& RK E
'E ~'
A E.
E S
E E
hE r
ES,
~ S 1
IJ EE( rCh>>
'gc
\\
"~
.-;-'6
~'S h ~ rJ E
APPENDIX CON CONSEOUENGES OF THE E'/ENT TO THE REACTOR The fact that the first scram resulted in an out-of-sequence rod configuration with the reactor still critical at a low but not precisely known power level has
~
raised the question of whether there could have been some part of the reactor which would have been running at a power density high enough to overstress the fuel at the existing cooling conditions.
A pragmatic answer to this question lies in the comparison of before-and-after levels of radioactivity in the primary coolant (Table CON-I) and the off-gas stream (Table CON-2).
It is seen that there is no systematic change fn primary coolant gross radioactivity or iodine.
The off-gas activity shows a generally increasing trend in June and August, but July levels did not show such a trend.
Also, July levels were not significantly different than those of June.
These data do not suggest any damage to the fuel.
CON-I
TABLE CON-1 PRIORY COOLANT CHEMISTRY BF-3 Gross Radioactivity (u Ci/ml)
JUNE 1980 JULY 1980 Solids High Low Avgas 9.12E-03 2.03E-04 1.27E-03 8.63E-03 2.37E-04 1.36E-03 Filtrate High Low Avgas 2.45E-01 2.64E-02 1.20E-01 1.48E-01 5.43E-03 6.85E-02 Iodine 131
{u Ci/ml)
High Low Avg.
1.55E-03 2.35E-05 7.46E-04 3.34E-04 0.75E-O5 1.94E-04 CON-2
TA8LE CON-2 OFF GAS SYSTEM PERFORMANCE*
BF-3 Off Gas Radioactivity, post-treatment monitor, u Ci/cc JUNE 1980 JULY 1980 AUGUST 1980 Start of Month 2E-03 3E-03 1.5E-03 End of Month 5E-03 3E-03 8E-03 Maximum 5E-03 3E-03 8E-03 Off Gas Flow Rate, SCFM, approx.
avg.
100 80-90 80-95
- The values given are the result of putting a straight line through the daily readings for the month and estimating the intercepts.
Hence the value at the end of one month does not necessarily agree with the value at the beginning of the next month.
Also, in the case of July, values for approximately the first half of the month were not used since the reactor was shut down for most of this time and the off gas radioactivity readings were either low or erratic.
- However, no July readings higher than about lE-02 were recorded.
Additionally, GE has calculated that no fuel operating limits were exceeded during the transient.
Copy attached.
(Also Ref. 7.)
CON-3
GENERAL "
ELECTRIC GENERAL ELECTRIC COMPANY, 17$ CURTNER AVESANJOSE, CALIFORNIA95125 NUCI.EAR FUEL AND SERVICES 0 I VIS I 0 N-FP F-80-271 August 22. 1980 Or. Hiles Leveritt Nuclear Safety Ana'lysis, Center 3412 Hillview Palo Alto, CA 94303 Oear Or. Leveritt:
The attached memo suInnarizes the work performed by General Electric to evaluate.
the condition of the Srowns Ferry<<3 fuel following the incomplete rod insertion on June 28.
The results indicated substantial margins to any fuel limits existed.
IIhile this work has not received the engineering reviews which-we would normally perform as part of our engineering evaluation efforts, the margins are so great that we believe the conclusions are accurate.
If you have any questions.
please call me.
Very truly yours, F.
P. Felini, Hanager Oomestic Product Service II/O 880, 925 1257 FPF'gs COIIA.
TO:
F.P. Felini FRoM:
J.A. McGrady
. THERMAL LIMITS EVALUATION OF BROMNS 'FCRRY 3 PARTIAL SCRAM EVENT oATE: Auoust 22, 1980 CFSO/80-237 REOVIREO RESPONSE OATE n/a FOR:
ACTION Q OECISION Q INFORMATION Itl 3-0 BMR simulator cases wIre run simulati~g the core conditions at Brcwns Ferry 3 just prior to and just after the first par ial scram.
The actual rod patt'erns and thermal-hydraulic conditions hat were a
combination of the best available data from the plant and best estimate (by engineering judgement) values were used in these two cases.
