ML19318B814
| ML19318B814 | |
| Person / Time | |
|---|---|
| Issue date: | 06/30/1977 |
| From: | Erdman C, Mary Johnson Office of Nuclear Reactor Regulation, VIRGINIA, UNIV. OF, CHARLOTTESVILLE, VA |
| To: | |
| References | |
| CON-AT(49-24)-0158, CON-AT(49-24)-158 NUREG-0282, NUREG-282, NUDOCS 8006300102 | |
| Download: ML19318B814 (26) | |
Text
_
NUREG-0282 SOME ASPECTS OF FUEL MOTION
]
IN THE UNPROTECTED TRANSIENT OVERPOWER ACCIDENT IN LMFBRs 4
l i
f i
i I
1 i
University of Virginia for U. S. Nuclear Regulatory Commission 8006300/02
NOTICE This report was prepared as an account of work sponsored by the United States Government. Neither the United States nor the United States Nuclear Regulatory Commission, nor any of their employees, nor any of their contractors, subcentractors, or their employees, makes any warranty, express or implied, nor assumes any legal liability or responsibility for the accuracy, completeness or usefulness of any information, apparatus, pro-duct.or process disclosed, nor represents that its use would not infringe privately owned rights.
Av2. ;~able from National Technical Information Service-Sprnigfield, Virginia 22161 Price:. Printed Copy $4.00 ; Microfiche' $3.00 The price of this document for requestors outside of the North American Continent can be obtained
- from the National Technical Information Service.
1 SOME ASPECTS OF FUEL MOTION IN THE UNPROTECTED TRANSIENT OVERPOWER ACCIDENT IN LMFBRs C. A. Erdman M. B. Johnson Manuscript Completed: April 1977 Date Published: June 1977 University of Virginia Schoolaf Engir.eering and Applied Science Charlottesville, VA 22901
'repared for Division of Project Management U. S. Nuclear Regulatory Commission Under Contract No. AT(49-24)-0158
ABSTRACT This paper addresses certain aspects of the fuel motion which would be associated with a hypothetical, unprotected, transient overpower accident (TOP) and loss-of-flow-driven TOP accident (LOF-D-TOP).
Experimental results are cited that suggest a time lag between failure and fuel ejection for irradiated pins. This time interval is needed for development of large rupture areas. When fuel does reach the channel, rapid upward dispersal appears likely, but so does eventual flow blockage.
Arguments are presenced for the availability of sufficient driving forces for sodium expelsion in the LOF-D-TOP with negligible axial fuel motion.
Results of calculations on plenum gas release are presented which indicate that gas driven voiding would not be a direct problem in either the TOP or LOF-D-TOP accident.
(
i
TABLE OF CONTENTS Page y
ABSTRACT.
i l.
FUEL EXPULSION DYNAMICS IN THE CLASSICAL TOP....
1 1.1 Pin to Channel 1
1.2 Expulsion from the Cora.
3 2.
VOIDING DUE TO FUEL-GA5 PRESSURIZATION IN THE LOF-D-TOP 6
2.1 Gas' Availability at Failure.
6 2.2 Fuel Motion in the Channel 7
3.
PLENUM GAS RELEASE.
11 3.1 TOP Plenum Gas Release 12 3.2 LOF-D-TOP Plenum Gas C, lease 17' 18 3.3 Summary.
REFERENCES.
20 1
1 ii
A i
1.
FUEL EXPULSION DYNAMICS IN THE CLASSICAL TOP 1.1 Pin to Channel Fuel expulsion from the fuel pin in either the classical transient overpower (TOP) accident or the loss-of-flow-driven TOP accident (LOF-D-TOP) is not well understood at this time.
The lack of under-standing is due to the complexity of events leading up to fuel ex-pulsion, in particular in the generation of driving forces for cladding failure and in the nature of cladding failure itself.
The following discussion focuses on TOP.
For the last few years most analysts have favored the concept of a high pressure " sealed-bottle" of fission gas and molten fuel which acts through surrounding strengthless solid fuel to stress the cladding to failure.
The mixture of gas and molten fuel is then expelled from the pin through the cladding rupture. Other investigators (e.g. Ref. 2) argue that cladding failure should be dominated by differential fuel-cladding thermal expansion and by fuel expansion j
upon melting. Arguments can also be made for intragranular fuel swelling due to fission gas bubbles or frictional interaction between pellets and cladding as the dominant mechanism.
