ML20065G706

Paper Entitled, Reactivity Accident Analysis of Univ of Missouri-Rolla Reactor Low Enriched U Core, Submitted for Natl Heat Transfer Conference in Philadelphia,Pa in Aug 1989

ML20065G706
Person / Time
Site:	University of Missouri-Rolla
Issue date:	01/31/1989
From:	Straka M MISSOURI, UNIV. OF, ROLLA, MO
To:
References
NUDOCS 9010230026
Download: ML20065G706 (15)

Text

s 1

[0-l13 .

l- ,

! I i

i Submitted for ,

{ National Heat Transfer Conference 1989 ,

j Philadelphia, PA, August 1989 -  !

l l

UMRR/88-2  ;

l REACTIVITY ACCIDENT ANALYSIS l

OF THE UMRR-LEU CORE ,

O.f MISs d n__ ,oU r.. u1,__ S 3

/ *

[

h5 hl

=

m

~

P'" I h

l

_ m o O oo 50o ooDD cooo oooo8oooo i SEACTOW  :

i i

M. Straka Rolla, January 1989 l

ih3 C '( i

)

9010230026 891231 ,

PDR ADOCK 05000123 '

_$ _ . .. _ .__ _ _ _ __ _ P D C , _ _ _ _ _ _ _ _ ,, _ , _ _ _ _ _ _ ,_ _ _ . , _

sr .

g

< 3; ,..

, , t ,

a 1- .u ' .j,

-* -; . , - 4 .

'e i

, i' ' i

}.. . . .

, c- ,

x , .. , .

Ng I

• 1 , ,. , )

. a d' a '< '

3 4 V

f. . y .

, 1.' ;I ^,

4 T

):

4

! 'l..

, x, -

UMRR/80-2 , ,

4 s

'8 i

q.  :

REACTIVIT ACCIDENT ANALYSIS .

('

v .-

OF THE UMRR-LEU CORE

' q '

T g

4 f (

. . i e '

L.

' 3 4

E -'

l i

[f%4 v

a 3 .'

V '

e g .

d- ,j, J O f;'. .

l  :. -

+

m .

4

,.....F a ,_..# .. +

i f o]

[.- m

=,-

, 1 -

F , - . R

• ;-v pe r>

i

-w t 5

_.3,

- i 7

. w- ,

j-Ns 4 Y

..i.;y. I

. h I.

.:s..= .,

w c.

(; 7 i-. - .

, s o-e-

a' 4

.. a ,

, m

.n,'

m _- g. ; -u 1 ,

s e- ,

4

,1 k '. - .~i I[' t y a 1

, s:

~

s .

,. w 0 . R.'. 5. j i.'u3' '

,z'y +

, , e , .

1 , v, J M . L. st r aka 'i ,

L , ., N-

- . ' . ' [, .

9y

, cRolla.cFall{1988 4 $ k T

. d [. _

' , . '-. y

. _3 - i 3 '

y , ,

4- _s -.

.)'

h 6 d '

y p3 i" ' i y1 s F c

' ~  ; r. 4 ' . : . ..,(;,..

[ / ..- '

, a.. . . s s a , t

' j.[k. p;Y- Y. - (

'l- ..-- g s <

,T. - f, - -

3.

W.\

._ g y

?'

.r l'

7 l_.

g"y , ,. 4 _ _ _

m

, g:s G

} {~\O.^ . i:} _f..'

e

. ' ( h . y _

1

• p- -,e

m-

~^lse

.t-e t .

{p-k

--r

g _}

-- , ~ vq- -.#%;4 e s t . .:.c

+

t

REACTIVITY ACCIDENT ANALYSIS OF THE UMRR-LEU CORE  ;

M. Straha Reactor facility University of Missouri-Rolla. Rolla. Missouri 65401 Unprotected reactivity induced power transients in a pool reactor have been investigated using a comouter code PARET. This code solves coupled thermal-hydraulics and ooint hinetics equations for a particular reactor channel geometrv. The course of a power transient is briefly described and the dominant thermal-hydraulic feedback is shown. Experimental data are used to compare results of the computer calculations. For a comparison, a protected power transient. i.e. with the reactor protection system activated, is shown, too.

• INTRODUCTION -

The University of Missouri-Rolla Reactor (UMRR) is' licensed for 200 kW steady-st at e thermal power . It is moderated.and cooled by l water circulating in the pool by natural convection. The reactor fuel consists of high-enriched uranium shaped as long, thin plates clad with aluminum. This fuel will be replaced in the near future with the low-enriched uranium (LEU) fuel of a.similar geometrv.

