ML20093H255

Rev 1 to Final Deficiency Rept DER 84-39 Re LPSI & Containment Spray Pumps Experiencing Abnormal Rumbling Noises.Initially Reported on 840605.Deficiency Not Reportable Under 10CFR50.55(e)

ML20093H255
Person / Time
Site:	Palo Verde
Issue date:	09/26/1984
From:	Van Brunt E ARIZONA PUBLIC SERVICE CO. (FORMERLY ARIZONA NUCLEAR
To:	Bishop T NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION V)
References
ANPP-30668-TDS, DER-84-39, NUDOCS 8410160216
Download: ML20093H255 (12)

Text

m

.,,. x EcayE9 Mc

- Arizona Public Service Company.

^

3 OCT -2 AN 8 05 September 26, 1984 ANPP-30668-TDS/TRB REG!0ilVi E U. S. Nuclear Regulatory Commission Region V Creekside Oaks Office Park 1450 Maria Lane - Suite 210 Walnut Creek,- California 94596-5368 Attention: Mr. - T. W. Bishop, Director Division of Resident Reactor Projects and Engineering Programs

Subject:

Final Report, Revision 1

-DER 84-39 A 50.55(e) Reportable Condition Relating To LPSI And Containment Spray Pumps Have Experienced Abnormal Rumbling Noises.

File 84-019-026; D.4.33.2 Reference A) Telephone Conversation between J. Ball and T. Bradish on June 5, 1984 B) ANPP-29866, dated June 29,1984 (Interim Report)

C) ANPP-30304, dated August 23,1984 (Final Report)

Dear Sir:

Attached is Revision 1 to our final written report of the deficiency ref erenced above, which has been determined to be Not Reportable under the requirements of 10CFR50.55(e). This revision provides further information to the Analysis of Safety Implications.

Very truly yours, L

Ct.L.L A

N E.. E. Van Brunt, Jr.

APS Vice President Nuclear Production ANPP Project Director EEVB/TRB/nj-Attachment cc See Page Two 0

.D gn3

r Mr. T. W. : Bishop DER 84-39, Revision 1 Page Two ec:

Richard DeYoung, Director Office of Inspection and Enforcement U. S. Nuclear Regulatory Commission Washington, D. C.

20555 T. G. Woods, Jr.

D. B.'Karner W. E. Ide D.. B. Fasnacht A. C. Rogers L. A. Scuza D. E. Fowler T. D. Shriver C. N. Russo J. Vorees J. R. Bynum J. M. Allen A. C. Gehr W. J. Stubblefield W. G. Bingham R. L. Patterson R. W. Welcher H. D. Foster D. R. Hawkinson R. P. Zimmerman L. Clyde M. Woods T. J. Bloom D. N. Stover D. Canady J. D. Houchen J. E. Kirby Records Center Institute of Nuclear Power Operations 1100 circle 75 Parkway, Suito 1500 Atlanta, GA 30339 l

l

' FINAL REPORT, REVISION 1 - DER 84-39 DEFICIENCY EVALUATION 50.55(e)

{

. ARIZONA PUBLIC SERVICE COMPANY (APS)-

PVNGS UNITS 1,L2, 3

,.u I.

- Description of Deficiency During performance verification tests on modified Low Pressure Safety Injection (LPSI) Pumps and modified Containment Spray (CS) Pumps on Unit 1 during May 1984 (reference DER 83-61 for.

details of modifications), an aperiodic " rumble" was evidenced in LPSI pumps 'IMSIA-P01 and IMSIB-P01 and CS pumps 1MSIA-P03 and-IMSIB-P03 and their adjacent suction piping. In the case 4

of the LPSI pumps the " rumble" was noted in the capacity range of 2800 to 3400 gpa.

In the case of the CS pumps the " rumble" was of lower magnitude and was noted in the capacity range of 1800 to 2800 gpm.

