ML19275A958
| ML19275A958 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 10/31/1979 |
| From: | English R PRESIDENT'S COMMISSION ON THE ACCIDENT AT THREE MILE |
| To: | |
| Shared Package | |
| ML19254E707 | List:
|
| References | |
| NUDOCS 7910300411 | |
| Download: ML19275A958 (18) | |
Text
.
TECHNICAL STAFF ANALYSIS REPORT ON THERMAL HYDRAULICS TO PRESIDENT'F CO.TIISSION ON THE ACCIDENT AT THREE MILE ISLAND ADVANCE COPY NOT FOR PUBLIC RELEASE BEFORE AMs, WEDNESDAY, OCTOBER 31, 1979 79/osocyr 1183 021
THERMAL HYDRAULICS BY ROBERT E. ENGLISH TECHNICAL ASSESSMENT TASK FORCE OCTOBER 1979 WASHINGTON, D.C.
1183 022
This document is solely the work of the Commission staff and does not necessarily represent the views of the President's Commission or any member of the Commission.
This pre-publication copy is a final document and will be subject only to minor editorial changes in its published form.
1183 023
SUMMARY
AND FINDINGS Thermal-hydraulic events and processes were analyzed in considerable detail for the first 3.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> of the accident by means of the TRAC computer code. Also considered were additional selected thermal-hydraulic problems affecting reactor cooling during the period
~
100-210 minutes from start of the accident. The following are the principal findings:
1.
Thermal-hydrautic analysis of the TMI-2 reactor ioops (or reactor coolant loops) by means of the TRAC computer code fairly accurately reproduces the observed operating conditions over the first 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />. At about this time, a peak fuel-rod temperature of about 3900 F was reached.
2.
At 101 minutes after start of the accident, the inability of the reactor coolant pumps to operate while pumping a water-steam mixture having a very high proportion of steam made it necessary to turn off the pumps. Stopping the pumps interrupted the reactor cooling provided by this two phase mixture, and the reactor fuel elements rose in temperature to 3500-4000 F.
3.
When the reactor coolant pumps were stopped, water was trapped in the lower portion of each steam generator. The geometry of the reactor loops prevented this water from draining into and cooling the reactor.
4.
During the period that the reactor coolant system lack d an adequate supply of circulating water to cool the reactor, natural circulation had the potential to cool the reactor by boiling water in the reactor, condensing the resulting steam in the steam generators and allowing the condensate to flow back into the reactor. This was not achieved for several reasons:
(a) The geometry of the reactor loops did not allow the condensate to drain back to the reactor, as cited above.
(b) The steam pressure on the secondary side of the steam gen-erators was not regulated so as to be lower than the steam pressure in the reactor loops.
))0b 0
(c) After 150 minutes from the start of the accident, hydrogen in the reactor loops prevented flow of steam into the steam generators.
Remotely operated vents at the high points in the loops m.ght have permitted venting this hyarogen to the containment building.
2 1183 025
INTRODUCTION This teport is concerned with both the flow of water and steam throughout the reactor loops (or reactor coolant loops) and with the ability (or inability) of these fluids to remove heat from the nuclear reactor. Fmphasis is placed on those situations causing trcuble or leading to overheating of the reactor.
A group of consultants led by Peter Griffith reviewed the thermal-hydraulic phenomesa at TMI-2 L af. 4), and what is reported here draws on their work.
Because the reactor was damaged severely during the period 100-210 minutes after start of the accident, the thermal history during the first 210 minutes was analyzed in considerable detail by means of the TRAC computer code (references 1 and 2).
Some thermal-hydraulic processes were examined for the period 100-210 minutes in order to evaluate some means by which the core damage might have been delayed, mitigated or avoided.
The principles fo, keeping the reactor cool after the reactor is keep the reactor full of water; (2) circulate shut down are simple:
1.,
that water throughout the reactor loops; the circulation can be produced by natural convection or by pumping; (3) provide a heat sink, that is, a place to dt p the heat. The heat sink can be sunplied by either the steam generators or by injecting water via High-Pressure Injection (HPI) that is boiled and then discharged through the relief valves (either the PORV or the safety relief valves). The term " water" as used herein always refers to liquid water.
The generation of decay heat by the fission products is yaantified in great detail in reference 3.
The amounts of decay heat at various times are illustrated by the following table.