Eighteen f uel plates are f astened by aluminum side plates such that a finished LEU fuel element assembly.has the overall dimensions of 7.6 x 7.6 x 90 cm. A typical reactor core consists of about 18-19 fuel elements. The coarse reactor control is accomplished by means of 1

1 . .

Severhl ht'utron hDBorbing Shim /S8f ety rods Which Can Shutdown the reactor in less than 100 msec in an emergency situation, e

]

Before the low-enriched fuel can be installed in the UMR Reactor a cafety analysis of the new reactor core must be ' periors,ed and i t s results documented in the revised Safety Analysis Report. A considerable part involves the analysis of two reactivity induced power transients.

Each power transient commences with a large reactivity insertion assumed to occur in a very conservatively postulated accident scenario. For example, it is assumed that the reactor protection system remains inactive during the entire power excursion. The more severe case, an instantaneous insertion of 1.5% .

reactivity into the LEU core will be discussed in this paper.

ACCIDENT SCENARIO AND ITS SIMULATION Initially, the UMR Reactor is assumed to operate at some low power . f or example 1 W. with a correspondingly low coolant flow rate.

The accident scenario postulates that a reactivity of 1.5% is inserted into the reactor in a step-wise fashion. This causes the reactor to become prompt supercritical with a steep power increase during which the power doubles approximately every 5 meec. As the power increases, the f uel and moderator / coolant (almost stagnating) heat up correspondingly. This affects the neutron multiplication and honce a further power increase through such feedback effects as the

' f uel expansion , resonance cross-section broadening (Doppler effect).

N cnd change in the moderator temperature and density. In a thermal cnd water moderated reactor such as the UMRR each of these ef f ects will give rise to a f eedback with a wide varying degree of magnitude.

A computer code used to describe this rather complicated 2

I

,. i l .

I i

behavior quantitatively must therefore include neutron kinetics and thermal-hydraulics equat ions. The PARET code satisfies these requirements. It has been recently revised and proposed (11 for use in predicting the course and consequences of reactivity gecidents in 1

l l

small reactors. The neutron kinetics equations are used in a point (i.e. lumped) f ashion, while the thermal-hydraulics equations are solved f or a one-dimensional reactor channel geometry. A reactor channel consists of a fuel plate and its associated moderator / coolant '

region. The thermal-hydraulics model is based on a conduction equation solved in the radial direction in the fuel region and a '

modified momentum integral method (MIM) to nolve incompressible equations of mass. momentum, and energy conservation in the coolant region.

The modified MIM defines the coolant density as a function of enthalpy and a local pressure [2] but retains the assumption that a channel averaged mass velocity can be used for the momentum equation. The channel averaged mass velocity G is defined as 1 /b (1)

G = 3- f Gdz where L is the reactor channel length. This enables to integrate the nomentum equation along the reactor channel to obtain an ordinary j differential equation

={(op-F) . (2) where Ap is the total channel pressure drop and F is a sum of the- 1 frictional, elevation and spatial acceleration pressure drop. The number of partial differential equations to be solved in the coolant region has been reduced by.one. In addition, the explicit time formulation has been'uned in finite difference equations.

3

l 1

Consequently, the computing time, required to solve a problem like the one presented in this paper has been considerably reduced.

Because of the M1M procedure, a disturbance propagating with the speed of sound (e.g. a sudden pressure reduction at the qhannel inlet) cannot be described well. Therefore, the PARET code is i

limited to pressure and velocity transients in which their duration is longer than the time it takes f or a sonic wave to pass through the reactor channel . This requirement is satisfied in the present investigation of transients induced by a change in power. '

Physical conditions of the coolant can range from subcooled liquid up to the superheated steam. In the case of saturated boiling voiding is determined by calculating the mass flow fraction of vapor and using the Martinelli-Nelson correlation. An option to calculate voiding due to subcooled boiling based on a simplified model (3) is provided in PARET too.

The boundary condition at th: clad-coolant interface is provided by one of the available heat transf er correlations. The code has been revised such that more than one correlation is now available f or oither the single or the two phase flow. The procedure to determine the heat transf er regime at a particular node is based upon consideration of the cladding surface temperature and the surface heat flux.

