ML19325C949

Requests Submission of Plan for Resolution of Problem Re Insufficient Justification for Extrapolating Test Results, Per NRC Bulletin 88-004

ML19325C949
Person / Time
Site:	Crystal River
Issue date:	10/10/1989
From:	Silver H Office of Nuclear Reactor Regulation
To:	Wilgus W FLORIDA POWER CORP.
References
IEB-88-004, IEB-88-4, TAC-69905, NUDOCS 8910180195
Download: ML19325C949 (9)

Text

,

~

4

_ m ;.

, ;. hl$

'l' Gfff $ S/8f

^

>p ['1

, z ;m

~

j

h I UNITEO STATES :

Q (f $

NUCLEAR REGULATORY COMMISSION'

'I

~ p % is

(J

. WASHINGTON, D. C. 20666 :

i k+e... )

~

Octsber.10,1989 1

s; 1

y

~Cocket;No.L50-302 q

'e 0

^

x MrfW.S.Eilgus-l J

' Vice' President, Nuclear. Operations 1

Florida ~ Power Corporation '.

ATTN:: : Manager,' Nuclear Operations 4

LLicensing.

P. 0. Box 219-NA-2I..

J

, Crystal: River, Florida. 32629

]

Dear Mr. Wilgus:

y

SUBJECT:

' EVALUATION OF CRYSTAL RIVER. UNIT 3 (CR-3): DECAY HEAT PUMP AT

.[

LLOW-FLOW OPERATION (TAC NO. 69905) g 3

L EAs the result of. potential low-flow operation: problems identified in your

?

m 3 response to NRC Bulletin 88 Florida' Power Corporatioh;(FPC) performed a ttestiof decay heat. pump 3B on April 14,1989. :The pump was subsequently 3

disassembled and' inspected.

l The test was. designed to run the pump at a nomina 1Lflow rate of 400 gpa, or-t about 100 gpm less: than.the expected minimum flow rate, for ten hours, which s

was'about twice the then_-expected maximur duration. Subsequent to the' testing,;

however, it was recognized that the test duration did not bound the potential:

imaximum duration of-three days, l

By letter: dated July' 11,19E9, FPC submitted a report of its consultant,.MPR S

t' i

Associates. Inc. -(MPR), which provided the results of its evaluation of the

.n ioperability of: the' decay heat pump:at low-flow conditions. Based on that' l:

reportj MPR and FPC concurred;that the decay heat' pumps.will; operate satis-l;

- factor 11y at:a-flow of 400 gpm for up to 3 days,

~

'W' WithLtechnihal assistance from our contractor, Oak Ridge National' Laboratory p

'(0RNL),.we have completed our review of the WR report,'as described in the 4

1 E

enclosure. Based on that review, we continue to believe:that there is not sufficient jusWfication for extrapolating the test results, for the following.

reasons,o There is no margin between the nominal test flow rate and the present a

demand condition flow rate. Also, there were uncertainties associated with the

~

. recorded' test data which served to aggravate this lack of flow rate margin.

.FinallydyMPR performed a static analysis which did not adequately represent.tiu extreme l adverse and complex (ynamic operating conditions associated with J

low-flow pump operation.-

FPC has indicated that various options are being considered to resolve this

. matter, including additional evaluation to define more precisely the required operating tise, further pump testing, and possible system modifications.

4 Ok p

\\\\

8910180195 091010 PDR ADOCK 05000302 o

FMM.

ggy i

~

TQpT e

s:

i n..

.:e e

m.:W :...

t

~

,x p

h~t.

lMr. W.JS. Wilgus,

-2 October -10,1989

+

a 1

s hYou have constitted to' implementation ofz an. acceptable. resolution by the end of.

M ic

. Refuel 7, and have also committed to provide us with your plan'for. resolution..

We understand = that. submittal of that' plan was delayed pending receipt of our j

comments on the PfR report, which should be. considered in your' plan and its, H

implementation.- We request that you submit youp plan no later than l

November:1, 1989.--

3 9

1,.

