ML20204J614

AEOD/E807, Pump Damage Due to Low Flow Cavitation

ML20204J614
Person / Time
Site:	Susquehanna, McGuire, 05000000
Issue date:	08/31/1988
From:	Caroline Hsu NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD)
To:
Shared Package
ML20204J591	List: ML20204J584 ML20204J614 ML20205L228
References
TASK-AE, TASK-E807 AEOD-E807, NUDOCS 8810250191
Download: ML20204J614 (18)

Text

l AE0D ENGINEERING EVALUATION REPORT PUMP DAMAGE DUE TO LOW FLOW CAVITATION August 1988 Prepared by: Chuck Hsu 8010250191 881012 PDR ADOCK 05000369 P

PDC Y

TABLE OF CONTENTS Page 1.

S UWA R Y............................

1 2.

INTRODUCTION 2

3.

DISCUSSION..........................

5 4.

FINDINGS AND CONCLUSIONS...................

13 5.

REFERENCES..........................

18

AE00 ENGINEERING EVALUATION REP 0PT UNIT:

Susquehanna 1 EE REFORT N0.: AE0D/E807 00CKET NO: 50-387 DATE: October 18, 1988 LICENSEE:

Penssylvania Power & Light Company EVALUATION / CONTACT:

C. Hsu WSSS/AE:

General Electric /Bechtel

SUBJECT:

PL'MP DAMAGE DUE TO LOW FLOW CAVITATION EVENT DATE: Pey 26, 1986 (LER 86-021-00)

SUYPARY Susquehanna 1 Licensee Event Report 86-021 describes an event in which the emergency service water (ESW) system was lost due to severe cavitation damage of the ESW pumps. The plant was at full power when an ESW pump failed. While investigating the cavitation damage, the bottom portion of the pump suction bell separated from the pump body and fell into the pump pit. The pump's impeller vanes were eroded through the wall. Similar but less severe damage was found on the three other ESW pumps. A subsequent inspection of the RHRSW pumps also found similar cavitation damage. The cavitation damage was not the result of an inadequate r.et positive suction head (NPSH) for the pump.

Rather, it was due to impeller suction recirculation which occurs when pump operates at flows significantly less than the design flow.

Erosion resulting from recirculation cavitation has also occurred on all four of the PHR pumps at Vermont Yankee.

This may be associated with low flow operation during the monthly surveillance testing. The mini-flow bypass lines for the RHR pumps limited flow to about 5 percent of the design ficw when the pump operates in the testing mode.

Based on these events an evaluation of pump damage due to low ficw cavitation was undertaken. The flew rate at which recirculation occurs is dependent on the design of the impeller. Several pump manufacturers have recently developed guidelines for establishing low flow limits on pump operation. However, the pump mini-flow bypass lines, in most operating plants, were sized solely on fluid temperature rise consideration. Generally, the minimum flows are in the order of 10 percent of design flows.

This rate may not be sufficient to avoid the damaging range of internal recirculation. Moreover, for pumps designed for a wide range of flows, the recirculation configuration may not have been fully considered. Major degradation of pump impellers and casings have occurred to centrifugal pumps that have been running continuously within recirculation regimes.

Many of the emergency core cooling systems in most operating plants are designed to operate with a wide range of flows and use a mini-flow bypass line for inservice testing of pumps. These operating conditions and events, therefore, irdicate that recirculation cavitation is a potential generic problem.

Operation of pumps at low flow conditions for extended periods of tine can cause recirculation cavitation damage to the pumps in spite of available NPSH.

Such damage induces slow deterioration of pump internals and, during early stage of cavitation, do nof Vfect the operation of the pump.

Hence, this type of damage is not eas1 i detectable.

Furthermore, cavitation indication on the pump internals cannot be observed without disassembly of the pump and the plant

l.

routine surveillance tests for pumps may not be capable of detecting early I

impeller degradation. There is the potential that recirculation cavitation on a pump impeller could go undetected until total failure of the pump occurs.

Such failure could prevent the system from performing its safety functions when needed.

L This report presents our findings and conclusions based on our evaluation.

NRC bulletin (NRCB 88-04, "Potential Safety-Related Pump Loss," May 5, 1988) has been issued notifying licensees of certain aspects of the potential prcblem.

The findinos of the AE00 investigation into the icw flow problems provided in this study are based on a wide scope of plant operating experiences and provide background infomation in establishing future guideline for evaluation of licensee response to the bulletin and as backup infomation for plant insr'ction regarding the problem.

l L

INTRODUCTION This engineering evaluation was initiated based on a report of an event at Susquehanna 1 described in LER 86-021, involving severe erosion damage of pump internals. The erosion of pump internals was caused by recirculation cavitation.

Pecirculation within centrifugal pumps is flow reversal at the inlet or discharge tips of the impeller when pumps are running in off-design regimes.

Flow reversal causes vortex action near the impeller vanes, inducing pressure surge and pulsation which cause rapid deterioration by cavitation of impeller metal in the inlet or outlet region even when adequate net positive suction head (NPSH) is provided (Ref.1).

Recirculation has been one of the most persistent and puzzling problems i

encountered in the operation of centrifugal pumps in recent years. Although serious failures had not been reported previously, direct evidence of low flow

^,

induced failure of pumps in nuclear plants and experience gained in both the laboratory and the field during the past decade, had shown that hydraulic insta-bilities and imbalance can occur in a pump running significantly below the design flow.