The " rumble condition" had not been reported prior to the DER 83-61 modifications as previous operation of these pumps did not include sufficient time in their respective " rumble"-

ranges for this phenomenon to be identified, i.e. :

The LPSI pumps would normally be started at 100 gpa fixed (continuous) minimum flow against a closed discharge 4

control valve,' and the discharge valve then opened to l

permit design flow of 4300 gym (or in the case of the 100 start test - reference DER 83 maintained at a 2000 to 2100 gpa valve setting).

In the case of the CS pumps, startup would normally be at 150 gpa fixed minimum flow against a closed discharge control valve which is then opened to permit. the 3900 gpm r

j design flow.

i i

The minor modifications to the pumps (reference DER 83-61) concerned the impeller running fits and have no influence on j

whether or not the pumps do or do not operate with a " rumble" in the flow range (off peak efficiency) between minimum flow and design capacity.

i This " rumble" phenomenon is the result of interaction between intermittent (aperiodic) inlet ' flow disturbances caused by the suction piping configuration and prerotation of the inlet

)

stream caused by the pump impeller while operating off peak efficiency. The noise source is due to cavitation f rom collapsing of bubbles in the flow stream about one foot below the pump casing.

The root.cause of the " rumble" phenomonon is explained in Reference 1 with respect to the LPSI pumps. Based on the cimilarity of puc9 designs and suction piping configurations, the save root cause is also applicable to the CS pumps.

i I

1 k

^

3 Mr> T.- W. ~ Bishop.

LDER 84-39, Revision 1; Page Two Should prolonged operation in the " rumble" range ever become' an operating. requirement, - this condition can be corrected by incorporating either a splitter at the pump inlet to negate the prerotation, or st alghtening vanes in the suction pipe below the pump to reduce the inlet flow turbulence to the-pump;-

The LPSI Pumps and the CS Pumps are' supplied by Combustion Engineering (C-E) and are manufactured by Ingersoll-Rand (IR). They are identified by_ tag numbers as follows:

' Unit 1-Unit 2 Unit 3-LPSI

'IMSIA-P01' 2MSIA-P01 3tSIA-P01 IMSIB-P01 2MSIB-P01 3MSIB-P01 CS IMcIA-P03 2MSIA-P03 3SMIA-P03

'IMSIB-P03 2MSIB-P03 3SMIB-P03 II.

Analysis of Safety Implications

.During shutdown cooling, the LPSI pumps are used to reduce the temperature of the Reactor Coolant System (RCS) in post-shutdown periods f rom an RCS temperature of 350'P to the refueling temperature of 125'P. Additionally, the LPSI pumps are used to provide emergency core cooling flow following a large break Loss of Coolant Accident (LOCA) up to initiation of a Recirculation Actuation Signal (RAS).

The shutdown cooling mode flow rate is 4500 gym and is controlled by the operator. Operation through the 2800 to 3400 gpa " rumble" range will be of a transient nature and of short duration. No impact resulting from the " rumble" phenomenon is anticipated as a result of operating in the shutdown cooling mode.

In the emergency core cooling mode of operation, the LPSI system could operate in the " rumble" flow range. A Probabilistic Risk Assesment (PRA) was employed to evaluate the operation of LPSI pumps in the range of flow where the

" rumble" occurs. The probability of operating in this flow range is very low, calculated to be a mean occurrence of 4.65 x 10-4 per operating year. Operation in the rumble range will depend on the combination of pumps running, LOCA break j

size and configuration, decay heat rate, andd operation of the L

steam generators as an alternate energy removal path, (reference 2).

I e

}

l 1

,, _ + - - ~ ~,

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

1

-y.

Mr. T. W. Bishop DR 84-39, Revisf on 1 Page Three.

The upper limit of = time for which the rumble could occur, assuming no corrective operator action, would be four hours.

This is based on a maximus Refueling Water Tank (RWT) volume of approximately 700,000 gallons with only one LPSI pump operating, until automatic shut-off by the RAS. -

The RWT is the safety injection system suction reservoir prior to RAS.

