ML20127A820
| ML20127A820 | |
| Person / Time | |
|---|---|
| Site: | 05000000, Limerick |
| Issue date: | 02/17/1983 |
| From: | Pasedag W, Postma A NRC |
| To: | |
| Shared Package | |
| ML19292B772 | List:
|
| References | |
| FOIA-84-624 NUDOCS 8308150070 | |
| Download: ML20127A820 (40) | |
Text
r e'
I I
AN ASSESSMENT OF FISSION PRODUCT TRANSPORT PROCESSES APPL.ICABLE TO THE LIMERICK PRA A.K. Postma W.F. Pasedag February 17i 1983 6
850tlCooJo O
+
I l
i CONTENTS i
1.
Summary and Conclusions....................
1 2.
Introduction 2
3.
Objectives 3
4.
Technical Discussion of Key Transport and Depletion Processes.
3 A.
Suppression Pool Scrubbing................
3 B.
Performance of the SGTS 33 5.
Literature Cited 35 e
l M
e 9
f a
e h
a a
f f
i t
. ~,.,,..,, -., -,
7-
.-.n.---
---,-------n
- + -
o 1.
SUMMARY
AND CONCLUSIONS An independent study was made of key fission product transport assumptions contained in the Limerick PRA. Two depletion processes were studied:
(a) aerosol scrubbing by a. saturated suppression pool and (b) performance of the standby gas treatment system (SGTS).
The performance of both of these systems can significantly affect calculated consequences. The satura-ted suppression pool was assumed to remove 90% of incident particles in the Limerick PRA. This represents a departure from WASH-1400 where scrubbing by a saturated pool was neglected. The performance of the SGTS was analyzed without account,ing for plugging by aerosol mass. While this treatment paralled that ustd in WASH-1400, other studies have pointed out that such filters would likely plug after trapping relatively small quantities of aerosol. ma'ss.
1 The work completed supports the following conclusions and summary statements.
F 1.
The assumption of a decontamination factor (DF) of 10 for a saturated pool was not supported by model calculations or by applicable experimental data. Therefore an adequate technical basis for a DF of 10 was not provided in the Limerick PRA.
2.
A mechanistic model of pool scrubbing was derived to predict DFs for saturated pools because a review of literature did not reveal applicable experimental data. Only one experimental study was
. uncovered where scrubbing efficiencies in a boiling pool were
measured. The lab-scale tests indicated a negligible scrubbing efficiency under boiling' conditions.
.3.
The model indicates that scrubbing efficiency is highly dependent on particle size..Both large (> 4 pm diameter) and small
(< 0.01 pm diameter) particles are predicted to be removed efficiently.
However for the intermediate size range. DFs as low as 1-2 are predictable.
4 For saturated pools, a significant steam flux is predicted for r4 sing bubbles.
The evaporating steam impedes particle deposition; its magnitude increases with pool temperature.
5.
Realistic analyses would show that scrubbing efficiencies would vary with time and accident sequence because numerous controlling parameters change with time. Therefore the assignment of a single value to pool DF is in itself unrealistic.
6.
A study of scrubbing efficiencies predicted by the model developed herein show that the use of DF = 1 for a saturated pool would lead to an appreciable overpredict\\on of the consequences of postukted accidents.
After considering the range of predicted DFs, the DF of 10 used in the Limerick PRA is judged to be a realistic esti-mate for risk assessment purposes.
e e
s 7.
.The published data base on pool scrubbing is inadequate to sup-port truly realistic predictions of pool scrubbing efficiences for accident-generated aerosols.
8.
Standby gas treatment systems of current design appear to be susceptible to plugging by aerosol masses that amount to only N 0.006 of the aerosol mass that could be generated by a severe accident. Therefore their ability to mitigate consequences is limited.
9.
Rialistic analyses of SGTS performance should account for the enange in pressure drop and flow rate as aerosol mass is accumulated.. The neglect of this factor, in both the Limerick _,
PRA and in WASH-1400 tends-to lead to an underprediction of accident source terms.
2 t
i
2.
INTRODUCTION This report presents an evaluation by the NRC staff of key fission product transport processes applicable to the Limerick PRA. The focus has been on key depletion processes, the treatment of which affects the source term to the environs and therefore the consequences of the various accident sequences.
The Limerick plant is located relatively close to a major population center.
and, in order to verify that the operation of the plant would not pose unwarranted ~ risk-to the population, the NRC required the applicant to pre-pare a probabilistic risk assessment (PRA) for the plant using WASH-1400 methodology.
The PRA was submitted to the Commission in March,1981 and_,
has been under review by the staff. -An earlier review, prepared for the NRC by Brookhaven National Laboratory, focussed on sequence probabilities, but did not deal extensively with details of fission product transport.
Several key aspects of fission product transport were treated differently in the PRA than ha'd been done in WASH-1400. Most important among these i
was the assumption of effective scrubbing (DF=10) for a saturated suppres-sion pool.
In WASH-1400 scrubbing by a saturated suppression pool was negl ected.