(Note that the first and third mechanisms listed above would not be applicable to very-low-burnup pins.)
Assuming for the moment that a failure mechanism and its timing can be identified, what can we say about the nature of the failure?
Is the rupture of sufficient araa that fuel and gas from inside the pin can ese ae 11y? This is one ideal situation that is modeled in currently svu11able codes.5,6 All resistance to flow is considered to be found within the-pin cavity and/or within the coolant channel.
In reality, solid materia) immediately behind the rupture may offer significant resistance to expulsion of molten fuel from the pin if the rupture is of small area. The resolution of such issues is currently being sought by various investigators (e.g. Ref. 7).
1
i The driving force for ejection of significant quantities of fuel from failed, irradiated pins, no matter shat the failure mechanism, is fission gas. Depending on the mechanism postulated, the gas would be gas initially in the central cavity in normal operation and/or transient released gas.
For very-low-burnup pins, fuel vapor pressure or fabrication / fill gas would be the only available driving forces.
The above discussion suggests several things, two of which can be partially checied by available experimental results.
First, of all, different failure mechanisms will be operative for different types of pins. This is certainly supported by the TREAT experiments, which show a clear failure threshold dependence on fuel burnup and rate of transient energy deposition.
Second, there may be a significant lag between initial failure and ejection of significant masses of fuel from the pin into the coolant channel. This may have been the situation in tests E3, E6, and E7, where delays of approximately 30, 70, and 300 ms were noted between initial pin failure and detectable fuel motion.
Also a delt; of the order of 100 ms was experienced in the R9 test 9 (only fresh pins) between initial failure and observed fuel motion.
We postulate that the delays observed between initial failure-and significant fuel motion from the pin are related to the time necessary to develop an easy access from the pin center to the coolant channel. We refer here to the time required to create a cladding breach of significant area at the failure site and to remove solid fuel directly behind the rupture which would otherwise impede the flow of molten fuel / gas from the pin cavity.
Indeed the initial failure might occur at an axial location above the molten fuel cavity, making the delay even greater.
The early post-failure pressurizations for the irradiated pin tests mentioned above could be explained by fission gas release from either the plenum or central void or solid fuel.
In R9 we have the first TOP test run on a full length FTR type pin, and a phenomenon not observed in earlier TREAT 2
tests on very-low-burnup pins may have been involved. For earlier J
TREAT tests on fresh pins, fuel motion seemed to immediately follow failure.
There is no inconsistency in having a delay between initial failure and fuel motion in irradiated pins but not in fresh pins.
This delay, remember, is postulated to be needed to develop large area flow paths. For fresh pins (e.g. H2, E4) the initial failure is preceded by local boiling lasting the order of 40 to 50 ms which can produce sodium film dryout and a large region of near-melting i
cladding.8 The resulting failure could be quite extensive. This s
is combined with very high fuel melt fractions to product good communication between the molten fuel and coolant channel immediately on failure. Very hot cladding can be achieved before failure in fresh pins because pressures inside the pin are too low to rupture the cladding at low temperatures. This is, of course, not the case for irradiated pins where, as a minimum, the plenum pressure would always be experienced at the cladding I.D.
l For irradiated pins, initial small ruptures would release gas which in turn could cause propagation of the rupture in one of the several ways suggested previously for pin-to-pin failure propagation.
(tiote :
Ref. 10 argues against such propagation at nominal power but not for the overpower situations of interest to us here.)
Eventually then, for either irradiated or fresh pins, molten fuel will be ejected from the pin through a rupture of substantial area into a coolant channel which contains fission gas and/or sodium vapor in addition to liquid sodium.
This leads us then to a dis-cussion of the dynamics of motion of. fuel within the channel.
i 1.2 Expulsion from the Core i
i A consideration of the forces available for moving fuel within the coolant channel indicates the ease with which fuel might be 1
3
expected to be removed from the active core region once it has entered the coolant channel. For example, the normal pin-bundle pressure gradient in high flow subassemblies of CRBR is approximately 0.108 MPa/m or N 1.1 atm/m. (This is larger than the pressure gradient in a standit.g column of molten fuel, which would be about 0.085 MPa/m.) One can show by performing calculations similar to those of Ref. 11, that the normal liquid sodium flow would produce 2
an initial upward acceleration of 600 m/s on a stationary particle 4
which is just small enough to fit into the coolant channel (radius
% 1.25 mm).