C-E has stated (reference 3) that if a LPSI pump operates for at least one minute, there are no break sizes where an LPSI failure would result in significant core damage. Also, IR has confirmed (reference 3, enclosure 2, item 4) that operation in the rumble range for up to four hours would not cause pump damage.

The ~ pumps were field tested with the same system conditions as would be encountered during -a LOCA. LPSI pump 1B was run in its " rumble" range durig tests for a duration of about two hours. Post-test inspection revealed no pump degradation.

The CS pumps are designed to remove heat from the containment atmosphere in the event of a LOCA while pumping at a rate of 3750 gpa. They are also used (below 200*F) to circulate reactor coolant, at a rate of 4500 gpa, to remove decay heat durig the latter stages of shutdown cooling.

The CS pumps will not have occasion to operate lin the range where rumbles occur.

The vibration effects on the piping were evaluated and found to be acceptable. The recorded displacements were compared with the analytical results of the seismic event analysis of the subject LPSI pipig. The recorded displacements during rumble were less than the analytical seismic displacements.

The' resulting maximum stress in the pipe is less than the endurance limit.

Based upon the above, the " rumble" condition for both the LPSI and CS pump systems 'are evaluated as not reportable under the requirements of 10 CFR part 50.55(e) since, if left uncorrected, it would not represent a significant safety condition.

Also, this condition is evaluated as not reportable under the requirements of 10 CFR Part 21 since it does not constitute a substantial saf ety hazard and, 'If lef t uncorrected, would not-

+

adversely affect the capability to safely shut down the reactor.

i t

w

-e d..m-.

-.e--v.a+

,m,

,er-.e-

-- - w e-,->.e=.

+m-wy1.----,=we-_,e

,w

,-we-

,+w-ww y.- r-se-v-w

..ms m-

' Mr. T. W. l Bishop

- DEL 84-39; Revision 1 Page Four III.

Corrective Action NCRs SM-4201 and SM-4229 have been dispositioned to use the pumps as is.

No corrective action is required as a result of this condition; (reference 3). The LPSI pump operating procedures are being revised to incorporate a warning not to operate in the 2500 to 3500 flow range during the shut-down cooling mode of operation.

IV.

References (1) Root Causes of " Rumble" in LPSI Pumps, Palo Verde Project. -

(2) Probability Calculations.

(3). V-CE-30530 dated July 10, 1984 with two enclosures.-

i l

l l

t;

{

DER 84-39 i

}

Ref;rence (1)

Root Causes of " Rumble" in LPSI Pumps Palo Verde Project i

The I. PSI pumps develop noise and vibration when operating in a flow range I

of 2800 to 3400 gpm.

Based on aural observations and on accelerometer and pre ssure data obtained by the startup personnel, the cause of this " rumble" (noise and associated vibration) has been determined to be as follows:

The fact that the problem only occurs in the 2800 to 3400 gpm range

)

)

which is at approximately 75% of rated flow, strongly suggests that pre-rotation induced by back flow frcm the impeller is part of the cause.

The intermittency of the events suggests that some other source of flow disturbance is combining with the prerotation to produce the noise and vibration.

The aural observation indicated that the noise was due to cavitation which occurs in the intake pipe about I foot below the pump casing.

The cavities collapse within the water column.

In such cases, the cavitation produces vibration and sound but no metal damage.

Aural observation of the intake piping at several locations disclosed that strong turbulence develops in the flow aperiodically.

The bends, tees and reducers in the system are sufficient to generate randou, large scale turbulence.

It is concluded, therefore, that cavitation conditions develop intermittently when the swirl, associated with a burst of turbulence, interacts with the prerotation induced in the intake pipe by opera-l tion of the pump at partial discharge. Justification for this con-k clusion is as follows:

Tests conducted at the California Institute of iechnology, for pumps of about the same specific speed as those at Palo Verde, demonstrated that backflow from the impeller induced prerotation in certain partial flow ranges depending on the impeller design.

The source of'the noise was determined from aural observations and was corroborated by the vibration data. The accelerometer data showed a definite time lag between the beginning of an for an accelerometer mounted on the intake pipe and event units mounted upstream.

