ML19289C462
| ML19289C462 | |
| Person / Time | |
|---|---|
| Issue date: | 12/27/1978 |
| From: | Eisenhut D Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML19289C461 | List: |
| References | |
| NUDOCS 7901120227 | |
| Download: ML19289C462 (12) | |
Text
ENCLOSURE
}l TOPICAL REPORT EVALUATI0ft REPORT N0.:
NEDO - 21052 REPORT TITLE: Maximum Discharge Rate of Licilid-Vapor Mixtures from Vessels REPORT DATE:
September,1975 ORIGINATItiG ORGANIZATION:
General Electric Company REVIEWED BY: Analysis Branch, DSS Tio//do&W/
SUM"AD.Y OF TOPICAL REPORT In tooical report NED0-21052 and the additional information provided November 8, 1977, and June 30, 1978, GE has pavided a proposed method for predicting break flow for use in containmen' response analyses.
T's containment, response analysis is used direct.ly in the load combina-tions for the containment structural assessment and it establishes the boundary conditions for the suppression pool hydrodynamic testing program for plants with Mark I containments.
Application of the break flow methods will be limited to double-ended break sizes in the recirculation piping of plants with Mark l containments.
These are BWR-3's and 4's with jet pumps and BWR-2's without jet pumps.
For containments of the Mark 1 design, the first second of blowdown is significant since this is the time period when the vents from the drywell to the suppression pool are clearing. At this time, the maximum structural pool loads are experienced. The firstten secor.ds of flow are also important since the peak drywell pressure is reachad at the end of this period. The pressure response of a typical Mark 1 containment to a postulated recirculation line break is given in Figure 1.
Topicai Recort NED0-21052 provides a comparison of the homogeneous ecuilibrium critical flow model (HEM) with experimental test data.
The medel was developed using the assumption that te flow process is isentropic and the report provides curves of mass flux as a function of the stagnation enthalpy and pressure.
The flow rates are essentially identical to the HEM flow tables contained in the RELAP-4 computer program.
For pipes longer than four inches, with low friction, GE concludes that HEM provides a best estimate for prediction of critical flow rates.
A slio flow model is also developed in NED0-21052 for use with long pipes, but this model will not be used for the prediction of flow rates in the Mark i test program. Therefore, this model is not considered in this topical report evaluation.
The additional infomation proviced November 8,1977 includes tables of HEM flow rates calculated by GE and additional justification for use of the HEM based on experimental test data comparisons.
The information provided June 30, 1978 discusses application of HEM in the M3CPT03 single node bicwdown code and provides a comparison of the break flow predicted by M3CPT03 to that predicted by RELAP-4 for a typical EWR with jet pumps. Since the one node M3CPT03 code does not consider local ressure variations when com:; ting the flow from the bro' Ken recirculation pipins, GE multiples the initial break flow cal-culated using HEM by a factor of 0.72 for the initial pipe decompression period. The basis for this factor is provided in topical report NED0-20533 (Ref.1) and is derived from solution of the mass, energy and momentum conservation equations assuming isentropic flow. The pipe decomoression period is about 50 milliseconds for pipes without a restriction between the break and the vessel, and is determined by the time recuired for a sonic pressure wave to traverse the distance between the break and the vessel and back. The sonic velocity at these conditions; is approximately 5000 feet /second.
For BWR-3's and 4's, the jet pump nozzles provide a large flow restric-tion at the vessel inlet nozzle.
For these plants, the 0.72 fac' v is utilized for the time required to exhaust the pice fluid inventory between the break and the jet pumo nozzles (about one sc ond).
The flow rates during the pipe decompression periods for each,ide of the break are asse"ed constant and determined from the HEM tLbles (flow vs stagnation enthalpy and initial pressure).
Following the initial pipe decompression period, flow rates are determined by HEM, which is programed into the M3CPT03 code.
The code assumes a constant input subcooled enthalpy ur^il the initial subcooled mass of water in the vessel is depleted. After this period, flow rates are detemined using the stagnation liquid enthalpy and pressure calculated by M3CPT03.
The break flow is assumeJ to be all liquid until the reactor system inventory is 80'; exhausted.
Since critical flow c :es for liquids are larger than those for two-phase mixtures, these assu".~tions act to maximize the release to the containment.
The switch to two-phase flow is made at about 20 seconds into the transient which is well beyond the times of peak drywell pressure and pressurization rate.
STAFF EVALUATION General Electric has presented the homogeneous equilibrium model as a best estimate calculation to be used as part of a method for predicting break flows.
They propose to introduce conservatism by use of the non-mechanistic one node blowdown model.