Based on these simulator cases, the therrIal limits at Browns Ferry 3
both before and after the first partial scram, were not even close to even the technical specification operating limits on
- MCPR, MLHGR, and R'PLHGR (See Table below).
- Thus, as not even normal ooerating thermal limits were violated during this event, thermal limit viola-ti.ns are of no concern for the actual Browns Ferry 3 partial scram event.
Technical Specification Thermal Limit 3-0 BWR Simulato.
Resul.s
'Before Scram After 1st Partial Scram MCPR RAPLHGR>>
>> 1.54 (1.28xK<)
c 13.4 Kw/ft
< 1.0
(
YAPLHGR II 2.21 6.49
.483 9.34
- 1. 81
.137 where MAPLHGR limit varies with fuel type and exposure.
These results have not been verified, but are believed to be representative of actual core conditions to the extent necessary to justify that thermal lirIits were of no concern during this event.
J.A.
McGrady Core 4 Fuel Systems Design Mail Code 740 - Ext.
56136 Concurrence:
K.L. Stark Senior Engineer Operating Core Oesign Management-2 CON-5
(
T
(
(
- , '
- :~'.~,:"'-'.-"'-'".;". 'jjh-. '.~'.(- '". =,.'JPPPSIX'PS
- ".~,. -
- .';~,'.';,:">>:;,,'"..."'";:.':;AKRRfOA.'ACTHNS N
O. PROCEDURE
S I
~
(
e t
(
\\
I I 'I
(
((
~
(h <
(
(
, I' i'
r
~.~
(I "(ji.
I I'
~,I
~ r
~ I
(( ~
,I,
~
I' r
( ~
r (
I
~
~
APPENDIX OPS OPERATOR ACTIONS AND PROCEDURES I.
PURPOSE The purpose of this section is to review the possibly applicable procedures available at the time of the event, the procedures and/or modifications to procedures as a result of the event, and to'omnent on the performance of the operational staff during the event.
II POSSIBLE APPLICABLE OPERATION PROCEDURES AVAILABLEAT THE TIME OF THE EVENT PROC.
NO.
85 63 TITLE Control Rod Drive Hydraulic System Section IV (Abnormal Operations)
Standby Liquid Control System
'A.
Operating Procedure No. 85.
This was the only existing procedure at the time of the event which actually addressed the situation of rods failing to fully insert after a
Basically, the operators adhered to this procedure.
It does,
- however, specify scramming the rods individually versus performing total core scrams.
The original intent of the procedure did not foresee such a
large number of rods not fully inserting and, therefore, it was reason-able for the operating personnel to opt for total core scrams using the manual scram buttons.
As a result of this event, the procedure has been modified to be applic-able when four or less rods fail to fully insert.
The failure of five or more rods to fully insert is now addressed in a new emergency operating instruction which is detailed under Item
( III) below.
B.
System Operating Instruction No. 63, Standby Liquid Control System This procedure is addressed to show only that it was available.
It states that the shift engineer or assistant shift engineer has the authority to order injection.
OPS-I
In the case of this event, operations personnel acted in accordance with their judgment as provided fn the then existing procedures.
- However, both the then existing and now existing procedures appear to lack adequate criteria for making such judgments.
Further review is advisable.
III.
NEW PROCEDURES PREPARED AS A RESULT OF THE EVENT The complete text of these procedures is provided below.
1.
Control Rod Drive Hydraulic System Operating Instruction No. 85, Section S, Mater in Discharge Header (By Ultrasonic Tests)
S.
Water in discharge header (by ultrasonic tests) 1.
Water level fn scram discharge piping less than 1".
a.
Dispatch personnel to visually verify open position of SDV vent and drain valves.
b.
Check for leaking scram discharge valves.
1.
Check CRD temp recorder - high drive temperatures may indicate leaking scram outlet valves.
2.
Check temperature of the scram discharge riser at each hydraulic control unit.
3.
If any riser is found to be warm to the touch, verify scram discharge valve leakage according to GEK-9582.
2.
Water level in scram discharge piping ) I,)/2" but
< 2".
a.
Immediately reduce load to minimum and scram the reactor.
OPS-2
3.
Mater level in scram discharge piping
> 2".
a.
Initiate a controlled shutdown to the cold shutdown condition.
2.
Failure of Control Rods to Insert Ouring a Scram (Emergency Operating Instruction No. 47)
SYMPTOMS A.