Sincerely,.

Original signed by-

.h.

i Harley Silver, Project Manager r

Project Directorate 11-2

+

i.

-Division of. Reactor Projects.-1/11-Office of Nuclear Reactor Regulation y

s L~

Enclosure:

)

cc w/ enclosure:

q L

.See next page i.

/,

l m

DISTRIBUTION-

~

a

ldE, ME&fl-e NRC-& LPDRs.

/

{ Wh:

JPD22:Rdg File L a

p 4, 9 j SVarga' 14/E/4 -

i'.

Y y Glainas 14/H/3-t.

s N-DMiller

~

].

lD

.. 'H$11ver-L

• x""

GWunder i c-

1 ngcy

- 1;1V p ;EJordan NNBB-3302 i, M : 4 'BGrimes 9/A/2

~,Ji j

q

~:.<'gACRS(10)

Li d ' HSinkule RII' 1

-p-i,;1TMarsh9/H/3'

)1 r

i t

0FC,:LA:PDII-2/1 :PM:P

'I-

D:PDII.2

.s.....:...... 4...:. y 7.j...::HBerk 7 :............:............:............:......... J" i

}NAMEl:DMil(

_.:HStW d

........:............:............:............:............:............:.........r y

K DATE - / _/89

10/gp/89-

10//o /89 t

.0TFICIAL RECORD COPY

Document Nane: LTR/CR3 DECAY HEAT PUMP i

,l 9,.

{

(*

1 t[

- Mr. W. $; Wilgus Crystal River Unit No. 3 Nuclear Florida Power Corporation.

Generating Plant cc:

i

~ Mr. A. H. Stephens State Planning and Development General Counsel Clearinghouse-Florida Power Corporation.

Office of Planning and Budget

- MAC - A50" Executive Office of the Governor P. 0. Box 14042:

The Capitol Building

- St'. Petersburg, Florida 33733 Tallahassee,' Florida 32301'

. Mr. - P. F. McKee, Director.

Chairseri helear Plant Operations Board of County Comissioners Florida Power: Corporation Citrus County 1

- P. 0; Box 219-NA-2C 110 North Apopka Avenue Crystal-River, Florida 32629 Inverness, Florida 32650 Mr. Robert B. Borsuit Mr. Rolf C. Widell, Director

- Babcock & Wilcox Nuclear Operations Site Support huclear Power Generatior.- Division Florida Power Corporation 1700 Rockville Pike, Suite 525 P.O. Box 219-W 21 Rockville, Maryland 20852 Crystal River, Florida 32629 Senior Resident Inspector-rir.. Gary L.- Boldt.

Crystal River Unit 3 -

Vice President, Nuclear Production

- U.S.: Nuclear Regulatory Comission Florida Power Corporation

'15760 West Powerline Street P. O. Box 219-SA-20 Crystal-River,. Florica 32629 Crystal River, Florida 32629 Regional Aca,inistrator, Region II U.S. ' Nuclear Regulatcry Comission 101 Marietta Street N.W., Suite 290C Atlanta, Georgia 30323

- Mr. Jacob Daniel Nash Office of Radiation. Control-Departinent of Health and Rehabilitative' Services 1317 Winewood Blvd.

Tallahassee, Florida 32399-0700 Administrator Department of Environmental Regulation l

. Power Plant. Siting Section e

State of Florida 2600 Blair Stone Road Tallahassee, Florida 32301 Attorney General Department of Legal Aff airs The Capitol Tallahassee, Florida 32304

~r

+-

• v

,.-,,w1-3-,

- -... -, -.-e e,

g

---

.=3,-,

m.e s wv e

e

--es

..==e-,w-e---a.-e-

---r--es.--..-e--..

.,,. - -_e_.--

--,-w,--- - -... -*

w

g; N

/

a

+

Vll n.

