ML20198K944
| ML20198K944 | |
| Person / Time | |
|---|---|
| Site: | University of Missouri-Columbia |
| Issue date: | 01/08/1998 |
| From: | Mckibben J, Meyer W MISSOURI, UNIV. OF, COLUMBIA, MO |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| References | |
| NUDOCS 9801150136 | |
| Download: ML20198K944 (10) | |
Text
I
- I .g Research Reactor Cent:r Research Park Co umbia. Mrssourl 65211 iI hiephone (573) 882-4211 FAX [573] 882=3443 UNIVERSIT.Y OF MJSSOURI-COLUMBIA January 8,1998 Document Control Desk U.S. Nuclear Regulatory Commission Washington,DC 20555
REFERENCE:
Docket No. 50-186 University of Missouri Research Reactor License R 103
SUBJECT:
Report as required by Technical Specification 6.1.h.(2) of an unanticipated reactivity insertion while the reactor was operating at full power Introduction The reactor was shutdown at 1215 on December 9,1997, by a high power scram initiated by the Channel 4 wide ange moniter of the Nuclear Instrument System.
The power icvel increase was indicated on the chart reconiers for Channels 4 and 6 (power ranges) and Channels 2 and 3 (intermediate ranges). The cause of the high power scram appears to be an unanticipated reactivity insertion of between 0.0009 and 0.0013 Ak/k. The power levelincrease was approximately 14.4% to a calculated thernml power of11.2 MW Primary and pool coolant flows, temperatures and pressures were all normal. None of the limiting safety system set points (LSSS) were reached. An unanticipated significant change in reactivity is an abnormal occurrence as defined in Technical Specification 1.1.
Descrintion end Evaluation On December 9,1997, at 1215 the reactor was shutdov n by a high power scram initiated by the wide range monitor (Channel 4) of the Nuclear Instrument System.
Before the scram, the reactor had operated at full power for 17 hours1.967593e-4 days <br />0.00472 hours <br />2.810847e-5 weeks <br />6.4685e-6 months <br />,58 minutes, following a normally scheduled maintenance period on December 8,1997. The nuclear inetrument power level readings taken 15 minutes before the scram showed Channel 4 indicated 104%, Channel 5 indicated 100% and Channel 6 indicated 104%.
The calorimetric for thermal power was 9.81 MW. Technical Specification 3.3 requires a high power seram at or before 125% of full power (10 MW). Tecimical
. Specification 3.4 requires a high power rod run-in at or before 115% of full power. The scram set point for each channel was conservatively set at 119%, and the rod run in .
. for each reactor channel startup routine checks. was ect at 114%. These were checked on December 8 a 99011W.36 900108 1 PDR ADOCK 05000186 8 PDR sl.la.lil.lm. ..l[lf.lrf, J
k .n.w w umn.=*m
. i Ider to Document Control Desk !
l January 8,1998 !
- Page 2 j e
)
. Ehe reactivity insertion was not believed to be associated with the control blades - l because they were not being shimmed at the time of the scran The regulating blade :
i in automatic control, prior to tfw Acram, was slowly moving out about 2 inches / hour ;
to compensate for xenon buildup in the core. %e regulating blade position indication, !
after the event, was abot 10.5 inches lower than the previously logged position 15 minutes earlier. It would have driven in responding to increased power level indication ;
on Channel 4. The agulating blade could travel 0.5 inches in 0.75 seconds.
Control room operators carefully logged relevant indications on the Unscheduled Shutdown Report for this event (#1072). A power spike was indicated on the chart
recottlers for power range Channels 4 and 6 and for intermediate range Channels 2 and 3 [see Figure 1, Relative Locations of Nuclear Instrument Detectors). The wide range monitor (Channel 4) was the only channel that had initiated a scram. The scram at 119% corresponds to a 14.4% increase in power, e.g. 119/104 = 1.144. It j
. i also indicated a rod run-in. Power range Channel 6 had only a rod run in indicated. .