Parameter changes 1.
Failure of all control rods to fully insert upon scram signal 2.
LPRM's indicate the reactor is critical after scram 3.
IRM's and SRM's indicate reactor is critical after the scram.
B.
Alarms 1.
Reactor auto scram channel A (B) 2.
Rod withdrawal block 3.
Scram discharge volume not drained 4.
Sequential events print-out would give additional annuncfations II.
AUTOMATIC ACTIONS Automatic actions directly associated with the event, causing the scram would occur (i.e.,
low water level, high drywell
- pressure, turbogenerator trip and etc.)
OPS-3
FAILURE OF ALL CONTROL RODS TO INSERT III. IMMEDIATE OPERATOR ACTIONS A.
Verify existing condition'y multiple indications (i.e.,
alarms, charts, indicating lights, gauges and other instruments).
B.
Verify all automatic actions have occurred.
If not, place controls in manual and make corrective manipulations.
CAUTION:
00 NOT PLACE CONTROLS IN MANUAL, UNNECESSARY WHEN AUTOMATIC IS FUNCTIONING PROPERLY UNLESS UNSAFE PLANT CONDITIONS WILL RESULT.
C.
Trip reactor recirculation pumps.
0.
Place mode switch in shutdown.
Place scram discharge volume high water" level bypass switch to bypass.
Verify scram discharge volume vent and drain valves open.
E.
Reset reactor scram and manually scram the reactor, reset and repeat if rod motion is observed until all control rods
- are fully inserted.
CAUTION:
CONTINUOUSLY MONITOR NEUTRON FLUX UNTIL ALL RODS ARE FULLY INSERTED.
F'.
Notify supervisor of events and actions taken.
OPS-4
IV.
SUBSEOUENT OPERATOR ACTIONS A.
If all scram valve position lights are not illuminated, isolate and vent the scram air header.
Restore air when all scram valves are open.
B.
Individually, from the auxiliary, instrument
- room, scram those control rods not fully inserted.
Reset each rod after it fully inserts and continue to scram rods until all rods are fully inserted.
Select those rods in areas of indicated high reactivity to be scrammed first.
C.
Select the control rods that did not fully insert and manually insert as RSCS permits; bypass the RWM; place the
- RSCS, sequence mode selector, in the insert position and select the rod group to be inserted.
From the auxiliary instrument
- room, bypass all rods to full in position, then manually drive the rods in one at a time.
0.
If the control rod system is unable to maintain the reactor in a subcritical condition; upon shift engineer's
- approval, initiate standby liquid control (SLC) and continue to inject until the entire contents of the standby liquid control tank is injected.
OPS-5
CAUTION:
IF AT ANYTIME THE FOLLOWING CONOITIONS EXIST, SLC INJECTION IS NNOATORY:
1.
Five (5) or more adjacent rods not inserted below 06 position and either reactor water level cannot be maintained,- or suppression pool water temperature limit of 110'F is reached.
2.
Thirty (30) or more rods not i~serted below 06 position and either reactor water level cannot be maintained or suppression poo1 water temperature.
limit of 110'F is reached.
E.
Confirm the reactor water cleanup system (RWCU) isolates or manually isolate the RWCU system.
F.
Restart the reactor recirculation pump(s) at minimum flow to aid in mixing of neutron absorber.
G.
Verify neutron flux decreases as SLC solution is injected into the core.
H.
Log events and actions taken in daily journal.
OPS-6
REFERENCES 1.
Evaluation of Incomplete Control,Rod Insertion Event at Browns Ferry 3.
Report NEDC-24276, General EIectric Co., October 1980.
2.
Report and Safety Evaluation of the Browns Ferry Unit 3 Partial Scram Failure of June 28, 1980.
T. D. Knight, Tennessee Valley Authority, Rev.
1, July 10, 1980.
3.
Scram Discharge System Special test BF STEAR No. 80-18.
Tennessee Val'ley Authority, Herbert Abercrombie, August 8, 1980.
4.
TVA Drawing 47W852-2 Flow Diagram, Clean Radwaste and Decon Drainage Browns Ferry Units 1-3.
5.
TVA Drawing 47W481-9 Mechanical
- Drains and Embedded Piping Stage III and IV.
Browns Ferry Units 1-3.