Time after shutdown Decay power,(megawatts) 1 second 168 1 minute 97 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> 36 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br /> 17 1 day 13 1 week 5.1 1 month 2.1 6 months 40 3
\\\\0b
In one second, decay heat was 6 percent of the rated power of 2772 megawatts. The decay heat droppea by almost half in the first minute and to 10 percent of its initial value in 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br />. The decay heat continued to decline but at a slower and slower rate as time went on.
A key problem in protecting the reactor is to properly dispose of the still fairly substantial amounts of heat over the first several hours.
As time goes on the heat removal is easier and easier, but heat must, of course, still be removed.
4 1183 027
ANALYSIS OF REACTOR TRANSIENTS AT TMI-2 References 1 and 2 are applications of the Transient heactor Analysis Code (TRAC) to analysis of the course of damage to the TMT-2 reactor. Picklesimer in reference 5 estimated reactor damage, but in making his estimates simply aol quickly, he introduced some serious compromises in thermal hydraulacs. For the the period of 100-180 minutes, he assumed a constant rate of fall of water level in the core of 12 feet per hour; instead the water 'avel should f all rapidly when th; core is fully covered, and the rate of fall should asymptotically approach zero as the water level approaches the bottom of the core. He also neglected convective heat transfer to the steam evolved by boiling water in the core; the result is that in his figures 3-7, he has a severe temperature inversion at the top of the core. Reference 4 shows that only small temperature differences occurred between the fuel rods and the steam. Thus, the hottest portions of the fuel rods were cooled somewhat by the steam; this heat was then transported upward by the steam, and some of it transferred to the cooler portions of the fuel rods. The result was an evening out of the spatial distribution of temperature in the core, the hottest parts being somewhat cooled and the cool upper portion being heated in comparison with the results in reference 5.
Damage to the core was accordingly also evened out.
Several attempts are under way to improve on reference 5 (references 1 and 6, for example).
The application of the TRAC code takes on a more ambitious effort, viz., analysis of the history of TMI-2 reactor operation during the whole first 3.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />. TRAC is a very complete, elaborate code designed for analyzing fairly brief reactor transients.
For this reason, its speed of computing was accelerated for the rather long transient at TMI-2 by using a small number of spatial nodes. For the first 81 minutes, 24 hypothetical cells were used in the reactor vessel and 42 in the two system loops; this permitted three levels vertically within the core.
At 81 minutes the reactor vessel was more finely dividad in order that the core might have 5 divisions vertically. The time increment for integration was chosen to be 0.1 second. The coolant was considered to be homogeneous within each cell.
In addition to two-phase flow, the TRAC calculations included the energy release f rom the zirconium-water reaction, which at high 5
1183 028
temperature produces more heat than radioactive decay of the fission products.
Input data for the calculations were the estimated values of high-pressure injection, letdown, and PORV flows into and out of the reactor loops.
Some representative results are shown in figures 1-4.
The close match of observed and calculated pressurizer levels (figs. 1 and 2) shows that the inventories and void (or bubble) volumes must also be close. Both pressurizer level and reactor-coolant pressure agree teasonably well with measured values out to three hours (fig. 2).
The temperature calculations out to three hours (ref. 1) in figure 3 have been extended to 3.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> in figure 4 (ref. 2).
A peak fuel-rod temperature of 3900 F was reached at 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> and 6 minutes after the accident started.
Finding: Thermal-hydraulic analysis of the TMI-2 reactor loops by means of the TRAC computer code fairly accurately reproduces the observed operating conditions over the first 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />. At about this time, a peak fuel-rod temperature of about 3900 F was reached.
6 1183 029
STOPPING THE REACTOR-COOLANT PUMPS At about 74 minutes the reactor-coolant pumps in the B loop were shut down (references 7 and 9).
At 101 minutes the reactor-coolant pumps in the A loop were also turned off. These two events were of considerable thermal-hydraulic significance.
At 101 minutes, when all four pumps had been turned off, the reactor began a period of heating up that severely damaged it.
At that time, the decay heat had declined to 1 percent of rated power (ref. 3),
and the heat fluxes in the reactor were accordingly low.
If the pumps had continued to operate and been able to circulate a two phase flow through the reactor, even a froth, the water carried along would have kept the fuel elements wet and, in that way, have kept them cool (reference 4).
Instead, the pumps were shut down because of vibration resulting from their inability to successfully operate on two-phase flow with high void fractions.
If the pumps had tolerated continued operation on a mixture of water and steam of continually declining water content, damage to the reactor could have been delayed and perhaps avoided.
Babcock and Wilcox engineers stated that, to their knowledge, full-scale reactor pumps have not been tested in this full range from all-water at one extreme to all-steam at the other (ref. 10).