In our evaluation, we consider both the comparisons of HEM to available experimental test data, and the apolication of HEM with GE's methodology to assess the overall con-servatism.
A.
Verification of HEM Flow Rates by Comparison with Experimental Data The GE justification for use of HEM in predicting break flow is based primarily on the data of Sozzi and Sutherland presented in Ref. 2.
These expercents involved the blowdown of a vessel through various i
nozzles of varying length and diameter.
The effect of increased nozzle length was found to decrease the flow rate. A large sensitivity was observed for nozzles less than four inches in length and a smaller sensitivity was observed for nozzles greater than four inches in length.
GE attributes the large sensitivity of short pipes to the non-equilibrium condition of '.he fluid at the po4t of discharge.
For pipe lengths longer than four inches, they conclude that the fluid will have the opportunity to reach equilibrium before leaving the test.section so that the flow rates could be predicted by HEM.
Pipe lengths longer than fJur inches would reduce the flow rate only by the reduced stagnation pressure resulting from the increased frictional pressure drop.,
. Flow rates predicted by HEM were found to agree with the Sozzi and Sutherland data in Ref. 2 for pipes longer than four inches. The HEM model was also compared in NED0-21052 to data taken by Uchida, Fauske, Henry, Allemann and Zaloudek. These comparisons also showed that general agreement was obtained for pipe lengths longer than four inches. Most of the data were for small diameter pipes of less than one inch ID. The Allemann data, however, included pipes up to 6.3 inches ID and also showed agreement with the HEM predictions.
The effect of nozzle diameter on break flow was evaluated by Sozzi and Sutherland for pipes less than one inch in diameter and 1.75 inches in length.
These results indicated that mass flux decreases as diameter increases.
Simon (Ref. 4) evaluated the effect of both length and diameter for nozzles of four inches and smaller.
The results are presented here as Figure 2.
In these studies, a complex relationship was observed on the effect of both nozzle length and diameter on the break flow.
The flow rate was observe'd to either
~
increase or decrease with increased nozzle diameter as a function of the nozzle length.
These studies indicate that small pipe data may not necessarily be applicable for predicting flows from large diameter pipes. The recirculation l}ne area for plants with MARK 1 containments range from 2 to 4 ft. while most of the test data is for pipe diameters in the order of a few inches.
Critical flow data for large area pipe sections from 1 to 2 ft.2 are currently being obtained at the Marviken facility (Ref. 5).
Preliminary comparisons of the HEM with data from the first two tests have been made by our consultants at the Brookhaven National Laboratory.
Comparison curves are attached as Figure 3 and 4.
These figures indicate that HEM underpredicts the data by as much as 40%. The results indicate equilibrium conditions may not be reached for large diameter pipes as was observed by Sozzi and Sutherland for small diameter pipes.
In one location the flow length that is available for choking in the BWR-3 and 4 type plarts does not appear to be sufficient to produce equilibrium condi: ions even for pipes of small diameter.
The jet pump nozzles provide a reduction in flow area resembling the geometry of an orifice.
For orifices, the data of Sozzi and Suther-land indicate flows in excess of HEM.
This is because the short transit time through the test section does not permit steam bubbles to form sufficiently for the equilibrium state to be reached. The fluid, is consequently discharged at a lower quality and higher density than would be predicted by equilibrium theory, and mass flow rates in excess of HEM are measured. For sharp-edged orifices, flows about 150% larger than HEM were measured for saturated water at 1000 psi.
Orifice flow data obtained by Silver, Bailey, and Schrock were compared by Collins in Ref. 3 to the predictions of HEM.
For flow of saturated water, the data was observed to be about 150% larger
. than HEM values.
Another experimental data comparison was made by Simon in Ref. 4 utilizing data taken by Uchida, Fauske, Friedrich, Burnell, Forster and Esthemer.
For flow of saturated water through an orifice at 1000 psi, flows 150% larger than HEM were also observed.
Flow rates were found to decrease as the nozzle lengths increased and converge on HEM for lengths of about eight inches.
The available ecperimental data indicates that HEM may significantly underpredict flow rates through the jet pump nozzles since they resemble an orifice. However, the jet pump nozzles represent only 20% of the total flow area, and would not produce a major portion of the total break flow.
B.
Application of HEM of Prediction of Break Flows Following a double-ended pipe break, the sudden discharge of fluid will produce a decompression wave which travels down the pipe to the vessel.
If the pipe is open to the vessel, a compression wave will be produced at the vessel which then travels to the break.
During the period of wave travel, the stagnation condition at the break will be reduced from the original state.
Using the isentropic flow assumption discussed in NED0-20533, Ref.1, GE calculated the flow rate during the initial wave propagation period to be 72% of the value obtained using HEM at the original stagnation condition. For the assumed condition of isentropic flow, we obtain similar results using the methods presented by our consultant at BNL in Ref. 6.
