ML20058P518

Accident Sequence Precursor Program Event Analysis 323/87-005 R2, Loss of RHR Cooling Results in Reactor Vessel Bulk Boiling

ML20058P518
Person / Time
Site:	Diablo Canyon
Issue date:	05/11/1990
From:	NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD)
To:
Shared Package
ML20058P504	List: ML20058P502 ML20058P513 ML20058P518 ML20058P522
References
NUDOCS 9008170142
Download: ML20058P518 (9)

Text

. .- .. -m mv :n maravr  ;

I PRELIMINARY 5/11/90 ACCIDENT SEQUENCE PRECURSOR PROGRAM EVENT ANALYSIS I E R N o.: 323/87-005 R2 )

Event Desiripdon: Loss of RHR cooling results in reactor vessel bulk boiling l Does of Evenc March 10,1987 Plant: Diablo Canyon 2 1

Summary:

i During the first refueling outage, the RCS wu drained to mid loop to facilitate the removal of the SG primary manways for nonle dam installation prior so 50 work. As a result of a l leaking valve during a penetration leak rate test, RCS inventory was lost. De resulting ]

low RCS level caused vortexing and air entralament and loss of both RHR pumps. RHR l cooling was lost for ~1.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />, during which balling ' occurred. After determining that the SO manways had not been removed, the RCS was flooded by gravity feed from the RWST 1 and an RIIR pump restarted. i i

ne conditional core damage probability point and-aw for this event is 7.$E 5. His valua is strongly influenced by assumptions concerning the operadon staff s ability to implement j non proceduralised recovery actions.

Event Descriptiont On Al"il 10,1987, the RCS was drained down to mid loop to facilitate the removal of prunary SO manways for norzle dam installation prior to SG work. De plant was in the seventh day of the first refueling outage. RCS temperature was being maintained at ~87'F.

Local leak rats esseing of containment building penetrations was also being performed.

! Temporary reactor vessel water level indication was being provided by a Tyson tube manometer inside containment and two level indicators in the contml room, ne level alarms on the reactor water level indication system (RVRLIS) had not yet been reset to alarm at the mid loop low level setpoint of 107'. i Reactor vessel level was being varied by draining to and feeding from the RWST via valves 8741,8805A, or 8805B, as appropriate.12tdown was fmm the RHR pump discharge via )

valve HCV 133, and charging was by flow from the VCT via the normal charging path (through a non operating centrifugal charging pump). Once the RCS had been drained  :

down to raid loop (107'), level was being maintained by balancing letdown flow and i makeup (chargmg) flow with the aid of VCT level changes. The allowed level range was j from 107'0" (below which RHR pump cavitadon was expected due to vortening and air entrainment) and 108'2" (at which water could enter the charmel head areas of the steam  !

generators).

9008170142 900815 PDR ADOCK 05000323 S PDC

- .. n ..... .w w - -- - - - -- m --- - - -

l PRELIMINARY 5/11/90

i RHR pump 2 2 was in service providing flow through both RHR heat exchangers (the j trains were cross tied). RHR pump 21 was operable but not in service. All RNR system

' instrurnentation was in se:vk*, )

f Addidormily: i

!

• 1he safety injection pumps were electrically isolated but available for service, if I

manual operstion of valves was performed.

• Centrifugal charging pump (CCP) 2 2 was operable and available for immediate

servios. CCP 21 and the nonsafety related posidve displacement charging pump were tagged out but were available for service. i

• he refueling water storage tank was available with level at appoximately 97E All four accu.nulators had been cleared and dmined.  !

All four steam generators had a secondary side water level of approximately 73%,

with the generators ventou to atmosphere through the open secondary pressure relief system.