Furthernere, it has been proven through analysis and tests that l

the effects of recirculation can be very damaging not cnly to the purp's

[

operation, but also to the life of the impeller and casing.

l DISCUSSION Licensee Event Report (LER) 86-021-00, dated May 24, 1986, for Susouehanna 1 describes an event involving pump damage due to erosion caused by recirculation cavitation. The pump damage were discovered in the emergency service water i

(ESW) and residual heat removal service water (RHRSW) systems. The event began on May 22, 1986, while the plant was operating at full power when an overcurrent l

a alam for the C ESW pump was received in the control room.

Investigation revealed the pump motor to be running at a low amperage, and the pump discharge check valve to be closed.

The pump was declared inoperable and the plant entered a limiting condition for operation (LCO).

Subsequer.t disassembly of the pump revealed that the bottom portien of the pump suction bell had separated j

from the purp body and had fallen into the pump pit.

Inspection of the damaged i

parts revealed that the suction bell had been penetrated around its entire circumference by cavitation.

.~

i,

On May 24, an inspection of the "A' pump by a dive y o 4.15 i ss seve w damage to the pump suction bell Me "A '

w3 cl..w 7

inoperable.

Since the condition o' tE

• ning E km thav wa ' also declared inoperable a.

tied -

. ~.

,,4 q u ei. ra t,..,,,

, 2d ' e. '1,

,.. e r '

the A, B and D pumps were nd.c.*0 wing continued to perforn at ne-t he i r.<. ' e i 1

the unit shutdowns.

l

, i damage sim!1er A subsequent inspection of (

' and D o. <

to the A and C pumps, The se' ion beli r siderably but had nc<

been penetrater' The impellers showed sign.

1e hich pressure side of their vc.nes at the suct1ur, ena. Ho:

.nes had..at been penc-trated and had retained their original shar.

.f e,J D pumps Were de0lGred operable on May 28. Due to a lack of spare p.-

oorary repi r wera '.ade to the C pump and the A ESW Icop was reste-snction:1 s m us on t.1e same day.

The ESW pumps norrally operate at about

-cent or less of their design j

flow of approximately 6000 gpm per one pur.

When the loop supplying cooling i

l water to the diesels is run with two cperati.ig pumps, each aump delivers approxi-i i

mately 3500-3900 gpm. The other loop that does not serve t1e diesels (usually the B loop) is nonnally run with only one pump at approximatel) 1000-1500 gpm.

Based on the inspection, the licensee concluded that impeller suction recircula-tion cavitation was the major contributing factor to the ESW pump failure.

The type of cavitation occurred when the pumps ran at ficw significantly lower than it design flow -- flow less than 60 percent.

i Following receipt of the required spare parts, the A and C ESW pumps were repaired, retested and declared operable on June 6,1986.

The B and D ESW pump were also repaired and retested on June 10.

Repair of all four ESW pumps was accomplished by the replacement of all suction bells and impellers. The replacement impeller is NiAl-Bronze which has a higher resistance to cavitation damage than the original impellers of carbon steel.

The replacement suction bells are made of the original carbon steel material. Stainless steel l

(nitronics 60) liners were installed en the suction bells of the A and C pumps.

l Some liners will also be, installed in the B and 0 pumps.

Due to their similar design, the Residual Heat Rer. oval Ser ice Water (RHRSW) pumps were also inspected.

These pumps are two stcge vertical circulator pumps.

Bryron Jackson type 28KXL. Cavitation damaga was found on the impeller liners on the Unit 1 pumps A and B, and the liners were replaced.

Cavitation damage i

to the Unit 2 pump A was minimal and impeller liner replacement was not warranted at that time. The Unit 2 pump B was not inspected until the rext refueling outage.

Although degraded, the RHRSW purps were capable of performing their design functions. The A and B pumps of Unit I have been able to pump 9000 gpm per pump to their respective heat exchangers (design ficw) during subsequent tests.

Subseouent inspections found that the impeller liner degrades before an/

significant darage can be seen on the inteller and the liner damace did not

\\

seem to cause a noticable drop-off in pump perfonnance as evidenced by the ficw data, There were no indications that the cav;+ation was attributed to flow

^

vortexing or inadequate NPSH. The cavitation was a result of flow recicrulation which was caused by opercing the pumps at low flow rates. These RHRSW pumps had operated at less than 50 percent of design flow met of the time. The licensee indicated that the cavitation damage can be avofded by operating the pump above 50 percent of design flow; specifically 75-100 percent of design flow is desirable. The RHRSW system design and method of operation would be reviewed to detennine what changes could be possible to avoid recirculation cavitation.

The damage to the ESW and RHRSW pumps was determined to have been caused by impeller suction recirculation cavitation.

The cavitation is caused by i

operating the pumps at flows which are significantly below the design flow.

The cavitation erodes the suction bell wall.

The impel.er was also eroded but at a slower rate. Once the suction bell wall is penetrated, erosion of both the suction bell wall and the impeller is accelerated as water is drawn through the l

sucticn bell penetrations.

+

This event suggests a comen cause failure mode for the pumps with low flow operation modes.