M

,[e,

s

[,

ENCLOSURE,

~i Yijel

\\

K i

i$

- COMMENTS ON THE FLORIDA POWER CORPORATION

', 1

- DECAY HEAT PUMP LOW FLOW TEST AND REPORT,

y i

O

' ( [1f a

by:

a

" i, D. A. Casada M. L. Adams

^

i

t 4

INTRODUCTION g

a As the result of potential low flow problems identified during the review of NRC Bulletin 88 04, Florida Power Corporation performed a test of Decay Heat Pump 3B. The pump

~

y

- was subsequently disassembled and inspected.

j y

The primary portion of the test was to run the pump at a nominal f' low rate of 400 ppm, or about 100 gpm less than the then expected mimmum flow rate for 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br />, which was about twice the expected maximum duration. The general approach taken, based upon

-knowledge available at that time relative to the conservatism m flow and duration, was

. appropriate, in ourjudgement. Note that there is no existing guidance, either in the form of regulation or accepted good practice, as to what is an appropriate test methodology for m

. demonstrating the acceptability of a pump forlow flow service.

' Subsequent to the testing, however, it was recognized that the test duration did not bound '

the potential maximum duration, and that there was no margin between the nominal test 1

~ flow rate r.nd the minimum anticipated flow rate that would exist in the piggyback mode following a small break LOCA. As a consequence of this recognition, additional analyses r

. were performed by Florida Power's consultant, MPR Associates,Inc.,in an attempt to E

justify that the pump could operate satisfactorily for a longer period at the flow rate

~

Ls expenenced during the test.

It is our opinion that there are two principal areas of weakness in the report submitted by Florida Power. 'Ihe first involves the uncertainty in exactly what the test conditions were, and specifically whether the test conditions bounded the conditions of concern. The second D

weakness relates to the inadequate modeling of the d the pump. These areas are addressed, respectively,ynamic forces that could exist in Sections I and II below and our overall conclusions are stated in Section III.

Lf l ':

l' t

r

,.m 1

b s' e-

-,e e

-e

-,--werv.r--,-ea, e,-

,,.ees.

..mm,

-.n,m

+-,,,-enw,,-a-o e-

.,,,-,w,,,,,

v n -e n -

we-,-

v-,----

^

FZy f i

1 r.e s

3[.

'I.

INSTRUMENTATION -

Minimum recirculation flow (through the restriction orifice) and return flow to the Borated Water Storage Tank (BWST), :uction and discharge pressure, svetion and discharge temperature, and motor amps (for each phase) were recorded during the test. De flow path used during the test is identified in Figure 1 (note that there was also r, small-4 unmonitored flow through a cyclone separator not shown on Figure 1). For comparison, Figure 2 indicates the f ow paths that would te used by the Decay He.at Pumps in the

• piggyback" mode, which is the low-flow mode of concem.

As noted earlier, when the test was performed,it was believed that margin in both flow and duration were being provided (100 gpm, or 25% margin in flow, and 5 hr, or 100%

margin in duration). MPR used data collected during the test as a part of the effort to demonstrate that the pump was operating as expectec:. It is important to recognize the-limitations associated with the monitoring of some of these conditions.

L FLOW Flow and pressure were monitored to comsare measured hydraulic performance with a vendor test curve for the same mode; pump. Since the flow dunng the 400 gpm (nominal) portion of the test is of the most significance, the.following comments are geared toward that condition.

i l

Pemunentiv Installed Flow Meter (FI2).

The retunt flow to the BWST was monitored by a permanently installed orifice and its associated instrument loop. The transmitter and signal processing string for this -

t orifice is scaled from 0 to 5000 gpm. The calibrauon indicators (SP-169C) specifies a tolerance ofi 100 gpm, procedure for the flow which is i 2% of span.

- This is a typical calibration tolerance, and normally represents the sum of the signal processing and indicator accuracies. Note that it does not include the accuracy of L

the primary element (orifice). If an orifice accuracy of i 0.5% of the loop 1

calibration span is assumed, the combined meter / loop accuracy would be 2.5%,

l-or 125 gpm. It should be noted that the loop calibration procedure only verifies L

the output devices are within 100 gpm of expected for delta P's corresponding to =

0,2500,3535,4330, and 5000 gpm.