Power range Channel 5 had no scram or rod run in indicated and no power spike indicated on its chart recorder (Channel 5 power level indication prior to the event _!
was 3 to 4% less than the' power level indication on Channels 4 and 6. This may 1 secount for its not initiating a rod run in or scram). Reactor temperatures, flows and 1 pmssures were all normal and had indicated no change. 6
- .Controi room operaiors assessed the sample evolutions that had occurred shortly before the scram. The pneumatic tube system was shutdown and a sample was loaded into reflector position R2 within 15 minutes of the shutdown. It was not 4 obvious that, either could have been the source of the event. .
Shortly after the shutdown, electronic instrument pmblems were investigated but ,
were not considered likely causes. The DC power supplies (2PS1 and 2PS2) for NI Channels 3,4,5 and 6 (General Electric) were tested and found to be stable. Channel 2 (Gamma-Metrics)is powered by 115 VAC directly from the UPS, as are the DC power supplies for the other NI's. The remainder of the reactor instrumentation and ,
control systems are also supplied by the 115 VAC from the UPS. This power source i if disrupted would likely have caused spurious indications on other reactor
, .instrumentstaon had it malfunctioned.
Lone of the first determinations made was that no safety limit was exceeded. In fact, no limiting safety system set point (LSSS) as defined in Technical Specification
. ; 2.2 was reached. The maximum thermal power level reached was calculated to be 11.2 MW. This is approximately 1.3 MW below the LSSS for reactor power and 17.6 MW bek,w the safety limit for the normal operating values of reactor flow,-
pressure and temperature as provided in Hazards Summary Report, Addendum 4,
- p. F.4, Raraty TJmit Ann 1vais for MURR.
he cause of the power spike was believed to be an unanticipated positive. '
reactivity insertion. The magnitude of the reactivity addition appeamd to be at least L0.0009 Ak/k,' but less than 0.0013 Ak/k. HSR, Addendum 5 (p.17) states " step 4 reactivity insertion of +0.001 will result in a prompt jump of 16.67% followed by a ,
lt , stable period of 63.8 seconds."; Equation No. 6 64 from Nuclear Reactor Annivsa
,. ~.
- 4 a~ _ a___ -_. __ . __ _ _.___o
i 14tter o Docunwnt Control Desk January 8,1998
- 1
, Page 3
- (Dpderstadt, p. 251) pmvides an approximation for the prompt jump expected from a '
j
. step reactivity : >!dition ('ar a 0.001 addition):-
" o, 3 _ ,ooi
= 1.15 0.e.1%)
"h_
The promptjump for a step insertion of +0.0013 is about 21% which is a magnitude t
we believe would have caused all power range channels to scram. Because there was no indication of pov/er overshoot on Chanral 4 while Channels 5 and 6 did not indicate ' ,
a serem, we believe the magnitude of the readvity insertion was approximately
- +0.001. -We belle'. the promptjump was just enough to cause the Channel 4 scram and the other powei range channels aever reached their scram tdp set points. The -
Channel 6 recorder did not indicate as high a spike as Channel 4, and may be due to -
- its detector being located farther from the core. The period scrams did not trip 3 i
becauso the period (rate of change) circuite do not have response times as fast as
' level cinuits and the indicated power level on Chennel 4 in this case was only 15%
from its trip set point. i The investigation of the source of the unanticipated reactivity addition initially ,
. focussed on three possible mechanisms: 1) a fuel element resenting while the reactor was at power; 2) the flux trap reinserting several inches while at power ifit had not been propuly latched; and 3) a failure of a flux trap sample that could cause voiding
- in the center tube hole, which has a positive void coefficient. Other scenarios were considemd involving the evolution of gas or airin the graphite afle: tor region that .
- could pass through the md gaps. An increase of neutron coupling to the detectors due to a void in the reflector could be a source ofincreased neutron power level indication without an actual reactivity addition. The plausibility of the graphite reflector scenarios causing the event was considered small because the reactivity worth is small outside the beryllium reflector.
The plausibility of the graphite reflector scenarios to introduce positive reactivity was dependent on whether or not the void coeflicient of the rod gaps between the pressure vessel and beryllium reflector la positive (when the blades are withdrawn and not occupying the rod gap), The reactor physicist was asked to model the reactor to determine the sign of the rod gap void coefficient.
