ML20043B256

Forwards Simulations of Lasalle 2 Incident W/Bnl Plant Analyzer. Rept Summarizes Effort to Study Recent Event Using Engineering Plant Analyzer

ML20043B256
Person / Time
Site:	LaSalle
Issue date:	06/06/1988
From:	Kato W BROOKHAVEN NATIONAL LABORATORY
To:	Jordan E NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD)
Shared Package
ML20042D069	List: ML20042D087 ML20043B252 ML20043B256 ML20043B257 ML20043B264 ML20043B280 ML20043B372 ML20043B379
References
FOIA-90-13 NUDOCS 9005250156
Download: ML20043B256 (2)

Text

- _ __

S ,,.

g .. s .

r.n.. u.,

/f ..,

cm m ,e r,. 4 t rg. . c c s c mst on.cs son i OAR hlDGE ftNNE$$Et 37s31 j

. CAK RIDGE NATIONAL LABORATORY '

- ortRATED Sv 64ARTM MARCTTA tesEROY SYSTEMS. WC. May 17, 1988 l

l l

l

~

Mr. Tai Huang Reactor Systems Branch Division of Engineering and Systems Technology U.S. Nuclear Regulatory Commission MS VFN-8E23 Washington, DC 20555

Dear Tai:

Enclosed are our comments on Larry Phillips' repor't about the LaSalle stability event. I think that he did an excellent investigation of the issue and I agree completely with his conclusions. I will be out of the country

, from May 18 through May 31. Please note our new Laboratory address, which includes the internal mail stop in the nine digit zip code.

Sincerely,

~.. .. K W % L-Lou ..

Joss March Leuba Oak Ridge National Laboratory P. O. Box 2008 Oak Rid5e Tennessee, 37831-6010 Enc.

L cc: w/ enc:

E. D. Blakeman-l N. E. Clapp l D. N. Fry 1 L. D. Phillips, NRC

9005250156 900328 l' y PDR FOIA L HIATT90-13 PDR b

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

j; '% '

" ; e' .,

COMMENTS ON AUGMENTED INSPECTION TEAM REPORT OF LASALLE'S MARCH 9, 1988, STABILITY EVENT j Joss March-Leuba ,

Oak Ridge National Laboratory 1

May 17, 1988 j Main conclusions From a quick review of the available ' documentation of the

' March 9, 1988, stability event in the LaSalle boiling water reacter (BWR), we draw similar conclusions about the event as the AIT report. A summary of these conclusions follows:

a) The event seems to be due to the establishment of undamped  ;

limit cycle oscillations in the neutron field. .,

b)- Given the frequency and shape of the oscillations as well as the reactor operating conditions, the observed oscillations were most probably due to a core

, thermohydraulic instability.

c) The clean, single-frequency, sinusoidal nature of the oscillations indicates that the oscillations were core-wide and-in-phase. Thus, and although local power measurements are not available, large amplitude local oscillations were most,probably not present during the ever.t.

d) Based on our previous experience and calculations, the neutron flux oscillations observed in the LaSalle BWR most probably did not result in large enough heat flux .

oscillations at the fuel cladding to produce fuel damage.

e) Calculations failed to predict LaSa11e's susceptibility to

• instabilities.

f) Operator action, although following regulations at all times, was not adequate to suppress the oscillations in a timely fashion. SIL 380 recommendations are too vague and place large demands on the operator at a time when he is under serious stress trying to diagnose / recover the event that place the reactor under natural circulation conditions.

I 4 d' 1

?

,a .

observations a) From our previous calculations, large amplitude limit cycles are possible as the power is increased along the natural circulation line. For instance, Fig. 1 shows the calculated oscillation amplitude for the Vermont Yankee reactor. The oscillation amplitudes observed in the

-Vermont Yankee reactor tests were small because the power was not increased any further once the limit cycle was reached, but if the power had been increased further (for instance because of feedwater heater failure), the oscillations could have been of large amplitude. ,

b) Calculations show that a large amplitude limit cycle (approximately 100% peak to peak) will develop in two to three minutes (see for instance Fig. 2). The second STARTREC recording (6 minutes into the transient) shows a L fully developed limit cycle with amplitude modulated due to changing operating conditions. The higher amplitude oscillations seem to coincide with a minimum of the water I level and natural circulation flow, that appears to oscillate with a low frequency (approx. 60 second period).