L During the 10 hour1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br /> portion of the test (with a nominal total flow rate of 400 gpm),

the average flow rate recorded for F12 was 255 gpm, or about 5% of the calibration -

span (note that the delta P across the orifice at 255 gpm would be only about %%

of that corresponding to 5000 gpm). Thus, the estimated loop accuracy would be about half of the average recorced flow.

Temocrarv Ultrasonic Flow Meter j

A strap on ultrasonic flow meter was used in the minimum flow line due to the nonexistence of permanent instrumentation in the line, ne accuracy of this meter is not known, however, it is suspected to be relatively poor. During the nominal 400 gpm portion of the test, the c igital indication from the ultrasonic flow meter was

. noted to range from about 130 to over 200 gpm.

2

o y

+

4 a

y i.y f.

l

([

.l j

ey * -

Combined Flow Innrument Accurnev

~

- As noted, the accuracy of the ultrasonic flow meter is not known. Ifit is assumed that actual flow through the miniflow line was somewhere in the range recorded -

- (this may or may not be technically supportable), total actual flow during the test rray have ranged from 260 gpm to 580 gpm, corepared to the average recorded i

flow through t te two lines of about 415 spm.-

}

PRESSURE Suction and discharge pressure were monitored with permanently installed (but normally valved out) gauges. The accuracy of these gauges is a relatively.

unimportant factor, What is important to recognize is that pump delta P is not a valid indicator of pump flow when the pum s is operating at flows below about 1000 to 1250 gpm. Note that during the test, t ie recorded delta P was either 189 or 190 psid at all measured flows less than the 1250 gpm (nominal) condition.' Thus, t

. pump delta P is not useful indicator of flow in the regime of interest.

4 MOTOR AMPS Motor amps were used in an attempt to estimate brake horsepower and compare this to a pump vendor test curve. MPR cited use of the following calculation (from Figure 1 of Appendix C):

P=4180V x IUNE x (3)o.5 x power factor

- Note that the above equation yields power in watts. An assumed value of 0.9 was used by MPR as a power factor. The above equation actually corresponds to power input to the motor, not brake horsepower, in order to determine power output, the rrotor efficiency must also be accounted for.

l p-The correct calculation of brake horsepower would be as follows:

p hp=[V x I x (3)0.5 x power factor x motor efficiency) / [746(w/hp))

Bus voltage was not recorded as a part of the test. - The 4180 used in the MPR calculation represents 20 volts above nominal bus voltage.

1 l'

The Decay Heat Pump motor vendor specifies afullload power factor of 0.921 and a fullload efficiency of 0.938 (multiplied together these terms yield 0.864). The full load power factor may be substantially different than the part load power factor.

Although vendor specific information was not obtained for power factor as a function of loading, some available generic data for similar size motors indicates that the power factor at half of the rated load may range from 77 to 93% of the full load power factor. Motor efficiency would also be expected to be slightly lower at lower loading. Note that the brake horsepower expected at 400 gpm (based on t-h pump vendor curves) would be about 40% of the motor rating.

in summation, there are several areas of significant error / uncertainty in the method used to compare the pump vendor's test brake horsepower with calculated values based on data from the Crystal River test. Use of motor current data in the absence L

of other important parameters, such as actual power factor and bus voltage, provides, at best, only a very rough indication of operating condition.

3

9 g

j g.n 97 i

1.