6.
TVA Drawing CNM-5-0024 CRO Scram Discharge Header Isometric.
Browns Ferry Nuclear Plant.
(Unit 3).
7.
GE Letter, J.
A. McGrady to F.
P; Felini, August 22, 1980.
"Thermal Limits Evaluation of Browns Ferry 3 Partial Scram Event" (Copy in Appendix CON).
8.
Internal GE report, "Analytical Studies to Simulate the Incomplete Control Rod Insertion Event at Browns Ferry 3 Using a Transient Model". OPE-0880.
0.
C. Rennels, August 1980.
GLOSSARY APRM CPR CRO CRW hp HCU Header Bank INPO IRM LPRM MCPR NSAC SCRAM SDIV SDV Average Power Range Monitor - Each APRM gives the average of 8 strings of LPRMs. There are 6
APRM channels Critical Power Ratio.
Bundle power at which transition boiling would start divided by actual fuel bundle power Control Rod Drives - Mechnical 8 Hydraulic Piston devices which insert/withdraw control rods (installed directly under reactor vessel)
Clean Radwaste Maste - Equipment leakage fs collected directly at source and sent back to storage prior to repeated usage Differentia'I pressure Hydraulic Control Units - Hydraulic, Pneumatic electrical devices which control insert/withdraw of CRO's (located outside containment)
The 4 pipes, 6" in I.D. about 50'ong which make up a
SDV.
There are 2 header banks - east and west Institute of Nuclear Power Operations (Atlanta, Georgia)
Intermediate Power Range Monitor.
Neutron detectors capable of being inserted into reactor to measure power at intermediate levels.
0.0001% to 30" of full power.
Reactor has 8
IRMs Local Power Range Monitor A neutron monitor for use in the power operating r ange.
The core contains 172 LPRMs arranged in 43 vertical strings of'4 each Lowest value of CPR in reactor Nuclear Safety Analysis Center (Palo Alto, California)
Insertion of control rods at maximum speed to terminate chain reaction abruptly.
Insertion time usually 2-3 seconds Scram Discharge Instrumentation Volume - Includes tanks and piping, fixtures and instruments which monitor water level Scram Discharge Volume - includes piping fixtures from individual HCU discharge lines thru to discharge into SOIV Source range power (neutron monitor).
Source to about 2
of full power.
Reactor has 4 SRMs.
Tennessee Valley Authority
0
~/
i',":
glINPO)'IGNIFICAN OPERATING EXPERIENCE REPORT 80-6 DECEMBER 19t 1980 PARTIALFAILURZOF CONTROL RODS TO INSERT
REFERENCE:
UNIT-BROIUNS PERRY 3 DOC NO/LZR NO:
50-296/80-24 1
DATE: 6/28/80 NSSS/AE - GE/TVA NSAC 20/INPO 3, ~alysis of Incomplete Control Rod Insertion at Browns Ferry 3 DESCRIPTION:
Nhile in the process of performing a normal shutdown for maintenance, a manual scram resulted in a partial failure to insert of '76 control rodst associated with the east Scram Discharge Volume (SDV).
Two additional manual scrams followed by an automatic scram were required prior to attaining all rods in.
The details of this event are described in the INPO/NSAC report referenced above.
SIGNIFICANCE:
The potential exists for water to accumulate within the Scram Discharge Volume (SDV) piping in such a way as. to remain undetected by the Scram Discharge Instrument Volume (SDIV) instrumentation.
This water accumulation
- may, in turn, result in premature pressurization of the SDV and thus affect control rod motion upon a scram.
RZCOMMENDATIONSt Since designs and conditions vary from plant to plant, it is recognized that any one set of specific recommendations can not apply to all BNR plants.
Therefore, these tecommendations are presented as objectives which should be attained, with examples of possible ways to achieve these objectives.
It is understood these etxamples do not necessarily represent all the acceptable ways oi achieving the objectives.
It should be noted that many of the examples are the same as specific fixes which have been proposed by others in the following documents:
NRC IE BULLETINNO. 80'17 ff tt it tf tf tt tl't ff SUPPLEMENT 1 SUPPLEMENT 2 SUPPLEMENT 3 GENERAL ELZCTRIC SIL NO. 331 SUPPLEMENT 4 UTILITYSUBCOMMITTEE RECOMMENDATIONS "Long-Term Evaluation of Scram Discharge System" RECPI~go Jun g~~a>
REGUlATORY STAFF.