Because realistic estimates of all relevant transport and depletion processes are needed for PRA anlayses, realistic credit for pool scrubbing is desirable.
The present report presents a staff evaluation of pool scrubbing and other 5
-a
- + - - - - -,
,,,-----n
,v,,
- a
,---,.-----------,--.n-s.
,,-,,p.
,-.e- -. ~,- - - - -,
-nww-
1 a
key fission product depeltion processes assumed in the Limerick PRA.
3.
OBJECTIVES The objective of this work 'is to provide independent technical evaluations of key assumptions used in the Limerick PRA that affect the predicted source terms. Areas addressed include:
(a) suppression pool scrubbin'g, and (b) performance of standby gas treatment systems (SGTS).
It is stressed that realistic, best estimate values are sought to describe fission product behavior herein.
4.
TECHNICAL DISCUSSION In this section, each of the areas discussed will be described, the approach adopted in the Limerick PRA will be noted, and independent analyses will be presented.
A.
Suppression Pool Scrubbing 1.
Limerick Assumptions Scrubbing efficiencies for suppression pools are ambiguously described in the PRA. However based on detailed discussions between the staff and representatives of the contractors who prepared the PRA, it appears that the following DFs (decontamination factors) were assumed:
a.
For sub-cooled pools, DF = 100 for all materials except noble gases and organic iodides.
b.
For saturated pools, DF = 10 for all materials except noble gases and organic iodides.
c.
DF = 1 for noble gases and organic iodides, d.- All radioactive materials, except noble gases and organic iodides, were coagglomerated into aerosol particles, and the fractional abundance of each material was the same for all particle size groups.
~
. Assumption (b) above represents the main departure from WASH-1400 method-ology. The use of a DF = 10 (an attenuation factor of 10 for the suppres-sion pool) reduces the source term by a factor of 10 for sequences where.
the main pathway goes through a saturated suppression pool.
The use of a DF of 10 for a saturated pool was justified by reference to a General Electric report by Rastler (Rastler,1981).
Careful review of the Rastler report reveals that the assumed DF is a judgment by the author.
No experimental data for a saturated pool were presented, nor was the DF justified by a theoretical model. Therefore it is concluded that an adequate technical basis for the assumed DF was not presented in the Limerick PRA.
A study of the literature of scrubbing of aerosol particles by water pools revealed just one reference that dealt with scrubbing by a boiling pool (Remy,1926). The experiments by Remy involved the scrubbing of chemi-cally generated aerosols (50 and NH Cl) in gas washing bottles. When 3
4 r -,
the water was cool, removal efficiencies were 90% or greater, but when the water was boiling the efficiency reportedly fell to zero.
Fuchs (Fuchs,1964), who reviews Remy's results, attributes the decrease in scrubbing efficiency to bubble growth and to a steam sweep effect (reverse diffusiophoresis) that opposes deposition of particles. While this small scale test result may not be applicable to suppression pools, neither does it support the assumption of 90% scrubbing efficiency for a boiling suppression pool.
Because information supplied by the Applicant and that available in the open literature-does not provide a technical basis for the assumed scrubbing efficiency of a boiling pool, an independent analysis was made.
2.
Analysis of Aerosol Scrubbing by Suppression Pools Two independent factors will control the scrubbing efficiency of the pool.
First, the properties of the pool determine the efficiency for a given particle size entering the pool.
Industrial air cleaning experience demonstrates that scrubber efficiency is always a strong function of particle size, with a minimum efficiency being encountered for particles having. diameters in the neighborhood of 0.3 pm.
The second factor is the particle size generated by the postulated accident.
Low efficiencies can be expected if a large fraction of the aerosol mass lies in the 0.05 -
1.0 pm size range. On the other hand, very high efficiencies can be ex-pected for particles larger than a few micrometers in diameter.
Prelimi-nary evaluations of both of these factors were made for the Limerick pRA, and the methodology and resulting conclusions are presented in the following.
a.
Model for Suppression Pool Scrubbing The model developed herein used as a point of departure the " bubbling" model described by Fuchs (Fuchs,1964).
In this model, the dominant scrubbing processes are envisioned to take place inside rising bubbles.
Convective Flow Effects The condensation of steam from the carrying gas represents a convective flow toward the water surface. This flow will enhance particle capture.
On the other hand the evaporation of water into the bubble represents a convective flow that retards particle deposition.
Generally, condensation, if any occurs, would do so very near the gas inlet.
A reasonable approximation to the removal efficiency of the con-densation process is:
fractional efficiency = fraction condensed Numerically, the fraction of inlet gas condensed can be expressed in terms of the mole fraction of non-condensible gas:
[X (1)
F=1-o where F = fraction of inlet gas condensed, X, = mole fraction of noncondensibles in inlet gas, X, = mole fraction of noncondensibles in gas bubble after it attains thermal equilibrium in the pool.
l L
A decontamination factor, applicable to all partilce sizes may be expressed as:
X DF = 1,p 5[i L
(2)
=
The value of X, is an input parameter whereas X, is fixed by pool tempera-ture (fixed water vapor pressure) and total pressure.
