ML19347E183
| ML19347E183 | |
| Person / Time | |
|---|---|
| Site: | Sequoyah  |
| Issue date: | 04/30/1981 |
| From: | Baer M, Berman M, Jamarl Cummings, Griffiths S, Sherman M SANDIA NATIONAL LABORATORIES |
| To: | NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| References | |
| CON-FIN-A-1246 NUREG-CR-1762, SAND80-2714, NUDOCS 8104240197 | |
| Download: ML19347E183 (133) | |
Text
..
NU REG /CR-1762 SAND 80-2714 R3 ANALYSIS OF HYDROGb;1 MITIGATION FOR DEGRADED CORC ACCIDENTS IN THE SEQUOYAH NUCLEAR l'OWER PLANT s
M. Berman M.
P.
Sherman J. C. Cummings M.
R.
Baer S.
K.
Griffiths Printed March 1981 Sandia National Laboratories Albuquerque, NM 87185 Operated by Sandia Corporation for the U.
S. Department of Energy Prepared for Division of Reactor Safety Research Of fice of Nuclear Regulatory Research U.
S. Nuclear Regulatory Commission Washington, DC 20555 Under Memorandum of Understanding DOE 40-550-75 NRC FIN No. A1246 g
6 9togzwf1
I t
-t ACKNOWLEDGEMENT Contemporary research has changed significau y from the days of Newton, Einstein and Edison. _ Present studies tend to be very broad '
and complex, frequently involving many dif ferent disciplin and spe--
cialties. _Research performed by one or two investigators has become The coordinated efforts of multi-faceted teams'are now commonly-rare.
required for the solution of complex problems.
Space'and propriety frequently limit the number of. contributors whose names can appear as-authors. The efforts of many other individuals frequently pass unno-ticed and unrecogrized, even though their contributions to the: success of the integrated effort'were quite significant.
We would like to acknowledge those ettorts here, and inform the readers of their important' contributions.
Bleine Burnham performed the MARCH calculations for this report, becoming a user and code modi-fier within a space'of 8-weeks.
Rupert Byers performed the CSQ deto-nation calculations.
Pat Rosario and Doris' Jackson typed the two ver-sions of this manuscript in record time and with a minimum of errors.
We also acknowledge the encouragement, guidance, and material support provided to us by our NRo program manager, Don Hoatson.
I 1
f 1
i i
4 j
-s I
l iii
SUMMARY
This report is intended to provide a preliminary assessment of three mitigation schemes proposed for hydrogen control in nuclear pow-er plcats: deliberate ignition, water fogging and Halon inerting.
We have been requested to respond to 21 questions addressing various as-pects of these mitigation schemes including efficacy, practicality, operational strategies, post-LOCA behavior, design concepts and costs, and, most importantly, whether overall plant safety will be improved or degraded by the deployment of su:h mitigation systems.
As a result of this study, we have concluded that no one of these mitigation schemes is clearly superior to the others t.nder all accident-conditions.
Advantages and disadvantages have been identified for all systems.
In addition, there are accidents for which all-schemes would provide improved safety.
However, accident scenarios can also be idan-tified for which deliberate ignition or.Halon inerting could be detri-mental to safety.
For scenarios. in which hydrogen is released from the primary sys-tem at moderate rates, and where mixing occurs rapidly, deliberate ig-nition of lean mixtures should be beneficial.
Preliminary calculations indicate that deliberate ignition in the lower compartment significant-I ly reduces the pressure rise in containment under the following condi-tions:
hydrogen release rates up to at least 35 lbs/ minute (the maxi-mum rates observed for MARCH calculations of small and large break LOCAs during the degraded core portion of the accident); and sufficient-ly rapid mixing to prevent high concentration pockets from forming.
Under certain accident conditions, however, the lower compartment could be inerted either by high concentrations of steam, or by low concentra-1 tions of oxygen.
If such inerting should occur, the interi= distributed ignition system (IDIS) as presently planned for Sequoyah has a serious shortcoming.
The inerted gas mixture entering the bottom of the ice iv 4
condensers will emerge as an extremely rich' mixture-at the top.
Con-centrations could approach or exceed the detonability limits in s
}
toroidal region around the periphery of the containment at the top of the ice condensers.
Four igniters are presently planned for this region.
We strongly recommend that those igniters be removed.
Instead, we sug-gest that the upper compartment deliberate ignition strategy should attempt to barn lean mixtures high in the upper ccmpartment away from the ice condenser outlet plenum.
Note that the upper compartment is much larger than the lower; furthermore, it does not hpve either the heat-removal capacity of the ice condensers or the'exp~ansion into the lower volume available to it.
Consequently, burns in the upper compart-ment can yield pressure rises approaching those computed for adiabatic, constant volume deflagrations.- Igniters placed high in the upper com-par tmen t (UC) will tend to burn more dilute mixtures than igniters immediately above the ice condensers.
Lean mixture combustion (below
- he downward-propagation limit of 9%) will pose a smaller threat to the UC since only part of the hydrogen will be burned. We therefore recom-mend placement of the igniters high in the upper compartment.
In the lower compartment, the present number and arrangement of igniters appears adequate for an interim (IDIS) system.
The 3equoyah IDIS presently involves the automatic initiation of all igniters simultaneously. While we believe that this system will improve safety, we also believe that an operator-controlled, computer-and detector-assisted system would be superior.
Such an advanced sys-tem would permit one or more igniters to be separately controlled, but would require a much more extensive monitoring system than presently 3
exists.
Important questions which remain to be quantitatively answered in determining the relative merits of the deliberate ignition syctems in-j clude: more accurate determinations of the rates, quantities and loca-tions of hydrogen generation for various accident scenarios; the prob-v
ability of producing locally high concentrations of hydrogen; the impact of the ice condensers and fans on hydrogen transport and combus-tion in the upper and lower compartments; the effects of combustion on i
safety equipment; and the fraction of hydrogen consumed in each combus-tion event.
Experiments to date indicate that combustion completeness is a sensitive function of geometry, igniter' strength, number and loca--
tion of igniters, and hydrogen concentration.
Conceptually, water fogging appears to be a very attractive miti-gation scheme. We have performed kinetic vaporization calculations which indicate that the droplets will be vaporized rapidly enough to validate the thermodynamic calculations; i.e., significant reductions in pressure and temperature resulting from the addition of 0.05 vol. %
liquid water will occur for droplets of all practical sizes (10-100 pm i
radii) during deflagrations in containment.
These same droplet sizes will not be vaporized in the flame front, but rather a short distance behind.
This implies'that the deflagration will not be quenched, per-mitting the combining of deliberate ignition with water fogging into an integrated system.
Our analysis, however, has uncovered a poten-tially severe limitation on the practicability of water fogs; i.e.,
j the requirements for its maintenance during the accident period.
In l
Sequoyah, only about 4500 gallons of water are required to reach 0.05 volume %.
To maintain this level, however, the spray system must provide water to make up for losses by settli.ig, agglomeration and collisions with surfaces. Under the assumption of gravitational settling and i
instant coagulation upon contact, calculations have indicated that losses may be severe.
Additives which increase surface tension might reduce agglomeration losses, but this is presently unproven.
It may also be possible to introduce the required volume of water by means of 5
a foam.
This is also being investigated.
From our studies, then, we conclude that fogging could be very beneficial if problems associated with its maintenance in the containment atmosphere can be solved.
vi 1
. m.
T Our investigations have uncovered a significant body of informa-tion concerning the use of.Halon as a hydrogen combustion preventive.
measure.
Halon'has been previously employed for such uses on a large scale (106 ft3).
At'known concentrations, Halon will inert hydrogen: air steam mixtures.
For Sequoyah, an adequate system would cost approxi-mately three million dollars and could probably be installed within a few months.
There are, however, some potentially detrimental aspects to the use of Halon as a hydrogen' control measure.
The addition of the Halon itself (264,000'lbs) will raise the containment pressure about 9 psi.
For accident scenarios where steam overpressurization poses a-greater threat than hydrogen combustion, this 9 psi Halon incremental pressure could be very detrimental.
If the Halon concentration falls
~
below the inerting level, the Halon itself could contribute energy to the combustion, and the high deflagration temperatures would decompose a
the Halon to produce highly corrosive halogens and halogen acid compounds.
Even if the Halon accomplishes its inerting objectives, some means would be required to remove the hydrogen from the post-accident environ-ment without decomposing the Halon.
Present recombiners would require modification to accomplish this.
Halon will also radiolytically decom-pose, and prolonged exposure could lead to significant' decomposition.
It is important to note that, unlike water fogs, Halon and deliberate ignition are incompatible mitigation schemes.
2 1
4 a
vii
TABLE OF CONTENTS l
Page
't 11 ABSTRACT iii Acknowledgement.........................
iv
~
SUMMARY
3 FOREWARD 5
1.0 DELIBERATE IGNITION 1.1 Introduction and Discussion.
5 Combustion of Lean Hirdrogen: Air Mixtures 6
21 Ignition Sequoyah Interim Distributed Ignition System 24 Local Detonations 30 35 Accident Analyses 1.2 Answers to t iestions 65 References for Deliberate Ignition 69 71 2.0 WATER FOGGING 2.1 Introduction and Discussion 71 Kinetics of Droplet Vaporization 74 82 Production of Water Fogs 83 Fog Maintenance and Stability.
The Ef fects of Fogging on Detonations 92 and Shock Waves 97 l
2.2 Answers to Questions 105 References for Water Fogging 3.0 HALON (CHEMICAL) INERTING 107 3.1 Introduction and Discussion.
107 3.2 Answers to Questions 109 122 References for Halon.
l
[
\\
123
' - NRC Work Scope 1-2 l
FOREWORD Late in August 1980, the U.S. Nuclear Regulatory Commission asked a
Sandia to perform a short term (2-3 months) analytic study to Investi-gate the behavior and utility of three hydrogen control mitigation schemes:
deliberate ignition, water fogging, and Halon injection.
At-tachment 1 describes the scope for this short term effort.
The urgency of the request was based on the following circumstances:
(1) Licensing decisions needed to be made on Sequoyah and other PWRs equipped with ice condensers and containments smaller and weaker ' aan' standard large, dry PWR containments; and (2) these containments are more vulnerable to hydrogen combustion damage, for quantities of ' hydrogen comparable to that produced during t'.ie TMI-2 accident.