Turning to a situation in which this large fuel particle is being moved by flowing vapor rather than liquid, assume that the pressure gradient in the vapor is twice the normal-flow gradient, i.e. 0.216 MPa/m. This assumption is based on a scenario of failure of the central pins within a subassembly with resulting flow diversion to the outer regions of the subassembly. Utilizing sodium vapor conditions characteristic of saturated sodium vapor at a pressure of 0.5 MPa and assuming applicability of the Blasius equation for the friction factor, we find the vapor velocity would be about 230 m/s. The initial acceleration for the 2.5 mm diameter 2
particle we referred to above would be about 970 m/s. Once again l
the upward acceir. ration is quite strong. What then keeps the ejected fuel from immediately being swept away? The principal early factor is, of course, the generation of large pressures in the coolant channel at the rupture site either by gas release or a fuel-coolant-interaction (FCI). A small impulse (%.01 MPa s) is sufficient to stop coolant flow, and a continuing pressure of the order of only
.8 MPa would establish and maintain downward voiding.
In Ref. 12 we discussed minimum requirements for producing voiding from an FCI and from plenum gas release.
In Section 3 of this report there is a further discussion of plenum gas voiding. An 4
efficient FCI involving only a few grams of fuel per pin could pro-duce flow reversal and extensive downward voiding. Plenum gas re-lease shows less potential for such voiding but would produce strong flow perturbations and probably induce secondary pin failures.
In addition to the above possibilities, a combined fuel / gas release from the active core region could produce significant downward voiding.
This scenario has been suggested for the E8 test.
The second major factor in preventing or reducing fuel expulsion from the core region is the creation of flow blockages by the plate-out of fuel or a fuel / steel mixture on the channel walls.
In virtually every flowing sodium test in TREAT, substantial blockages have been observed.
Formation of these blockages is probably enhanced by significant early channel voiding such as discussed above. The key question is: would we see these blockages in real reactors? The only attempts to mechanistically model the fuel freezing problem (e.g. Ref. 14) in prototypic geometry have so far produced zero-order models which are still in the development stage.
The most substantial piece of experimental evidence obtained so far comes from the R9 TREAT test.
Here a fresh, full length, FTR-type pin with prototypic flow conditions (i.e. flow rates, driving head, pin length, etc.) was subjected to a 500/sec TOP. Although there appears to be substantial upward fuel motion following pin failures, a complete flow blockage was observed.
In light of the above discussion and that of Ref. 12, our conclusions at this point on expulsion of fuel from the core region are as follows:
general upward dispersal will be effected initially partial or complete blockages are likely due to fuel or fuel /
steel freezing, but the location and timing of these blockages is still in question significant early voiding probably enhances development of blockages.
5 i
t 2.
VOIDING DUE TO FUEL-GAS PRESSURIZATICN IN THE LOF-D-TOP We have previously discussed three potential mechanisms for voiding in the LOF-D-TOP accident: FCI, plenum gas release, and fuel-gas pressurization. This section deals with the th.rd mechanism.
The term squeeze-out has sometimes been used to ide itify this mechanism for lack of a better term. We are referiing to voiding associated with sodium pressurization resulting from fuel and/or gas attempting to enter the coolant channel through a cladding rip.
The term fuel-gas refers to steady-state central void gas plus gas released from the fuel during the transient - not gas from the fission gas plenum.
Reference 12 provides some simple treatments of this mechanism. Two things will be added here: a further discussion of driving force, i.e. gas pressures, and a treatment of relative fuel /
sodium motion in the channel.
2.1 Gas Availability at Failure i
As indicated previously,12 a consistent treatment of fuel i
cracking, crack healing, and restructuring based on actual operating history would result in a prediction of larger central void volumes than the SSFUEL module of SAS3A predicts. 5 Including SAS predicted 1
gap volumes (or some reasonable fraction thereof) in the central void volume would give central void volumes of 0.5 to 1.0 cc at steady state for the high burnup-low power channels of interest in the LOF-D-TOP.