X, may be expressed as:
P P"
X, = 1 7
1 P + ph (3)
=
where._
P, = vapor pressure of water, P = total pressure at pool inlet, t
P = atmospheric pressure above pool,
~
p = water density, h = downcomer submergence height.
Condensation will occur if X, is greater than X ; otherwise evaporation g
will occur, and the vapor flux will retard particle capture.
For X, I X g it is assumed that no deposition occurs, i.e. that DF = 1.
Once the bubble begins to rise, evaporation will begin because the pressure inside the bubble decreases with height.
The number of moles of non-con-densible gas that are contained within a single bubble may be expresses as:
M
=
g R
i w
i j
moles of non-c ndensible gas, where M
=
G l
l I
a M
P
". total pressure, t
bubble volume, V
=
R gas constant,
=
T, gas temperature in bubble, assumed to be the same
=
as th'e water temperature.
To a good approximation, the moles of gas (H, CO, etc.) do not change 2
2 significantly during.the bubble rise period. The bubble volume at any time can be calculated from Equation (4) to be:
~
P V
o to (5) 7- " (1-[p
)Pt t
In Equation (5) the zero subscripts denote conditions applicable to the original bubble at a depth equal to the downcomer submergence.
Inspection of Equation (5) shows that V/V, will be greater than unity because P '
t the total gas pressure, decreases with height.
The amount of growth, and the amount of water vapor added depends strongly on P,/P.
An example t
0 calculation illustrates bubble growth:
for inlet gas at 500 C containing 20% hydrogen, the pool would reach a steady state temperature of 95.5 C (one atmosphere pressure above the pool); for an initial pool depth of 7 feet Equation (5) predicts a volume ratio of 3.8.
Most of the increased volume is water vapor evaporating into the bubble. The water vapor flux i
retards particle deposition.
l Stepwise calculations show that the steam evaporation velocity is a minimum L
at the entry point into the pool and increases with time.
For a small change in height, the average steam velocity may be detennined from changes in volume:
P, (Ds - D )
8 2-i (6) v =
2 2
3Pt (D + D ) At 1
2 where v =
average steam velocity at interface, cm/sec, P,
water vapor pressure, atm,
=
average total pressure, atm, P
=
t D1 bubble diameter at beginning of time step, cm,
=
bubble diameter at end of time step, cm, D2
=
duration of time step, sec.
at
=
The steam evaporation velocity, v, is directed inwardly, and hence acts to oppose particle deposition onto bubble walls..
Sedimentation Gravitational forces cause all particles to settle downward. The net downward velocity is:
V - v sin 0 (7)
V
=
y 3
l where V
downward vertical velocity, cm/sec.
=
y V,
sedimentation velocity, em/sec,
=
1 l
l
. ~ -.
steam evaporation velocity, cm/sec, v =
angle, measured from horizontal.
0
=
The horizontal projection of bubble surface also varys with the angle G.
For a spherical bubble, the' differential surface area. was found to be expressible as:
2 2HR Cos e sin e d 0 (8) dA
=
differential horizontal projection of surface area.
where dA
=
'R bubble radius.
=
The net deposition rate constant, the integral of V dA, was found to bei.
y HR (y, j. y)
(g) 2 (deposition velocity)(area)
=
If the bubble is assumed to have a constant size and be well mixed through-out the bulk of its volume, the airborne concentration decays exponentially with time.
The concentration decay, expressed as a decontamination factor, is:
l' 3
7V
-v 5
exp At (10)
=
0 where DF decontamination factor for sedimentation,
=
V, terminal settling velocity, cm/sec,
=
water' evaporation velocity, cm/sec, i
v =
l 1
.. _ _. m.
bubble diameter, cm, D
=
time period of bubble rise, sec.
At
=
Particle settling velocity is related to particle and gas properties by the standard Stokes-Cunningham law:
p d2 g C" D D (11)
V
=
s 18 tt where V
settling velocity, cm/sec,
=
s 3
p particle density, g/cm,
=
p 2
~
acceleration due to gravity, em/sec,
g
=
C, Cunningham s, lip factor,
=
11 = gas viscosity, g/cm sec.
9 C, depends on particle size and the mean free path of the gas molecules:
-br C,
1+Af+Qfe (12)
A
=
where A
mean free path, cm,
=
particle radius, r =
A,Q,b constants.
=
According to Fuchs (1964), the constants have values of A = 1.246, Q = 0.42, and b = 0.87 when A is assigned a value of 0.653 X 10-5 cm for air at standard conditions (one atmosphere, 273 K). The mean free path
3
- r, s.
s depends on gas properties and conditions.
Kinetic theory (Moore, 1955) leads to the following expression'for A:
1.245 X 10-2-( T)p (13)
A
=
2 MP where-A mean free path, cm,
=
absolute temperature, K,
T
=
molecular weight, M
=
gas pressure, atm,
{
P
=
gas viscosity, poise.
i p
=
The constant in Equati:n (12) was selected to yield A = 0.0653 pm for air at STP.