Sandia staff began to investigate these schemes during the first week in September 1980.
This report summarizes the results of eight weeks of intensive study.
Although we believe that significant progress has been made to date, the reader must keep in mind the extremely short duration of the study. We have attempted to ensure that no major er-rors were made, that no significant omissions cccurred, and that the is an accurate evaluation of the three mitigation schemes in report l
terms of present knowledge and experimental information.
L The three mitigation schemes were investigated separately.
The three major sections of this report reflect that separation.
Each sec-f tion is divided into two parts.
The first part is an introduction con-In the cerning the general nature of the specific mitigation system.
cases of deliberate ignition and water fogging, a more detailed discus-sion is presented concerning the important phenomena and conclusions.
Following the introduction and discussion, each of the questions asked I
in the work scope (see Attachment 1) is specifically answered, based on our best engineering judgment at the present time.
l t
i 3-4 f
)
J I
1.0 DELIBERATE IGNITION e
1.1 Introduction and Discussion
" Failure" pressures for the Sequoyah containment have been esti-mated by seve,ral groups (1-3) with the results shown in Table I.
Table I
" Failure" Pressures fcr sr? Sequoyah Containment Pressure, psig Group Yield Ultimate TVA (Ref. 1) 33 43.5 TVA (Ref. 3) 22.7 43.2 NASA-Ames 35.6 66.7 R&D Assoc.
27 NRR 34 The highest value in the table, 66.7 psig, is admitted to be not very accurate.
The NRC recommended the value of 31 psig as a conserva-tive " failure" pressure in Ref. 1.
In subsequent conversations, how-ever, we learned that 45 psig is now accepted (4) as a conservative maximum.
For the purposes of this report, we have performed calcula-tions for a range of failure pressures which encompass both the 31 and 45 psig values.
It should be noted that during informal conversations, I
structural engineers at Sandia indicated that failure often begins at local stress concentrations and weaknesses (flaws), and propagates throughout the structure.
Idealize'd analyses, such as those shown in Table I, may be too optimistic.
i e
We can conservatively predict the pressure rise due to the defla-gration of homogeneous hydrogen: air or hydrogen: air: steam mixtures in containment by considering the combustion as a complete, ad iabatic,
i 5
constant-volume process.
In reality, anticipated deflagrations during possible accidents in Sequoyah will not be adiabatic, nor will they oc-cur at constant volume.
In Sequoyah, the upper and lower compartments y
are quite distinct and separated by paths running through the ice con-densers and recirculation fans, Figure 1.
Burns which take place in the lower compartment can expand ' into the upper co'mpartment and, in the process, transfer large quantities of heat by cooling and condensation in the ice condensers.
These mechanisms greatly reduce the pressure and temperature rises compared to adiabatic, isochoric deflagrations.
Burns in the larger upper' compartment, however, can approach the adia-batic, isochoric results.
Pressure or temperature reductions result only from heat transfer to the walls (and cooling by the sprays if they have been pre-activated), and expansion through the return-air fans; these mechanisms are much less effective than those available for lower compartment burns.
Results for adiabatic, isochoric, complete combustion are shown in Figures 2 and 3.
Thirty-one psig overpressure corresponds to a combus-tion of a little less than 6% hydrogen by volume; 45 psig corresponds to about 8% hydrogen.
Table II summarizes some pressures which would result from deflagrations of various concentrations of hydrogen for certain initial conditions.
The percentage of metal-water reaction (MWR) is based on a total 100% yield of 2250 lbs (Sequoyah FSAR).
(Note that other references consider 1950 lbs to represent 100% yield; see discussion at end of Section 1. )
COMBUSTION OF LEAN HYDROGEN: AIR MIXTURES Assumptions less conservative than those of no heat transfer, con-stant volume, and combustion completeness would reduce the pressure and temperature rises illustrated in Figures 2 and 3.
As Table II shows, combustion may be incomplete for lean mixtures of hydrogen in air.
Below about 9% hydrogen (the downward propagation limit) the combustion of 6
e 4.N SM/ ELD BU/LD/MG DCME I
STEEL CONTAINWENT VESSEL EL 429 08 I'
p
?
sti$ TON POL AR CRANE $.A,,
gyjggy
,y,to,g
{ POLAR CRANE
?
WALL m I,
i p.
e_
i TOP OFDECK
\\!
ypp op cppyt pyyg
- a3PMING LIN.
I#
KE LOADING BRIDGE CRANE ktNTILAT/0N {
F A-yngfyg TOP OF KE BED
}y' \\W N
UNITW[
UNITS 41*-6 R l
ll~
h l
EL 780.0 %
?l (DryrDER gARRIER l gg yyg_gg Wr ji'i p,i e-i.
e.
g f
!t i t
'[
l:lf: ;
gf
- ES ~
ICE CONDEN$ER --~
.T a\\ f' ll
'( \\
ll i e
\\1, l
':S,
t CONTAIN,ENT s
s
\\
l
?
t..t.Jt.1 k s srf,
i cm
!,y y; i i i
GE Euro,.
\\
CONrROL RCD DRivt
+,
aissiit snino l. a>. u. i
~
t.
Y E ##
'f KE COND
- s 1
,-a nu, A-p-
\\
/,. &p
' " "'0' A
gm l
-u--
e
- pJs.
p
... ;. -M il vAROR n
.. fCOO 7
U d
0'##ff" J'0*-
-# 0 '
N l
,au-aL,70R~
R
,R
\\
r
+
[
GRADE
}&,LOCR El 702 0 f
_f El7C30's. - - -
h l n es3.0
- f f- ( $
g)k*-
g '.
g f
o I'
N
+
A Eca7 t
.... ' k i.
U EL 6910 m
s
~
-+-
u+ t '.
Y--
21
'r 1
.r f.
.' a'
"" !)
l e
e fEL 679 7#
j fg
'l i !
g
._k.
4
[
g,
x, _,
e.
", h...
5 hSTEEL'LIIVER t *RE CTOR -. 'e e
y
.,p r
N ANCMOR BOLT
.e, I
a/tc0.so+e m n estas
' e' y
soap 3
.e.
Figure 1.
Reactor Building Elevation 7
m i
i i
i INITIAL CONDITIONS 3000 Ti = 298 K= 25'C STOICHIOMETRIC 2800 P, = 1 ATM = 0.1013 MPa -
RATIO 100% RELATIVE HUMIDITY 2600 x
I
& 2400 i
l 2200 W
i 1 2000
\\
D
[
k 1800
=
DETONATION LIMITS
$ 1600 c.
2 1400 I
W H
1200
.J h 1000 4
DOWNWARD PROPAGATION LIMITS i
800 UPWARD PROPAGATION LIMITS
=
600 I
I I
I 400 O
8 16 24 32 40 48 56 64 72 INITIAL HYDROGEN CONCENTRATION, VOLUME PERCENT Figure 2.
Theoretical Adiabatic, Constant-Volume Combustion Temperatures of Ilydrogen: Air Mixtures
I 5
t 9
i 1
INITIAL CONDITIONS a-I
'[ STOICHIOMETRIC T = 296 K = 25'C i
RATIO i
N 8 Pi = IATM = 0.1013 MPa
~
a
\\
100% RELATIVE HUMIDITY 3 7 4
W i
D W
t e
C Q.
6 DETONATION LIMITS
=
c 4
g "5
F-E5 I
w EU 4
=
=
~
]
m DOWNWARD PROPAGATION LIMITS e
w C
1 1 3
- -l i
j Z
P UPWARD PROPAGATION LIMITS
=
2 0
8 16 24 32 40 48 56 64 72 INITIAL HYDROGEN CONCETRATION, VOLUME PERCENT 4
Figure 3.
Theoretical Adiabatic, Constant-Volume Combustion Temperatures of Ilydrogen: Air Mixtures i
4
hydrogen by weak igniters, such as spark' plugs and glow plugs, is found to be incomplete in large chambers.
Results for the ignition of hydrogen: air mixtures by a spark plug at the center of a 12-foot sphere 6
by Furno, Cook, Kuchta and Burgess,(5) and for similar experiments in a 6-foot-diameter sphere by Slifer and Peterson(6) are shown in Figare 4.
These results show very incomplete consumption of hydrogen up to 8% in-icial hydrogen volume fraction, rapidly approaching complete consumption for volume fractions over 9% hydrogen.
When the spark plugs were re-placed b;' much more intense igniters, pyrofuses, a much higher fraction of the hydrogen was consumed below 8% initial hydrogen volume fraction, as shown in Figure 5.
These _ experiments indicate that the strength of the ignition source influences combustion contpleteness for hydrogen concentrations below about 9 volume % hydrogen.
Table II Hydrogen-Air Deflagrations Assumptions Homogeneous, Constant-Volume,-Adiabatic, Air Partial Pressure 1 Atmosphere, 298 K, 100% Relative Humidity Overpressure, psig Hydrogen Concentration
- Metal-Water Reaction Complete Incomplete in Air (by volume)
Equivalent (%)
Combustion Combustion 6%
17 34 1 - 22 8%
24 44 4 ~ 42 9%
27 50 38 - 50 16%
52 86 86 19%
64 98 98
-The size and shape of the container in which the lean hydrogen: air:
steam mixture is burned also have a significant effect on the fraction of hydrogen consumed.
Recent work by Harrison, Tamm, MacFarlane and i
- Data from Sequoyah FSAR:
Zirconium inventory--50,913 lbs (equivalent of 2250 lbs H2) and containment volume--1,191,500 cu f t.
10 I
no
/
7 p
INIT I A L 0
(*Cl fatml 2
tus h
100 13 3e j'
e
=
V 38 1A 5
d 71 1.2 17-21 60
/
O FURNO4OATA
/
THEORETICAL O P FOR CouPLETE COUBUSTION
/
/
/
/
1
~
/
5
/
j' e
/
v
=
f
/
/
/
/
/
/
/
/
A
/
ec-e^ /dt
,/
i I
12 0
2 4
6 8
10 MYOROGEN CONCENTR ATION fv0L %I Pressure Rise Versus Hydrogen Concentration, Figure 4.
Spark Igniter at Center of 12-foot Sphere (Data are from Refs. 5 and 6)
I l
1 11
19 e
V
/
/s vd*/,.