Cal.culation of the apparent acoustic wave speed gave values of approximately 3750 feet per second, a value to be expected.

It is apparent, therefore, that the noise and vibration starts in the intake pipe.

During early startup a strainer was installed in the spoolpiece upstream from the intake. This strainer was concentric but was placed in an eccentric reducer.

It, therefore, presented an eccentric configuration to the flow. k'hile it was in place the

" rumbles" occurred at from 2800 to 3200 gpm. After removal of the strainer, the " rumbles" moved to the 3200 - 3400 gpm range.

This substantiates the conclusion that the piping configuratian is contributind to the event and that minor changes can cause appreciable changes in results.

7

.g

-(

(.

{,

6' DER 84-39' Reference (2)

Page 1 of 2 4

_ PROBABILITY CALCULATIONS' The probability of a LPSI pump running at 2500 to 3500 gpm during safety injection is determined by the probability that a' leak would occur which would require the LPSI pump to deliver 2500-3500 gpm, and.the probability that the number and configuration of running pumps requires the LPSI pump (s) to deliver 2500-3500 gpm. The " rumble" was determined by tests to occur in the 2800 to 3400 gpm flowrange.

2500 to 3500 gpm for these calculations.. For conservatism, this range was expanded to P Rumble = P Leak X P Matching Configuration Number Running-Gallons / Minute Case LPSI HPSI-LPSI HPSI Total 1

2 2

5000 2400 7400 7000 2400 9400 2

2

'1' '

5000 1200 6200 7000 1200 8200 3

2 0

5000 0

5000 7000 0

7000 4

1 1

2500 1200 3700 3500 1200.

4700 5

1 0

2500 0

2500 3500 0

3500 A range of leaks between 2500 gpm and 9400 spm could thus cause the rumble.

The LPSI pumps shutoff at 475 feet of head (204 psi) and run out at 5100 gpm at 285' (124 psi). They will run at 2500-3500 spm in the SI mode at about 145-165 psi.

i Flows of 2500-9400 gpm at 145-165 psi would come from reactor coolant system leaks with equivalent hole diameters of 2-1/2" to 5".

Such leaks are classed as medium LOCA's.

10 ' per year, Reference A.The mean probability of a medium LOCA is 4.65 X

~

the required flow for each specific case is equal to this number.The sum of th We are conservative if we use the medium LOCA frequency for all cases.

Y The LPSI pumps run for about 40 minutes in the injection mode before the recirculation actuation signal (RAS), triggered by low RWT level, will shut them off.

This shortest run time applies when both HPSI pumps, both LPSI pumps, and both CS pumps are running.

2,500 gpm, it would take 3.2 hourt If only one LpSI pump were running at only'one LPSI pump running is 3.32 X 10to regch RAS cutoff. The probability of lower than having all pumps running.

r.'

...'c

(

c DER 84-39

,~,..,

Reference (2)

'Page 2 of 2 PROBABILITY CALCULATIONS (continued) determined as follows:The probability of having a given combination of pumps runnin

c P Running + P Failure = 1 or P Running = 1-P Failure Case Comb.

Probability 1

2L 2H (1-PL)

X (1-PH) 31 2

2L IH PH (1-PH) X (1-PL)

EPH 3

2L 011 (PH)

X (1-PL)

E(PH) 4

- lb 1H (PII) X (PL) X (1-Pil) X (1-PL) EPHPL 5

IL CH

,,(Pil) (PL) X (1-PL) EPH PL Where PH = Probability of failure to start or run (llPSI)

PL = Probability of failure to start or run (LPSI)

!! PSI MSI Failure to start 3.29 X 10 3 3.29 X 10 3 Reference A Failure to run for-one hour

.03 X 10-

.03 X 10-Reference A l

PH i

3.32 X 10 3 PL 3.32 X 10-l Evaluating the cases:

P1 = (1-3.32 X 10- )

X (1-3.32 X 10- )

E1.0 l

P2 = PH = 3.32 X 10-P3 = (PH)2 = 1 X 10-P4 = PH X PL = 1 X 10 $

i PS = (Pil) (PL) = 3.32 X 10-probability of leak is conservatively equal to the probability range where rumbles will occur is 4.65 X 10 g/yr or les l

ow years of plant operation.