In PARET. up to four channels can be solved simultaneously f or a reactor: Two channels have been used in this analysis:

a " hot" channel simulating a fuel plate with the maximum heat generation rate and an " average" channel representing the remaining fuel plates. The ratio of power generation in the hot and average channel is about 1.8.

N l

l

• QISCUSSION OF RESULTS i In Figure 1.

a characteristic trace is shown as calculated f or '

the nower excursion due to a large ttepware reactivitv incertion ei 1.5% described in the previous section. The maximum reactor power of about 600 MW is reached at 0.144 sec . The full-width half-maximum of the trace is about 10 msec.

In the analvsis of short nower excursions the t ot al ener gy

• released and the resultinF fuel and clad temperatures are the most important par amet er s . In Figure 1. the t emperature of fuel, elad and coolant in the not channel are shown for this power transient. As expected, the coolant temperature lags f ar behind the clad and fuel temperatures. The maximum temperature of claddina is reached at about 0.16 see and amount s to 435 + C which is st ill distinct ly below i it s melt ing t emperature.

In the reactor averate channel a fuel temperature of 245 eC is indicated while the fission energy releared during the transient amount s t o about 8 MWsee. Subcooled boiling occurs in the central region of both channels as early as at 0.137 sec. A t' some nodes, transition and film boilinF is later observed.

Boiling causes moderator / coolant to void which in turn produces a strong negative feedback (about -0.70%) in this region. This effect and the Doppler effect in the fuel (about

-0 .1 M 1 reverse the t:ien of the power tranuient and bring the reactor to delaved supercritical at the time 0.144 sec. From this time on. therefore, the reactor power rapidly decreases and the transient quickly dies awav.

A number of model parameters have been ident2fied as being ouife important for the simulation. While some of them, for example, the 1

bubble life time, have been adiusted by using t he dat a available in the literature, others have been det ermined -in the reactor phvsies 5

calculations t5).

The influence of the moderator / coolant coefficient .. on the simulation results is shown in Figure 2. The clad temperature as calculated in the hot channel of the reactor undergoing e power transient is shown for two different values of .. (in units reactivity /% of core voided). The value of 0.3 has been det ermined for the UMR Reactor in the reactor physics calculations and has been ,

used in the safety analysis. Reducing this value by about 50% eauses the moderator / coolant density feedback to decrease and while the maximum power reached in the reactor t r an si en t with this reduced parameter is about the same as the one shown in Figure 1. its width has increased. (The full-width half-maximum is 0.13 msec.)

Therefore, the released energy has increased accordingly and manifests itself by driving the clad temperature hieher. Noticeab1v.

the plateau at the maximum temperature became much broader t oo f or the case of reduced v (Figure 2).

The results of calculations have been compared with the experimental data collected at the SPERT facility (61. A series of power excursion tests has bec;. carried out at this facility using various reactor cores. The data measured include the reactor power cnd the clad temperature. They are shown in Figure 3 together with the released energy versus the reciprocal reactor period. ,The broken lines show the calculational results obtained with the PARET code.

All of them seem t o be consistently overest imated. It is. suspected that this is due to a value of the syst em pressure used in our calculations. It was selected 170 kPa but it was learned later that

! it should have been 100 kPa. It is felt that a higher system pressure used in our calculations suppressed moderator / coolant 6

l

a * ..

• boiling and hence reduced the negat ive f eedback thus driving the calculated reactor power higher than the observed one. The clad l

temperature and the released energy results follow a similar pattern.

It is instructive to see how this reactivity induced transient.  !

as postulated above, would look like in the UKR Reactor if a credit for the reactor protection system were to be claimed. Figure 4 shows 1

corresponding transient traces as calculated for this case. Both the 1

reactor power and the fuel and clad temperature s are significantly reduced. The energy released in this protected transient does not exceed 6 MWsec at any time. Nevertheless, it is sufficient to cause a short period of boiling in both reactor channels. For example. in the average channel come modest nucleate boiling occurs in the central region, however. it concurrently ceases with the rapidly decreasing power.

CONCLUSIONS Results of the present computer simulation show that the UKR Reactor can sustain an unprotected power excursion caused by a severe cudden reactivity insertion. The transient is terminated by a few inherent and strong negative feedbacks. The feedback mainly responsible for the self-termination of a power excursion is due to the early occurrence of subcooled boiling. Its regime include nucleate, transition, and film boiling as indicated in the calculations. This is a rather surprising result considering the speed of power increase. However, the short time constant of the thin f uel plates allows for a large amount of energy to be transf erred to the clad surf ace and into the thermal layer of the coolant in a very short period of time, thus causing an early 7

4

! . . a 1-  :

i boiling.