The evaluations and determinations are summarized below by category:
- 1) An unaanted fual elament resentina while at nower '
This scenado requires several fuel elements to be unseated to achieve a positive reactivity insertion ofgreater than 0.001 Ak. One of the last steps of the refueling '
- procedure (SOP II.2.2)is a veri 6 cation that all. elements are seated. This involves use of a remote toolindexed to the upper bridge. The Senior Reactor Operators i as_sociated with performing this task during the refueling before the event were certain they had performed this verification. Prior to the com defueling, after the .
event, the fuel elements vyre veriSed to be seated. Subsequent review of the drawing i W -swh., a+v'r a gr. agrygv -- p m ,us, e ,-ip,p - m vp b. ,7 v,- qu- ?=- .y g. p % -.
op--- -
14tter to Document Control Desk January 8,1998 Page 4 of the fuel support spider (MURR #56) shows the maximum distance an element could move fmm unseated to seated is 0.3 inches. At the rod height when the event occurred,0.2 inches of rod bank would be mquired to insert greater than 0.001 Ak, This would be equivalent to all eight fuel elements moving 0.2 inches. It is unlikely a single element unseated could have initiated this event. Another reason this scenario I
was not likely is because the force on each element due to primary coolant flow is approximately 150 pounds / element. This force far exceeds the force needed to seat a fuel element with the fuel tool. l I
- 2) h finr tran ==mnla hairiar innardna while at rower _
I This scenario would require the flux trap holder to not have been latched, which is ,
required by SOP VIII.2.3. The Senior Operators who performed this task wem '
certain tl:e flux trap had been pmperly inseded and latched as per SOP. Subsequent ;
- eview of the reactivity that could be inserted by up to 4 inches of flux trap sample '
holder inwani movement indicated this was not the source of a mactivity insertion
- approaching +0.001 Ak/k. Four inches was chosen as the test height difference !
because it is a distance greater than the length of the flux trap holder latching ears.
Even with the flux trap sample holder raised up 4 inches, the sample holder exttnds below the core bottom. Therefore, any reactivity inserted by inward movement.of the sample holder is limited to the ralative variation in the worth of samples over a given distance (i.e., it is not the reactivity difference from displacement of water by solid l samples). l l
4 3) Failure of a flux tran aamnle can. creating a void in the center test hole This scenario was exhaustively pursued becaus,e a failure of a flux trap sample was a likely source (the center test hole has a positive void coeflicient). All flux trap samples were unloaded and visually incpected. When an obvious failed experime nt :
sample could not be identified some were examined with binoculars and some with an underwater camera. All flux trap samples were then transferred by cask to the facility hot cell for leak testing (the cans were submerged in an alcohol bath in a bell jar and then vacuum was applied to the jar). When this technique did not reveal an !
obvious rample can failure, additional methods of testing / inspection were used. All multicarrier cans (aluminum seal welded 1 inch (nominal) diameter can with multiple 1/4 inch (nominal) diameter seal welded cans inside) were opened to look for evidence ,
of water. All other cans that were not opened were heated to 100'C on a hot plate in the hot cell to see if they off gassed or showed evidence of a leak. Only one sample can (a sulfur target) indicated a small leak. No sample can in the flux trap was found to have failed in a manner that could intmduce enough void in one tube of the flux trap
! holder to approach an insertion of +0.001 reactivity.
- A gas bubble greater than 50 cc.would have been required at the peck flux position in the flux trap to have created a positive reactivity addition of 0.001. The '
~ total volume of a 2 inch long,1 inch diameter sample can is about 26 cc. The one
' sulfur sample that was found to be leaking would have had much less free space -
inside the can (approximately 8 cc).. j The water volume of a flux trap sample tube that could be voided is limited to the p
^
annulus of water that surrounds the 1 inch diameter cans (i.e., the volume of water ,
i
< l' 14tter to Docament Contro1 Desk i Jar.uary 8,1.998 j Pase 5 ;
. {
between the outside diameter of the can and the inside diameter of the flux trap -l sample holder tube). This volume is approximately 50 cc, but is not concentrated at i the peak flux position that would produce the maximum reactivity for the minimum i void. The reactivity associated with this void distributed'over the length of a flux trap - ;
tube is less than half the reactivity insertion required to initiate this event. !