3t w

"M:

TEMPERATURE Suction and. discharge temperatures were measured using surface mounted -

elements' For all practical purposes, no temperature changes of significance

~

occurred during the test, with the exception of the short time when operating on;

' minimum flow only, During operation on minimum flow only, both the suction and discharge temperamres rose fairly quickly by about 6 7 'F, and then stabilized,-

i indicating the reaching of equilibrium between the tube and shell side of the Decay Heat HX. De measured temperature nse across the pump was,in all cases, either equal to zero (0) or minus one ( 1) 'F (the negative temperature differential -

indicates either cooler surface temperatures at the discharge than at the suction due to room ventilation conditions or the general level of instrumentation accuracy, or both).-

a

' A process temperature related figure included in the MPR report (Figure 2 of Appendix C)is misleading. De MPR figure provides a plot of expected discharge temperature assuming a constant suction temperature of 75 'F and also plots measured suction and discharge temperatures. While the data points seem to follow the predicted curve fairly well in the MPR figure, in reality, this figure is a J

comparison of " apples and oranges" in that the suction temperature was not a

~

- constant 75 oF.

A more relevant comparison between theoretical and measured temperature rise across the pump is provided in Figure 3. It can be seen that there is no apparent relation between measured and theoretical temperature rise. - What Figure 3 does indicate are the limitations inherent in the accuracy of the instrumentation used to monitor suction and discharge temperatures'and the general test configuration.

During the duration of the 400 gpm (nominal) portion of the test,IEt1h the suction and discharge temperatures grac ually increased (over several hours) from about 76

'F to 78 79 oF.

These temperature rises were probably more a reflection of the

~

general environmental condinons than an indication of pump operating conditions.

[

11is apparent that the monitoring of tem no inriication of operating conditions. perature differential during the test provid 1

L

SUMMARY

This test was conducted under conditions and with the use ofinstrumentation that are consistent with normal practices for pre-operational testing of pumps and a

systems across the potential range of flows in nuclear power plants.

However,in light of the fact that the condition of concern during the test was at a flow.which was such a small percentage of normal flow, and at an even lower percentage of the process instrument span, the utilized flow instrumentation did not provide results that are appropriate for the intended use. For comparison, ASME Section XI requires that instruments used in Inservice Testing of pumps have a range less than or equal to three times the reference value. The range of the o

permanently installed flow instrument loop used during the test is about twenty times the recorded flow rate during the ten hour test duranon.

- It is also clear that there are considerable uncertainties and/or weaknesses associated with other data collected (which could have potentially provided an indirect

' indication of flow) and calculations made based upon tiat data It is now recognized that there is no margin between the nominal test flow rate and that 4

5 e...m.-

-.m m..-

m

..m

. e.+-

-.-e--%--.-

.- - - -, -- + - -

~

~

,w j

/

[%

.i associated with the demand condition (post small break LOCA piggyback mode),:

and that the duration of the test was roughly an order of magnitude less than the -

m soiential demand operadng time. Based upon the review of available indications, it '

s concluded that there is too great an uncensinty as to exactly what the test l

conditions were to sup > ort the argument that the test adequately represented the-demand conditions, anc could therefore form the basis for the extrapolation of the.

test results out in time.

h IL MODEt ING OF THE PUMP FOR(T3 MPR's analysis regarding the fatigue evaluatio' of the pump shaft simply does not -

n account for the considerable uncenainties inherent in evaluating, through analyses, the operability of the pump under the extremely adverse and complex operating conditions associated with such low flow o>eration. The main thrust of their analysis was the use of maximum static shaft deflection, based upon wear ring.

bushing clearances, and using the associated shaft bending stresses thereby calculated as a basis for a fatigue failure evaluation. As was established at the August 17 meeting, they used the measured wear bushing clearance of the tested

. pump, not the maximum clearance possible per manufacturing dimensions and tolerances, and with no allowance for any in service bushing wear. Note that the as measured stuffing box ring diametral clearance was.025 mch (also note that the a

pump tested was a relatively new pump), compared to an allowable clearance per t procedure MP-131, " Maintenance of Building Spray and Decay Heat Pumps" of

.054 inch. Use of the allowable clearance instead of the as measured clearance would reduce the calculated factor of safety to less than half that documented in the MPR repon. Funhermore, a centered im >eller was assumed (i.e., there was no allowance for the fact that the potential ceflection could be more than half the diametral clearance). While these analysis flaws reduce considerably the factor of safety they could state from their approach, it is still not the fundamental flaw which is simply that no economically feasible analysis could provide a reliable assessment of this situation.