~

The second import ant feedback is due to broadening of the uranium-238 resonance cross-sect ion . This is an effect which takes i place instantaneously. It contributes about 20% of the qverall feedback.  !

One of the model parameters identified as being sensitive in the simulation is the moderator / coolant temperature coefficient. Its margin is shown to be f airly narrow and could, ultimately, decide about the acceptance of the accident consequences. It further demonstrates the importance of reliable reactor physics calculations for the purpose to determine this coefficient.

The simulation results are in a f airly good agreement with the measured data. Although some parameters in the boiling model had to be adjusted, it seems that consistent results are obtained once a proper adjustment has been found. It appears, therefore. that the PARET code is suitable for calculating power transients in research reactors such as the UMR Reactor.

REFERENCES (1) Woodruf f W. L. , "A Kinetics and Thermal-Hydraulics Capability I

f or the Analysis of Research Reactors". J. of Nuclear l

Technolomv. Vol. 64 February 1984 (2) Obenchain F., "PARET - A Program for the Analysis of Reactor Transients". IDO-17282 Idaho National Engineering Laboratory.

Janu ary 1969 1

I 8 e --

(3) Zuber N. et al . " Vapor Void Franction in Subcooled Boiling and in Saturated Boiling Systems". Proc . of the Third International Brat Transfer Conferengg. Chicago. Illinois. August ly66 (5) Straka M. , Covingto's L. . " Study of Neutron Physics: Conversion of the University of Missouri-Rolla Reactor to Low Enriched Fu e l " . Tr an s . Am. Nucl. Soc., 11 (1987) 594 (6) Miller R.W. et al. " Experimental Results and Damage Effects of Destructive Tests". Trans. Am. Nucl. Soc.. E (1963) 138 This work has been supported by a grant . DE FG02 86ER75272,

" Conversion of the University of Missouri-Rolla Reactor from High-Enriched' Uranium to Low-Enriched Uranium- Fuel" f rom the U.S.

l Department of Energy.

1 l

l I

9

- 1 o . e-I t w.

500- P0vt R i

L000

('C)

(MW)

.$00 400 -

400 300 Power 1 300 200 Fuel 200 Clad

'00' Moderator / coolant 100-0- --

i i i i i i i i 0

i i i 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0 18 0,20 11ME (Sec)

Fig. 1. LEU Core: Step-wise Reactivity Insertion (Unprotected) l i

)

7 a

e 1( P188 000'

(*C) "v  : 0.15 500 -

j ay = 0.3 400 .

300-200-Onset of Boiling l00-l 0-i e i i i i i i i i i 0.00 0.02 0.04 0.06 0.08 0.10 0.I2 0.I4 0.16 0.I8 0.20 ttrE (sec)

Fig. 2.

Clad Temperature in Hot Channel Shown for 'No Values of the Moderator / Coolant Coefficient

, 1

l

\

4

_ l l 1 I 3II I I .

PUK .- - \'

l POWER l 10

( Mw) -

/ e ..  ;

- ,/ , -

/ "

s -

js

' ;g T~~~~~~,'

MAX. SURF,

-. /

/ TEMP.(CC ) _: _

,/ -

/

' TOTAL ,A -

-f

. ENERGY <

(Mw-sec) ,

10 -

=.

e E.:-~~'~

- .

• PRESS ~

_ . ( p sig ) -

DESTRUCTIVE 1 _

TEST J l > > i linin' e i

10 10 0 500 RECIPROCAL PERIOD ( sec-8 )

Fig. 3. Comparison with SPERT - Excursion Data

(- SPERT -- - - PARET) I

s. 1 1

flHP. '

poggp

, 400' l tro

( ('C) (MW) 160 1 1

150 140 i

300 .

-l 0' Power #*

110 1

Fuel 90 1 200' Clad 80 to l

60 k

50 100-40

. 30 Moderator / Coolant 20 l

j

~

10 0

i i i i c i e i i 0

i 0.00 0 05 0 10 0.15 0.20 0.25. 0 30 0.35 0.40 tit 1E (sec)

Fig. 11, LEU Core: Step-wise Reactivity Insertion (Protected) i l

1 e

i i

i

% l -]

. _.