The application of Equations (10) through (13) permits the scrubbing DF arising from sedimentation to be predicted.
Particle size and bubble diameter are obviously important variables because they enter into the argument of the exponential expression for DF. The steam velocity v is important for parti-les whose settling velocity is of comparable magnitude.
Deposit 'on Due to Centrifugal Forces i
l Rising bubbles circula :e as a result of the differential velocity between l
l the gas and the interf.ce. The centrifugal force is expressible as:
.~
.~
- ~... _.
i, (14)
F c
r where F
centrifugal force, dynes,
=
g mass of particle, g, m =
peripheral ve'locity, cm/sec, y =
radius of curvature, cm.
r =
This force produces a particle drift velocity, whose magnitude is such that centrifugal force is balanced by drag force.
From Stokes law, the drag force,is:
3 II d, p V Fd (15)-
=
cm where F
drag force, dynes,
=
d V
drift velocity due to centrifugal force, cm/sec.
=
c The periferal velocity, V, can be related to rise velocity by assuming that an element of bubble surface. moves fr.om the top stagnation point to the bottont stagnation point in the time a bubble rises through one diameter (Li,1965).
V is simply fV (16)
V
=
b where Vb rise velocity of bubble compared to surrounding
=
water, cm/sec.
,.e-
^
Combining Equations (14) through (16) leads to the following expression for particle drift velocity:
11 p d2 V Cm 2
D (17)
V
=
c 36 D y This drift velocity is measured relative to the gas, and since the steam velocity moves inwardly, the net velocity relative to the bubble interface is:
~
V - v.
(18) deposition velocity
=
c If the deposition velo' city identified in Equation (18) is assumed to deplete particles over the total int'erfacial area, the DF is expressible as:
N P d V[C, 2
6 p
-v at (19)
DF exp y
=
36 D y For sufficiently small particles V will be smaller than v, resulting c
in the argument of the exponent becoming negative.
For such cases the DF is assigned a value of unity.
h
-a
s
, Diffusional Deposition Particles have a finite diffusivity as a result of momentum exchanges i
with surrounding gas molecules. The particle diffusivity is related to gas and particle properties by (Fuchs,1964):
kTC 6
3n y d (20)
=
p
~
2 where 6
particle diffusion coefficient, cm /sec,
=
k Boltzmann's constant.
=
The rate of transfer to the bubble wall may be computed from the penetra-tion theory of mass transfer (Dird et al., 1960; Crank,1967). The mass transfer coefficient, or deposition velocity may be expressed as:
6 2( g t, v
=
D where v
=
0 deposition velocity, due to diffusion, cm/sec, t,
exposure time, sec.
=
Exposure time for a circulating bubble may be computed as the ratio of bubble diameter to rise velocity:
0 t,
y (22)
=
b
~
Combining Equations (22) and (21) lead to V
2(6 g )8 (23) b v
g This depcsition velocity applies when the vapor flux at the wall is.
negligible.
The net deposition is reduced by the steam flux at the bubble wall. A correction' factor for high mass transfer rates can be estimated from data presented by Bird (Bird et al., 1960).
For the present case, using the penetration theory, a correction factor 0 can be related to a flux ratio:
(1 + erf 4)-1 exp (.4 )
(24) 2 0
=
where 0
correction factor for high mass transfer rates,
=
a flux ratio.
4
=
4 is equal to 4 /
H, where AB
- AB (25) 6 b
2(H0)h For small values of 4AB (low v or high 6), e approaches unity.
O approaches zero for high values of 4 By correcting the deposition AB.
velocity defined by Equation (23) with 0, the DF due to diffusion can be expressed as:
h 4
, - _ = _
t exp
(-
b.)1 5 At (26) l DF
=
A conunon term in the three DF expressions Equations (26), (19), and (10) is the bubble residence time, At.
This time is equal to the rise distance
[,
divided by the swarm rise v'elocity:
f (27)
At
=
bubble rise distance, ft.,
where Ah
=
U =~ swarm rise velocity, ft/sec.
~
Overall DF The overall DF may be expressed as the product of the individual DFs given by Equations (26), (19), (10), and (2):
[X exp(K
+K
+ K ) At (28)
=
3 c
D i
DF for steam condensation (DF 3 1),
where X,/X$
=
first order. rate constant for sedimentation, K
=
3 first order rate constant for centrifugal deposition, K
=
c first order rate constant for diffusional deposition.
K
=
D Parametric Numerical Calculations Several cases have been analyzed with the scrubbing model described in
~~'
,.-,.e e
.c.,
,m
.-._._z_
4 the foregoing pages in order to illustrate the importance of key parameters.
Results are described as follows.
Case 1.