• All core exit 16ireples had been disconnected in preparation for reactor vessel head removel. >

l 1

• The contsinment equipment hatch and personnel air lock were open. The emergency personnel hatch was closed. Various jobs were in progress inalde of containment, and a continuous purge was in progress with the containment vendladon exhaust fan discharging in the plant vent.

l At approximately 2010 hours0.0233 days <br />0.558 hours <br />0.00332 weeks <br />7.64805e-4 months <br /> a plant engineer emered containment to begin draining a containment penetration to conduct a local leak rate test. De penetration had been previously isolated, but one of the isolation valves did not properly seat. De plant engineer did not notify the control room that he was draining the penetradon. Due to the leaking isolation valve, a drain path was created between the VCT and the reactor coolant day tank (RCDT). VCT levelimmediately began to decrease. De operators attempted to ,

restore VCT level by increasing letdown flow to the VCT. This action resulted in a slow decrease in the reactor vessel water level, as indicated on the temporary RVRLIS.

Due to the apparent loss of inventory from the RCS, plant operators isolated charging and letdown flow paths at approximately 2122 hours0.0246 days <br />0.589 hours <br />0.00351 weeks <br />8.07421e-4 months <br />, ne resulting loss of flow to the VCT '

caused the VCT level to decrease rapidly. De decrease in the level in RCS stopped at an indicated level on the RVLIS of 107'4".

m ,

. PRELIMINARY 5/I1/90 At 2125 hours0.0246 days <br />0.59 hours <br />0.00351 weeks <br />8.085625e-4 months <br /> control room operators nodced that the amperage on the 2 2 RHR pung began to fluctuate ne pi.mp was shut down, and RHR pump 2-1 was started. Amperags on the 21 RHR pump alm fluctuated and it was shut down. Plant operators suspected vortexing or caviwian of the pumps as the cause of the pump motor amperage fluctuations.

At this point both RHR pumps wese stopped, RHR oooling capability was lost, and RC5 heatup began. Since tht, core exit thermocouples had been decoupled in preparadon ibt subsequent reactor head removal, no RCS temperature indicadon was available to the plaat operators.

Since the apparent vortexing or cavitation of the RHR pumps was unexpected, plac operators suspected the validity of the temporary RVRLIS indicadon in the control room, and an operator was dispatched into the containment building to verify level indicadon on the Tygon tube manometer which was being used for RCS level indicadon inside containment.

'Ihe Shift Forsman, being uncertain of the status of acdvides involving the removal of primary side manways on the steam generaton, requested that the status of this work be verified. This was necessary to assure that no personnel were inside or in the vicinity of the steam generator channel heads or manways before he opened valves in either of two paths to allow gravity flow of water from the RWST to the RCS.

At appmximately 2210 hours0.0256 days <br />0.614 hours <br />0.00365 weeks <br />8.40905e-4 months <br />, the control room recorder for the temporary RVRLIS began to show an increase from 107'4". (Plant operators subsequently, at approximately 2241 hours0.0259 days <br />0.623 hours <br />0.00371 weeks <br />8.527005e-4 months <br />, attributed the indicated increase in RYRLIS Indicadon to steam formadon in the reactor vessel head area.) Eleven minutes later, the control room operators received notification that the Tygon tube manometer inside containment !adicated a level of between 106'9" and 10T0". At this time an attempt was made to restart RHR pump 21. The pump was immahely shut down due to amperage fluctuations.

At approximately 2241 hours0.0259 days <br />0.623 hours <br />0.00371 weeks <br />8.527005e-4 months <br />, the control room operators were notified that the steam generator manways had not been removed, although bolts s,ecuring some of the manways had been de tensioned. Valves west then opened from the RWST to establish makeup to the RCS. 'Ihlrteen minutes later, with RCS water level indicating 1117", plant operators successfully restarted RHR pump 2 2. Shortly following the pump start, the RHR pump discharge temperature on the control board recorder rose to approximately 220*F. Within 5 minutes, the pump discharge temperature had dropped to less than 200T.

Event Related Plant Information RHR Design. The Diablo Canyon 2 RHR system consists of one suction pipe which 3 draws water from one RCS hot leg, two RHR pumps, two heat exchangers, and return  !

lines which direct cooled water back to the RCS cold legs. At Diablo Canyon, water is

j wa wnx ne n5gg .w l

l PRELIMINARY 5/11/90  !

i j normally retumed to all four cold legs.