(Remember that these were once higher power pins and were restructured accordingly.) Gas masses for a steady state pressure of 5 MPa (% 50 atm) and a temperature of 2000 K would be about 20 to 40 mg.
If this gas were to still occupy 1 cc but at a temperature of 3100 K (% T f r fuel) at pin failure, the pressure would be 7.75 melt MPa.
Allowing the gas to expand isentropically to fill one third of the active core coolant volume associated with a single pin would drop the pressure to the range of.19 to.38 MPa.
Note that the inlet 6
pressures at this point in the accident would be only the order of
.3 to.4 MPa.
Carrying the analysis a step further and allowing for pre-failure transient gas release to aid in the pressurization, we see that the voiding potential is significant, especially in the period immediately following failure. This really would also be true of most classical TOP cases, but the important difference in the LOF-D-TOP is the possibility of extended rips.
2.2 Fuel Motion in the Channel The calculations of Ref. 12 assumed sodium slug expulsion based on pressurization of sodium in the regions of the failure. If the pressurization involves motion of fuel out of the cladding (either l
by an extrusion process or a hinging of the cladding away from the rupture), the fuel motion must be considered.
It is important to note that we are most interested in the first 5 to 10 ms following failure, because SAS calculations suggest that sodium motion would be most critical in this period.
On such a time scale, fragmentation cannot be demonstrated, especially if coolant velocities are initially very low.
Therefore, fuel motion will mostly be governed by the same pressure gradient which moves the adjacent sodium.
We have performed a series of calculations which examine the relative motion of sodium and fuel.
The calculational sequence is as follows. At time zero (failure) a stagnant column of sodium is assumed to exist in the coolant channel.
The driving pressure is suddenly applied to the sodium slugs which extend above and below the rupture site. Motion of the sodium slugs is calculated for the first time step (using an acoustic constraint).
Fuel is assumed to leave the pin and fill the entire volume made available by the slug motion. During the next time step, volume is again made available l
7
to the failure zone sodium and fuel. The fuel already in the channel moves away from the failure site at a volumetric flow rate t p, Ag equal t times the rate of sodium motion, where p, and pg A
L f si are sodium and fuel densities respectively, and A and A m&
f g
channel flow areas occupied by fuel and sodium at the failure site.
Fuel again enters the channel to fill the available volume.
This process is repeated. The equations for the ith step are shown below:
1-1
[ A AL i
AV =AL +
f j=1 AL
= 2uAtiA si j,1-1 f
I i-1[A A
=
ff j=1 AP f
(acoustic) l u=
I ss or du, O i (inertial) dt p,L A = area occupied by fuel introduced in jth step A ) = total fuel area at the failure site at jth step f
A,3 = channel area occupied by codium at the failure site at the jth step L, = rupture length L = length of constraining sodium slug 8
AV = volume available for fuel ejection at ith step AL
= motion of jth fuel segment in ith step AP = driving pressure for ith step g
u = velocity of sodium slug c = velocity of sound in sodium A series of calculations was carried out which looked at various rip lengths (.05,.15, and.3 m) for two different values (4.2 and 1.05 MPa) of initial pressure and 1 x 10 m of gas.
The table below gives i
6 3 values for the maximum length of coolant channel which contains fuel at a time of 10 ms following failure.
Maximum Length Occupied by Fuel (m)
Rip PVY = 4 x 10~4 PVY = 1 x 10~4(Pa*m )
5 Length (m)
.05
.113
.051
.15
.156
.150
.30
.302
.300 Axial motion of the fuel is obviously very small during the first 10 ms for cases where a long rip is considered. The other quantity of interest is the amount of sodium which has been squeezed out beyond Y
4 the fuel. For the case with rip length =.15 m and P,V, = 4 x 10 Pa m,about 65% of the sodium initially at the rip (% 1.55g) has 5
been pushed ahead of the fuel. The total fuel mass outside the original rip region is about 1/3 g, but the reactivity associated with the small axial motion (an average distance of %.038 m) for an axially centered break would be insignificant and is partially balanced by internal fuel motion. Moreover, the in-channel motion effect is even smaller 9-
for longer ruptures and/or smaller values of P V.