/
W 9
- /
/
/
/..
v
/
7//
/
f1 0
'O l*'I
/
f*Cl terme MO 2
I sI o
<00 3.
n
.J
/
/
o
.0 33, C
70 to 10 h
I 400 31 25-31 j
O
.0 si E
t d
34 1.0 0.6 7
30 10 5
f d
Tl 1.2 17-21 g#
o e
i ICNiton O
' van'un h/(
r b
f DETONATOR I
IGNITOR LOCATION O
CiNiin 02 e
TO.
A f f
/
1
, Q BOTTOaA 3
O SIDI I
ga FumNO D AT A'
~3s w,o 7
,d mes l
d d
'O,""
gg ee e f,
y*c 4
8 8
to 32 NYDMOGEN CO*eCENTRATION f%)
Figure 5.
Nature of Pressure Rise Curves Obtained Experimentally (Ref. 6) 12
/
i7I Clegg on the spark ignition of lean mixtures in a vertical tube of 10.16 cm diameter and 26.53 cm height, shown in Figure 6, is in agree-ment with the much older work of Coward and Brinsley(8) on the combus-tion of hydrogen: air mixtures in a 4.5 cm diameter tube.
The results e
of Harrison, et al., Figure 7, show a much higher fraction of hydrogen consumed, for mixtures with initial hydrogen concentrations below 8%,
than the results of similar experiments in large spheres.(5,6)
If we assume the data for spark ignition in.'arge spheres are typi-cal of the fraction of hydrogen that would be consumed for deliberate ignition in reactor containments, then a 31 psig pressure rise corre-sponds to less than 8% initial hydrogen concentration.
Because of the rapid rise in pressure expected as the initial hydrogen concentration goes from 8 to 9%, the uncertainty in the failure pressure of the con-tainment, shown in Table I, has little effect on the maxi =um hydrogen concentration that can be burned without containment failure.
An 8%
hydrogen concentration in containment corresponds to about 45 psig (for complete combustion) and 27% oxidation of the zirconium cladding in the reactor core.
Without the addition of a second mitigation scheme, such as fog-ging, the combustion of hydrogen mixtures of over 8% could result in containment overpressurizing.
If the generation of hydrogen is slow enough, the hydrogen could be slowly consu=ed in multiple burns with interim cooling of the atmosphere by heat transfer and water sprays, preventing a continuous buildup of pressure.
Hence, deliberate igni-tion is a mitigation scheme which is limited in the amount of hydrogen it can control for a single ignition, and limited in the rate of hydro-gen it can control for multiple ignitions with interim cooling.
In su= mary, the completeness of combustion of lean hydrogen mix-tures, below 9% hydrogen (the downward propagatic' limit) but above 4%
hydrogen ( the upward propagation limit), for a gi.n homogeneous mixture can depend on the following:
13
Pressure Transducer Heater CJ l
5 I
$??
l
??
l
.______A_________________
_ _ _. i._._ _ _ _
l __
l 2 LITRE i.__._ Vessel l._.
l._.
_ __o___ c_ ___
i
- Hot Water:
i
_ Tank i __.
i i _-
i j
__.3______________
______t.._
J
[
Ignitor Thermo-g X
Couple i
To Vacuum X
l System l
JUQ l
I I Bellows Pump i
I Figure 6.
The Combustion Apparatus i
(Ref. 7) 14 c
p 7
m w
--y-w
-y,.
- 7y 4aw y
--m
-w--.-,.----y
*---r
-se a
+,ye-e-
,w-mi.-%e
+-ee.
w-
l00 g
'O ov 90 f
/
O 80
/
70 00 O
/ O O
/
e O % STEAM 60 coward et al-F
'O o O % STEAM V 5 -7 % STEAM m
O 7-10% STEAM 50
/
5 O
oO O TO-15 % STEAM o
O I
I
? 40 x
I u.
I I
g 30
/
5 V
I
(
I 4
cc u.
20
/
I I
I l
lO ov D l
3 I
5 h
7 h
h 10 ll 12 0
HYDROGEN CONCENTRATION (%-VOLUME)
Figure 7.
Degree of Combustion in flydrogen: Air: Steam Mixtures (Ref. 7).
Solid circle data are from Ref.
8.
15
9
- 1) Container size, shape, and orientation
- 2) Method of ignition, energy, and duration 3)
Location and number of igniters 4)
Initial turbulence and flow velocities.
We will now attempt to give a possible explanation for the disparate ex-perimental results and present our thinking on the application to hydro-gen combustion in containment.
The work reported by Slifer and Peterson(6) indicates that the frac-tion of hydrogen consumed using strong ignition sources, such as pyro-i fuses, is much h'.gher than the fraction consumed using spark igniters, as shown in Figure 5.
This is not surprising since the pyrofuse is a longer lasting source of larger size, igniting more hydrogen.
The dif-ferences between the results of tube experiments,(7,8) and those con-ducted in large spheres (5,6) can be explained as follows.
The upwardly propagating flames in lean hydrogen: air mixtures ap-pear in the form of bu.ning globules.
Below 6% hydrogen concentration, the flame temperature is of the order of 500-700 K, below the spontane-ous ignition temperature of approximately 800 K.
It has been observed that hydrogen rapidly dif fuses into the flame zone giving locally rich-1 er and, hence, hotter flames and depleting the region between the glob-ules.(9)
Part of the reason for the incompleteness of combustion of hydrogen, then, is the failure to burn the residual hydrogen in the re-gion between the globules.
Pres umably, in tube experiments the glob-ules fill the cross section of the tube.
The results shown in Figure 7 may be typical of the fraction of hydrogen consumed in the region tra-versed by the globules.
^
Upwardly propagating flames are expected to propagate in a roughly conical volume, surrounded by a region of zero combustion, when the in-itial atmosphere is quiescent.
Thie result was found by Sapko, Furno and Kuchta(10) for methane: air: nitrogen near-flammability-limit mixtures,
'16
Figure 8.
The idea of conical-type upward flame propagation was mentioned by Carlson, Knight and Henrie(11) for lean hydrogen flames.
One can visualize the combustion of lean hydrogen: air steam mix-tures as shown schematically in Figure 9 for a single weak igniter and a quiescent atmosphere.
If this model of lean hydrogen combustion is a valid approximation for ignition of a homogeneous mixture in containment, then the pressure rise can be computed given two pieces of informations the completeness of combustion of hydrogen as a function of initial hy-dryjen con entration in the zone of partial combustion, and the cone angle or other quantity specifying the increase in the partial combus-tion zone area versus height from the igniter.
The completeness of com-bustion in the partial combustion zone may be approximated by the data The main uncertainty then would be in the taken in tubes, Figure 7.
rate of growth of the partial combustion zone with height.
The upper compartment in Sequoyah is relatively uncluttered, and the model in Fig-ure 9 may be appropriate.
The lower compartment, however, is extremely cluttered and crowded.
The complex geometry and large number of obsta-cles might tend to increase the degree of combustion completeness.
If the combustion takes place in the upper compartment of the Se-quoyah containment, the steam present in the atmosphere of the lower com-partment presumably would have been condensed in passing through the ice condenser.
However, it is possible that hydrogen may continue to evolve after the ice has melted.
Combustion in the lower chamber may take place The effect of steam in in the presence of a large fraction of steam.
hydrogen combustion is to act as a diluent, more ef fective than nitrogen or excess oxygen in lowering flame temperature because of its higher The effect of steam on the flammability limits (12) is specific heat.
shown in Figure 10.
At low steam concentrations, there is little a
effect on the lower flammability limit of hydrogen.
It takes about 56% steam to inert hydrogen: air: steam mixtures at high hydrogen concentrations.
The presence of up to 15% steam seems to have 17
TIM E, msec 900 720
- g. -
-q l
1 I
\\
'op Q3f l
\\
p g
f 540
\\
/
F-q g
/
I
\\
\\
%d I
\\_s 3
360 f
s_
f%
/
\\
'Q35 s'
\\s.
Sparke em O 2,0 4,0 a
i i
in O 10 20 SCALE Figure 8.
Side View of Flame Profile History for Central Ignition of a 6.9% CH -65.8% Air-27.3% N2 Mixture in a 12-Foot-4 Diameter Sphere (Uncorrected for Parallax).
These data are from Ref. 10.
18
l ZONE OF INCOMPLETE BURNING NO BURN NO BURN I
IGNITION SOURCE l
I l
1 Figure 19.
Possible Model for Lean Ilydrogen Combustion by Weak
- y Ignition in Large Chambers with Quiescent Atmosphere 19
100". Air 10". Iteaction ('ontainer Ambient Temperature 216* V
- o 10*'. Itcuction (bntainer S
~
Ambient 280* F
/
100". Itcaction (%nt.ainer 100". Itcaction Container Ambient 216* F
. Ambient 250*F "o
8
/
- s e,+
/
I e
Assumed k
1)ctyna(lon,1,imits fr/i V
\\\\
- o
/\\
o4!
Flammability Limits e
/
k
/
D a,
+o i
I 100".
H0 60 40 20 100".
tum 113 go,.rrent II.j Flamto:.ble 1.t:uit.<
73* F - 0 l'Sl(i C>- - - - - 300
- F - O l'Sl(i 300* l' - 100 l'SK; i
~
i l
Figure 10.
Flammability and Detonability Limits of Hydrogen: Air Steam tiixtures (Ref. 12).
i l
x 20
little effect on the completeness of hydrogen consumption for lean mixtures.(7)
IGNITION Experiments have been performed to determine the minimum ignition energy (13,14 ) required to ignite hydrogen: air mixtures.
The results
/ are shown in Figure 11.
For near stoichiometric mixtures at atmos-pheric pressure, energies of less than one-tenth of a millijoule are required, but the energy required increases rapidly as the flammability limits are approached.
At approximately 7% hydrogen, an energy of 0.6 millijoules is required. Drell and Belles (13) caution that these energies are minimums attained in laboratory settings and that gas motion, turbulence, and spark duration may increase the required energy.
There is a lack of' experimental information on the minimum spark energy required for ignition very near the flammability limit.
The gap between the electrodes of a spark plug used for lean hydrogen combustion must be larger than that used in automotive spark plugs.
The gap must be larger than the quench distance, the distance to cool down an incipient flame.
The quench distance for lean hydrogen mixtures increases rapidly as the flammability limit is approached in a fashion similar to the minimum required spark energy.