The citec damage, which caused the pumps to be inoperable, included eroded impellers and suction bells. The pump damage is due to suction recirculation cavitation which resulted from running the pumps at a low flow rate.

(Approximately 60 percent of its design flow of 6000 gpm).

Operation at icwer flow creates mismatches of flow angles within the pump and causes water to recirculate back towards the suction. The recirculating currents cause local pressure zones which are below the vapor pressure of the water. This causes vapor bubbles to form which collapse when a high pressure zor.e is reached, eroding the local material. The erosion begins on the 1

high pressure side of each impeller range at its suction end.

Prolonged opera-tion of a pump at its low flow can result in cavitation damage on impellers.

J When running a pump at a low flow, flow recirculation can occur at the discharge regions as well.

This is called discharge recirculation. Discharge recircula-tion also creates surges and local deterioration by cavitation at the impeller tips. Recirculation in the suction and discharge regicns does not necessarily occur at the same flow rate.

The recirculation cavitation is not related to inadequate pump NPSH. Since inadequate NPSH would also cause pump cavitation, the similarity between patterns of cavitation damage from recirculation and from inadequate NPSH may often lead to an erroneous conclusion as to the cause i

of the damage. However, the mechanisms that cause the damage are entirely different. The cavitation damage from recirculation proceeds from the high pressure side of the inlet edge of the vane through the metal towards the low pressure side, while the damage from inadequate NPSH starts in the opposite direction, from the low pressure side of the vane and proceeding through the metal toward the high pressure side.

Pecirculation characteristics are dependent on the design of the impeller.

It is inherent in the dynamics of the pressure field that every impeller design must begin to recirculate at some point of flow.

Recirculation becomes progressively pronounced as a pump is operated further away from tht, design flow.

The percentage of design flow rate at which recirculation will begin is dependent on many factors. The most critical factors which have influences en i

i

~

. low flow pump performance and minimum continuous flow are: pr e r intensity (pump size), suction specific speed, and specific speed. Altuvugh pump manu-facturers (Ref. 3 and 4) have recently developed guidelines for establishing i

low flow limits on pump operation, studies on low flow aspects are still continuing. Fost of the guidelines which have been published (Ref. 5) stipula-ting recomended minimum flow for pumps, present minimum flow rates as a func-tion of suction specific speed; the lower the suction specific speed a pump is designed for, the lower the flow rate the pump can operate without recirculation.

Low flow operations are generally reouired for the standby systems when j

performing inservice surveillance testing of pumps through the miniflow bypass line, and for systems with a wide range of flows when operating pumps in low flow modes. These low flow operations are general design configurations for emergency core ecoling system testing in ni.*: lear power plants.

In most plants, j

the pump bypass lines were sized only on the basis of limiting the temperature rise of the pump when operated in the testing or minimum flow mode. Typically, this temperature rise based minimum flow is in the order of 10 percent of the best efficiency point (BEP) flow.

In response to the concern by pump manu-i facturers that testing pumps at low flow on the order of 10 percent of BEP flow, may lead to premature failure of pump components as a result of higher vibration during low flow testing, EPRI (Ref. 2) conducted a study on surveil-lance testing of standby pumps in operating nuclear power plants in 1985 to determine whether test-related failures were caused by some aspects of the tests. Although this study neither provides conclusive evidence against nor vindicates the use of low flow testing practices, it does support the expect-ation that icw flow test operation will lead to degradation and premature failure of pump internals and concludes that prolon very low flow (in the range of 10 percent BEP flow)ged operation of pumps at can cause high vibration l

which is a hydraulic instability manifested by flow recirculation in the pumps.

Lately, it is increasingly recognized by pump designers and manufacturers that factors other than tenperature rise, such as energy level, suction specific i

speed, developed head per pump stage, and impeller design details, should play l

a role in establishing the reference value for flow rates during inservice i

tests. Several pump manufacturers are now reconvending that standby pumps be l

tested at a flow no less than 25 percent of BEP flow (s).

[

Five additional events involving either pump failure or potential for pump degradation resulting f rom low ficw operations were found in a search of opera-tional experience data base files. The data search included the Sequence Coding)and Search System (SCSS), the Nuclear Plant Reliability Data System i

(NPRDS, and the foreign event files. These five events occurred at Vement Yankee, H. B. Robinson 2, Turkey Point 3, a foreign reactor and at Haddam Neck.

The pump damage at Vermont Yankee was attributable to insufficient miniflow recirculation line capacity. The potential for pump degradation identified at H.B. Robinson and Turkey Point are also associated with the inadequacy in the l

original design of miniflow recirculation line and that of the last two plants i

was caused by prolonged operation of pumps in low flow modes.

The effects of recirculation resulting from low flow operation manifest themselves i

not only in material degradation -- cavitatien, but also in the form of pressure pulsations and vibrations. Hydraulic pressure pulsation and pump vibration are also significant contributors to deterioration of pump components because of

---

r-.n

. -,,,, - - - - - ~.,,,, - - - -

l

^

l 1

1

• l the high amptitude dynamic forces that they produce. The pump damage at Vermont Yankee was a result of cavitatien, while that at the foreign reactor and at l

Haddam Neck were attributed to pressure pulsation and pump vibration.