For open pipes connected to a vessel, the period of reduced flow is of short duration since the wave propagation speed is approximately the speed of sound for liquids (5000 ft/sec). At this velocity, the time required for the pressure wave to traverse a BWR recircula-tion pipe would be about 50 milliseconds.
For a pipe which has a blockage at the vessel such as the jet pump nozzles, a wave of reduced magnitude woub be reflected from the vessel so that the flow rate will decrease from the initial value.
This situation would occur for the recirculation piping of BWR-3's and 4's which enter the reactor vessel through the jet pump nozzles.
Instead of decreasing the flow rate during the initial blowdown period as the pressure in the pipe is reduced, GE proposes to assume that the flow remains constant at the initial value of 0.72 times HEM until the initial pipe inventory is exhausted. This requires about 1.2 seconds. Following this time, the flow is based on 1.0 times HEM using the flow area of jet pump nozzles.
The 0.72 factor is larger than the value actually predicted using the methodology of NED0-20533 since it is based on the assumption that
' the discharged fluid is saturt ted.
If the actual subcooled state of the fluid in the recirculation piping were utilized, a slightly 1cwer value would be obtained.
For the piping section connected to the vessel at the vessel outlet location, GE will use the 0.72 multiplier only fu the brief amount of time required for the acoustic wave to traverse through the piping to the vessel and return.
Following this period, a flow rate of 1.0 times HEM and the pipe cross sectional area will be used to compute flow for the duration of the blowdown.
As justification for the reduced flow rate during the pipe decompression period, GE has provided a comparison of break flows using the RELAP-4 code for a typical BWR with jet pumps. The RELAP-4 analysis utilized the Henry-Fauske model to predict break flows when the flow was subcooled and the Moody slip flow model was used to predict flow for saturated fluid conditions.
The flow rates calculated by these models are about 60% highe Mn HEM for saturated and slightly subcooled conditions typical o.'
BWR.
Comparisons of the RELAP-4 flow models to test data from 'ie Marviken experiments were made by the staff in Ref. 7 and by INEL to Semiscale test data in Ref. 8.
These comparisons indicate that the models are conservative.
The BWR RELAP-4 model included a multinode description of the reactor vessel piping. The multinode piping description permits RELAP to calculate the acoustic wave propagation following the break. Since the GE model does not take credit for the depressurization of the line between the break and the jet pumps until the line has been evacuated, the model produced 20% higher flows for this period than RELAP. The comparison of the integrated break flow between RELAP-4 and the GE'model is attached as Figure 5.
Following the End of the pipe blowdown period, the GE results continued to be more conservative than the RELAP-4 predictions. This results primarily because GE assumes the fluid leaving the vessel is at the liquid stagnatior, enthalpy.
This enthalpy is lower than the twc-phase stagnation enthalpy calculated by the RELAP-4 code.
The assumption of an all liquid blowdown increases the break flow calculated using HEM so that by the end of 10 seconds, which is about the time of the peak drywell pressure, the GE prediction still exceed RELAP by 15%. The GE results continued to be higher than RELAP for the remainder of the blowdown. The total mass release in the GE model is higher than the RELAP prediction for the total bicwdown because of the conservative treatment of feedwater in the GE model. The feedwaeer is assumed to be within the reactor vessel at an elevated temperature rather than in the system piping.
STAFF CONCLUSIONS Based on our comparisons of the HEM to experimental data as discussed in ti.
preceeding evaluation, we cannot conclude that HEM is either a conservative or best estimate method for predicting break flow.
The Marviden tests provide a break geometry similar to the vessel outlet side of the postulated break.
The evaluations of our consultant at BNL indicate that for these tests flow rates are under-predicted by as much as 40% using HEM.
For the vessel inlet side of the break that contains the jet pump nozzles, the flow geometry resembles an orifice.
The data in Nferences 2, 3 and 4 indicate that for orifice geometry the flow rates could be in excess of HEM.
by as much as 150%
GE has utili:ed HEM in a non-mechanistic reactor system model which does not take credit for pressure reduction in the piping during the early portion of blowdown and conservatively assumes all liquid flow during most of the remainder of the blowdown. By comparison of the mass and energy predictions of the GE model to those of a conservative RELAP 4 analysis, we have concluded that the GE model is conservative for prediction of critical flow rates for a postulated double-ended recirculation line break for BWRs with MARK 1 containments.
The GE methodology on the application of HEM to reactor blowdown is presented in the form of answers to the NRC questions. We require.
that this and the other supporting material in the letters of November 8, 1977 and June 30, 1978 be incorporated into the approved version of topical report NED0-21052.