For a 7% hydrogen mixture the quench distance is nearly 0.3 cm.
The failure by Carlson, Knight, and Henrie to obtain ignition in mixtures usually considered flammable using automotive spark plugs may be due to quenching by the electrodes.(ll) l l
Glowplugs were selected for ignition in the Sequoyah containment because of concern about possible radio frequency interference caused by sparks.
There is a smaller body of literature on the ignition of combustible mixtures by hot surfaces than for spark ignition.
The tem-perature of the glowplug required to ignite the mixture will be near the spontaneous ignition temperature.
The spontaneous ignition temper-ature data of Shapiro and Moffette(12) are shown in Figure 12.
There 21
6 i
i 4
2 gw
.J D
PRESSURE, P,
,o 1
3
.8 atm
'2 2
.6
.'y
.4
.3 a:wz*
.2 9h
.5 z
.1 e~2 08 3
2 06 E
- 04 2
1
.02 0
t i
i t
i O
10 20 30 40 50 60 70 HYDROGEN IN AIR, PERCENT BY VOLUME Figure 11.
Spark Ignition Energies for Hydrogen: Air Mixtures (Data are from Refs. 13 and 14) 22
. ~ _
i 590 i
580 b
g o
30% H O y
2 o 570 D
cc i A w
1 o.
560
'(
O 2
)
20% H O 2
2 550 9t-z o 540 10% H O 2
m D
8 530 Z<
Z 520 0%HO 2
O o.
m 510 500 O
10 20 30 40 50 60 PERCENT H2 s
Figure 12.
Minimum Spontaneous Ignition Temperatures of Hydrogen:
Air: Steam Mixtures at 100 psig (Ref. 12) i 23
are a lack of data near the flammability limit.
It would appear from extrapolating the curves given that temperatures of the order of.600*C are required.
SEQUOYAH INTERIM DISTRIBUTED IGNITION SYSTEM (IDIS)
TVA has proposed a distributed ignition system (IDIS) for hydrogen control in Sequoyah.
Thirty-two glowplugs have been installed in light-ing fixtures throughout the containment. TVA plans to have the operator activate all the igniters simultaneously, based on some accident signal (e.g., ECCS activation).. We will refer to this IDIS system as " automat-ic" deliberate ignition (DI), in contrast to an operator-controlled de-liberate ignition scheme which would permit each igniter to be activated separately by operator action.
These two philosophies of deliberate ig-nition are discusse'd in more detail in the question-answer section fol-lowing this section. We'will assume for this discussion, however, that TVA is constrained to use the " automatic" DI system in the near term, and the following remarks concern that system Basic questions dealing with DI include.he ef ficacy and reliabil-ity of the igniters under accident conditions.
These issues are pres-ently being addressed experimentally in a program at Lawrence Livermore National Laboratory.
In this investigation, we assumed that the igniters would work; i.e., be capable of igniting mixtures of hydrogen in the range of 5-10% in the presence of large volumetric fractions of steam, and not be adversely affected by sprays or other environmental factors.
Deliberate ignition will be beneficial if the hydrogen can be con-sumed slowly enough so that neither containment integrity nor equipment survival is jeopardized.* This can be accomplished in several ways:
burn lean mixtures; burn in the lower compartment; burn slowly enough to
- We have tacI:1y assumed that DI would not lead to the combustion of other materials (e.g., flammable insulation).
TVA must insure that such combustible materials do not become secondary sources of deflagra-tions.
24
prevent pressure and temperature buildup.
Lean mixture combustion has already been discussed extensively.
Burning in the lower compartment will generate pressures much lower than would occur for an adiabatic, isochoric combustion.
Pressure reduction results from the expansion of gases through the ice condenser into the larger upper compartment, and from the condensation of steam and cooling of the other gases as they pass through the ice.
Deliberate ignition in the lower compartment could only be harmful if it led to detonations, pseudo-detonations, or, possi-bly, if all the ice has been melted.
Extremely high concentrations are unlikely to develop in the lower compartment for most accident scenarios, except perhaps TMLB (loss of main feedwater and all power).
TVA has presently allocated 25 glowplug igniters to the lower com-par tme n t.
(We have enccuntered conflicting statements concerning the number and location of the igniters; we believe this is due to revisions reflecting TVA's continuing evaluation.
However, it is possible that the following data no longer reflect the present situation.)
The number and location of these igniters are shown in Table III.
Table III Glowplug Locations in Sequoyah Lower Compartment Entrance of Ice Condensers 5
Outer Eqavment Annular Space 9
" Raceway" Annular Space 4
Main Subcompartment (including primary pumps 7
and steam generators)
The locations are indicated in Figures 13 and 14.
This distribution cov-ers all subcompartments except the small central compartment above the reactor vessel and below the control rod drive essembly missile shield.
Because of the possibility of a local detonation in this compartment, we are, at present, unable to determine whether or not an igni.ter should be 25
~
e i
- (5) 818.o f t.
792.0 ft.
C4) i i
i n
n I
~
\\
U 7 31.o n..
i
- (5[
[7)
Ait RcruRn
_- FAR 700.0ft.
- (8) i :
l
' ~ ~ *l*
6 89.0f4.
_b I
O m3
\\
.J 6 69.'o fi
\\
Figure 13 Number and Elevation of Igniters for Sequoyah IDIS 26
270' j
AIR ErfURN TAN
.(
g,_ _- D
~D DC C
50 4
(
f so-3 I
y_ //.
_ 4..
Y~
~~~'
l g.;
g y#fja Te,g e'
a E
Elev.
LOC.
E 731 Ice condan**r A = ACCufvuTOR
- 0.3 Ou E
s 699
" Raceway *
-SG =' stb)4 GDUATOR g
Figure 14.
Cross-sectional view of Sequoyah lower compartment indicating igniter locations 27
installed in this subcompartment.
We intend to study this question fur-ther.
Otherwise, for the interim system, we believe that the present
~
scheme provides adequate coverage in the lower compartment.
The question of igniter redundancy can be more profitably addressed af ter the results of the LLNL tests become available.
Burns in the upper compartment (UC) can occur for two reasons:
the entire containment has been overwhelmed by an extremely rapid and large release of hydrogen from the primary system (a very unlikely scenario),
or the lower compartment (LC) has been inerted by excess steam and/or deficient oxygen.
Since DI will result in repetitive deflagrations in the lower compartment, fresh oxygen needs to be supplied from the upper compar tment.
This oxygen supply can be impeded if the LC burns occur more frequently tha'n the time required to resupply the oxygen (order of minutes) or if the f ans are partially or completely inoperative.
Steau inerting can occur whenever the steam fraction in the LC exceeds about 30-56%, depending on hydrogen concentration.
This can occur either be-cause of large steam releases accompanying the hydrogen, increased steam concentrations from previous burns, or failure of the air-return fans.
Conditions for deliberate ignition in the upper compartment are very different than in the lower compartment.
The pressure rise due to com-bustion can approach that for an adiabatic, constant-volume combustion.
As a result, mole fractions much above 84 hydrogen should not be burned in the UC, based on a "f ailure" pressure estimate of 45 psig.
The volume of the lower compartment is only about 43% of the upper compar tment.
It is doubtful if significant flow can occur from the upper to the lower compartment in time to diminish the combustion pressure rise.
The doors to the ice condenser normally permit flow only one way, from lower to upper compartments.
There are two air-return fans of about 40,000 CFM capacity.
In an upper compartment deflagration, the higher pressure in the upper compartment would presumably cause an increase in 28
. - _ _ _ _. ~ _ _ _ - - -
this flow rate.
Nevertheless, the flow rate through the fans is too low to significantly reduce the upper compartment pressure rise.
There are seven glowplug igniters in the upper compartments four at the exit of the ice condenser and three high up in the dome near the
~
water spray nozzles.
The action of the water sprays in reducing the upper compartment gas temperature may permit the use of repeated burns in the upper compartment provided each burn consumes less than a given amount of hydrogen (less than 8% hydrogen), the exact amount depending on the ini-tial pressure, and provided there is sufficient time between burns for the water sprays to cool the gases (typically several minutes).
The concentration of hydrogen in the upper compartment will be a maximum at the exit of the ice condensets.
It is possible for a steam-inerted mixture entering the ice condenser to become detonable at the exit.
The mole fraction of hydrogen at the exit, XH2ou t' the mole fraction in, XH2in' densed in the ice condenser, XH 0, by 2
H2in XH out "
1-X 2
HO 2
For example, consider a modest hydrogen mole fraction of 9% inerted by a 60% steam fraction in the lower compartment.
If all the steam is con-densed in passing through the ice condensers, as expected, the hydrogen concentration at the ice condenser exit will be 22.5%.
This concentra-tion is extremely detonable, and would form a 300* partial torus adja-cent to the containment wall.
Under such conditions, the rationale for locating four igniters in this region is very questionable.
We would e
strongly recommend that TVA remove these igniters for the IDIS.
In our early analysis of deliberate ignition, we were inclined to place igniters near the top of a volume to benefit from the dif ference in upward and downward flammability limits.
This strategy also appears desirable for Sequoyah upper compartment DI.
Igniters placed near the 29
top would tend to generate lean mixture deflagrations which would not propagate downward, and which would only involve a fraction of the hy-drogen contained in the UC.
Such deflagrations could gradually remove the hydrogen, while still allowing the normal heat removal mechanisms (sprays) to limit pressure and temperature rises.
LOCAL DETONATIONS With 32 igniters distributed throughout containment, the possibility of developing uniform hydrogen concentrations in excess of 18% (the det-onability limit) seems very remote, if not impossible.
It is also of little interest in terms of reactor safety for Sequoyab, since a uniform 18% deflagration, by itself, has a very high probability of failing containment.
Local hydrogen accumulations, however, could conceivably reach detonable concentrations.
The previous section indicated that the plenum at the top of the ice condensers is such a region.
Another sus-ceptible region might be the subcompartment above the reactor vessel I
head, Figure 1.
A local detonation will produce a shock wave propagating outward j
from the detonating region and decaying with distance.