H. B.

Robinson and Turkey Point identified the existence of the potential for pump degradation and failure due to the insufficient flow rate designed for the recirculation lines.

l Low flow testing can be a possible source of impeller cavitation damage. The resulting hydraulic recirculation present inside the pump at low flow can create cavitation damage. Vemont Yankee was notified by the pump vendor, Bingham /

Willamette, on November 13, 1986 that the minimum flow rates for the RHR pumps should be made higher than previously indicated to Yemont Yankee. Similar notifications for increase of mini-flow rates have also been sent to four other plants: Cooper, Pilgrim, Browns Ferry and Peach Bottom.

The pump vendor indicated that the minimum flow requirements established for the RHR pumps at Vermont Yankee since plant startup may not be adequate for all pump operating modes. Specifically, the pump vender recomended that the value for continuous minimum flow for the pumps should be 2700 gpm, or about 38 percent of the pump design flew of 7200 gpm, and the value for intennittent operation (less than two hours of operation within a 24-hour period) should be 2075 gpm.

The PHR pump has a flow orifice in the recirculation line designed to limit i

flow to about 350 gpm. This orifice sizing was based solely on the pump flow required to limit the temperature rise of pump when operated in the minimum flow mode. The pump vendor apparently has since determined that additional I

factors must be considered in determining the minimum flow requirements, including pump inlet and outlet circulation flow patterns that will occur at l

lower flow modes. Recirculation flow patterns can occur and result in i

component damage 6 a if there is sufficient available NPSH.

In a similar j

letter dated November 21, 1986, the vender also recomended a continuous minimum flow of 1500 gpm (versus about 350 gpm) for the Bingham 12 x 16 x 41/2 CVDS core spray pumps. A minimum flow of 1350 gpm was recommended for intermittent operation.

As a result of this advice, Vemont Yankee initiated a Potential Peportable Occurrence and its subsequent review. This review detemined that this matter dces not pose a substantial safety hazard for the plant because of plant specific application of the pumps. The length of time that the RHR and core l

I spray pumps would be required to operate in the minimum flow mode over the 40-year design life would be far less than the maximum times at minimun flow i

recommended by the pump vendor.

Bingham /Willamette defines "intermittent operation" es less than two hours of operation in a 24-hour period over the 40-year design life. This translates to a value of up to a total 29,200 hours0.00231 days <br />0.0556 hours <br />3.306878e-4 weeks <br />7.61e-5 months <br /> of operation. Vermont Yankee has no significant accumulated time in the ninimum flow operating mode to date (other than successful pre-operational testing).

Monthly surveillance testing does not utilize the minimum flow path for more than 15 to 30 seconds per month and therefore is considered to be negligible with respect to the 2075 hours0.024 days <br />0.576 hours <br />0.00343 weeks <br />7.895375e-4 months <br /> allowable.

In the event of a small break LOCA, FPR and Core Spray would te required to operate in the minimum flow mcde for a maximum of 4 or 5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />.

Vemont Yankee's evaluation estimated a total of 5 to 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> of operation in the minimum flow rode for the life of the plant. As can be seen, these operating durations are far below the 29,200 hours0.00231 days <br />0.0556 hours <br />3.306878e-4 weeks <br />7.61e-5 months <br /> of opera-tion that Bingham /Willamette considers to be in the "intemitNnt" operating l

I

range, and that sufficient operating times available in the minimum flow mode for either the PHR or the Core Spray purps would not be attained for recircula-tion cavitation failures to develop.

However, this matter could create a substantial safety hazard at another nuclear i

facility, depending on the application of this manufacturer's pumps and the length of time that a pump would be required to operate in the minimum flow rode. The

[

l licensee determined the matter was potentially reportable under 10 CFR 21 and

[

notified the NRC of the potential design defect on March 20, 1987. As an added precaution, the licensee would incorporate a caution statement into the appro-i priate procedure to alert the operators of the need to minimize time in the minimum flow mode, i

i NRC Information Notice 86-39, "Failure of RHR Pump Motors and Purp Internal,"

discussed the failure of a RHR pump at Peach Bottom Unit 3 due to impeller wear ring failure as a result of intergranular stress corrosion cracking (IGSCC).

4 1

After receiving this notification, the Vertnant Yankee licensee performed an l

inspection of purp internals on the RHR pumps during the period from April to l

May 1987, since they are the same model and of the same manufacturer (Bingham -

l Willamette single-stage vertical mounted, model CIVIC, centrifugal purps).

In acdition to the findings of through wall impeller cracks on two of the RHR pumps l

and wear ring cracks en one RHR pump, the inspection coincidently found evidence f

of erosion resulting from suction recirculation on all four of the RHR pumps.

I The licensee also indicated that the wear ring with cracks in the "A" PHP pump was a stationary ring, cast from a material not susceptible to IGSCC cracking.

j However, the cause of impeller and wear ring cracks was not determined, pending l

}

the completion of the licensee's destruction examination being performed at Brookhaven National Laboratories. Based on the recirculation flow erosion

[

effect observed on the suction side of several impellers, the licensee would L

re-evaluate the previous engineering conclusions regarding the adequacy of the l

RhR minimum flow lines in light of the results of the pump internal inspection.

l The impeller and wear ring cracks may also appear related to excessive flow l

turbulence as a result of suction recirculation. The flow reversal at the impeller eye under the condition of suction recirculation creates a vortex

]

action which induces pressure surges and pulsations, causing rapid deterioration by erosion of impeller metal.