Ignoring frictional forces between the accelerating sodium and fuel can be, simply argued on the basis of small values of the quantity vt/Dj (analogous to the Fourier modulus for heat transfer), where v = dynamic viscosity of liquid sodium D = hydraulic diameter for sodium flow H
t = time since pin failure For the cases discussed above, the quantity never exceeds % 2 x 104 The results presented above and the discussion of Section 2.1 give support to the occurrence of sodium voiding due to fuel-gas pressurization in the LOF-D-TOP if the basic premise of long rips is accepted. The question of failure mode and nature in the LOF-D-TOP remains to be resolved. As indicated in Section 3, the only voiding phenomena which now seem to us capable of producing fast voiding in the LOF-D-TOP are fuel gas pressurization and FCI, and both of these are critically dependent on the nature of the pin failure.
4 1
10
3.
PLENUM GAS RELEASE The fission gas accumulated in high burnup fuel pins constitutes another potential source for subassembly voiding following pin failure.
Unlike the FCI and the fuel gas pressurization effects which are associated with failures well down into the active core region, signiricant plenum gas releases should result only from cladding failures above the active core.
The positive reactivity effects due to this voiding mechanism are therefore limited to 'ownward voiding of the coolant channel, and are expected to be milder than reactivity effects generated by the FCI or fuel gas expansion. High pressures and relatively large masses of gas available in the plenum, however, might be expected to yield extensive and sustain'ed subassembly voiding prior to ejection of molten fuel from the pin.
(This scenario should favor fuel plateout and plugging of the coolant channela.)
With these possibilities in mind, we have examined the voiding histories of plenum gas release in the TOP and in the LOF-D-TOP scenarios as modeled by the two-slug inertial coolant constraint /
isothermal gas flow coding in GASFLO. We have concluded that plenum gas release alone is not likely to cause extensive voiding of the active core in the classical TOP accident. For the LOF-D-TOP accident, gas release can produce extensive long-term voiding of the ective core if pessimistic initial conditions are assumed; however, voiding rates immediately following failure are not large enoagh to give the positive react!vity ramp rates that are of interest in the LOF-D-TOP.
Since subsequent events in high-power LOF channels dominate the accident history, we have also concluded that plenum gas release voiding is not likely to be important to the LOF-D-TOP.
The mechanism considered here for plenum gas escape is flow through a thin annular gap between the upper blanket pellets and the cladding to a rupture at the top of the active core.
The highest cladding temperatures found above the active core occur here; also, l
11 f
O fission-product attack may preferentially weaken the cladding at this location. Ruptures at higher axial locations resulting in direct communication between plenum and channel can follow major accident events in the active core (for example, a strong FCI).
Such events, however, would most probably also severely damage the subassembly geometry, and render both the models we use and the phenomena we are considering here irrelevant.
On the other hand, release of plenum gas from ruptures. lower than the top of the active core, which would increase the reactivity consequences of equivalent gas release rates, is too slow to be important. The annular flow path for gas is effectively closed in the active core for irradiated fuel. Also, during the accident, the retained fission gas in the fuel and in the central void will attain pressures higher than that of the plenum.
A gas flow from the plenum to a low cladding rupture under these i
conditions is improbabic.
As in the rest of this report, our consideration is limited to independent voiding events and their direct effects.
Our analysis cannot include some secondary effects of plenum gas release that may f
be important to accident histories.
Due to coding difficulties, we have airo not considered the ameliorating effects of noncoherent pin failure in the subassembly.
However, the nature of our conclusions does not suggest a need for 6
more realistic evaluation of noncoherence except in extreme cases, I
which. are mentioned below.
3.1 TOP Plenum Cas Release As indicated in Section 1.2, significant voiding of a CRBR subassembly under transient overpower conditions requires coolant channel overpressures imparting over 0.01 MPa s (0.1 atm s) to the lower sodius slug. The overpressures also must remain significantly higher than the 0.76 MPa subassembly inlet pressure. Although plenum gas release overpressures can meet both of these criteria, they do so only marginally and under assumed conditions which we consider un-12
.d realistic for the E0EC core..
In earlier work on plenum gas release by ANL, both the initial plenum gas pressure (1750 psia, or 12 MPa) and gap size (0.165 mm) which were utilized exceed our estimates of reasonable values for these parameters. The steady state pressure of 1750 psia used in Reference 17 is the upper limit value (2o confidence level) given in the CRBR PSAR for the highest burnup pin in the design E0E0 core.