A rough estimate can be made of the distance within which the shock wave can be damaging j
by using the data for high explosives. (15,16)
For a given energy re-lease of explosive, E,
the ratio of overpressure to ambient pressure, (p - po)/po, is usually given as a decreasing function of the dimension-less distance from the center of the explosion, R/Ro, where R is the distance f rom the center of the explosion and Ro is a scale distance given by i
(
Bo = (E/po)l/3,
l t
The energy released in a hydrogen detonation can be considered as the product of the molar energy of reaction and the number of moles of hy-drogen consumed.
i 30
. ~..
X po H E= (4sr3/3) 2 o
.WTo where r is the radius of the detonable cloud, XH is the mole fraction 2
of hydrogen in the cloud, JPis the universal gas constant, To the ini-tial gas temperature, and Q the energy release per mole of hydrogen con-sumed, 2.4 x 105 kj/kmole.
For an initial temperature of 300 K and a hydrogen mole fraction of 0.28, we have Ro = 4.8 r.
The blast overpressure is a weak function of explosive size, speed of energy release and other parameters.
However, we can take the following values for rough approximation.(15) i 3
R/Sg (p - po)/po Impulse /r, n-s/m 0.5 2.0 139 1.0 0.5 69 2.0 0.2 42 i
Normal reflection from a rigid wall will increase the table values of peak pressure and impulse by a f actor of about 2.3.
If.we consider a i
l peak (unreflected) pressure of 30 psig (2.0 po) as nondamaging, then the shock wave pressure will decay.to acceptable values in approximately i
l 2.5 cloud radii.
A more complete analysis would have to consider the effect of impulsive load on walls, reflections, finite detonable cloud l
size, etc.
Nevertheless, the above analysis indicates that the detona-6 tion of small detonable clouds-may not damage containment, if they occur i
(
sufficiently far from the walls.
The effects of larger detonating clouds would have to be examined more carefully.
l l
i 31 i
4 A detonation could also threaten containment by the generation of missiles from explosively-produced debris.
Referring to Figure 1, the control rod drive missile shield (or pieces thereof) might become such a missile if a detonation occurred in the subcompartment below it.
We performed a 2-dimensional axially symmetric calculation of a detonation
- hin this compartment using the CSQ hydrodynamic computer code. (17)
The calculational geometry is shown in Figure 15, with the reactor head below and the missile shield above.
Other initial conditions included a hydrogen mole fraction of 20%, and initial temperature and pressure of 375 K and 1.66 atmospheres, respectively.
The ignition source for CSQ can be arbitrarily specified, but must be axially symmetric.
- Here, we assumed that ignition occurred in a ring around the upper part of the compartment (resulting, perhaps, from an external deflagration). We in-tend to repeat the calculation with a point source, but do not anticipate significant dif ferences in resultant pressure histories.
Figure 15 also illustrates the detonation waves (pressure contours) proceeding inward and downward from the ignition source region at 1.4 ms.
The pressure history at top center for the full 50 ms calculation is shown in Figure 16.
The highest pressure of about 49 bars occurred about 2.5 ms after ignition.
Pressures continue to oscillate due to i
multiple internal reflections of the shock waves.
A very simple analysis of the missile shield was performed.
The complex pressure behavior was approximated by a constant 10 bar pres-i sure pulse lasting 50 ms.
The impulse delivered to an intact missile i
shield was insufficient to make it strike the containment dome.
A more l
complex structural analysis, which would also treat the possibility of 1
spallation and fragmentation, is beyond the scope of this short term effort.
l 32
JL
\\
A a
in lp 3.6 m "q
__ L REACTOR III;AD a
ca
_l '___
~
W h.h a I
t 8.0 m, O
Figure 15.
Subcompartment Detonation Calculation with CSQ 33
50.00 i
i i
i e
i i
<a o
E 45.00 -
40.00 35.00 I
30.00 N
- r M
25.00 mwz>
3 20.00 w
E 3
15.00 wa Q.
.0.00 5.000 0.000 0.00 10.0 20.0 30.0 40.0 50.0 cm
~
TIME (SEC) x10 1
SEQUOYAH TOP RPV COMPARTMENT (0.2 H2. DRY)
Figure 16.
Pressure IIistory at Top of Compartment 34
ACCIDENT ANALYSES One method of assessing the performance of the Sequoyah interim dis-tributed ignition system would be to perform accident calculations with and without the IDIS, and compare the threats to containment survival.
Two computer codes have been used to perform such calculations:
the Westinghouse CLASIX code, and MARCH from Battelle Columbus.
Both codes are unverified, and are essentially still under development.
Results of CLASIX calculations were reported in Ref. 1.
For a small break LOCA, SD (see WASH-1400 for symbol definitions and usage), they showed that 2
multiple burns would occur in the lower compartment resulting in pressure rises of only a few psi per burn.
The burns were separated generally by two or more minutes.
Containment failure was calculated to occur only if heat removal in the ice condenser was artificially reduced, or if both fans were inoperative.
The results were also relatively insensitive to the ignition criterion employed.
MARCH S D calculations were qualitatively similar to the CLASIX re-2 s ui ts. (1,3)
Peak pressures were generally below anticipated failure conditions prior to the core slumping into the lower plenum.
Calcula-tions carried beyond core slump always predicted containment failure.
Sandia has obtained a copy of the MARCH code and the S D~ input deck 2
from Battelle (BCL).
With their assistance, we have set up and run many MARCH calculations investigating both small and large break LOCAs.
We will briefly review some of our results for this report.
More extensive documentation will be provided in subsequent Sandia quarterly and topical reports.
Figures 17-21 illustrate the threat to Sequoyah from hydrogen gen-(S D).
Hydrogen and steam are released eration during a small break LOCA 2
from the primary system into containment.
The hydrogen is assumed to ac-
=
cumulate without burning.
Hydrogen production begins approximately 61 minutes after the break (Figure 17).
About 800 lbs of hydrogen are gen-erated prior to core slumping, which occurs at 93 minutes.
Hydrogen is 35
- - _ _ - ~. _
w*
nt.50 n
+o 2
e-4 i
X 4-2.25
,i G 2.00 B
i-m 1,75 m
r l
1.50 z
W e
@ 1.25 i
i o
i i
c 1
1.00
_J i
g o.
750 2
1 H
1 J
.500 1
.250 3*
.l I
0.00 l
0.00.
20.0 40.0 60.0 80.0 100.
120~.
i TIME (MIN) j Figure 17.
Total H2 Generation for Small' Break IDCA (S D) - No Burn-2 t
4 i
e e.
0 02 2
D 0
e02 0
0 a0 1
nru 0
B T
0 e0 o
S N
1 E
T s0 0
)
D 41 )E 2S D
T
(
2
- 0. U C
0N L
S s2 I
1 n
M i
(
n H
0 -
o 0
i s0 t
A E
c 1
Y M
ra F
O
- 0. I T
s0 e
U 0
lo Q
M E
2 0
H S
a0 0
8 1
0 e
s0 r
4 u
g i
F 0
s02 0
0 O
M 6
4 0
L 0
0 0
0 dZ EZN x<C2OO Z.- QZ O. ZO
<%. NAO L
1 C
r 2
.l i!
I:i l
.ij'
- 14 4
- 1
<ll
0 022 0
,002 0
0
,0 1
nru B
0 T
,0 o
0 N
1 S
E
)
0 D
T 0
2
,4 )
S 1 E
(
D T
C 2
- 0. U U
0N S
,2 n
I 1
i M
n
(
o 0 -
t i
H 0
,0 c
A 1 E a
r r
Y M
F O
- 0. I e
T l
,0 U
0 oM Q
12 E
0 1
,0 S
6 9
1 e
0 r
,0 u
4 g
i F
0
,0 2
0 0
o 8
0 4
2 0
t 0
0 0
0 0
"gH E
<0^OU Z CZ go ZO o4Zk O2 Q
I 4
l:
ll'.
4 i
l4
t
~
2
~h c
o
-h
~
w M
H 8
n m
N h
E*
O N
eD a
M
.5 3
8 Z
_@ i i
ca 2
>4
- E o
ep
.s
-8 e
D O'
s 2
N e
w
-g e
5 o
-4 g
a
-5 m*
s s
s o
e o
y q
o 4
d d
d o
d I 'ON IN3NIHVdM00 NI MV31S 30 NOIIOVH3 3' ION 39
6
~
u
.g
=
1v
=
N
-B nr jt e
o d
-g "3=t 2
jt 4
n oe C
e m
r -92
=
E i
e O
eI
.c Z
-5 k
4>
K o
on X
-s -
o a
jrt 2
-B W
~
5
_g 2
i
-B L
I 6
,O 5
a' a
E g
g 2.ON TNEMTRAPMOC NI MAETS FO NOITCARF ELOM e
tO a
,.A.~.a h.
e.
-s..,--.4
_-.*k A.-
r 4.-
aes e -wse 4
.*,_m p
generated extremely rapidly following core slump, but for this interim study, significant core melt will not be considered.* The average: rate of hydrogen production during the " degraded core" fraction of the acci-dent is about 25 lbs/ min.
Figures 18 and 19 show the mole fractions of hydrogen in the lower and upper compartments, respectively.
We see that detonable concentrations are reached in the lower compartment, and con-centrations of about 9% in the upper compartment, prior to core slump.
Figure 20 illustrates the very large steam fractions reached in the LC; Figure 21 shows that the ice condensers have been very effective in re-ducing steam concentrations in the UC.
Figures 22-26 illustrate a similar no-burn calculation for a large break LOCA (AB).
The hydrogen release begins much earlier in the acci-dent, 2.5 min (Figure 22).
Hydrogen is generated for about 20 min until core slump.
Seven hundred pounds of hydrogen are released during this period, with an average generation rate of 35 lbs/ min.
As with the.small break, detonable concentrations of hydrogen develop in the LC for the degraded core portion of the accident.
Hydrogen concentrations in the UC, however, are lower than for S D (Figures 23 and 24).
The efficacy 2
of the ice condensers in reducing steam fractions is again evident in Figures 25 and 26.
Many calculations were performed where the hydrogen was burned when its concentration exceeded. given number, for example 8 or 10%.
Fig-ures 27-32 illustrate such a calculation for an ignition concentration of 10%.
Multiple burns occurred in the LC, spaced approximately 5 min-utes or less apart, Figure 27.
Pressure rises were very small, Figure 31.
No burn occurred in the UC, since the concentration there never reached the 10% ignition point (prior to core slump, of course).