The presture surge and pulsation can produce I

dynamic forces which may rise high enough to cause vibration and add undue stress on the wear ring,'resulting in the cracking failure.

l 1

The event at H.B. Robinson was reported in LER 67-026. On October 30, 1987 I

I the potential for degraded recirculation flow for the RHR purps was identified 2

by the licensee during a review in response to the Festinghouse (NSSS Vendor) i concerns of inadequate miniflow design for the RHR pumps. The N3SS designer

[

]

had identified two concerns recently involving the potential for dead heading t

of one of two RHR pumps in systems that have a connon miniflow recirculation l

i 1

line serving both pumps, and, the potential for insufficient miniflow I

recirculation line capacity even for single pump operation. Based on the l

licensee evaluation, the potential for insufficient miniflow recirculation for l

the RHR pump is due to inadequacies in the original design based on today's i

i criteria.

The miniflow recirculation line was designed on the assumption that l

l the two pumps have equal flew / head curves and that each would achieve a flow cf i

1 about 250 gpm while both were operated simultaneously.

Recently, however, the i

1 pump vendor reconnended a minimum flow of 500 gpm for each pump to prevent excessive vibration and pump binding caused by heat up of the recirculated fluid.

I i

I t

,-n,

---

,r---,---m,-..--,--------an--,

w--

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

. i A similar case at Turkey Point 3 was described in LER 88-030. On October 27, 1987, during the design basis reconstitution of the RHR system, it was 1

discovered that the existing minimum recurculation design configuration was potentially faadequate. The miniflow recirculation line, which was shared by the two RHR pumps, was insufficient and had potential for dead heading one of the two pumps in the RHR system. The licensee's corrective action was to i

install independent minimum recirculation lines for each pump. The modified recirculation system would allow operation of both pumps for at least 30 minutes without affecting pump operability.

In response to the concerns of inadequate miniflow design for RHP pumps, NRR issued an Infortnatien Notice on November 17, 1987 (NRC IN 87-059. Potential RHR Pump Loss) which indicates that the problem may be generic to all water-cooled reactor designs, regardless of the pump application or the NSSS manufacturer.

This is based on the belief that miniflow lines have traditionally been designed for only 5 percent to 15 percent of pump design flow, while some pump manufac-turers are advising that their pumps should have minimum flow capacities of 25 percent to over 50 percent of best efficiency flow for extended operation.

]

Another event occurred at a foreign reactor. During a periodic inspection, j

maintenance personnel heard a loud noise from an RHR pump. The pump was irine-diately stopped. Upon disassembly, the shaft wes founded broken, and slight j

contact marks were found on the wear rings. The markincs on the broken area of the shaft indicate that failure was due to a low cycle fatigue fracture.

Subsequent evaluations concluded that the pressure distribution around the 4

impeller become:; uneven durine low flow operation increasing the radial thrust, and th? bending stress becomes approximately three times that at rated flow l

j ope ra tion. The RHR pumps perform low flow operations during monthly surveillance testing and cleanup during plant shutdown.

The pumps had run approximately 5963 accumulated operating hours in low flow operation when the failure occurred.

In addition to the main shaft replacament, the licensee mccified its long term operation management of low flow operation.

Previously, both modes of operation of the RHR pump, the clean-up and the cooling modes, i

were perforred by one pump for each purpose. Modified operation is to operate one pump for both purposes to minimize accumulated low flow operations of one pump.

The event reported in LER 88-003 at Haddam Neck occurred on February 4,1988.

With the plant in mode 6'and the reactor core off loaded, the electric driven fire pump was ceclared inoperable due to a high amperage condition noted after a

j manual start during a routine iurveillance. The nortral indication of 200 amps increase to 340 - a60 amps. The indication increased to 1000 amps during the 1

following manual restart. The cause of the inoperability was physical damage to the stuffing box brass bushing located in the upper shaft area of the i

electric driven fire pump. This caused the bushing to shear, resulting in a

]

locked rotor condition. Based on the licensee's evaluation and their discussion i

with the manufacturer, it was concluded that prolonged low flow operation of the pump may have caused the problen. Operation of the pump at or near shut of f head had occurred during the Containment Integrated Leak Rate Test (CILRT) in which the fire pump operates at low flow mode only to provide cooling water to the air compressors.

I i

-9 These events illustrate that recirculation cavitations were caused by operating pumps at flow significantly below the design flow rates, either at low flow modes or at: miniflow testing through a bypass line.

The damage was a result of slow deterioration accumulated over a long period of time during which the pumps were still functional and remained operable at early degradation. Cavitation indica-tion on the pumps' internals can cnly be observed by disassembly of the pumps.

[

The routine surveillance tests of pumps provided in the plant inservice test prograns may not be capable of detecting early impeller degradation.

In addi-tion, since the pumps operated in the specified operating range, the plants were i

not aware of the problem until the occurrences of pump failure. There is the l

potential that recirculation cavitation on a pump impeller could go undetected until total failure of the pump occurs.