This pressure does not realistically characterize whole subassemblies.
Furthermore, it does not reflect the available pressures at the lower burnups present in the early CRBR cores.
i' By assuming complete release to the plenum of all fission gas generated in the fuel, we can conservatively estimate the pressure of the plenum gas as follows:
0.274B x 0.153 x 103 T/ pin
= 1.81 x 10 10 kg mole n
=
8 2.32 x 105 mwd /kg mole fissions n RT 8 #
= 0.072B x T P P
=
p 20.88 x 106 3
Pa m
where B is an average burnup in mwd /T for the subassembly, and Tp is the initial plenum temperature. A fission gas yield fraction of 0.274 is used..In the case of the maximum burnup CRBR subassembly
-for early cycles (80000 mwd /T) and an initial plenum temperature of 1100 K.(a temperature. representative of transient failure conditions),
the predicted pressure is 6.3 MPa (919 psia); again, this assumes zero fission gas retention in the fuel.. Our interest in the early CRBR cores led us.to select'7 MPa as a reference initial plenum i
pressure. At higher burnups, slightly higher pressures should be i
't 13
used for conservative analysis.
Figure 1 compares predicted voiding histories for plenum gas release at the top of the active core and at initial pressures of 12 MPa and 7 MPa.
The gap size is 0.165 mm diametral.
Significant long-term voiding does not occur for the initial plenum pressure of 7 MPa which we consider more realistic, even for the large gap size used. A more realistic gap size of 0.1 mm determined from SAS3A pre-failure calculations was used for the case TOP-1, which is 4
compared with the large-gap 7 MPa case (TOPANL-3) in Figure 2.
Here the two-slug coolant model fails to provide quantitative information on voiding history, since the lower slug does not remain below the failure point. The effects due to plenum gas release in this case are clearly small.
A second case (TOP-2) was run using the small gap size with an adiabatic treatment of gas in the coolant channel (versus TOP-1, isothermal treatment) to determine in part the sensitivity of the process to heat transfer assumptions. Without the cladding heat sink, channel gas temperatures and pressures could have remained aLove the values predicted in TOP-1.
This was not observed. Voiding histories for the two cases are very similar (Figure 3).
Case TOP-2 and information we have previously reported on the invariance of gas flow rates in the gap lead us to predict a relative insensitivity to heat transfer assumptions for ideal gas flow.
We conclude that plenum gas release is not an important mechanism in itself for TOP voiding in initial CRBR cores.
The non-condensable gas can produce extended voiding of the active core only if extremely high pienum pressures and large gap sizes are assumed.
Its potential for promoting molten fuel entry into a voided channel should be considered minimal. Equilibrium cycle cores, for which the highest fission gas pressures occur, can experience major voiding effects from TOP gas release under the conservative assumptions l
l i
14
k.4 05:3 00 3 0s:2 00 2 0s.2 oo 1 oS o oN B
\\
\\
I
%*4 o
o op
\\
@y l n
\\
\\
n p
\\
\\
_P n
.v e_
g g
. i
\\
$ O $.
rH
\\
L
\\
M g
o N
\\
o 2
N w
N
/
g O
w N
.s P_
s/
\\
o
.P "
'B ! "g 3
s
\\
m
=
s
\\
2
- M T~$
\\
~
~
g g--- h ---*
$N
\\
a s
\\
ga s
mg.
N
\\
5 2
N I
.E d% $
1
?.
\\
\\
P
\\g g
-8 8
g 00 gQ.4 05 3 00 3 05;2 00 2 05 1 00 1 05.D 00.J
)M( SNOITISOP ECAFRETNI h.G 00l5 00l4 00 3 00 2 00 1 009k l
l 0
o l
N l
N
?
t P_
\\
.P N
tr e
\\
38
$ s '
g 05 N
N o
P.
g g
N P
$ o.
g.
i rm
/N 1
l 2%
~
N b
2 l
G b
g
\\
l m
d o
o@
SE G
\\
i I
\\
re 5
s
/
5 t-P_
\\
/
P gp
\\
l /
Np
\\
l oc
/$l
-S er N
5"-
s
/8 N
s
/N I 2 l
'4 y
8
-gL 8,
y g.G 00 5 00 4 00 3 00 2 00 1 00.J M( SNOITISOP ECAFRETNI IS
1 9T
(
INTERFACE POSITIONS (H)
?