Figure 20 shows that the steam concentrations frequently reached concentrations high enough to inert (approximately 30 to 56%, depending on hydrogen
- We have reason to believe that MARCH may be overly conservative in the post-slump period, both for numerical reasons and because of physi-cal assumptions.
i 41
4 0
m 2.50 i
+o X 2.25 -
c-t i
C; 2.00 -
m M 1.75' m
i i
r 1.50 z
i Lao 1.25
. o I
l 1'.00
_J Ho.750 H
i
.500
.250 4
i 0.00O.00 5.00 10'.0 15.0 20~.0 25.0 30.0 35.0 40.0 i
-TIME
-(MIN).
Figure 22.
'Ibtal th. Generation for Large Break 'IDCA (AB)
.J t
0 02 2
0
,002 0
0
,0 1
nr 0
u T
,0 B
0 S
o 1
E N
T 0
,0
)
4 B
1 )E BA T
(
- 0. U C
A 0N L
,2 I
1 n
M i
(
n 0 -
o H
0 i
,0 t
A E
c 1
a Y
M r
I P
O
- 0. T
,0 e
U 0
loM Q
2 E
0 H
,0 S
0 3
2 0
e
,0 r
4 ug i
F 0
,0 2
0 0
0 8
0 4
2 0
t 0
0 0
0 0
dZ &ZdNm4rMOO Z OX O ZOeL*0CNN E 3
L 4
C O
I i;'
.i lll
'.',i, l!
, i a
1
0 022 0
0
,0 2
0 0
,8 1
nru 0
B T
0
,0 1
o S
N E
T 0
)
,04 B
1 )E A
B T
(
C
- 0. U U
0N A
,2 n
I 1
i M
(
no 0
H i
0 t
,0 c
1 A
E ar Y
M F
I O
- 0. T e
,0 l
0 o
U M
Q 2
1 E
0 1
,0 S
0 4
2 e
0 r
I
,0 u
4 g
i F
(
0
,0 2
0 0
o 0
6 A
2 0
t 0
0 D
0 0
m g E-sg4 OU Zm QZ gO ZoEo<%N o;3 C
I l
5 iIi<
1
,!il:l
'j' i!i1i,l l
l l
l
0 0
22 0
,002 0
0
,0 1
0 T
,00 n
S 1
ru E
B T
0 o
,0 N
41 )E B
T C
- 0. U L
0N n
A
,2 I
i 1 M n
(
o i
0 -
t H
0
,0 c
1 a
A E
r Y
M F
I m
O
- 0. T a
,0 0
e U
t S
QE 0
,0 5
S 0
2 eru 0
g
,0 i
4 F
0
,0 2
0
.i,p 0
o 0
0 4
2 0
t 0
0 0
0 0
O z E~ (j M r.
AMOO g MC NO W<O~HO4 k
[211- @
aU 1
1 i.
ljI ji*
ij.
l-i?i' l
D e
~
D C
_s o
9 R
-8 E
w 0=L E
8 8
_T D
N 5
l C
~
Z h
o 9
C 3
i Z
-8 5
'a d>
b
,E a-t o
8a C.
e O
rI 0:t 6
M
_g o
2
- 8' t
5 e
O
-B l
D 1
,O o
e o
A N
6 e
P P
P o
O R.ON TNEMTRAPMOC NI MAETS FO NOITCARF ELOM t9
-._._-e m
.am.a4.
.4 a
-.a aa 4.aha
-.-a w-.a,.m_.___
-,a_-
.-w
i i
SEQUOYAH S2D TEST d
1.0 Z
E-ZN h
0.8 -
4
%2 t
oO 0.6 -
Z
~
4 N
l Z
O.4 -
O ZO 02 -
M t
i
/
N
, - -[-
4 0.0 O
0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 100.0 180.0 200.02 220.0 TIME - (MINUTE) i Figure 27.
H2 Mole Fraction in LC (S2D) - 10% Burn D
SEQUOYAH S2D TEST m
Y 1"
i 1
E*
b i
2 i
H 0E -
h O
l C
0.6 -
z.
N 1
X I
N 0.4 -
o Z
O 3
{
b02-4 i
A o
0.0 e
i i
i i
t y
0.0 20.0 40.0 60.0 80.0 100.0 12.0 140.0 100.0 180.0 200.0 - 220.0 I
TIME - (MINUTE) l Figure 28.
H2 Mole Fraction in UC (S D) - 10% Burn 2
l i
l l
l 1
i
?.
022 0
,0 x.
02 0
0
,0 n
1 ru B
0 T
0
,0 0
1 1
S ET 0
)
0 D
,4 )
2 1
S E
D T
(
C 2
- 0. U L
S 0N
,2 n
I 1
i M
(
no H
0 -
i 0
t
,0 c
A E
a 1
Y M
Fr O
,\\
- 0. I T
e
,0 l
U 0
o M
Q
~
m E
0 ea S
,0 t
6 S
0 9
2
,0 4
e t
~
ug 0
i
,0 F
2 0
0 o
8 6
4 2
0 t
0 D
a 0
0 M ez H mE a%oO l2-N<hm%oZ3sU4E M4$
l i
!;l
-I; i
I iI jli i!
- i!
I.!
l b
~
6 w
.C
~$
R
~
~
w i
- 1. t
=
D 3
.M d
O b
O C
w
.N ~
G z
m Z
E p
g q
m E
2 o
ed
-8 2
C 2
O 1IL 3
g DC
-8 6
8
.g 1r c
o
-8 L
I 6g f
f f
I o
6 e
A M
O a
p 9
o P
2.ON TNEMTRAPMOC NI MAETS FO NCITCARF ELOM l
l 50
0 0
22 0
,002 0
0
,8 1
0 0
T
,6 1
S nr E
u 0
B T
,04 1 )E 0
1 D
T 2
- 0. U 0N
)
S
,2 D
I 1
2 M
S
(
(
v 0
H C
0 L
,0 A
1 Y
E n
J i
A M
L I
e O
- 0. T r
J
,0 u
U 8
ss Q
er E
0 P
,0 S
6 1
3 0
e
,0 r
4 u
g i
F 0
,0 2
g 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 4
2 0
8 6
4 2
1 1
1 dmm &
BAD 40> E2M Z.o z~ a%DD t
C C
s ij;'
i i.1:
4 l!
l
0 022 0
0 d 2 0
0 d i 0
T 0
d n
1 S
ru E
B T
0 0
0 4) 1 1
E D
T 2
- 0. U
)
S ON D
N 2
I 1
S M
(
(
C 0
U H
0 d
n A
E 1
i Y
U M
e J
r I
O A
- 0. T u
b s
U a
se Q
r P
E G.
S 0
6 2
3 er 0
u 0
g
'4 i
F 0
02 g-0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
4 2
0 8
6 4
2 1
1 1
$AaO m2 PAO > gm5Z B z8 z. gg
$A 0
B I!
concentration, initial temperature and pressure; see Figure 10).
The highest steam concentrations, however,-tended to occur both before and after the major hydrogen production period for degraded cores.
The capability to recognize steam inerting and prevent burning under those conditions was programmed by Sandia and incorporated into the MARCH log-ic.
For these calculations, 56% was selected (input number) as the steam inerting concentration.
Referring to Figure 10 and the discussion on IDIS, we can conclude that this selection was not necessarily conserva-The S D calculation with 10% burn indicated that steam inerting tive.
2 did not occur.
This conclusion needs to be regarded with skepticism, since it requires a significant precision in the MARCH calculations of hydrogen and steam inventories as functi,ons of time - a precision which MARCH may not possess.
Furthermore, the calculation did not employ the lower steam-inerting concentrations that would prevail at high ambient temperatures and low hydrogen concentrations.
Figures 33-38 illustrate the AB accident behavior under the assump-tion of burning at 10%.
Again, no significant pressure rises occur due to combustion in the LC prior to core slump.
One final calculation will be reported here; i.e.,
a small break LOCA with both f ans inoperative, and 10% burn criterion.
Figure 39 shows the hydrogen concentration in the lower compartment.
Note that it has reached nearly 30% prior to core slump; i.e.,
although burns were permitted at 10%, they did not occur.
The reason for this is evi-dent in Figure 41 - for this calculation, the LC was stt am inerted (con-centration in excess of 56%).
This LC inertino nar *.ttei the concentra-tion in the UC to steadily increase, as shown in Figure 40.
Because of the ice condensers, the UC still had a relatively low steam fraction (Figure 42).
In summary, our MARCH calculations qualitatively support the previous work performed by BCL and Westinghouse-TVA.
The 17)ortant conclusion is that deliberate ignition in the lower compartment appears 53
e
~
~
1 m
_g C
Q 3
E-
-8
=
~
w g
N
~
E-
_a i
w O
-a 5 3
- s C
e4 y
O C
-h
.e
~N t
3 0
a
_e o
)
GT r
N q
r w
-8 A
q
~O 8
?
-r4 O
l
\\
~
N\\
qo l
d I 'ON IN3NIHVdN00 NI 2H do NOIl0VH3 3' ION 54
l l
SEQUOYAH A-B TEST m
g t0 4
b 4
i x
i E-0.8 -
o, U
0.0-Z w
CQZ 0.4 -
g O
Z o
l E
O 02 -
b l
0.0 -
0y 0.0 20.0 40.0 00.0 80.0 100.0 12 0.0 140.0 100.0 100.0 200.0 220.0 TIME - (MINUTE)
Figure 34.
H2 Mole Fraction in UC (AB) - 10% Burn
0 0
22 0
0 d 2 0o d i nru 0
B T
0 1
0 S
1 E
T 0
0
)
'4)
B 1 E A
B T
(
- 0. U CL 0N A
d I
n 1 M i
(
no 0 -
H i
0
,0 t
1 c
A E
a Y
M r
F I
O
- 0. T e
,0 l
8 U
o M
Q m
E 0
a
,0 e
S 0
t S
0 5
,0 3
g 4
erug 0
i
,0 F
2 0
0 0
2 0
4 2
n i
0 0
0 0
o oZ s msQa%oO Z.
sg %O 2sO<
w4 2 i-i<
4!