- While continuing the evaluations of the reactivity insertion, the Reactor Action ~ !
Subcommittee meeting was scheduled for 8:00 a.m. on December 10,1997. ;
Enhanamittee Meetina j i
On December 10,1997, the Reactor Action Subcommittee (a special quorum of' l the Reactor Safety Subcommittee) was convened to inform them of the apparent j reactivity insertion event. The subcommittee members were T. Storvick (Chairman, Reactor Safety Subcommittee), W. Miller (Chairman, Nuclear Engineering Program), i J. S. Morris (Senior Research Scientist at MT mR), J. C. McKibben (Associate Director) and W. Meyer (Reactor Manager). Ce , Reactor Physicist, Operations Engineer and Shift Supervisor were present and participated in the discussion.
The subcommittee discussed the event, potential causes, and the investigations i completed pdor to the meeting. The estimated magnitude of the mactivity insertion :
was discussed as well as the relative magnitude compared to Technical Specification i limits for experiment reactivity.
The subcommitteo deliberated on a conservative path to follow to rs3 tart the ;
- mactor. The path decided upon included additional investigative steps and an administrative step to reinforce the operator's authority to immediately shutdown :
the reactor if any abnormal conditions or indications occurred. The path recommended included the following steps: '
- 1) Examine the R2 reflector can and samples that was inserted in the graphite mflectcr shortly before the scram for evidence of failure.
- 2) Start up the reactor without the flux trap sample holder to check base reactivity ;
associated with the fuel, the control rods and the reflector.
- 3) If the base reactivity is satisfactory,i.e., the initial critical position is close to ECP, start up the reactor with the fully loaded flux trap sample holder installed but with the following samples,'which were in on 199/97 removed: the sulfur target that was known to have leaked; one other sulfur target that had been .
. installed on the Monday 12/8/97 startup; the contents of multicarrier sample can ;
.- #54; and the bomn nitdde-shielded can that had been installed on Monday. The bomn nitride-shielded can is the sample with the highest individual mactivity that ;
isloaded in the flux trap.- !
- 4) If the mactivity balance with the flux trap sample holder inserted meets ,
specifications, take the reactor to 10 MW. !
- 5) Provide training to crews regarding conservative decision making. Reinforce to 1
- the operators their authority in shut down the reactor immediately if any *
- abnormal conditions or indications occurmd.
9 ,
a
_m. __.m__ . _ . ..
Letteri Document Cont. . Desk January 8,1998 Page 6.
The subcommittee recommended additional investigation into Channel 5 to determine why it did not indicate the transient on the recorder and decided to reconvene at 1600 to review the findings.
The actions recomniended were completed by 1500 on December 10,1997. The Actior. Subcommittee reconvened at 1610 on that date to review the results and follow up from their earlier recomm.mdation which are summarized here.
- 1) The reflector sample (R2) was opened and inspected and no failure of the experiment had occurred.
- 2) The reactor was started up at 1200, December 10, with the strainer cover in place of the flux trop sample holder in the center test hole. The critical position for this startup was 0.2 inches above the ECP, but well within the administrative allowance in SOP I.4.3.G of 0.7 inches.
- 3) The flux trap sample holder was loaded as described earlier and installed for the second reactor startup. The reactor critical position was 0.4 inches above the initial ECP (or an additional 0.2 inches above the ECP for the strainer startup).
This critical posit. ion is within admin.strative limits of SOP I.4.3.G. (For informaticn, the 0.2 inches difference in the control blade ECP corresponds to about 0.0012 Ap. The administrative limit 0.7 inches corresponds to about 0.0044 Ap in the rod height range 17 inches to 18 inches where the reactor went critical).
- 4) Reactor power vas taken to 10 MW at 1439 on December 10,1997, with normal conditions except some random bubbles emitting from between the #2 and #3 graphite wedge (south side of the reactor) during the first 10 minutes after startup. [See Figure 1] The reactor has operated with no other indication of an unanticipated reactivity insertion event recurring since the December 9 ovent.