Such failure could prevent the r

associated system from perfcrming its safety function.

i in 'he ongoing RES effort on aging and service wear of the auxiliary feedwater (AFW) pumps for PWR nuclear plants, testing of AFW pumps at flow less than 25 i

percent of BEP has been identified as a source of hydraulic instability and unbalance which will acctlerate component wear and lead to premature failure I

of pump internals. The FES study found that the miniflow line for AFW purrps was typically establishrd only to prevent pump overheating and thus is nomally 10-15 percent of BEP flow. Although it is suggested that the surveillance testing of AFW pumps be completed at increased flow, in rnost cases it is not practical to test the ptmp at higher flows without either modifying the existing miniflow circuit to increase its flow capacity or to perform testing while tha plant is shutdown (In response to these concerns, one of the contractors of the RES study has developed improved auxiliary feedwater pump I

testing guidelires). The preliminary findings and conclusions of the RES studies correlate well with the operating experience evaluated in this report.

]

FINDINGS AND CONCLUSIONS Cased on the preceding discussion and related follow-up activities ccnducted for the study, the following findings and conclusions are provided:

1.

Operation of centrifugal pumps at low flew conditions for extended periods of time can cause cavitation damage in spite of available NPSH. A centri-i fugal pump is designed for best performanet; at a specific cerrbination of 1

capacity, head, and speed, that is, the best efficiency point (EEP). At the design or BEP llow rate, the fluid motion is compatible with the L

physical contours of the hydraulic passage and is *.herefore vell behaved.

Once deviating from design flow, the operation starts to create mismatches t

of flow engles within the passage and diverts part of flow to recirculate within the pump at certain low flow rates.

The circulating currents cause local pressure zones which are below the vapor pressure of the water. This causes vapor bubbles to form which collapse when a high pressure zone is 1

reached, leading to the erosion of the local material.

Such flow recir-culations can occur at the impeller eye and exit as well as outside the t

irtpeller shroud and huh.

i

]

2.

Low flow operations are generally required for the standby systems when j

performing inservice surveillance tests of pumps by restricting discharge 1

i flow through the mini-flow bypass line, and for systems designed for wide range of flows when operating the pumps in the low flow regime. Many of j

i I

j l

, the energency core cooling systems in most operating plants are designed to operate with wide range of flows and use a mini-flow bypass line for inservice testing of pumps during the standby mode.

3.

Recirculation cavitation bas caused damage to the pumps of the ESW and RHRSW systems at Susquehanna 1 and the pumps of the RHR system at Vermont Yankee.

The recirculation cavitations were caused by operating pumps at flows significantly less than their design flow iates, and the damaces were the result of prolonged operations of these pumps at the low flows.

The cavitation darage of the ESW pumps at the Susquehanna plant was very severe. The impeller vanes were eroded through the wall, and suction bells were penetrated around most of the circumference. One of the suction i

bells had separated frem the pump body and fell into the pump pit. The damages to the RHRSW pumps were less severe. The ESW pumps nomally operate at about 60 percent or less of the design flow and had run approxi-mately 18,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br /> when the failure occurred. The 18,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br /> of opera-tien is only a small fraction of the design life. The RHRSW pumps had operated at less than 50 percent of design flow for most of the time and had run approximately 9,000 hrs. The recirculation cavitation of the RHR pumps at Vermont Yankee could be associated with iow flow operation during t

the monthly surveillance tests. The RHR pump miniflow bypast line is a single line sized to bypass about 5 percent of the design flow. The I

accumulated time in operation at this low flow is not known.

4.

The effects of recirculatio.; manifest themselves not only in material l

degradation--cavitation, but also in the fom of pressure pulsations and vibrations. Hydraulic pressure pulsations and pump vibrations are also f

significant contributors to deterioration of pump components because of the high arplitude dynamic forces that they produce.

Excessive forces r

on the impeller and pump vibration have caused darage to the RHp pump at a foreign reactor and the fire pump at Haddam Neck. The licensees for H. B.

L Robinson 2 and Turkey Point 3 have identified the existence of the j

potential for pump failure due to insufficient flow rate designed for the t

recirculation lines of their RHR pumps.

4 5.

It is inherent in the dynamics of the pressure field that every impeller design must recirculate at some point of flow -- it cannot be avoided. The flow rate at which recirculation occurs is dependent of the design of impeller. Although pump manufacturers have recently developed guidelines for establishing low ficw limits on pump operation, the pump bypass lines, in most operating plants, were sized solely on the basis of limiting the temperature rise of the pump when operated in the testing mode or minimum ficw mode. Generally, the flows are in the order of 10 percent of design 1

flow.

I 6.

In response to a concern by pump manufacturers that testing purps at low flows on the order of 10 percent BEP flow, may lead to premature failure of pump components EPRI conducted a study in 1985 on surveillance testing of i

standby pumps in operating nuclear power plants. The result of this

(

study does support the expectation that low flow test operation will lead to degradation and premature failure of pump internals and concludes that L

prolonged operation of pumps at very low ficw (in the range of 10 percent 4

to BEP flow) can cause high vibration which is an hydraulic instability.

i

~

~

. 11 -

7.

Several pump manufacturers hau recently recocrended the standby pump be tested at a flow no less than 25 percent of BEP flow (s). One pump vendor, Bingham /Willamette, has inforned Vermont Yankee of inadequacy in the minimum flows designed for the RHR pumps and the core spray pumps. The mininum flow requirements wera established for the pumps at plant startup.