Cool Pool Input Parameters:
1.0 (no steam)
X
=
9 10 ft H
=
1.0 (unit particle density)
=
pp D,
0.5 cm (initial bubble diameter)
=
-X, 0.9923 (water pressure = 0.01 atm)
=
1.0 atm (pressure above pool)
P
=
U =.3.8 ft/sec 0.263 sec per step (10 height increments) at
=
Calculated Parameters 9.9 X 10-5 cm/sec v at top of pool
=
D at top of pool 0.54 cm
=
Scrubbing Efficiencies d,, um DF d,, um DF 0.001
>l (6)*
0.40 1.5 0.005 4.6 (5) 0.5 1.5 0.01 710 1.0 2.2 0.02 28 2.0 10.7 0.,05 4.0 3.0 157 0.10 2.3 4.0 6.6 (3) 0.20 1.6 5.0 8.0 (5) 6
- 1 (6) means 1 X 10
~
6..
Case 2.
Warm Pool i-Input Parameters l
i 0
Same as Case 1 except pool temp. = 82 C Calculated Parameters
' v at top of pool 0.011 cm/sec
=
'.58 cm D at top of pool O
=
Scrubbing Efficiencies d, in DF d, um DF p
p 0.001
>1'(6) 0.4 1.2
~
0.005 2/6 (5) 0.70 1.7 0.01 510 1.0 1.5 0.02 23 2.0 7.2 0.05 3.5 3.0 95 0.10 1.9 4.0 3.5(3) y 0.2 1.4 5.0 3.5 (5)
Case 3.
Hot Pool Input Parameters Same as Case 1 except pool temp.
99.58 C
=
i k
I, O
l t
l'
.,,-,-,w--,-
m
,, - -. -..,,.,....,. - ~, -, - -, - -, -, _...,, -,,., -.,,, -, -, - -. -,, -, - - - -, - - - ~ ~...
c.
Calculated Parameters v at top of pool 0.74 cm/sec
=
D at-top of pool 1.35 cm
=
Scrubbing Efficiencies d,,
>l (6) 0.40 1.0 0.005 7.9(3) 1.0 1.0 0.01 62 2.0 1.6 0.02 6.1 3.0 7.1 0.05 1.7 4.0 82 0.10 1.22 5.0 2.3 (3) 0.20 1.1 6.0 1.5 (5)
Several conclusions and summary statements may be drawn from a review of 6ne results for these three arbitrarily selected cases.
Among these are:
1.
Particle size is a dominant parameter.
The pool DF will be
.m-
-limited by the fraction of the aerosol that is present in particles in the penetrating size range, d.05 to 2 pm.
2.
Steam evaporation velocities increase rapidly with pool tempera-
' 'ture, and for po61' temperature close to the boiling point steam
- s evaporation significantly impedes the trapping of particles having aerodynamic diameters between 0.05 and 3 pm.
3.
Particles ~ larger than 4 pm AMMO are calculated to be efficiently trapped even in a ' hot (saturated) pool. Their inertia is able to overcome the inward steam flux at least for a significant fraction of the bubble residence time.
Likewise, diffusional capture remains effective for particles smaller than 0.01 um.
~
b.
Growth of Soluble Particles by Water Vapor Uptake The gas inside rising" bubbles will be nearly saturated with water vapore This atmosphere is conducive to the growth of soluble particles by water vapor uptake.
In view of the critically important effect of particle size on scrubbing efficiency, preliminary calculations were performed to estimate the amount of growth that would be expected.
The equilibrium drop size reached in a humid atmosphere is governed by the degree to which the vapor pressure of water is lowered by the soluble material, and the degree to which curvature effects increase the vapor pressure.
Both effects are well understood and are calculable from classical physics and chemistry.
The saturation ratio, s, is related to drop size by an equation presented by Fletcher (Fletcher,1966).
+
L
exp n kTa
~
~
s =
(29) imM 1+
M(hHa p,,)
3 relative humidity,
~
where s =
surface tension of solution, c =
3 no. of molecules /cm of solution (solvent + solute) n,
=
Boltzmann's constant.
k
=
T temperature, K,
=
radius of drop, a =
~i vant Hoff ionization factor.
=
density of solution, p =
molecular weight of solute, M
=
M, molecular weight of solvent,
=
mass of solute in the drop.
m =
Equation (29) was evaluated under the assumption that the solute was cesium hydroxide, that the solvent was water, and that the temperature was 100"C.
Results are summarized in Table 1.
9 4
a e
m e
y:.~
Table 1.
Growth of Cs0H Particles in Humid Atmospheres at 100*C Dry Pt.rticle Droolet Radius in Stated Humidity Radius, um 5 = 0.9 5 = 0.95 5 = 0.99 5 = 0.999 0.01 0.0195 O.0225 0.0295 0.0345 0.10 0.195 0.255 0.425 0.775 1.0 1.95 2.55 4.45 9.35 10.0 19.5 25.5 44.5 95.5 The data of. Table 1 illustrates that particle growth depends on relative humidity, s, and also that significant growth factors are predicted.
Because the drop size / particle size ratio depends strongly on the relatfve humidity,. a first estimate of relative humidity inside a bubble was made.
This was done by computing the concentration gradient required to produce the water vapor flux that would be required to mainta' n saturation in a i
rising bubble.