Now consider the effect of turning off the B-loop pumps 27 minutes before the A-loop.
EPRI (reference 8, app. TH, pp. 49-50) postulates the following consequences of this sequence: As the A-loop pumps continued to pump a two phase mixture into the reactor, some of this fluid would also flow from the reactor and into the idle B-loop pumps.
In this region of essentially zero flow, liquid would drop to the bottom and vapor rise to the top of whatever space they occupied.
Perhaps in this way, water accumulated in the idle steam generator B and completely filled its lower portion.
In that event, the steam generator B might have contained considerably more water than steam generator A just af ter the A-loop coolant pumps were turned off at 101 minutes (see figure TH 14, reference 8).
During the operation of one set of pumps on 2 phase flow, the idle steam generator is thus a reservoir ia which a considerable portion of the liquid inventory might be trapped.
7 1183 030
With both sets of coolant pamps turned of f at 101 minutes, water was trapped in each steam generator. The geometry of the reactor loops prevented any of that trapped water from flowing back into the reactor.
The staff has concluded that if the steam generators had been elevated relative to the reactor and the piping arranged to permit drainage of that water into the reactor, damage to the core could have been delayed and perhaps prevented.
Findings:
1.
At 101 minutes after start of the accident, the inability of the reactor coolant pumps to pump a water-steam mixture having a very high proportion of steam made it necessary to turn off the pumps.
Stopping the pumps interrupted the reactor cooling provided by this two-phase mixture, and the reactor fuel elements rose in temperature to 3500-4000 F.
2.
When the reactor coolant pumps were stopped, water was trapped in the lower portion of each steam generator. The geometry of the reactor loops prevented this water from draining int) and cooling the reactor.
8 1183 031
NATURAL CIRCULATION During the period from 101 to 175 minutes (ref. 9), no water was pumped through the reactor, the reactor heated up, and considerable damage occurred; throughout this period, several high-radiation alarms were recorded, indicative of release of fission products.
If effective natural circulation had been established at the beginning of this period, perhaps serious damage to the reactor could have been avoided.
Reference 8 (app. TH, p. 54) makes the following judgment concerning the conditions following shutdown of the A Icop reactor-coolant pumps, thereby producing natural circulation that was effective in cooling the reactor:
The behavior of the primary coolant system following the trip of the loop A pumps indicates that steam occupied a substantial fraction of the system volume at that time. The behavior also illustrates that heat was effectively removed from the primary system by steam condensation on the primary side of the steam generators when the auxiliary feedwater was flowing.
But conditions were not maintained that would sustain this two-phase natural circulation.
Although the presence of steam would preclude the usual natural circulation in a water-filled loop, natur.1 circulation with both water and steam can also provide effective reactor cooling if the proper operating conditions are established in loop design and in operation.
In order that the reactor might be cooled by this two phase natural circulation and that heat be transported to the steam generators, the following conditions must prevail:
(1)
In order that there may be a temperature difference for producing heat transfer in the steam generator, the steam pressure in the secondary loop must be lower than in the primary, or reactor, loops. At TMI-2 the operators did not consister,tly maintain this condition.
(2) The steam in the reactor loops should be able to reach those porticr: of the steam generator tubes having cold water on the t
i183 032
secondary side. Either a liquid level higher on the primary side than on the secondary or non-condensible gas in the reactor loops could block the steam's path to the cold section of the tubes in the steam generators.
(3) Water condensed in the steam generators should be able to flow back to the reactor for reuse.
(4) For preservation of the reactor loops' inventory and, therefore, to permit continuation of the natural circulation, openings in the loops, such as the PORV, should be closed.
None of these conditions was fully met during the period 101-210 minutes. The PORV, of course, was cpen until about 140 minutes and the operators did not always maintain secondary pressure lower than the primary.
Closed valves on the secondary side of the steam generator B prevented any heat removal by that steam generator. But feedwater was being supplied to steam generator A, and it therefore offered the potential for heat removal.
The crucial factor was that the steam generator A may have had only a small inventory of primary water after the period of operating the coolant pumps A on 2 phase flow, as discussed above.
In that event, condensate formed in the steam generator could flow to the bottom of the steam generator, but it would there be trapped and unable to' flow back to the rea cor.
If the reactor loops had been designed so that the steam gencrators were elevated relative to the reactor and the piping arranged to permit drainage of that water into the reactor, damage to the core would have been delayed and perhaps prevented. Again, the design of the reactor loops appears to have impeded the process of cooling the victors.