- i l:
o 1
o-
~
6
-8 o
E m
o R
-g 5
w 8
6
-8 s
-s g
w I
8
-c
=
Z
~!I
-[
g
$p ot 1
W E
o ed a
-8 2
C 2
D N
o 2
.e o
y O
b On i
o
-s o
b b
b M
o y
o o
P O
P
~
2.ON TNEMTRAPMOC NI MAETS FO NOITCARF ELOM i
l l
SL S
,A m.__,
4 m
4...
e 4
m,,
.4
.w m
a
..e.Am.-_
A m_-. - - -
._--.a=--
m-m a_
m_
e
_m.
m
.Am
o e
M o
-8 CJ qoe o
E-
- se a
u cn E
E-q o
2
-n l
e b
2 i
C
(
_g Z a--
i 6
S Z
i
'$I
.5
_* r=a
>4 2
s
=
O m.
-o D
e Cf M
q g
CO
-8 m
I p
3 o
w l
.v w
l O
'N 4
t
,o e
o e
o e
q q
q l
?
8 8
8 8
8 VISd 'I 'ON Eitn'iOA INZNNIVINO3 NI EERSSIEd l
l l
58 1
o b
9
~$
qo
~S q
H
-8 e
w s
~
N&
8 8
m CQ E-I l
cD E
_6 Z
~ 3 o
v a
o Z
'_ g l e
m 8-2 8
7 m
O
-8~"
a u
O' M
8 d
w
=
L 3
o w
I 6
A y
a O'h
^
o.o e
o o
q q
q q
q N
N 8
8 O
O VISd '8 'ON 2KfrICA INEiNIVINO3 NI SMRSSIMd 59
0 022 0
0 i02 0
s 0
n i81 a
F o
0 N
0 T
i8 1
nr S
u E
B 0
T i0 4
0 1 )E 1
D T
U 2
- 0. N
)
0 D
S
.i 2
_=_
I 2
1 S
M
(
(
C 0
L H
0 i0 n
1 A
E i
M Y
n I
o O
- 0. T i
i0 t
8 c
U ar Q
F E
0 e
i0 l
S 0
oM 2
1 0
1 i04 9
3 0
e i0 r
2 ug i
F 0
-i0 0
8 0
4 2
0 1
0 0
0 0
0 dZ EZW%9M<OMoO Zm NN No ZO f<%r N43'"
4 r
l I'
1; i'
l I
j'
0 022 0
0 d 2 s
0 n
0 a
1 F
o N
0 0
i0 n
T 1
ru S
B E
0 0
0 T
i4) 1 s
1 E T
D U
)
2
- 0. N D
0 2
i2 S
I S
1
~
M
(
(
C U
0 -
H 0
i0 n
i 1 E A
M n
Y o
I i
0 O
i 0. T tc 8
a U
r F
QE 0
e l
,0 0
o S
M 2
1 1
0
,0 4
0 4
e 0
r 0
u
,2 g
i F
0 i0 0
8 0
4 2
0 1
0 0
0 0
0 s8O Z* CN go ZoEo4M% boy Q
m
,z E
x I,
l 4
l!
,i!
d
> i
0 022 0
0
,0 2
s 0
n 0
a
,0 F
1 o
N 0
T 0
n
,0 r
1 S
u B
E T
0 0
0 1
,4 )
1 E D
T
)
2
- 0. U D
S 0N 2
,2 S
I 1
(
M C
(
L H
0 -
n 0
,0 i
A E
1 n
Y M
o i
I O
- 0. T t
,0 c
U 8
ar Q
F E
e 0
l S
,0 o
0 M
mae 0
t
,0 S
4 1
4 0
,0 e
2 r
ug i
0 F
,0 0
8 0
A 2
0 1
0 0
O 0
0 fO Z3EO4k T r$
l1 e
E-
%E-a % O O Z
% C 1 ff
<P j
r z
e
1 0
022 0
002 sn 0
a 0
F
'01 o
N 0
n 0
r i0 T
u 1
B S
E 0
0 1
0 T
e4) 1 ET
)
D D
- 0. U 2
2 0N S
i2 S
(
I 1
O M
C
(
U
~ 0 -
n H-0 i
Y0
~ 1 E n
A-o M
Y i
I t
- 0. T c
O--
s0 a
3 r
U-F Q~ -
e l
E 0
o T0 M
0 S-
~
m
~
a e
t
- 0 S
r0
- 1 2
4 o
iO e
I
- L r
u g
i F
0
_ - O i-O R
6 A
2 0
L O
0 O
0 0
Eh [t O Z O YA U'<$ <
94#
j N CZ EY,H,.EY.;,%oO y,e M<1 i
P I' i
4 p.
L nW c
to be beneficial for large and small break LOCAs for degraded core cccidents.
We should note, at this point, that the agreement is qualitative.
We believe that all accident calculations to date, including ours, have been performed under the constraints of short deadlines.
For example, we have not performed time-step sensitivity studies, although we believe that the results may indeed be sensitive to time-step sizes.
We would also like to note some important discrepancies which we uncovered during this work.
From various sources, we have seen zirconium inventories quoted for Sequoyah of 41,993, 43,000 and 50,913 lbs.
Such differences lead to large differences in anticipated hydrogen generation.
These dif ferences may be due to heated and unheated sections, but no reference has ever stated this.
They may also be due to 17x17 bundles replacing 15x15.
We are also not convinced that the Sequoyah FSAR provides the most reliable information, since recent TVA reports sometimes disagree with the FSAR.
Containment volumes also vary significantly depending on source.
PSARs, FS ARs, I&E documents and various references have provided us with the following discrepant information concerning volumes:
upper compartment including ice condenser:
897900, 855000, 808500 cu. ft.
ice condenser volume:
181880, 157500, 125000 cu. ft.
lower compartment:
387700, 383000, 289000, 287000 cu. ft.
total containment:
1.286, 1.284, 1.192 million cu. ft.
Part of the discrepancy may be due to the ambiguous definition of " dead-ended" volumes.
This could not explain the dif ferences in ice conden-ser or total volumes.
We believe that it is essential for TVA to pro-duce a document providing the latest information on Sequoyah including volumes, geometries, zirconium masses, final number and location of glowplugs, and additional pertinent data.
If the Sequoyah FSAR is indeed the most reliable source of data, then all researchers should be persuaded to use the same set of initial conditions.
64
1.2 Answers to Questions 1)
What ignition strategy should be followed:
on continu-ously? turn on for accident? or turn on for specific
~
times?
An extrapolation of a simple core boildown calculation indicated t
that it would be very difficult to generate hydrogen earlier than 10 minutes after a break.
However, a large break LOCA calculation with MARCH indicated that hydrogen would begin to appear in less than 3 min-utes.
Therefore, for an all-or-nothing automatic deliberate ignition system, the igniters should be activated very early in an accident -
perhaps at trip time or ECCS activationf There is no reason to keep i
the glowplug igniters on continuously during normal operation.
It is our understanding that the Sequoyah IDIS system 13 an interim mitigation scheme.
At this stage of our research, however, we feel that an operator-controlled system would be preferable in the long term.
The system should permit the selective activation of some igniters and the deactivation of others.
If the operator can monitor, in real time, the hydrogen concentration as a function of position and' time, selective lo-cal ignition of low concentration pockets could greatly improve the safe-l ty and ef ficacy of deliberate ignition and decrease the possibility of accidental overpressurization.
A computer could be programmed to prevent the deliberate ignition of dangerous hydrogen concentrations.
- However, it would also be desirable for the reactor operator to understand the principles of deliberate ignition, the hydrogen concentration limits for which upward and downward propagation are possible, and the desirability i
of lean mixture combustion over an extended period of time.
I I
2)
For degraded core accident scenarios (short of core melt),
will ignition avoid containment threat?
l i
I I
I 65
For almost all degraded core accident scenarios in the large, dry R4R containments, we believe that deliberate ignition of hydrogen would be beneficial, even for 100% metal-water reaction.. For Sequoyah, how-ever, the smaller free volume and the weaker containment combine to make deliberate ignition an imperfect mitigation scheme (assuming it would be possible to minimize or eliminate accidental ignition sources).
As-suming a failure pressure of 45 psig, the containment may not survive the adiabatic, isochoric, complete combustion of about 8% hydrogen by volume in air (about 24% zirconium oxidation).
For deliberate ignition to be successful, uniform concentrations less than 8-9% need to be burned, and at a rate and frequency sufficient to permit the normal heat removal nechanisms to operate (to prevent pressure buildup).
Lower compartment burns enhance the e,fficacy of deliberate ignition because of pressure rise reductions from heat removal in the ice condensers and expansion into the upper compartment.
Upper compartment burns, however, can ap-proach the maximum adiabatic, isochoric pressure rises.
A real-time hydrogen concentration monitoring scheme coupled with the intelligent application of deliberate ignition by computers and well-informed operators can reduce the probability of accidental, unwanted ignition.
The system could also be augmented with an intentional flaring system near high point vents on the primary system.
Although we believe that the IDIS is not the complete solution to the hydrogen problem, we do expect it to be beneficial for most types of degraded core accidents.
3)
Are there negative aspectL to deliberate ignition versus existing potential ignition sources?
To the best of our knowledge, no engineering study has been per-formed to evaluate the possibility of accidental ignition, nor to con-sider means to reduce that possibility.
The absence of such informa-tion has not prevented strong opinions f rom being formed both for and 66
against the likelihood of unintentional (non-deliberate) ignition.
In the short period of this study, we have not investigated-this problem, but strongly suggest that this matter be pursued.
The question we can address is:
Are there situations under which deliberate ignition could be harmful? The answer is yes.
If the lower compartment is either steam-inerted or oxygen-depleted, then hydrogen concentrations could build in the upper compartment, where combustion could be more dangerous.
If the hydrogen is released from the primary system at an extremely rapid rate, the igniters in the lower compartment could be overwhelmed and large concentrations could develop.
It is also possible that pockets of high concentration could develop, which might detonate.
One especially dangerous scenario could be a re-covered loss-of-power accideat.
That is, for a period of time, the de-liberate igniters would be deactivated, while very high concentrations of hydrogen develop and uniformly fill the containment.
When power is restored, all the igniters could come on simultaneously, thereby guar--
anteeing a large deflagration or detonation.
(Note that an operator-controlled system would not be forced into this situation.)
If accidental ignition is both inevitable and unavoidable, the IDIS system is clearly desirable.
However, a future study may show that there are only a small number of potential ignition sources in containment, and that these sources can be eliminated (perhaps by spark arrestors or-simi-lar devices or enclosures).