Reference A - Pickard, lowe and Garrick, Inc., Generic Database for PWR's t

~ ' * ' ' - *

(

[

$ EfE12EHcef C E P!w:r Sy;t:ms TII. 203/6881911 Combustion Engineering, Inc.

Tal;x: 99297 t

100o Prospect Hill Road l.

Post Office Box Soo y *Rgg Windsor. Connecticut 06095-0500 g

pV. Y. O L POWER SYSTEMS July 10, 1984 JLA.16 %

Y-CE-30530

-I ". c Y.

Mr. V. G. Bingham u

• a-Bechtel Power Corporation P

07-12400 East Imperial Highway l_

jf,"

Norwalk, CA 90650

_ g_ g_

4i=

Subject:

Low Pressure Safety Injection (LPSI) l System Disturbance

".f mu.

s

Reference:

(A) B/CE-E-48279 dated May 30, 1984

] '[

\\ _ __

(B) B/CE-E-48429 dated Juni: 20, 1984 gg w

Enclosure:

(1) LPSI Pump Probability Risk Assessment (PRA)

~h "'

~

(2) LPSI System Noise Anomaly

{ - --

Dear Mr. Bingham:

This letter addresses the LSP1 system disturbance / noise / rumble found to occur when a LPSI pump is operated in the 2800 to 3400 gpm range. An evaluation has been made to determine if the safety functions of the system would be impactec by this rumble phenomenon. The LPSI pump / system is used to provide safe cooldown to cold shutdown in a shutdown cooling mode of operation and also may provide emergency core cooling flow in the short term (.p to RAS actuation) if there is a LOCA.

The shutdown cooling flow rate is 4500 gpm and is controlled by the operator.

Operation through the 2500 - 3500 gpm will be of a transient nature and of very short duration. No impact resulting from the rurble phenomenon is anticipated as a result of operating in the shutdown cooling mode.

In tmergency core cool'iMg mode of operation, one can expect the system to operate at the " rumble" flow rates.

It is felt that the upper bound of the time period for which rumble could occur assuming no corrective action would be four hours. This is based on a maximum RWT volume of approximately 700,000 gallons with one LPSI pump operating. " Enclosure (1) provides a more detailed analysis of required run times for various break sizes.

Operation in the rumble mode will depend on the combination of pumps running, break size and configuration, decay heat rate, and operation of the steam J

generators as an alternate energy removal path. The approximate break size range where one would anticipate flows in the rumble region fo combinajions would be breaks of 3 to 8 inch diameters (0.05 ft{ differe to 0.35 ft ).

1 The pumps have been field tested in the sane configuration and the same system conditions as would be encountered during a LOCA. One pump was run during testing in the rumble mode for a duration of about 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />.

No rump degradation was observed as a result.

..3..

c c

<Y Further, IR has advised that operation in this mode for 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> was not expected to be detrimental to pump performance or its expected life.

An analysis was also performed from a survey of LOCA analyses.

Specifics of the survey are shown in Enclosure (1). The reference to "significant fuel damage" pertains to fuel clad temperatures in excess of 2200 F.

The results of the survey show that if the LPSI pump operates for at least one minute or more, there are no break sizes where a LSPI failure would result in significant core damage, consequently, from a PRA standpoint, no increase in risk would be predicted.

Of the two assessments, the latter is considered the more realistic situation; one LPSI pump operating requirement between 1 minute and 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br />.

Reference (B) asked for Ingersoll Rand input to three questions. Their replies plus their reply to a fourth C-E question are included in Enclosure (2).

Reference (A) requested that C-E have Ingersoll Rand (IR) design and fabricate a flow straightening device for the LPSI pump suction. C-E und,erstands that this effort has been put on hold. Please be advised that IR has completed the design portion of this effort.