If the arrangement of the st :~ generators and their pipes had permitted condensate flow back to the reactor, perhaps non-condensible gases might still have precluded natural circulation. A large amount of non-condensible gas can always stop 2-phase natural circulatior., and a tiny amount will very likely always be tolerated. A key issue bearing on this question is the tolerance that the steam generator has of non-condensible gas.
Consider that natural circulation was established at the time the pumps were stepped.
If modest amounts of non-condensible gases were produced, they would be swept along by the flow of steam and then accumulate in the steam generator.
In general, the gases would be carried toward the outlet er.d of the steam generator, where they would collect and effectively block the flow of steam to the cold end of the tubes.
If only a small portion of the length of the stea:n generator tubes were immersed in the cold feedwater, then a modest amount of non-condensible gas could block the access of steam to that cold segment of tubing. So a crucial factor is the way in which the steam generator was operated.
Reference 9 on page 19 states that the water level was at 51 percent of the operating range, as required for establishing natural circulation. Figure OTSG-2 of reference 8 shows the full range, or height, of the steam generator tubes as 600 inches and the operating and 10 1183 033
startup range as extending 250 inches from the bottom of the tubes.
Non-condensible gases sufficient to fill the bottom 175 inenes of the steam generator would block heat removal, if operated as described in reference 8.
Raising the water level in the steam generator to say, the 525-inch level seems simple enough to do in a 600-inch steam generator, under the emergency conditions for which it would be required. This would at least triple the volume of non-condensible gas that could be tolerated.
If the required water level had been, sa!, 525 inches, the system would have been more tolerant of non-condensible gas when striving for natural circulation on 2-phase flow. Water-level indication on the secondary side for the full 0 to 600-inch range is standard instrumentation for the steam generators (ref. 8, app. OTSG,
- p. 2).
In reference 8, figures TH4 to TH7 show that the water level in the steam generators never exceeded the 250-inch level at TMI-2.
B&W's once-through steam generator has a provision for spraying auxiliary feedwater onto the upper end of the tubes in the steam generator (FSAR, fig. 5/5-3).
This teature, if utilized, should improve natural circulation and permit the steam generator to tolerate a considerable quantity of non-condensible gas in spite of a low level of water on the secondary side of the steam generator.
Even in the absence of non-condensible gases, another factor must be considered in setting up natural circulation of 2 phase flow: The liquid level of the condensate in the steam generator must be at least as high as the liquid level sought in the reactor vessel, usually above the top of the core.
In figure 5.l-5 of the FSAR, the 250-inch level in the steam generator appears to be at just about the same elevation as the pipes for flow to enter and leave the reactor.
Because the condensate level in the steam generator will always be somewhat lower than the secondary-water level and because a goal should always be to keep the core covered, it appears that prudence dictates maintaining a water level above 250 inches for natural circulation with both water and steam in the reactor loops.
For natural circulation with water throughout the reactor loops, circulation and heat-removal capability will both be greatest if tall columns of cold water can be established in the steam generators. This facto also argues in favor of high feedwater levels in the steam generator for natural circulation. Elevation of the steam generator relative to the reactor would also improve natural circulatica in a water-filled loop, as already incorporated into the design of the Davis-Besse powerplant.
Perhaps the absence of such an approach as this prevented natural circulation from being effective and thereby contributed to the reactor damage at TMI-2.
Non-condensible gas in the reactor loops also interfered with effective cooling of the loops.
During the period 150-210 minutes from start of the accident, a large amount of hydrogen was produced within the reactor. Remotely operated vent valves in the head of the reactor 11 1183 034
vessel and at the top of each candy cane could have vented this gas to the containment building and thereby have aided in maintaining natural circulation.
Findings:
1.
Failure to always maintain a pressure lower on the secondary side of the steam generator than on the primary was one of several factors preventing natural circulation to cool tha reactor during the period 100-150 minutes from the start of the accident.
2.
The low elevation of the steam generators and the piping strangement between the steam generators and the reactor trapped water in the steam gererator rather than permitting it to flow back to the reactor. This was one of several factors preventing natural circulation during the period 100-150 minutes from start of the accidee..
3.
Restriction of water level on the secondary eide of the steam generc*. ors to levels not exceeding 250. inches impaireJ their
.pacity for natural circulation and made them senaitive to blockage of steam flow by non-condensible gases.
4.
The design feature in the steam generators that permits spraying auxiliary feedwater onto the upper portion of the steam generator tubes improves natural circulation and improves their tolerance of noncondensible gases.
5.
