ML19246A366
| ML19246A366 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 05/07/1979 |
| From: | Etherington H Advisory Committee on Reactor Safeguards |
| To: | Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-M-0016, ACRS-M-16, NUDOCS 7906180649 | |
| Download: ML19246A366 (9) | |
Text
4..
/"" %.
kcf5g;gojf,,
+
f *'f7,i, ;t/ i k',
NUCLEAll HEGULAlOliY COMMISSION
[
ADViSOltY CO, 7.1 f l L L OfJ H1' AC10f ? S Afl.GUAHDS 8
Et e
k
'; '[' [
VlAStilNG10f J. D. C. 2RsLS fiay 7, 1979 m
ACRS I M bers ACI'S Technical Staff PRFLI:lli:APY APiMISAL OF Tile Tl:REE l'ILE ISLAt:D ACCIDE:lT The attached is for your information and use.
- d)~ 's
- 11. E t he ri n e, co\\l si Atta ch:wnt :
- 'm ;. m. by H. Etherington re Preliminary Appraisal cf the Till Accident 7906180619 228 272
. y..
~
Preliminary _/graisal o f the Three flile Island Accident In t ro d uc ti o n.
The Committee has revicued sone of the data recorded during the Threc !!ile Island accident - the most impcrtant data are shown in attached Fig. 1.
It seems clear that material balances and heat balances, in conjunction with recorded data not yet available to the Com-nittee, will lead to an accurate history of the condition in the reector, at least up to the tirne the reactor coolant pumps were tripped.
The follouing c.
- lusions are supported by rough calculations which can be made available.
It is evident, as discussed later and as suqqested by Curve F of Fiq. 1, that, after the initial blouda.1n, the hot leg (at least) was at saturation temperature throughout the first 100 minutes.
The void fraction is determined by the volu:actric loss of primary system invente y, With no llPI or makeup pump, the loss of water and steam through the E!'. / (electro-T natic relief valve) may be about 1 perccnt per minute of the systen volume.
This woul d also be the rate of void forration.
The heat belance will reflect temperature changes necessary to generate the required steam.
Thermal expansion and contraction is reflected in a change of void fraction of about 0.15 percent per OF (depending on amount of subcooling).
TheJressurizer l evel.
There has been speculation that the level instrumentation may be reading incorrectly.
Ilouever, I believe the level indications are consistent with what is happening in the system, and it sceus nare reasonable to try to interpret what the instrument is seeing than to question its accuracy.
228 273
=
. 1,' hen vaids are present in the system and the ["RV is open, all of the steam and water entering the pressurizer must be discharged through the val ve.
The pressurizer is thus filled to the_ to2 with a tuo phase mixture.
In this conditior, what the level indicator is (probably) sensing is the weight o f a two-phase mixture relative to the weight of water, e.g., a reading of 300 (Fig. 1 Curve A) indicates a void fraction in the pressurizer of 330/400, or 5 percent.
To estiraatt the probably much greater void fraction in the reactor systen, it would be necessary to know the electric heater input and the rate of rise of bubbles in the pressurizer.
The presu.aptions that arise from these considerations are that, with the EMRV open, 1.
the level indication cannot exceed 400 2.
the dif forence between 400 and the indicated levels is a valuable, but uncalibrated, measure of the void fraction in the system On this basis, Curve A of Fig.1 indicates the presence of voids throughout it.ast of the accident, increasing from 40 to 100 min.
The decreasing coolant flou shoten in Curve C confirms this by the evident progressively increasing pump cavitation.
Jhe perio d_o f _ forced ci rcul ation, 0-100 min.
1.
The transients during the first 15 sec., Nioving loss of main feedwater supply, follow a noroal course, i.e., pr essure rise, actuation of the DIRV, opening and closing of the sa fety valves, and other transients shoun in Fig. 1.
Failure of the URV to close is the coverning factor in subsequent events.
The e f fect o f the delay in auxiliary feedwater flow was transient.
228 274 s
3-With the LimV open, the system blows doun to saturation pressure 2.
Curve F of Fig.1 shows that this occurs in 6 min.,
in a few minutes.