OtTgassing from between the #2 and #3 wedge has occurred for about 10 minutes following each of four subsequent reactor startups since December 10 The bubbles do not appear to affect reactivity or nuclear instrument indications.
They do produce a slight elevation of the gas monitor channel of the stack monitor due to Ao41.
- 5) The Reactor Manager provided the recommended training to the shift starting up the reactor on December 10. This training was augmented by a written letter to the control room operators reinforcing the policy that they are authorized to shutdown immediately and secure the reactor if a non normal reactor condition ar '
indication was detected.
Further testing of Channel 5 with simulated input changes produced normal recorder pen response.
The subcommittee asked the Reactor Manager to keep them updated regarding further reactor operations.
14tbr te Document Contr61 Desk -
= January 8,1998 ;
Page 7 >
n- :. l M
i After corddering a number of reactivity addition scenarios,it appears that l
- failum of one of the south graphite wedges (#2 or #3) could account for the indications ;
observed during the event. 1 A number of reflector scenarios were postulated and discussed by the Action l Subcommittee. Their plausibility depended on the rod gaps having a positive void !
reactivity coefficient. This was verified after MCNP modelling was completed i December 29,1997; The rod gaps were calculated to have an average positive l' reactivity c; efficient of about 1/5 that of the center test hole.
- A possible cause of this event could be a massive void or gas bubble fmm a failed :
wedge on the south side of the reactor. This event could cause an increase in the :
L neutron current or coupling to the south side detectors (Channels 2,4 and 6) which could cause them to indicate a step increase in neutron indication without a positive :
. ~ L reactivity addition._ This could account for little or no recortled increase in Channel 5 ;
- (north side), but was initially ruled out because of the power spike recorded on i L Channel 3 (also north side). A positive reactivity insertion of some magnitude would ;
have been required for Channel 3 to spike from a failure on the south side of the s reactor, although a scenario postulating a positive reactivity insertion of considerabl/
less than +0.001 Ak/k is possible. The reason for the exaggerated response on the ,
south side nuclear instruments would be due to changing the neutmn coupling to l these detectors by replacing the water between the reactor and the detectors with gas bubbles. This could account for a considerable portion of the Channel 4 power .
levelinemase. We are currently modeling this scenario to quantify the non reactivity :'
4 related contdbution to the indicated powerincrease.
The graphite wedges at MURR are sealed in aluminum cladding. The original !
wedges have all been replaced over the past 15 years and many wem leaking during -
operation. The failure mode previously documented involves slow leaks, indicated by streams of small bubbles during full power operation with the heaviest bubble stream ;
occurring immediately after reaching full power. The mechanism for this bubbling appears to be that a hole or crack has formed that allows for flow into and out of the [
c wedge. During full power operation there is gamma heating inside the wedges of >
around 1 watt /gm for the portion nearest the beryllium reflector. The wedges will cool L down right after a reactor shutdown and this will draw pool water into the wedge. This ,
t ' water has absorbed air 'which includes Ar-40)in solution in equilibrium with the pool !
- surface temperature (105*F). When the reactor is approaching full power, the gamma heating raises the temperature of this contained pool water causing it to -
degas. The bubbles are heaviest dudng this initial heating while the wedge and the enclosed pool water are approaching the higher equilibrium 10 MW temperatures, s
. One 16 sue in this scenaria which is different from previously documented wedge leaks is them was no indication of elevated Ar-41 gas indicated on the stack monitor >
during or after this event. This may have been because the wedge had been back
- filled with helium when it was fabricated. The absence of off gas activity may be i consistent with the initialleaking of a canned graphite wedge. The stack monitor
~: 1 l- t 1
.- . -. .- -_ - --..- .- . - ~ , - . _ . . _ _ . - . -. - . ... .. - . - .
- Letter to Document Cor.th! Desk .