Following 150 minutes from the start of the accident, a large amount of hydrogen in the reactor 1. oops prevented natural circulation from cooling the reactor. Remottiy operated vents at the top of the candy canes may have permitted venting this gas to the containment building.
12 1183 035
REFERENCES 1.
Anon.: Preliminary Calculations Related to the Accident at Three Mile Island.
LASL LA-UR-79-2425, Sept. 1979.
2.
W. R. Stratton et alia: Alternative Event Sequences, Appendix B.
Staff report for President's Commission on Three Mile Island.
October, 1979.
3.
T. R. England and W. B. Wilson: TMI-2 Decay Power:
LASL Fission Product and Actinide Decay Power Calculations for the President's Commission on the Accident at Three Mile Island. LASL LA-8041-MS, Sept. 1979.
4.
Peter Griffith and Neil Todreas: Review of Thermal-Hydraulic Events in TMI-2.
Aug. 16, 1979.
5.
M. L. Picklesier:. Bounding Estimates of Damage to Zircaloy Fuel Rod Cladding in the TMI-2 Core at Three Hours After Start of the Accident, March 28, 1979. Memo for File, NRC, June 20, 1979.
6.
C. E. Hendrix and.S. R. Behling: An Analysis of the Accident at Three Mile Island Using RELAP4/ MOD 7.
Aug. 30, 1979.
7.
Jasper L. Tew: Sequence of Events. Staff report for President's Commission on Three Mile Island, October 1979.
8.
Analysis of Three Mile Island - Unit 2 Accident. Nuclear Safety Analysis Center, EP;I, r:p::t 'ZAC 1, July 1979.
9.
Jasper L. Tew: Catalog of Events Related to the Accident at the Three Mile Island Unit-2 Nuclear Power Plant on March 28, 1979.
Staff report for President's Commission on Three Mile Island, October 1979.
10.
D. LaBelle: Comment on August 2, 1979 during visit to Babcock and Wilcox, Lynchburg, Va. by R. English, J. Hench and S. Levy.
13 1183 036
i 4
,4
,ij4 4
g g
g g
400 -
W~^
1 n 350 g
i g
o was g
4 l
i g
o c
1 I
s t
Ii
\\
300 -
J l
1i s
W ll
\\,
11 l
y
! 250
)'
TRAC - - - -
y
,Jl TMI DATA l
a m.
i I I E 200 3m I
m 1
W ll
[ 150 is II q
L 9
t I
t t
,,,I
-20 0
20 40 60 80 100 120 TIME AFTER TURB!NL TRIP (min)
O Fig. 19. Pressurizer water level ccq:arise.s cut to 12C =i:=tes.
1183 037
(584ou!) 13ATI W32180SS38d R
8 8
8 8
g s
v n
u u
s s
s a
s.,
,5 I
! h_
Yl 4
y is 8 I E
,' 15-wa W 8 al C
o e 3
e 0
I 5 0$
8 c3 ww a
ze o cc
-~
<3
-am H
E com a-
+ow w
w v
<ue
,z x
>s o
a.
e
_w w c
c 2o2 c:-
5 s ~ h.. -_ __ __ _.g _ _, _ _ _ _ _.
e
-=
r tu*
D y
e R~__;..______.------'
gy cm m&
N A
w >.
g e _-_
em e
c 228 s' % } N'
%'~
w 8
m H
2 e.ya.,..
,),.,
n <u-e.
e W
r/
2
%p g
F-
/
t
.%A w.-
,f f g_2_r-- ---,
"7c-_---}3
= = -
e s
1 i
E i
j 8
8 8
8 8
9 e
0 b
v
~
u (5 sd) 3EnSS3Hd 1NVlOO,' B010738 1183 038
i i
i i
i i.i i
i i O
O O
_c e
~
u O
f.
1' O
O=
- 5 wr O.,
g-
]
a ew 2
1 s
.s O
3 8
s s
v t
]
O
{
8 we 4
N
+x m
i I
I I
I I
I I
l l
l 1
I I
I I
I C
O O
O O
O O
O O
o O
O O
O O
O O
O O
C
_e v
N O
c c
v N
( M ) d W3100810H WnWlXVW 1183 039~
a 6
(
%v ihZ s
C.
N "N
0-O g,
g{g g
go
=c c
c c
c c
c
- c..
- f#
gg O:
=
.J "rf b
.tw4 4b T
.g
-r e
a a
5
,g 5,
cf.
i x
o b
4
$e
.gs m
m 2
e <:
m o
bh gg g
e
=
4' 4
1 a
D b
1 H
H E
a (y) a u:. con w 1133 040