d).00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.@
b a
w P-P EEEo 2
l 2
mmmN i
o
- E $ <
O somo o
o 0Y$$d-
\\
d
~~
o o
4 e F M*
__.,fi -
\\
N a
=
sna s
ND m b O
m ieo - p_
O
\\
N s o
- c. s a 6 2
\\
A N
m
\\
/\\
u y,gg
\\
\\
- J e
ar m+
m
\\
c E$$E
\\
nn 8
\\
I s
o o
s g g' $ @
E
\\
E E
s
\\
E s
ogan
." $ o a
\\
E
\\
Po$
,\\
\\
o o
o H"
@.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
- 4. @
$n INTERFACE POSITIONS (M)
J).00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4,@
y O
s n
s
\\ 'N 4-
- o. -
i/
\\N o
e s
c
$@P@o
//
\\
' ~/ j
\\\\s e
emN n
e s
vi E
'*- d fI l
\\ \\
- 3 g 'P A d -
o
\\ N. 'N /
~o 7.
o mso go m,/ / I
\\
-\\
e s
N o
m N m
/
's e$ 7. l <
0 l
\\
i o m O
\\
N.
u td m +
.e-2 I
o g
n M$ S U
M l
N r
U U
[g f
li l
r N
h.
o'f
~
5
\\
P P-emt E
c x.
a 9u s
\\
u m
u i 1
N?
"@w" i
\\"
.o.m.
-o I
o
,i x,
e u
@.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.@
H w
PE*o 8 0."
(notably gap size) developed here. The inclusion of noncoheren*ce effects (only 50% of pins failing, for instance) in the analysis should be sufficient to reduce such predictions to mild voiding effects similar to the 7 MPa cases presented here.
3.2 LOF-D-TOP Plenum Gas Release The LOF-D-TOP events generated in unvoided or partially voided low-power' subassemblies, vhich experience cladding failures about a
the time of fuel motion onset in the high power core regions, can be important to the accident scenario. The smaller coolant velocities and inlet driving head in the LOF accident allow more pronounced and more symmetric voiding following failure than is the case for
)
the classical TOP.
Failures at the top of the active core and the resultant plenum gas release can therefore lead to more significant downward voiding if the coolant is not already boiled out of the upper subassembly region. Here, though, the phenomenon of interest is the reactivity insertion rate due to early voiding, rather than the long-term voiding. SAS calculations have shown that the LOF accident j
history can be very sensitive to the magnitude of such ramp rates.
Positive voiding reactivities cannot develop from plenum gas-driven voiding until the void extends further than several centimeters I
into the active core. Up to that point, the negative reactivity-effects of blanket voiding by the plenum gas dominate, and resulting i
[
reactivity ramp rates are negative. One might consider plenum gas release into an upper core region already voided by boiling (which would permit generation of positive ramp rates by the first downward slug motion) as a more likely scenario than the release into an un-voided channel.- In this case, though, plenum gas release does not pressurize the large void quickly, and no early downward motion occurs.
. Figure 4 compares several cases run with the isothermal gas flow model.using conditions'obtained from a SAS LOF run made by NRC 117-
i personnel. The channel was considered unvoided. Cap size selection is similar to that of the TOP cases.
Case LFT-1 is a reference case, c
with initial' plenum pressure P = 7.0 MPa and diametral gap size P
a j
26 = 0.1 mm.
Cabe LFT-2 was run with P = 3.5 MPa, which is closer P
i to the SAS-channel average pressure. Case LFT-3 used P = 3.5 MPa P
and 26 = 0.15 mm.
Case LF-5 used P = 12.0 MPa and 26 = 0.1 mm.
l P
(Inlet pressure was assumed to be 0.3 MPa and outlet pressure to be 4
1 O.13 MPa for all LFT cases.) None of the cases indicate early voiding of the active. core, and none exhibit rapid downward voiding rates at any time.
l-A number of other LOF-D-TOP cases run with the GASFLO code (comparisons not shown) demonstrated the relative independence of l
the voiding predictions to assumptions on inlet and outlet pressure, J
and initial coolant velocities. A calculation was also made with a 0.5 m initial void (simulating sodium boiling) in the upper blanket region. As expected, the delay in channel pressurization precluded early voiding effects. All the calculations using GASFLO argue j
against any major role for plenum gas voiding as a source of significant reactivity insertions in'the LOF-D-TOP channels even without consideration of non-coherence effects.