' January 8,1998 - :
Page d l l
j indicatior) accompanying previously documented leaks may only occur after the j
, . initial event. ;
Original HSR, Section 13.2.3, Boiling Within Other Ragians. concludes that boiling :
of water in the rod gaps "with expulsion of all the water amund the beryllium is -!
calculated to result in the insertion ofless than a prompt critical amount of i' reactivity." The release of gas from a 30' graphite wedge would not likely affect more than the equivalent'of one quadrant or one rod gap of the reactor. Any further event like this one, which has occurred only once in 30 years, would thus be selflimiting to a reactivity insertion ofless than +0.002 Ak/k. ]
FaHmr.Up Actions i One of the first determinations made was that no safety limit was exceeded. In fact, none of the limiting safety system set points as defined in Technical i l Speci6 cation 2.2 were reached. i An exhaustive search for the cause of the apparent positive reactivity addition was initiated. The Action Subcommittee of the Reactor Advisory Committee was .
convened on December 10,1997, to review the investigation of the event and to I recommend a conservative path to restart of the reactor. These subcommittee .
members continue to review ongoing investigations as to the cause. The parent ,
committee, the Reactor Advisory Committee, was apprised of the event at the ,
. December 16,1997 meeting. The parent committee approved the actions of the
' Action Subcommittee. i Another early evaluation of this event, when experiment failure was considered a leading source of the reactivity addition, was to evaluate the magnitude of the i reactivity addition. The evaluations have concluded that a minimum reactivity 1 addition approaching 0.001 Ak/k was required to create the 14.4% power increase -
that caused the Channel 4 scram. To add perspective, the maximum possible reactivity insertion was bounded by the Technical Specification reactivity limit for an unsecured experiment (0.0025 Ak from T.S. 3.1J) and most likely was equal to or
- slightly greater than the Technical Specification limit for a movable experiment (0.001 Ak fmm T.S. 3.1.1).
The scenario of a failum of reflector wedge #2 or #3 may be difficult to absolutely verify, but appears to be the likely candidate by elimination of other possible sources.
Evidence of a breach of wedge #2 or #3 was found on the first startup to full power -
c.fter this event.-- If one of these wedges breached and released gas, the magnitude of the reactivity addition may be considerablyless than 0.001 Ak/k. A significant part of the 15%l change in Charmel 4 could be due to increased neutron current to this and the other south side detectors caused by a mass of bubbles from a wedge rupture.
. (For example, the design void in wedge #3 is approximately 4,000 cc.
This scenario was postulated early in the search for the cause of the December 9 ,
shutdown, but its plausibility awaited results of MCNP computer modelling of the L control blade gap completed December 29,1997.
k3 j g
.4..m e-- , 4 ,, e ----. ..- & ,-w--r ,-,_p i ,.,,s m p. , -e , ..w-,,,, .s. e-4 < , .ye .-- , ,,.w,-m-,m4 - , , , - v,m-h,
Letter to Document Control Desk January 8,1998 :
Page 9 Another possible soume of gas bubble generation in the rod gap is the release of helium gas trapped in a boral control blade. This is not a likely source of a large volume of gas, but we will investigate nonetheless. The control blades and their dnve >
mechanisms will be pulled and inspected during the maintenance periods in January and February to remove any doubt.
No new potential mechanisms for inserting positive reactivity were identified that would exceed the bounds of the current safety analysis. Another wedge related failure similar to this occurrence would be limited to less than the reactivity limit set for an unsecured experiment.
Sincerely, ENDORSEMENT:
Reviewed and Appmved J }hk '
n w
Y Walt A. Meyer Jr. J. Charles McKibben Reactor Manager Associate Director xc: Mr. Alexander Adams, Jr., USNRC Mr. Tom Burdick, NRC Region III Dr. Ehdne Charlson Dr. Edward Deutsch Reactor Advisory Committee Reactor Safety Subcommittee
[ , lhL } { 1( (l, Vjilh CHRISTINE M.ERRANTE Notary Public-Notary Seal STATE OF MISSOURI Boone County My Commis:wn EgL" s: Apd! !4. Im
I
\
f i
\
re
- o. '
=
se ,
IluCLIM pet 1JherTS NUCLEM BISTuftIDfig mes-1 N
NO PLAN VIEW Figure 1. helative Locations of Nuclear Instrument Detectors, h.' .