RCS Level Indication and Contml. When the RCS is partially drained. water level is measured by making two -a-daas to the RCS and determining a pressure difference, no first connecdon is an RCS drain on the crossover pipe of Loop 4, and the second is at the top of she pressuriser. Two types of level instrumenudon are used - a Tyson tube for I

local level ladicadaa and two difflerendal pressure transmitters which display level in the control room on a recalibrated and relabled accumulator level Instrument. He level

, observable in the Tyson tube was assumed so be RCS level. De Tyson tube rnanometer in l

use dudng this event suffered form a number of deficiencies:

the tube was of small diameter (which slowed response) and its installation was poorly controlled.

• the level of interest was in a high radiadon ares and was difficult to read. '

• the Tygon tube was marked with a marking pen at approximately one foot graduadons. Water level had to be estirnated by alghting structural elevadon markings and transposing by eye across available cat walks, etc. to the Tyson tube.

RVRLis level indicadon is influenced by RHR flow, the extent of air entralament and temperature differentials. level indicadon in the Tyson tube was further impacted by the small diameter of the tubing, which introduced algnificant delays in response. De udlity estimated that two inches was added to indicated RVRLIS level by pinnping 10% entrained air at 3000 gym RHR flow.

l RCS drain down in preparation for 50 maintenance requires very close control of RCS level. Rapid draining of SG tubes requires RCS level be maintained below 107'5.5" but above 10T3.5", at whkh vortexing in the vicinity of the RNR suedon piping naamion is fully developed with an RHR flow of 3000 gpm (Wesdnghouse calculation). At 1500 gpm, vortexing is fully developed at 10T1.2".

f"'nre Hw? Bulk boiling was estimated to have occurred 4$ minutes after loss of RHR. '

This was twice as fast as indicated in information available to the operators at the time of l the event. Since the RCS was essentially intact, little inventory was lost, and it has been ,

concluded (NUREG 1269," Loss of Residual Heat Removal System") that the core would have remained covered for an extended period of time because of condensation of steam in the SGs. If the SG primary manways had been removed at the time of the event, thereby providing a vent path for the RCS, time to core uncovery is estimated to be 1.6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> after initiation of bolling, or 2.4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> total.

KHR Recoverv and Sunnlemental RCS Makeup. Diablo Canyon plocedure OP AP 16, <

Rev. O, " Malfunction of the RHR System," provided no informadon specifically conceming loss of RHR during mid loop operation. General guidance was provided for loss of RHR with the reactor head in plar c (repressudre the RCS with the charging pumps.

. ~ .

m PRELIMINARY 5/11/90 start a seactor coolant pump or establish natural circulation and utilize the sos for decay heat removal).

' For this event, the RWST was full and had been used earlier to provide RCS makeup water. In addidon, the SI pumps and charging purnps could be used for RCS makeup.

Analysis Approach:

Care Damage Model. The core damage model considers the possibliity that the loss of RHR could have occuned either with the RCS intact (which was the case during the event) or with the RCS vented to the containment through openings such as the 30 nunways.

In the event the RCS is intact, core coollag is assumed to be provided if (1) water is maintained in the SGs (SGs not drained and open prior to the event) or (2) if RCS injection is provided prior to RHR pump restart using the Si or charging pumps or by gravity feed from the RWST Por item (1) to be effecdve for core cooling, steam from the reactor vessel must travel to the sos, and condensate must flow back to the vessel, as described in NUREG 1269. l If the RCS is open, then RCS injecdon prior to RHR pump restart is required to prevent core damage. The event tree model is shown in Pigure 1.

Pour core damage aquences are shown. Sequence 1 involves the case in which the RCS is open, RCS makeup is provided, but an RHR train is not vented and restarted. This sequence is considered very unlikely since the operators were aware the pumps were cavitating and tripped them (no pump damage) and, since RCS makeup is effective in the wquence, had to be aware of the need to restore RHR. Sequence 2 also involves an open RCS, but with failure to provide RCS makeup. Numerous makeup sources (RWST gravity feed, SI pumps, charging pumps) are available, and failure to provide RCS makeup '

is assumed to be dominated by operator error in diagnosis of the event. Sequences 3 and 4 are similar to sequences 1 and 2, but involve an intact RCS with unavailable sos. No core damsge is assumed to occur with an intact RCS and operable SGs. 'the time to core uncovery in this case is very long, and can be increased by adding water to the sos.