1 i
3.3 Summary a
i The potential'of plenum fission gas as a voiding agent has been evaluated for parameters-we consider to be characteristic of the transient overpower and LOF-driven transient overpower subassembly environments. In the TOP channel, preferentially upward voiding patterns i
prevent major voiding of the active core.
In the LOF-D-TOP channel, although strong downward voiding may occur, appreciable positive reactivity is not generated on a timescale of interest to the accident t
sequence.
We do not infer from these results that plenum gas release is un-important to the accident analysis of these channels. The interaction 18
of the gas release with other voiding processes and the promotion of additional cladding failure are two of the many possible multiple interactions outside the scope os our analyses, i
2 19
REFERENCES 1.
L. L. Smith and M. G. Stevenson, "Ef fect of Reactivity Insertion Rate on Fuel Pin Failure Threshold," TANS,17, 284 (Nov. 1973).
2.
D. B. Atcheson, R. R. Sherry, and K. J. Shimane, " Cladding Failure in TREAT Overpower Experiments: A Mechanistic Interpretation and its Implications for LMFBR Safety Analysis,"
Int. Meetirg on Fast Reactor Safety and Related Physics, Chicago, Ill., Sessioq 10 (Oct.
7 1976).
3.
H. G. Bogensberger and C. Ronchi, " Calculation of the Effects of Fission Gas in an LMFBR, for the Analysis of an Unprotected Overpower Transient," EURFNR-1203 (1974).
4.
J. H. Gittus, " Theoretical Analysis of the Strains Produced in Nu-clear Fuel Cladding Tubes by the Expansion of Cracked Cylindrical Fuel Pellets," NED, 18, 69-82 (1972).
5.
H. U. Wider et al., "An Improved Analysis of Fuel Motion During an Overpower Excursion," Proc. ANS Fast Reactor Safety Conf., Beverly Hills, California, CONF-740401, pp.1541-1555.
6.
P. A. Pizzica and P. B. Abramson, " EPIC--A Computer Progra? for Fuel-Coolant Interactions," loc. cit. 2, Session 7.
7.
C. C. Meek, ANL-RDP-48, pp. 7. 7-7.10.
8.
A. B. Rothman et al., " Review of TREAT Experiments In Support of Transient Overpower (TOP) Analysis for Fast Reactor Safety," loc.
cit. 5, pp. 205-219.
9.
R. N. Koopman and B. W. Spencer, ANL-RDP-50, pp. 7.21, 7.22.
10.
J. B. van Erp, T. C. Chawla, and H. K. Fauske, "An Evaluation of Pin-to-Pin Failure Propagation in EHFBR Fuel Subassemblies," loc.
cit. 5, pp. 615-640.
11.
C. A. Erdman, R. L. Garnett, and A. B. Reynolds, " Fuel Sweeping Assumptions in LMFBR Overpower Transients," TANS,1][, 360 (Nov.1973).
i 12.
C. A. Erdman, M. B. Johnson, and A. B. Reynolds, " Post-Failure Phenomena in LMFBR TOP Accidents," Annual Report, NRC Contract No.
AT(49-24)-0158, University of Virginia, October 1976 (Draft).
13.
A. B. Rothman et al., "Results of Recent TOP and LOF Experiments in TREAT," loc. cit. 2, Session 10.
20'
i 14.
K. Wong, V. K. Shir, and W. E. Kastenberg, "A Study of Fuel Freezing and Channel Plugging During Hypothetical Overpower Transients in a Prototypic LMFBR," loc. cit. 2, Session 22.
15.
M. G. Stevenson et al., " Current Status and Experimental Basis for the SAS LHFBR Accident Analysis Code," loc. cit. 5, pp. 1303-1321.
16.
Clinch River Breeder Reactor Project, Preliminary Safety Analysis Report, p. F6.2-48.
17.
T. C. Chawla et al., "The Rate of Coolant Recovery Following Release of Fission Gas from a Postulated Multiple Pin Failure in an LMFBR Assembly," TANS, M, 315 Oune 1975).
21