For a bubble submerged initially 10 ft. in a water pool at 99'C..after one second, an average steam evaporation velocity of 0.0304 cm/sec was calculated.
For a bubble containing H as the irfert gas, 2
s in the bulk of the bubble was calculated to be 0.993 and for CO s was 2
estimated to have.a value of 0.985.
Based on this estimate it appears that _an s value of @.99 would be a reasonable ball park value.
From Tchle 1, this s corresponds to particle growth factors of 4.4.
This W
amount of growth could be very important because it could move particles from the most penetrating range (0.1 - 1.0 pm) into larger size ranges 4
which are more efficiently collected.
c.
Application to Limerick PRA The scrubbing models. developed herein are applied to a saturated pool
' in a Mark II BWR, the Limerick design. This application will illustrate the factors that affect scrubbing efficiencies and provide estimated DF's for selected conditions.
Pool Temperature A heat balance on the suppression pool was made to determine its tempera-ture because the vapor pressure of water importantly affects the steam evaporation velocity into rising bubbles.
It was found that starting with a boiling pool, a slightly lower temperature is quickly reached when steam / gas mixtures are introduced.
Noncondensible gases (H, CO '-
2 2
noble gases, etc.) appear to always be present in the aerosol carrying gases for severe accidents, and hence the pool is expected to cool to a temperatt're slightly below the normal boiling point. Results of the heat balance, which accounted for decay heat ' amounting to 0.1% of full core power, and which assumed 1 atmosphere above the pool, are summarized in Table 2.
e 6
m 4 e.
a
7 e
Table 2.
Equilibrium Temperature of Suppression Pool Inlet Gas Mole Fraction of Gas
- Equilibrium Temperature. OC Inlet Stream Pool Temperature. OC 1000
- 1. 0-77.4 500 1.0 68.0 100 1.0 49.5 1000
'O.5 90.7 500 0.5 87.5 100 0.5 83.2
~
1000 0.2 96.6
~
500 0.2 95.5 100 0.2 94.1 1000 0.1 98.3 500 0.1 97.8 100 0.1 97.1 Gas
- means non-condensible gas as opposed to steam The data of Table 2 illustrates how pool temperatures would vary with gas inlet temperature and the mole fraction of non-condensibles. While the temperature variations may at first glance appear to be minor (mostly 0
0 95 C to 100 C) steam evaporation velocities, v, and the fraction of gas initially condensed vary significantly for the various cases. The con-l l
ditions that actually developed in an accident would depend on the detailed sequence of events that applied.
l 4
~
g n-
...._u.
=.-..._;
s Pool Depth According to the FSAR for the Limerick Station, the downcomer submergence depth is 10 ft. Some increase in depth, arising from the discharge of water from the RCS, can be expected under accident conditions.
If the 3
b
.11,770.9 ft of RCS volume is added to the suppression pool (surface area 2
f5700ft) the submergence depth would be computed to be 12 ft.
o Particle Size Distribution As noted previoGsly pool efficiencies predicted by bubble scrubbing models are highly sensitive to particle size..Therefore a realistic prediction can be made only if realistic particle size distriubtions are inputted to i
the model.
The particles produced by severe accidents are typically -
submicron in size initially, and then grow to larger sizes by various l'
agglomeration processes.
Thus particle size is expected to vary with both the sequence of events chosen and with time in any particular accident sequence.
l Recent studies by NRC contractors (Battelle,1983) have encompassed cal-culations of particles size for accident sequences applicable to pressurized l
water reactors.
Future woric will focus on BWR sequences, but results for-the BWR are not yet available. Becausa particle size exiting from the RCS t
j or produced by core-concrete interactions is not expected to be markedly
'different for the BWR, available information generated for the PWR can be l
used for first estimates of' suppression pool efficiency. App 1,1 cable infor-i s
motion is summarized in Table 4 L
f 1 L
Table 4 Size of Particles Produced by Severe Accidents Geometric Number Median Standard Description of Particle Source Diameter, un Deviation Exit from RCS, AB Sequ'ence 0.3 - 0.8 2
Exit from RSC TML8' 0.5 - 1.5 2
Core-Concrete Interaction 0.15 - 0.5 2
Example Calculations of Pool Scrubbino Efficiencies Several example, cases are analyzed to provide a basis for choosing DFs applicable to saturated pools. Conditions were chosen to bound those that could occur to illustrate a rhnge of suppression pool DFs.
Case 1.
Inlet Gas at 1000'C, No Steam This hypothetical case represents an extreme in terms of inlet gas composi-tion and. temperature.
Several particle size distributions are used, b
Number median is assumed to be 0.3 un, the particles are presumed to be Cs0H, and be in equilibrium with water vapor saturation ratio, s = 0.99.
Ib Number median is assumed to be 1.5 un and the particles are assumed to be 3
insoluble with a density of 10 g/cm,
~-
s Ac,c Number median diameter is assumed to be 0.15 pm, and the particles are 3
assumed to be insoluble with a density of 4g/cm,
The particle size distributions noted above were converted to aerodynamic mass distributions using the following equation, applicable to log normal distributions.