. PRELIMINARY 5/11/90 d O RCS Open RHR DudnD AvaRaNe RCS RHR Train to g 9,g g

1. Contairant n Provided Restarted End Seq.

St.e .

OK CD 1 CD 2 OK OK CD 3 CD 4 Figure 1. Core Damage Event Tme for Loss of RHR Cooling During Mid-

1. cop Operation at Diablo Canyon RCS Makenn. The likelihood of failing to initiate RCS makeup was estimated based on caw enur pmbabilldes developed Emm Time Reliability Correladons and shown in Figute

2. Four types of crew response are addressed: (1) response based on detailed operating procedures, (2) trained knowledge based performance, (3) typical knowledge based performance, and (4) knowledge-based performance during very unusual events. Figure 2 was developed hem curves appropriate to in<ontrol room medon, and the response time was skewed 20 minutes to account for recovery outside the control room. Typical knowledge based response was assumed for the event (the operating procedure provided no informadon concerning mid loop operation). For the estimated 2,4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> to care uncovery, a etcw ermr probability of 1.0E 4 is indicated.

RCS and SO Statue; '!he likelihood of the RCS being open at the time RHR was lost was assumed to be 0.5. The likelihood of the sos being unavailable was also assumed to be 0.5.

l

ItAY tt '90 10:58 SAIC 265 PAGE.08 PRElfMINARY 5/11/90 Analysis Resultst 1he esdmated core damage probability associated with the loss of RHR cooling at Diablo Canyon is 7.58 5. This value is stmngly influenced by assumptions concermng opemtor aedon during the event.

Substandal uncertainty is also associated with this esdmate. Provided the RCS was intact and the SGs were available for decay heat removal, an extended period of time was available to effect recovay. If the MCS was open,2.4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> were still available for ,

recovery. However, recovery acdons was not praedumllzed at the time of the event.

1he impact of different assumptions concerning the dme aner shutdown, the status of the RCS, and ability to cool the core using SO: as described in NUREO.1269 are shown below.

Revised Core Asanzados Damage Prnhability Event occurs two days after shutdown (time to bou esdmated to 9.8E 4 be 0.13 h, time to cost uncovery with open RCS esdmated to be 1.0 h.)

SO manways removed. 1.0E 4 Natural circuladon coollag using SO ineffective. 1.0E 4

s .

o .......eti e I

s.e -

1 I

..i -

)

o..: .

t i

.. i -

u n w u.t e.4 (ica.w k A p . l. O

~

.1 0.e en -

T3 pl<~t ve nt. (u n.wt< 4 6 .

o r.,

s e.. u

' Tmnc A c ut ( t.w ale dy .L.a -;

g i 80 i too tee, "I

W 6Wk%

Fi 3e t.

Peaba.Wlih eR ec.+ I ., pt, m e,,An Res rn a q f

_ m _ _ _ _ . _ _ _ _ _ _s.,_, ,. . a m ., . . __e., .-.w. _ ., - , - - _ , . -, 2-.- -.m,.-~w. -- .--w.-

I o y .

_ . . _ _ -. - 1 i

M--. .--

TELECOPY TO: M d. MCAALM vvu ~

0 AT: L3O >

1 FROM: LQS a@ , .-

et 11 L e s

AT: DIVISION 275, OAK RIDGE (ITG)  ;

l FAX (615-482-6745) ,.

FAX: ,

VERIFY: -

1 CHARGE NO: S69 ,

f NOTES 6}\\lQO Qu & w & M SA' $ i

( O

T9 \ Q 89 i

l i

I 4

L-

~ ~~~^~']___ _

.__~,g.

4

- . _ _ _ _ . . . . . _ _ . - _ _ _ _ _ _ . . _ . _ _ _ _ _ _ . . _ _ . - . . _ , _ . _ _ _ _ _ _ _ _ , , . , _ . . . . , _ _ _ , _ , . , . _ . . - _ . _ . . _ _ _ _ _ . - _ , . . _