During and calculation from the pressurizer steam inventory gives 7 min.
the first ninute, the pressurizer level falls abcut 100 in, from the peak (Curve A). About 70 in, of this is accounted for by thermal contraction and about 30 in by loss of pressurizer water by flashing during the blou-down.
The rise of 220 in. in the presserizer level between 1 min, and 4 nin.
occurs before the tenperattre has dropped to saturation (Curve E).
Ccneral voiding of the system is therefore precluded as the cause of the I;o completely satisfactory explanation for this rise is offered.
ri se.
and There has been a rise inthe average system temperature (average of Tg T, not adequately shown in Fig. F) of 2 to 3 F.
This could account for c
45 in. rise in pressurizer level.
The strain relaxation of t' e steel boundary under a 400 psi pressure drop would cccount for only 1-1/2 in.
Each of the three high pressure injection pumos woule raise the icvel at The the rate of 20 in./nin. or 80 in. during the four ninuu period.
pumps could therefore easily account for the rise, but information on the pump history is con'osin3 and the curve fron which Curve A is drawn does not show discontinuities such as would be expected in starting and Recent information suggests that one pump was running stopping pumps.
at capacity and one in a contrailed mode.
Voiding in the core is possible, but, with full flow circulation of subcooled water to remove decay heat, 228 275 s
4_
a large amount of core voiding does not appear likely.
(I have made no cal c ul a t ion s ).
3.
lherea f ter, with the IIRV still open, the system will follow either of two courses:
(a )
It will remain at saturation, with the pressurizer level indication less than full, or (b)
It will be pressurized to above saturation by "gaing solid" with the high pressure injection (HPI) pur:ps, in which case the pressurizer level will Le over the 40C inch mark (with en adjustmen' Cor steam generated by the heaters).
It would be possible to opertte in different modes at different times, out Fig.1 shows that with the possible exception of the period 11-10 min.
the system remained at saturation pressure throughout (Curves A and F).
The following conments are therefore addressed to the saturated condition.
4.
liith the system at saturation, the fluid in the hot leg is two-phase, i.e., saturated water and steam.
The steaa voids increase progressively, tith time, and cavitation of the reactor coolant pumps increased correspondinD y (Curves A and C).
It is concluded that nis-l interpretation ( f the level iridication led to continued loss of invento y from the system.
5.
The hot leg is at saturation ter.perature.
After the first few ninutes, the heat t;alance shows that, with all RC pu:rps running, the col'd 0
leg temperature is, as an upper limit, only about 1 F subcooled.
It is reasonably certain that, after the first few minutes, at least part of the core was functioning as a forced circulation CUR.
It does not seem likely, however, that, at the low rate o f decay, the core would have suffered danage during this period, even at the high probable void fraction.
(I have uade r.o calculations).
228 276 7.,
5 6.
TM systen at r.o time appeared to suffer frora serious loss of 4
cooling capacity, except possibly during the period between 5 and 8 minutes.
Curve F shows, with the steau generators dry, a rapid rise o f temperature, and Curve B shows a correspan fing rise of pressure.
If the auxilliary feed.zater supply had not been started, the pressure rise would presumably have continued catil the sa fety valves opened.
7.
Infor.;ation available to us ' c'oes not show when Hpi (" makeup")
pu::ps were started and stopped.
It appears that the pumps can easily rm'a up the E".RV loss and keep the system pressurized in a " solid" condition.
The period of " natural circulation."
1.
Natural circulation in a P'..'R has been generally understood to nean circulation of liquid uater.
The conservetive conditions for natural circulation in this mode are:
(a ) The water nust be subcooled (b) The heat sink in the steam generators must be at a suf ficient elevation above the core.
In Three Mile Island 2. the water was at saturation tempe.ature, i.e., there was no subcooling.
It should be noted that, when a reactor goer, into natural circulatica, T will increase and TC will decrease, in order to g
give a temperat ure rise consistent with the reduced rate o f circulatior..
It is not suf ficient that the saturation pressure of the system be above the initial hot-leg temperature - it raust be above the new hot-leg tenperature.