If you should have any questions please contact me.

Very truly ours, C. Fergu i

Proje nager CF/JDI/CDB: sic F73181 cc: Messrs:

E. E. Van Brunt, J r. w/e J. Vorees w/e W. H. Wilson R. H. Holm J. W. Dilk G. A. Butterworth S. N. Mager

'D. B. Amerine w/e W. L. MacDonald J. R. Bynum O

p-q,. c, o 7

.g

(

Enciesure.(1)to V-CE-30530 Page 1 of 2 ANPP/FALO VERDE-LPSI SYSTEM NOISE AhCMALY PRCBASILITY RISK ASSESSMENT (FRA) g5005510N i

A r.oise', in itself, of cour:,e will not contribute to risk.

The extent to J.ich the "roeble" is indicative of over-stressing of pump or ;ipir.; co.;onents is currently unknown.

The discussien belos is based on cor.sieeration of too alterr.ative cases:

Cese'1: A rumble that persists for r., ore than one minute is assuned to lea: te a co.plet': less of*p..p ficw.

l C6se 2:. A'r :ble that persists f or rere than one hcur is assumed to lead to l

co.plete loss of p np fle..

i C-E perfer.ed a survey of LOCA a Elyses in crder te dettr-ine.htch apprestrate break sizes cc.,1c lead to LPSI nna flow in the rar.;e of E0 + 500 gpm f or sufficier.t tine to cause pump dar. ige (See Cases 1 & 2 at:.e). ~The results of the sursey incicate that:

For Case 1 (i.e., one minute rumble tolerance) LPSI pu p f ailure would not be 2

predictedforbreaksizeggreaterthanabout0.2ft For breaks in the range of 0.005 to 0.2 ft, LP51 pump f ailure ceuld occur but the LPSI pur would not be needed for either ir.2,ection or pcst-LOCA shutd

,F or d

breaks snaller than about.005 f t the LPS! pump would not produce flows

(

in the 3000 gpm rangd' prior to post-LOCA shutdown cooling, in post-LOCA shutdown cooling the LPSI punp ficw can be cor. trolled to avoid the 3000 g;m l

range.

For Case 2 (i.e., one hour rumble tolerance) no break sizes would be expecteo to lead to LPSI punp f ailure. Operation in the injection r.sode is not'reautred for longer than one hour.

In post-LOCA shutdown cooltr.g the LP51 pump ficw can be cor.trelled to avoid the 3000 gpm range.

RISL IMPACT The recent Calvert Clif fs IREP assuned LOCA f requencies of approximately:

2(<.02ft}

3 x 10-2 per year 5..a11, itall LOCA, 5

Sna11 LOCA, 53 (.02 - 0.1) 3 x 10-4 r,er year large LOCA, A

(>0.1) 1x 10*4 per year l' sing thase frequencies as a basis, we can dras the follo.ing preliminary conclusions:

_.______2_

________________._._______________________.______._______._.___.._._________._-________________._.________._______________________._____m

-s

(

(

y v

Enclosure (1) to Y-CE-30530 Page 2 of 2 CASE 1 8 2 Given a cr.e r.inute, or one ho;r rur.ble tolerance, there are no break si:es 3.at wos1d result in LPSI failure lending to significant core ts.. age.

The r ef ere,'

b.ased on traditional FRA cc-titeratien, no increase ir, risk would be precicted.

Heaever, there :ust be assarance that the ;; ps can really tcic ate a

the rurtle at the ti..e that iney are de ar.ded.

To be sure of this would rquire* making sure that the ru':ble is not a s/c.pto. of continual c'.Ulathe l

degredation of the p.,"f or piping such that ficw delbery reliability is decreasing.ith age. The tests cor. ducted at PVZS :nd t'.e subsequent inspection pr0vida this assur;.nce.

F 9

e 4

e.'

se ll' 1

6 4

l l

?

I I

I i

e

. -