(p)b NMD exp 3 in e (30) 8 AMMD
=
where THt0 -= aerodynamic mass median diameter, particle density, p =
NMD number median diameter,
=
geometric standard deviation = 2.
a =
Results of Case 1 calculations where the efficiency was integrated over the particle size distribution are presented in Table 5.
Table 5.
Predicted DFs for Case 1 Assumptions Case AMMO. um E
la 5.52 44 4
lb 20 1.6 x 10 le 1.3 2.8 l
i r
i i
c 0-o The data of Table 5 show, as expected, the variation in DF with particle size. Noteworthy is the significant DF predicted for two of the three cases.
Case 2.
Inlet Gas at 1000'Cc 50% Steam This hypothetical case represents an intermediate steam / gas ratio. The three particle size subcases that were used in Case 1 were also used here.
X,, the mole fraction of non-condensibles in the initial bubble is computed to be 0.477 which is less than X.
Therefore evaporation will occur j
j initially ar,d the initial DF, X,/X, is assigned a value of 1.0.
Predicted g
DFs for the three particle size subcases are listed in Table 6.
t Table 6.
Predicted DFs for Case 2 Assumptions l
Case AMMD. um Of, 2a 5.52 28 3
2b 20.
8.7 x 10,
2c 1.3 2.0 The DFs listed in Table 6 for Case 2 are comparable to, but slightly i
smaller than those for Case 1.
The reduced DFs for this case aris's from higher steam evaporation velocities.
The equilibrium pool temperature for Case 2 was 90.7'C as compared to 77.4'C for Case 1.
The corresponding higher water vapor pressure led to higher steam evaporation velocities for Case 2.
I
6 0
Case 3.
Inlet Gas at 500 C. 80% Steam This case is representative of moderate gas temperatures and higher steam fr' actions.' Also, X, is predicted to be 0.372 (pool temperature 0
of 95.5 C) so the initial DF is calculated to be 0.372/0.2 = 1.86.
This DF is applied to all particle size ranges. Predicted DFs for this case are listed in Table 7.
Table 7.
Predictdd DFs for Case 3 Assumptions Case AMMD. un
'01 3a
5. 52 35 3
3b,
20 7.9 x 10 3c 1.3 3.2 The scrubbing efficiencies for this case are higher by approximately a factor of 1.8 than those of Case 2.
This is the result of the initial DF due to condensation that occurs during the short time interval that the bubble comes to thermal equilibrium with the surroundings.
Case 4.
Pool temperature of IdO'C. 50% steam in inlet aas This final case uses a pool temperature of,100 C, the boiling point.
It 0
represents the highest pool temperature that could occur (for 1 atm. pressure) and therefore maximize the steam evaporation velocity. This thermal 4
L
condition would be representatiive for the initial ste'am/H / aerosol mixture 2
that would enter a saturated pool following failure of the containment.
Predicted DFs for this case are listed in Table 8.
0 Table 8.
Predicted 0Fs for a pool at 100 C
^
Case AMMD. um D1 4a 5.52 11 4b 20 1170 4c 1.3 1.4 The scrubbing efficiencies established in Table 8 are significantly lower than those for the lower temperatures, reflecting the increased staanflux[PerhapsmoreimportantisthatDFsignificantlylarger than unity are predicted for this h,ighest steam flux case.
- d. Sunnary of Scrubbina Efficiencies for a Saturated Suppression Pool The foregoing study of particle scrubbing efficiency by a saturated suppression pool supports the following conclusions and summary statements.
1.
Scrubbing efficiency is highly dependent on particle size. Both large particles (d um) and small particles (<0.01 um) are predicated to be very efficiently captured by the pool. However, for an intermediate size range DFs dose to unity are predictable.
I l
l 2.
Significant steam evaporation velocities will occur into bubbles rising in hot pools. The steam flux retards the capture of particles in the intermediate size range, but is too small to prevent the capture of large particles.
3.
Pool scrubbing efficiencies would vary with time in an accident sequence because controlling parameters would be time snd sequence dependent.
Important parameters include: particle size and density, fraction of aerosol that is soluble, steam fraction in
' carrying gas, molecular weight of non-condensible gas, pool temperature, and scrubbing height. Therefore realistic, mechanistic predictions of pool scrubbing efficiency should account for all of these factors.
4.
A study of the DFs predicted herein show that the use of DF = 1 for a saturated pool would lead to an overprediction of the consequences of postulated accidents. After considering the range of likely DFs, it is concluded that a DF of 10 would be more realistic than a DF of unity. Therefore, the DF of 10 used in the Limerick PRA is judged to be a realistic estimate for risk assessment purposes.
5.
ThelowDFs(N2)predictedforthevaporizationrelease(.13AMMD
?
o s
i particles) are judged to err on the low side because the particle size used in the calculation applies to particles very close to the source (%1 m) a point where little agglomeration would have occurred.