Therefore, the automatic scram was probably caused by the water level oscillations, not by diverging neutron flux oscillations. If the water level stayed constant, the neutron flux oscillation would not have reached the scram point and the event might have continued until operator l

l action was taken (a maximum of 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> according to plant

. technical specifications). ,

l c) The control room monitoring equipment does not seem to be ad.eq'uate to detect this type of events. Unless the

- oscillations become large enough to sound the LPRM downscale alarms, limit cycles might pass undetected by operators that are busy attempting to recover from a pump trip. It is disturbing to realize that the control room APRM recording equipment does not have the frequency response necessary to observe this type of oscillations.

Thus, it seems possible that similar events might have l

[

happened before but, since they did not trip the reactor, I they were unrecognized / unreported. It would seem a good idea to review past pump trip events to determine if a general problem exists.

d) Calculations failed to predict susceptibility to instabilities in-the LaSalle BWR. In our previous reviews, we have pointed out that the most significant source of errors in predictive decay ratio calculations is the determination of the operating conditions at which calculations are to be performed. Once the reactor conditions and cross sections are well known, the codes workwithgoodaccuracy,butpyedictingthemostunstable 2

f,.. ..

w. .- ,

condition for a new core is a difficult and unprecise task. Another major unresolved problem is related to flow measurement. In all the tests we have been involved with, we have never been able to determine the flow accurately below 45%. For instance, in Browns Ferry, the flow instrumentation measures a value of 18% at natural .

circulation when this value,-according to GE, should be 1 close to 30%. BWR core-wide stability is extremely sensitive to flow level; thus calculations can be affected L

by this errors.

e) We have performed a quick LAPUR run using Grand Gulf cross sections and end of cycle axial power shape under the LaSalle event conditions (43% power 27% flow). The calculated decay ratio is 0.89 with an oscillating frequency of 0.4 Hz. Preliminary calculations with the new version of LAPUR capable of modelling out-of-phase instabilities show that the second (out-of-phase) mode is i more stable than the fundamental and, thus, coretwide instabilities should have been expect.ed for LaSalle.

Recommendations .

a) The proposed actions by LaSalle technical staff seem adequate to detect and suppress limit cycle oscillations.

In particular, the labeling of the high worth control rods for prompt insertion should be considered for all reactors susceptible to instabilities. When designing technical specifications it has to be recognized that a stability event.will most probably occur on top of one or two malfunctions as it did in LaSalle. The operator action should be automatic and he should not have to rely on on-the-spot-calculations or log books.

b) If possible, it would be appropriate to review old recirculation pump trip events to establish whether limit cycle oscillations of magnitude not large enough to cause a scram were present and undetected due to instrumentation deficiencies.

..,o . ' .c 3

_ _ __ _ _ _ _ . _ . _ _ _ - m _ _ . _

1

. \ :- ~ . ; '

150 i. i M  :  :

W  :  :

8  :  :  :

100 --

** i* -

y .

.CL . l -

3: . .

q . .

2 -

o m

50 - - - --

- ~ . .

~

U . .

y) . .

O  :  :  :

I I I '

O 50 55 60 -

65 70 POWER %

i.

1. Figure M Sensitivity of limit cycle esplitude to power along the natural circulation line.

l 120 i i i 120 t

'a' .

l ~

, 90 -

90 - -

1

! H cr w

s O

• wee "v,",yllfi))  ! l 60 -

l' I r

', 60 - -

I t

h l

hI I h W 30 ' ' ' ' -' - 30 0 25 50 75 100 125 150 175 160 175 TIME (s) T!HE (s)

(s) Limit cycle development (b) Detail of limit cycle oscillation

,,,_s. 'am:

2.

Figure Pf1 Development of a large amplitude limit cycle.

.i . ,

M *t o'Fict son 5 i

CAK RIDGE NATIONAL LABORATORY l opewto ov umn wwcru tw=ov systtus. we May 17, 1988 1

s Mr. Tai Huang

• Reactor Systems Branch Division of Engineering and Systems Technology U.S. Nuclear Regulatory Comaission WFN 8E23 Washington, DC 20555

Dear Tai:

bility A quick review of the NRC information notice rega'rding the LaSalle sta event has not shown any relevant technical problems.

positions taken in the document.