228 277
\\
eien e -
\\
6-The conclusion is that natural circulation did not, and probably could not, occur with the high void fraction, and the reactor sat in a pool-boiling stcte for 14 hours1.62037e-4 days <br />0.00389 hours <br />2.314815e-5 weeks <br />5.327e-6 months <br /> with no cooling except that supplied by the HPI pu"?s.
I have not nade any calculations to see whether this should have been adequate.
The heat capacity :hould be adaquate, but distribution and continuit.y of pu.ging would be controlling factors.
2.
L' hon the EMRV ucs isolated hal f an hour late, the pressere (Curve B) and suturation temperature (Curve F ) rose gradually to a condition in 1 hich natural circulation night possibly have been induced.
3.
Items that could have alerted the operator to the condition include:
and Tg.
In forcod circulation (a ) The wide spread between TC with all four pumps, this shoul d be about 1 F at 100 min.
If the natural recirculation rate were 5 percent of this, the spread would be only 20 F as compared with the wide 0
spread indicated in Curve F.
(b) Availability of stcau-table saturation te"perature/ pressure data proposed by Mr. I'athis.
This would have shown, not only that natural circulation si.ould not have been attenpted, but also that T started to indicato superheat i mediately, g
shawing atsence of water at the hot-leg temperature sensor.
(c )
I:r.nediate loss of steau generator pressure (Corve E).
4.
These are other modes of natural circulaticn than the presumed design rnoda, but these should not be relied on without thorough analysis and testing.
hplicat ion 1_for Onera_tino P@l Plants The problen o f the suall break has been addressed by !!r. Michelson, Mr. Eberscle (Pebble Springs), and, more recently, Mr. Streeter o f the URC Staf f in a memo that was still under review at the time of the accident.
BM! has always responded with analyses that shnu the syteu to be capable 228 278 (w.y
\\
of handling the accident.
The analyses, however, assune no operator ill-advised intervention, and it is not clear that a con'plete spectrum o f breaks has been covered.
The "l:: C Status nei' ort on feedaater Transients in Bfd! Plants, April 25, 1979", su:,iarizes the problem and status.
- 11. Etherin gton April 27, 1979 28 7
i s
w F J 1, I Tl 'ii;.T M it t IS f./ f 'D /h.c tl2L.;T
!!s,',' I - t:.70
._- +. -
-s,. ~.. ~.,
I f
/
w f
u
' /
i t
N 1,
Js(
s'N _-
g, 4.. s..,.,.., g., g._
t g
__...e_.--.---
o _.,
g,/
. '\\
N,.N v
~
-,.o c!);n O
R.c. I F.r. 3 m r e
o.---.-.....
c-(' N I
W t
cf y RJ,1 I. 0' V g
c 't : '/ E C c;
- j
i i
g f'
M.,
, ~g j. ' ' ^^
j '~r, g,'n rf^,'ey-n P {'.,} d I,';. -f Of D I. t y Ge P'
. g r. / 6/.#-J 'q-).W
, t.
qT\\,m Lt 0(' ~
9_ v,,, 7. a. t.M i'..,.,v,
'r
...Y
, _ g, 0.,-gia n?.,
.,c..,
(-
upp v
~
~
_____..~..
_s' l
', ~,
g 4-4 ---,
- y
- s
-f,
's
"(
\\*[ \\,
[
i b
w_ ), p e
i I
s
\\,
/.,_.. -
i s
e f
/
.o n
' b.
.d,,.$
4 i
g
' * ~
( /,
,'_4
/
L.,
b, i 7.x r b..
g i
i' l
^
s
(
O 300 s
i
\\
l t
- ~ ss_/
'k.
4 s...*
If g','
y
( I Y,Yf.
- s. b, bf
. I 5. I U E.
\\
i c
\\
l j
i s
O j
i.
l l
l t
.j o ( *
,,, y.
i l
'(
i l
8
(/;sT i.
l e
i i
i i _.
t.
...o.--.,-..
,._....I.-..a._t.....x.
3,.
w.,
o y
a 3ac, Q v
ar=
- s s
a o
Hwi.S p,
f.p.
.5 L s e.. t *,
ri t tw1 L ',
t 0ojf
-u
\\