Significant agglomeration is expected between the particle source and the pool so,the particles entering the pool will be larger than the cited, measured size. For the particle 3
concentrationsexpected(10-100g/m)agglomerationwouldbe unavoidable.
Future evaluations should quantify this agglomeration.
6.
The very low 0Fs reported by Remy (Remy, 1926) for a boiling pool are exNainable by the model developed herein; the particles were N 1 um in diameter a size for which steam flux suppresses deposition. Alsothescrubbing'heightsweresmall(s0.5ft) compared to those for suppression pools.
7.
The model employed herein used a number of assumptions that need to be verified experimentally. Of particular importance is the degree to which bubbles circulate, considering the effect of interfacial impurities.
Experiments that investigate particle capture mechanisms under realistic conditions are needed to support scrubbing models.
l t
.n.
---a
r o9 l
B.
Performance of the SGT5 The standby gas treatment system ($4TS) is assumed to operate effectively l
in a number of leak path cases. The calculational method appears to be based on the assumption that pressure drop,across the SGTS does not increase significantly as a result of influent aerosol.
In reality, the aerosol loads would be likely to plug the SGTS in a relatively short time.
l Plugging of SGTS filters by aerosol loads was neglected in WASH-1400, and this neglect represents an area of non-conservatism in WASH-1400.
Realis-tic treatent of SGTS performance would require that aerosol loads (mass deposited), humidity, flow rates, etc., be followed as a function of time for.each sequence that involves the SGTS.
Such a treatment was not done in the Limerick PRA nor was it.;done.in WASH-1400. The following order of magnitude analysis provides a measure of perspective.
I l
l HEPA filter systems are generally not used for air cleaning applications where large aerosol masses are encountered because they are relatively l
easily plugged.
Experiments by McCormack (McCormack,1978) indicate that for fine, wet aerosols, the loading capacity for a standard 1000 cfm filter unit was approximately 0,6.
Thus for the SGTS, the loading capacity would be of the order of g X no. filters I'(0.6)(20)=12kg.
mass collected a
4 4
D u
This aerosol mass, which would include sorbed water, represents a small fraction of the aerosol mass which would be produced by a severe accident.
Using an aerosol mass of 2000 kg (NRC,1981), the SGTS would be able to trap only N12/2000 or 0.006 of the total.
If the standard co-agglomera-tion assumption is used, then only @.006 of the radioactive material could be collected by the SGTS. This limitation on mass collection cap-ability limits the mitigation capability of the SGTS.
It appears that the SGTS would perform effectively only for accidents in which retention factors greater then s1/0.006 = 167 limited the aerosol mass entering it.
In summary, the potential plugging of the SGTS by aerosol mass was neg-1ected in both the Limerick pRA and in WASH-1400.
Because the SGTS could capture less than 1% of aerosol mass generated by severe core melt acci,,
dents, this neglect tends to lead to an underprediction of accident source terms.
Realistic analyses should account for the time rate of accumulation of aerosol mass by the SGTS, the change in pressure drop, and aerosol flow rates in parallel leak paths.
4-
o ca.. s 5.
LITERATURE CITED Battelle Columbus Laboratories and Sandia National Laboratory.
" Radio-nuclide Release Under Specific LWR Accident Conditions (Volume I, A PRW Analysis"" NUREG-0956 (Draft)
U.S. Nuclear Regulatory Commission, Washington, DC, January 14, 1983.
R.B. Bird, et al..
Transport Phenomena, pp. 495-676 John Wiley and Sons, Inc., New York, New York.1960.
J. Crank.
The Mathematics of D1,ffusion, pp.1-98, Oxford University Press, London, 1967.
N.H. Fletcher. The Physics of Rainclouds, pp. 37-63, Cambridge Univercity Press, London, 1962.
N.A. Fuchs.
The Mechanics of Aerosols, pp. 240-245, 25-30 The Macmillian Company, New York, New York,1964.
P. Li, et-a1... " Unsteady State Mass Transfer from Gas Bubbles-Liquid l
Phase Resistance" AlchE Journal, 11 (41, pp. 581-587, 1965.
l J.D. McCormack, et al..
" Loading. Capacity of Various Filters for Sodium Oxide / Hydroxide Aerosols" CONF-780819, pp. 1018-1043, August, 1978.
W.J. Moore.
Physical Chemistry, Second Ed., pp. 160-199, Prentice-Hall, Inc., Englewood Cliffs, New Jersey,1955.
Nuclear Regulatory Commission.
" Technical Bases for Estimating Fission Product Behavior During LWR Accidents" NUREG-0772 Washington, DC, March l
1981.
D.M. Rastler.
" Suppression Pool Scrubbing Factors for Postulated Boiling Water Reactor Accident Conditions" NEDO-25420 General Electric Company, San Jose, California, June 1931.
H. Remy and H. Finnern.
" Absorption Chemisher Nebel durch Flussigkeiten and durch feste Stoffe" Z. Anora. Chem., 159, pp. 241-255, 1926.
i
(
G
- - - - - - - - - - - - - - - - -