If properly implemented (i.e., with appropriate bushing clearance), the MPR approach is at best a "necessary condition" but not a " sufficient condition". That is, failure to pass the "MPR test" would surely be an indication of a problem, but passing it does not ensure that a problem does not exist.

At very low flow operation, the dynamic forces produced from the violent unsteady flow in and around the pump impeller are both axial as well as radial, ne MPR analysis focused on radial forces. The magnitudes and frequencies of such forces -

are always of considerable uncertainty, but typically very abusive to even the most robust centrifugal pump design. The specific pump in this case has a panicularly unfavorable configuranon from the point of view of rotor vibration.

Specifically, the impeller overhang is of the same order of magnitude as the bearing span. The possibilities for shaft failure from vibration phenomena are expanded f

considerably with such a relatively long overhang. This pump design, when i

originally configured for process service was supplied with shaft packing overinost of the overhang length, and such packing can have the added ac, vantage of acting like an extra bearing. The decay heat pump version, however, has the packing replaced with a mechanical seal, but with the same overhang length, now l

unsupponed.

t 5

..g

• l}

1 1

S? $f y

( ;f'[

y1

+'

No analyses or assessments have been provided to quantify the critical speeds,'

A instability threshold speeds'and unbalance response and sensitivity.1 Tnese are standard items to investigate in critical evaluations of machine reliability.- Not -

l knowing the operating parameters under which vibration resonance will occur and.

not knowing to what de such resonances are damped, leaves the integrity of the ?

pump under the de conditions quite uncenaln ' Since rolling contact bearings are employed in this pump, vinually no dampin contrast to oil film beanngs (e.g., journal bean,g is available from the bearings,I ngs, pivoted pad thrust bearings).

Thus, whatever dam ping is available must come from the pumped liquid, primarily at the wear ring bushings. The clearance and condition of these bushings is quite critical to their dam ing capacity. An acceleraad wear rate of these bushinJs would '

not necessarily uce an approximately linear vibration growth, but cou d in fact bring on a sudd increase in vibration after continual wear for some time.

It would be quite interesting to delve deeper into the vibration characteristics of this I

y pump-using-state-of the an computational tools and prior experience and p

. experimental data. However, it is our firm belief that the additional insights sol 4

pained would only funher strengthen the case for retesting; of the decay heat pump L

m c uestion.- While providing potentially valuable insigits, even such vibration L

ana yses could not form a legitimate substitute for an appropriate retest of the pump.

p

[

III. CONCLUSIONS.

L It is our judgement that the effon made by Florida Power to address concerns identified by Bulletin 88-04 on a shon turnaround time and in the narrow window -

of opponunity that became available during their outage this spring was made in u,

good faith. Unfonunately, the limited amount of time available to thoroughly review the situation has resulted in the performance of a test which does not bound v

(.

the condition of concem.

While we applaud the well intended approach used by Florida Power, we are not able to concur with the attempt to extrapolate the test results. First of all, the duration of operation at low flow was not bounded by the test. Secondly, there was no margin between the nominal test flow rate and the demand condition flow l

g:

rate, plus the fact that there was considerable uncenainty as to what the actual test I

flow rate was (which served to aggravate the lack of margin). Thirdly, the fatigue evaluation'of the pump shaft through static analysis does not account for-p uncenainties and complex conditions extant during operation at low-flow.

L It is recommended that a retest be conducted. In light ofinherent weaknesses in the ability to accurately monitor operating conditions at the plant, the schedular disadvantages associated with conduct of such tests, plus some conditions that are simply not reproducible at the plant (such as temperature and available NPSH),

consideration should be given to testing of a similar model pump under more controlled conditions.

t 6

.