The value for minimum flows for the pumps, according to the vendor, should be increased to about 38 percent of the pump design flows. Similar notifi-cations for increase of minirum flow rate for the RHR pumps also have been sent to four other plants: Cooper, Pilgrim, Browns Ferry and Peach Bottem.

4 8.

In the ongoing RES program on aging and service wear of the auxiliary feedwater (AFW) pumps for PWR plants, surveillance testing of AFW pumps at flows below 25 percent of BEP has been identified as a source of hydraulic instability and unbalance which will accelerate component wear and lead to premature failure of pumps.

9.

It appears that degradation caused by recirculation cavitation due to low-flow operation will require a great length of time to cause 1

catastrophic failure of a pump. Such damage induces slow deterioration of i

pump internals and during early stages of cavitation do not affect the operation of the pump. Hence these types of damage are not easily detect-able.

In addition, the degradation will also be difficult to detect in the basis of measurements taken during inservice surveillance testing. This follows from the fact that, in most plants, the bypass flow test provides neither the proper cperating range of flow rcr sufficient running time to ccmprehensively trend and predict degrading condition.

pecirculation cavitation damage of the pump impeller could go undetected until total failure of the pump occurs.

10. Although the mechanism that causes cavitation damage from recirculation is i

entirely different from that of inadequate NPSH, the similarity between patterns of the cavitation damage from both may often lead to an erroneous conclusion as to the cause of the damage. This may be one of the reasons that the concern of recirculation cavitation has not been widely recognized.

i SUGGESTIONS 1.

The effects of pump recirculation can cause not only operational problem, i

such as vibration, but also lead to damage and loss of life of the impeller and casing. Major degradation of pump impellers and casings have occurred to centrifugal purps running continuously at low flows. Pany of these i

problems can be avoided by specifying and designing pumps for lower suction specific speeds and limiting the range of operation to capacities above the point of recirculation.

For example, low flow operation of centrifugal purps could be avnided in a syttem in which multiple pumps operate continuously at low flows by reducing the number of pumps running and thus j

increasing the flow thrcugh each pump, or by realigning the system to increase flow through the pumps.

It is suggested that pumps specified for operation in a wide range of flows should be checked to detennine whether any point in the operating range fall in the recirculation zone of the pumps.

I

2.

The bypass flow rates for the pumps with bypass line should be reconsidered and acceptable values established to avoid operating pump in the recircu-lation zone.

If the condition of running a pump in a recirculating zone cannot be avoided, a procedural control should be established to limit the length of operation in the bypass mode such that premature failure of the pump can be prevented.

1 3.

For pumps having the potential for recirculation cavitation (i.e., pumps not meeting items 1 & 2 above) appropriate inspection intervals should be established to facilitate early detection of recirculation damage of the pumps.

Near the completion of this study, FRR issued Bulletin 88-04, "Potential Safety-Related Pump Loss," as a result of the staff's evaluation of the miniflow design problems which were initially identified by Westinghouse and subsequently confimed by several licensees. The bulletin requests all licensees to investi-gate and correct, as applicable, the pump miniflow design problems in all safety-related systems. The actions addressed in the bulletin appear to be limited to the miniflow problem. However, this study had found that pump damage induced by internal recirculation can be caused not only by insufficient miniflow capacity t'ut also by operating purps at low flow modes.

In view of this concern, we suggest that NRR review the operating experience to detennine whether further actions are warranted, i

REFERENCES 1.

W. H. Fraser, "Recirculation in Centrifugal Purps," presented at the ASME Winter Annual Meeting, Washington, D.C. November 15-20, 1981.

2.

EPRI NP-4264 Volume 1 Emergency Pumps "Failure Related to Surveillance Testing of Standby Ecuipment," October 1985.

3.

C. C. Heald and R. Palgrave, "Backflow Control Improves Pump Perforr:ance,"

011 and Gas Jcurnal, February 25, 1985, pp.96-105.

4 Bingham - Willamette Company, "Guide for minimum Flow Operation of Purps," Technical Bulletin No. 68, Rev. O, November 18, 1982.

4 5.

W. Stanly Tinney, "How to obtain Trouble Free-Performance from Centrifugal j

.eumps," Chenical Engineering, June 5,1978, pp.178-181.

a a

t 6

i

1 r

SSItiS NO.

l IN 88-XX l

l UtlITED STATES fiUCLEAR REGULATORY C0!' MISSION OFFICE OF NUCLEAR pEACTOR REGULATI0fl WASHINGTON, D.C.

20555 August

, 1088 l

NRC INFORMATION tt0TICE fl0. 88-XX:

Purp Damage Cue to Low Flow Operatinn 1

i l

Addressees:

1 1

All holders of operating licenses or construction permits for nuclear power reactors.

Purpose:

l This information notice is provided to alert addressees to potential prcblems l

resultino from operation of contrifugal pumps at low ficws vthich may cause l

severe component damage to the pumps.

It is expected that recipients will l

review the information for applicability to their facilities and consider actions, l

as appropriate, to avoid similar problems. However, suggestions crntained in this infonnatien notice do not constitute NRC requirements; therefore, nc j

specific action or written response is required.

l Description if Circumstances:

1 Described Perein are three events involving pump damage resulting from low l

flow operations.