The strongest point appears in page 16 that says " ". . . the staf f beli I have performed

, d dI the ultimate power level without scram is unknown, ...severa if the reactor is made unstable agree in principle with that statement: In reality, however, enough, it can reach any amplitude of oscillation. lly limits the there is a range of reasonable parameters that oscillatLons. d hopefuTo my ith l

addressed this point of maximum believable oscillation amplitu e w reasonable calculations to back their position.

in the information notice you suggest the possibility Although ofthat installing is a ORNL I j d sound L online stability monitoring system in operating BWRs.

an alarm when an instability is reached.

based on simpler principles that the stability monitor,- that needs is stable.

to bility estimate the numerical value of the decay ratio while t in a reliable and fast manner.

l I

l

g.  ;~ - ~

.e With respect to GE's response letter, they just confirm our statement in all our previous reviews that the most difficult part of predicting the reactor  !

- stability is defining the most unstable operating condition during a cycle.

CE indicates that their main source of error was duo'to-the modeling of the whole core as a single channel, Based on our previous experience, we found that approximation to be a problem (we normally use 6 channels)'and estimated i the errors to be of the order of 104. - The 30% difference found in LaSalle  !

could be due to extreme radial power titling under natural circulation, ,

I I hope ~this review is of some help. I will call'you beek June 1.

Sincerely ,

i

~3 g ' .A>ck- M t .

Josd March Leuba cc: enc /w- l

, E. D. Blakeman I N. E. Clapp

. D. N. Fry

-i e

i

. . j i

4 8

4 #* , 3 e

C-2

., f DATE: May 9, 1988 ,

TO: J. G. Guppy ok "w f'

FROM: H. S. Cheng 4 ,fg

SUBJECT:

Simulations of the LaSalle Incident with BNL Plant Anal _yzer ABSTRACT The BNL Plant Analyzer has been used to simulate the recent incident occurred at LaSalle 2 Nuclear Power Plant (BWR/5). The observed growing power oscillation is of concern to NRC because of the implication that the reactor might be unstable at the time of the incident. The transient was initiated from the 85% power and 75% flow condition by a recirculation pump trip together with partial feedwater heater loss. The Plant Analyzer simulation showed a limit cycle oscillation of 3. 5-second period and 13% amplitude (26% peak-co-peak) .

The observed power oscillation exhibited a period of 2.3 seconds and diverging amplitude so as to eventually scram the reactor. It was postulated that a boiling instability due to a power / flow mismatch was responsible for the growing power oscillation. Provision was made to the BNL Plant Analyzer such that a power / flow mismatch could be produced interactively during the transient by applying a multiplier to the form loss at the core exit. Using this option, we were able to reproduce the diverging power oscillation as observed in the recent LaSalle event.

i.

L

1.0 INTRODUCTION

On March 9, 1988 an Instrument Maintenance (IM) technician at LaSalle Unit 2, while performing a functional test on Differential Pressure l

Switch, caused both recirculation pumps to trip off due to a valving l error. Due to the large and rapid power reduction, feedwater heater high' level alarms caused a partial isolation of feeadwater heaters, resulting in a total loss of 70 F of feedwater heating. About 5 minutes into the event, Local Power Range Monitor (LPRM) up- and down-scale alarms began annunciating and the Average Power Range Monitor (APRM) were observed to be oscillating between 25% and 50% power with an approximate 2-3 second period. Cognizant of the unit's location on the power-to-flow map, the shift was preparing to manually scram the reactor, when an automatic scram occurred on high flux trip (118% trip on APRM). Prior to the scram, the operators attempted to remedy the situation by trying to restart the recirculation pumps but failed.

The diverging power oscillation observed in this incident is of concern to NRC because of the implication that the reactor might be unstable at the time of , incident. sc Although the reactor was automatically shut down 'by the high-flux scram and no serious consequences resulted from this incident, important questions do I

M

e ,

2 arise: Why did it occur? Was the reactor unstable? What if the operator did restart the recirculation pumps? What if the Hain Steac Isolation Valves (MSIV) were inadvertently closed right after the pump restart?

In order to find some answers to these questions, we have simulated the LaSalle 2 incident using the BNL Plant Analyzer (BPA). Prior te the incident, the reactor was operating at 85% power and 75% flow with all rods out near the end of cycle 2. The transient was initiated by a recirculation pump trip (RPT) together with a partial feedwater heater failure due to steam isolation. This memo is issued to document the results of the BPA simulations of the incident as well as the important findings of the present analysis.

l l

l

<* A.