RED I'i'IVEDIATE ATTE ITIO'4 YELLOW - PROMPT ATTENTION GREEN
- NORMAL ATiENTION
. l. guard Aeainst Accumulation of Undetected Water In the SDY The SDIV instrumentation can indicate the absence of water in the SDV; when, in fact, the SDV is essentially full. The SDIV water detec.ion instrumentation should provide reliable, direct indication of water in the SDV.
2.
Ensure Adeauac of Procedures on Detection of Water Ensure that adequate procedures exist regarding action to be taken if water is detected in the SDV system when it should be free of water.
3.
Guard A ainst Obstruction Or Traooihm In The SDV-SDIV Connec;or Pipe.
Any obstruc.ion of the SDV-SDIV line, whether by solid matter or by a water trap, should be guarded against.
Possible ways of guarding against a solid obstruction or
~ater trap in the Une include:
Enlarge the diameter of the line connecting the SDV to the SDIV as close as practical to the internal diameter of the SDV headers and repipe with care to eliminate locations where solids would be expected to accumulate such as dips or humps in the line, internal roughness at weld points, diameter changes, etc.
Consider relocation of the SDIV to a lower level in the plant to allow for increased pitch in the line between the remote SDV and the SDIV.
or Install separate SDIVs for (he two header banks, so that the line from each header bank to its SDIV wiH be relatively short, straight, and as close as practical to The same internal diameter as the SDV headers themselves.
Guard Azainst Traooin In The Vent Line Any location or configuration favorable to trapping water in the vent line should be avoided.
Possible ways of doing so include:
Put vent valves in vertical, not horizontal pipe runs, and ensure that all horizontal portions of vent lines actually have sufficient slope to assure drainage without trapping.
5.
Guard Aeainst Interference From CRW Drain System The possibility that steam or water from other drains into the CRW drain system could inter;ere with operation of the SDV-SDIV system by inducing a suction or producing a pressure should be guarded against.
A possible way of doing this is to:
Isolate the vent and drain lines from each other and from other CRW
- drainage, by running them as dedicated lines direc:ly to the reactor building equipment drain. sump.
6.
Guard Aaainst The Possibilit Of The Vent Not 0 enin
~
Failure of the SDV vent valve to open should be guarded against since such failure could interfere with the SDV draining freely.
Possible. ways of doing this include:
Install a redundant vent valve
-or-Cross connect the vents on the two header system
~
~
7.
Guard Against Excessivel Slow SDV Draina e
Excessively slow drainage of the SDY should be guarded against since, if successive and close-spaced scrams are necessary as in the 8rowns Ferry 3
case, it may be difficult to achieve them if the SDVs are essentially full of water from the preceding scram.
After one scram has been attempted, the readiness of the system for a possibly needed next scram should not be limited by the SDV-SDIV drain time.
One way of achieving this objective is:
Ensure that the SDV vent valves open when the scram is re-set by installing redundant valves and ensure that the SDV vent lines are clear.
Ensure that the SDIV drain line capacity is sized to allow prompt drainage and install redundant SDIV drain valves.
8.
Guard Against Trapping In Individual Headers In some plants each SDY header bank consists of 2 or more header pipes connected
- together, in parallel, to form the desired volume.
The possibility that one header could fail to drain even though the other parallel header has drained should be guarded against.
Ways of achieving this include:
Ensure that the design provides an adequate slope =or all headers in each bank.
9.
Ensure That Failure To Scram Procedures Are Adeouate Procedures, adequate for guidance of operators and shift engineers in the event of failure to scram, should be provided.
These procedures should speci y under what circumstances specific measures should be taken in the event of delayed scrams or incomplete scrams.
10.
Guard A ainst Uncontrolled Release Of Reactor Coolant To the Reactor Buildin Failure to close on signal of a single vent valve or single drain valve can result in loss of reactor coolant to the reactor building sumo.
Redundant means to close off the drain and vent pathways during a scram should be considered.
INFORMATION CONTACT:
Dick 8aker (INPO)
(404) 953-7616
-or-Hi 1 es Leverett (NSAC)
(415) 855-2936
cF e
~
~ 1E