E
References 1.
W. Bilanim, "The Genral Electric Mark III Pressure Suppression Containment System Analytical Model", General Electric Report NED0-20533, June 1974.
2.
G. L. Sozzi and W. A. Sutherland, " Critical Flow of Saturated and Subcooled Water at High Pressure", General Electric Report NED0-13418, July 1975.
3.
R. L. Collins, " Choked Expansion of Subcooled Water and the I.H.E.
Flow Model, ASME Journal of Heat Transfer, Vol.100, May 1978.
4.
U. Simon, " Blowdown Flow Rates of Initially Subcooled Water" ANS Topical Meeting on Water Reactor Safety, CONF-730304, March 1973.
5.
L. Ericson et al, "The Marviken Full Scale Critical Flow Tests Interim Reports", Results from tests 1, 2, 3, 4, and 5,1978, Unpublished.
6.
P. G. Kroeger, "The Propagation of Phase-Change Fronts in Moving Fluids," BNL-NUREG-50687, August 1977.
7.
W. L. Jensen, NRC Memo, " Preliminary Investigation of Marviken Critical Flow Data", May 1978.
8.
Douglas G. Hall, "A Study of Critical Flow Prediction for Semiscale MOD-1 Loss-of-Coolant Accident Experiments, TREE-NUREG-1006, December 1976.
I
63 i
i 50 _
- 1. DRYWELL PRESSURE
- 2. SUPPRESSION CHAMBER PRESSURE 4c 1
6 1
2 g 30 e
E 2
20 1
2 10 1.2 0.'1 1.0 10'O 100.0 10
.0 TIM E (SEC)
Figure 1 TYPICAL MARK I CONTAINMENT RESPONSE TO A RECIRCULATION LINE BREAK
i
' p O = 50 ata
! cst rem!!s by Fau:,ke, Forster, Fdeddch. Etheimer nl data reduced to: sharp-edged inlet Fv/Fn ~ 60C0 steady state conditions 4g k=w m:
3.0 t/mm = 10 2.5 p 20 -(
bW%-
2.0 Pdcody del, f.d = 0 1 = 100 mm,
~
I
+
d
,, d = 100 mm-1.5
[
"(Ungerer, extrapo!sted) 1.0 0
10 20 30 d/mm DIAMETER, mm Note: The flow rate k is the measured rate normalized to the HEM value.
Figure 2 FLOW RATE FACTOR RELATED TO N0ZZLE GEGMETRY (SHARP-EDGED INLET)
e
-O I
l 1
l
=
I
=
I i
,e
.=
U e
I e
d l'
em e
w<
-c
-=
e n
=:
e I
C- <
e>
m
-e
==
a
-=
C XmO v =
=
.f
$ *.1:
=.*
a =
-c v =
me
/
e=.
f_w xe v f
l
/
5-
=EE wuc e
=
=
==u I
=
mem
/
i e
I f
u uve e
=
l l
/
e m-
=-w
~
/
w
/
=
=
.u_ e.u j
=
=
=
-
_c
/
=;
=m-e
>=6.
m m
/
c=-
u
= x >.
,/
c-=
/
m><
-==
<
<=-
/
=
==<
ec-l u-m
/
E oa=
/
- _n
/
/
/
,l/
i sc
=
=
=
c e
o e
=
~
e o
e n
~
(:ss
=/5'4.Olx) X3'ii SSYW c
8
.e s
o
~
/
s
/
I
/
f l
e c_*
I eee j
E
~c l
e me s
u e
w
~ c-e c
rc m-e
]
==w ac c,
.e,
(
5 L :=
e u.
=
t u ^ a_
ec e
)
"E-c c v f
~e EdN l
/
ocE
^
wus l
/
x
==-
=
==u
/
eew e
c
(
v uv=
l 1
=
w en e e. a
/
-
~
y Nw
/
c W
- C
/
-e-ES
/
ce r-=
CZ" uw=
>- w
\\
u.
e
==-
6
/
= x >-
c-=
/
w.
= <
-==
/
~<=
<r-
/
e r=<
==~
v-m f
-eN w
eu
/
==
/
-u.
t 1
{
c i.
e o
e e
e e
o e
N c
c e
m N
(cas g /Sa otx) XA'id SSYn' e
6.-
S.-
.Fi
- 4,-
noo 9
~
LAI
/
y 3..
a
'ce s'
/
mq 2. -
1.
/
M3CPT03/IIEM (GE)
/
- - - - RELAP 1.-
i l
l 0.
i O.
10.
20.
30.
40.
50.
TIME (seconds)
Figure 5 INTEGRATED MASS RELEASE