The low ficw operations created hydraulic instability which resulted in pump danage from cavitation, pressure pulsation and/or vibration l

following extended periods of operation. These events occurred at Haddam Neck, Susquehanna 1, and a foreigr reactor. The pump damace at Susquehanna 1 l

was a result of cavitation, while that at the foreign reactor and at Haddan Meck l

were attributed to pressure pulsation and pt.rp vibration.

i i

l l

Hadden l'eck Event:

l l

On February 4,1988 with the plant in Mode 6 and the reactor core off leaded, 1

t 1

t

~ _ - _ _ -

2 the electric driven fire pump was declared inoperable due to a high amperage condition noted after a manual start during a routine surveillance. The normal indication of 200 amps increase to 340 - 360 amps.

The indication increased to 1000 amps during the following manual restart.

The cause of the inoper-i ability was physical damage to the stuffing box brass bushing located i

in the upper shaft area of the electric driven fire pump. This caused the j

bushing to shear, resulting in a locked rotor condition. Based on the

[

l licensee's evaluation and their discussion with the manufacturer, it was l

concluded that prolonged low flow operation of the pump may have caused the I

problem. Operation of the pump at or near shut off head had occurred during i

]

the Containment Integrated Leak Rate Test (CILRT) in which the fire pump operates at low flew mode only to provide cooling water to the air ccmpressors, j

g Susquehanna 1 Event:

j i

i l

The plant was at full power when an emergency service water (ESW) pump failed on l

May 22, 1986. The pump failure was discovered when an overcurrent alam for the l

pump was received in the control room.

The pump was declared inoperable and the I

1 i

plant entered a limiting condition for operation.

Subsequent disassembly of the

(

pump revealed that the bottom portion of the pump suction bell had separated l

from the pump body and had fallen into the pump pit, and the pump's impeller i

vanes were eroded through the wall. Similar, but less severe damage was also i

I found on the three other ESW pumps. A subsequent insr9ction of the RHRSW pumps I

also found similar damage.

The danage to the ESW ar

'JiRSW purps was determined f

i to have been caused by recirculation cavitation.

The recirculation cavitation l

was caused by operating the pumps significantly below their design flow rates.

l l

[

j The ESW pumps noreally operate at about 60 percent or less of ti,eir design flow l

of approximately 6000 gpm per pump. When the loop supplying cooling water to the diesels is run with two operating pumps, each purp delivers approximately i

3 j

3000 - 3900 gpn.

Pump operation in this renge is likely to cause recirculatten j

]

ca*Jvation. The other loop that does not serve the diesels is nonnally run with only one pump at approxinately 1000 - 1500 gpm. This operation causes l

5 recirculation cavitation. The pHRSW pumps had operated at less than 50 percent l

1 1

i i

i t

3-I l

of design flow most of the time. The 'icensee indicated that the cavitation i

damage can be avoided by operating the pumps above 50 percent of design ficw; j

specifically 75 - 100 percent of design flow is desirabic.

[

]

t 1

l Foreign Event:

l 4

Another event cccurred at a foreign reactor in October 1986. During a periodic l

inspection caintenance personnel heard a loud noise from an pHR punp. The j

pump was imediately stopped. Upon disassembly, the shaft was founded broken, j

and slight contact marks were found on the wear rings. The markings on the j

broken area of the shaft indicate that failure was due to a low cycle fatigue l

I fracture. Subsequent evaluations concluded that the pressure distribution I

around the impeller becomes uneven during low flow operation increasing the f

1 radial thrust, and the bending stress becomes approximately three times that i

I at rated flow operation. The RHR pumps perform low ficw operations during l

monthly surveillance testing and cleanup during plant shutdewn. The pumps had

)

run approwiately 5963 accumulated operating hours in low flow operation when l

l the failure occurred.

In addition to the main shaft replacement, the licensee

{

)

modified its long term operation management of low flow operation.

Previously, l

t l

both modes of operation of the RPR pump, the cleanup and the cooling modes, were performed by one pump for each purpose. Odified operation is to operate j

one purp for both purposes to minimize accumulated low flow operations of one f

pump.

i l

j Discussion:

I 1

1 These events illustrate that purn damages were caused by operating pumps at j

ficws significantly t'elow the design flow rates. The damage was a result of I

slow deterioration of purp internals accumulated over a long period of time during which the pumps were still functional and remained operable at early I

l degradation. Damage indication on the purps' internals can enly be observed l

by disassembly of the pumps.

The routine surveillance tests of pumps provided i

in the plant inservice test prograns may not be capable of detecting early i

component degradation.

In addition, since the pumps operated in the specified i

i operating ranges, the plants were not aware of the problem until the occurrences

{

}

l l

)

F' 4

of pump failure. There is the potential thet pump degradation due to low flow operation could go undetected until total failure of the pump occurs.

Such failures could prevent the associated system from performing its safety function.

No specific action or written response is required by this information notice.

If you hava any cuestions about this matter, please contact the technical contact listed below, the Regional Administrator of the appropriate regional office, or this office.

Charles E. Rossi, Director Division of Operational Events Assessment Office of Nuclear Reactor Regulation Technical Centact: Chuck Hsu, AE00 (301) 492-4443 i

f I

6 b