The new question, then, would be whether deliberate ignition is preferable to no ignition.
We cannot answer this question now.
4)
Calculate pressure rise for partial combustion in H /
2 air / steam mixtures and compare this to literature data to estimate completeness of combustion as a function of 112 concentration.
67
At present, there is no way to compute the pressure rise due to par-tial combustion because Sne cannot compute the fraction of hydrogen that will be consumed.
Given the fraction of hydrogen consumed, computation of the pressure rise is easy.
Results are presented in Figures 2 and 3 for 100% combustion.
We have discussed a model of lean hydrogen burning in which we divide the volume into a conical region of partial combustion and an external region of no burning.
If this model is correct for glowplug ignition in a quiescent atmosphere, it may be possible to estimate the pressure rise after the details of the model are determined.
l t,
e e
l l
l 68
REFERENCES FOR DELIBERATE IGNITION l.
" Hydrogen Control for Sequoyah Nuclear Plant, Units 1 and 2,"
N.?.C.
(Aug. 13, 1980).
2.
T.
E. Blejwas, " Yield and Ultimate Pressures of Sequoyah and McGuire Containment Vessels," Letter to W. A. Von Riesemann, Sandia Laboratories (July 10, 1980).
3.
Sequoyah Nuclear Plant Hydrogen Study, Vol.
I, April 15, 1980, Re-port Issued September 2, 1980, Tennessee Valley Authority.
4.
Statement by ACRS Chairman during ACRS meeting with NRC Chairman and Commissioners on September 5, 1980.
5.
A.
L. Furno, E. B. Cook, J. M. Kuchta and D. S. Burgess, "Some Observations on Near-Limit Flames," 13 Sym. on Comb., Pittsburgh, Comb. Inst., 593-599 (1971).
G.
B. C. Slifer and T. G.
Peterson, Hydrogen Flammability and Burn-ing Characteristic in BWR ContaiyEents, NEDO-10812, 73 N ED 49, General Electr,1c (April 1973).
7.
W. C. Harrison, M. Tamm, n. MacFarlane, L. S. Clegg, " Canadian Hydrogen Combustion Studies Related to Nuclear Reactor Sa.'aty Assessment," 1980 Western States Section Meeting, Combustion.
Institute (1980).
8.
H. F. Coward and F. Britisley, "The Dilution Limits of Inflamma-bility of Gaseous Mixtures.
Part I.
The Determination of Dilu-tion Limits.
Part II.
The Lower Limits for Hydrogen, Methane, and Carbon Monoxide in Air," J. Chem. Soc. 105, 1859-1885 (1914).
9.
B. Bregeon, A. S. Gordon and F. A. Williams, "Near-Limit Downward Propagation of Hydrogen and Methane Flames in Oxygen-Nitrogen Mix-tures," Comb. & Flame 33, 33-45 (1978).
10.
M. J.
Sapko, A.
L.
Furno, and J. M.
Kuchta, Flame and Pressure Development of Large-Scale CH -Air-N2 Explosions, Buoyancy Effects 4
and Venting Requirements, Bureau of Mines, Report of Investiga-tions 8176 (1976).
11.
L. W. Carlson, R. M. Knight, and J. O. Henrie, Flame and Detona-tion Initiation and Propagation in Various Hydrogen-Air Mixtures, i
l With and Without Water Spray, AI-73-29, Atom. Int. Div., Rockwell l
Int. (May 1973).
12.
Z. M. Shapiro and T. R. Moffette, Hydrogen Flammability Data and Application to PWR Loss-of-Coolant Accident, WAPD-SC-545, Bettis Plant (September 1957).
13.
I.
L. Drell and F. E. Belles, Survey of Hydrogen Combustion Prop-j erties, National Advisory Committee for Aeronautics, NACA R1383 (1958).
l 14.
B. Lewis and G. von Elbe, Combustion, Flames and Explosions of Gases, 2nd Ed., Academic Press, New York (1961).
l 69
_~_____ _ _ _____- - _____ _ __ _ - _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ _ - _
i i
T REFERENCES (cont.)
15.
G. F. Kinney, Explosive Shocks in Air, Macmillan (1963).
j 16.
W. E. Baker, P. A. Cox, P. S. Westine, J. J.
Kulesz, R. A. Strehlow, A Short Course on Explosion Hazards Evaluation, Southwest Research 4
Inst. (1979).
17.
S. L. Thompson, CSO -- A Two Dimensional Hydrodynamic Program with Energy Flow and Material Strength, Sandia National Laboratories, SAND 74-0122 (August, 1975).
i 1
d j
I 1
l
). -
4 4
i J
l a
'I.
t 4
i l
4 70-
2.0 WATER FOGGING 2.1 Introduction and Discussion 4
Fogging is the suspension of a large mass of liquid water in the form of small droplets in the containment atmosphere.
The fog acts as o
a large thermal capacitance, greatly reducing the temperature rise which would otherwise occur as a result of hydrogen combustion or steam re-
? ease from the reactor vessel.
Figures 1 and 2 show the theoretical d
temperature and pressure expected for the complete, adiabatic, constant-volume, combustion of hydrogen: air mixtures including the evaporation of added droplets.
(The calculations neglec+ combustion-product dissocia-tion, and hence somewhat overestimate the temperature and pressure at high concentrations, a region of no interest in Sequoyah safety studies since the high pressures there will surely overpressurize containment.)
Table I illustrates the large pressure reductions which would occur if 0.05% water droplets (fog) were suspended in the hydrogen: air mixture prior to combustion.
Table I Hydrogen Concentration Required to Attain Given Pressures Hydrogen Volume Percent No Fog Pressure Comple te Incomplete 0.05% Water (psig)
Combustion Combustion Droplets j
31 5.5 6-9 13.5 45 8.2 8-9 15.7 The complete, adiabatic, isochoric combustion of 8 volume % of hydrogen homogeneously distributed throughout containment would raise the pres-sure from 0 to 45 psig, as shown in the table.
Lean mixture, incomplete o
combustion would increase this to no more than 9%, since combustion is quite complete at this concentration.
However, the addition of a l
71
INITIAL CONDITION $:
'g
~
T = 298 K = 25 C P = 100 hPa a 1 ATM.
pSTOCH10 METRIC 100% RELATIVE HUMIDITY RATIO 7
Ya
$E6 I
5 t-z55 I
uj4 0.00% NO DROPS z
C 0.01% VOLUME F* TACTION OF DROPS 0.02%
3
~
/-
0.03 %
0.04 %
e 2
1 0.05%
I i
I 4
I e
t i
t i
I i
I e
I i
g 0
4 8
12 16 20 24 28 INITIAL HYDROGEN CONCENTRATION. Vlo Figure 1.
Ratio of final pressure to initial pressure as a function of initial hydrogen concentration and volume fraction of water drops 72 l
e INITIAL CONDITIONS:
T - 298 K = 25 C 2500 P = 100 kPa a 1 ATM.
100% RELATIVE HUMIDITY
- STOCHIOMETRIC i
RATIO 2000 g
o e
a e#i e
o-5
$~
- a. 1500 0-O o
A o$
z 1000 o!
O i
500
-- INITIAL TEMPERATURE i
i t
i I
00 8
12 16 20 24 28 INITIAL HYDROGEN CONCENTRATION, Vlo Figure 2.
Final temperature as a function of initial hydrogen concentration and volume fraction of water drops G
73
.... -.. - - - -,. - - - -... - -. - -... -. -..... - -.., - - - - -, -,. ~
-- _ _ ~
1 fog (0.05% by volume) prevents that large a pressure rise (45 psig until the hydrogen concentration reaches as high as 16%, corresponding to more than 50% metal-water reaction.
The benefits of water fogging are both obvious and significant.
For fogs with concentrations lower than 0.05%
by volume, the benefits would, of course, be proportionally lower, as shown in Figure 1.
i An experimental study of the effects of 0.05% water drops on hydro-gentair combustion showed the expected reductions in pressure rise duc to deflagrations, and the suppression of detonations up to the highest hydrogen concentration tested, i.e., stoichiometric.(1)
The suppression of detonations could be very important for large dry PWRs, where the strength of the containment is much higher.
For the ice condenser con-
)
tainments, only a local detonation in a hydrogen-rich pocket would be l
of interest.
All other detonations will lead to containment f ailure.
The fog would be generated by spraying sufficient droplets into the l
containment atmosphere to suspend a large mass of water in a finely divided state.
Methods are available for doing this.
We have also just begun to address the use of foams.
Foams may have advantages over drop-lets in terms of engineering and design requirements for replenishment.
3 i
KINETICS OF DROPLET VAPORIZATION To achieve the benefits of pressure and temperature reduction from
}
fogging (Figures 1 and 2), the droplets must be small enough to vaporize
[
not far behind the flame front.
If the droplets are too large, their vaporization would be delayed.
This could result in a large region of higher-temperature burned gas corresponding to the curves in Figures 1 i
and 2 for no droplets.
For these oversize droplets, pressure rise dur-ing combustion would initially overshoot the equilibrium value, and even-tually achieve equilibrium as the droplets vaporized.
The overshoot in pressure could last long enough to generate higher static loads on the containment structure.
On the other hand, however, if the droplets are l
74
~, - -..
too small, they could vaporize inside the thin (~1 mm)(2) flame front and inhibit or even quench the flama.
For these reasons, the optimum droplet size is bounded both above and below.
The time required to burn the atmosphere in containment will be of the order of the characteristic length of the flame path divided by the effective flame speed.
Typical flame paths could vary froa 10 to 40 me-tt. r s.
A discussion of flame speeds to be expected in hydrogen: air com-bustion can be found in Refs. 3 and 4.
The effective mean flame speed is expected to be higher than the laminar flame speed which ia below 1 m/s for lean hydrogen mixtures.
Assuming a flame speed of about 2 m/s, th e time for combustion would range f rom about 5 to 20 second3.
For this analysis, we will assume that the fog would be effective in re-ducing pressure if it vaporized in less than 1 second.
The rate of vaporization of the fog is primarily dependent on the following quantities:
1)
Mean droplet diameter and size distribution, 2)
Ambient tempera ture,
3)
Speed of ambient gas relative to droplets, 4)
Composition of ambient gas.
We examined a theoretical model of the rate of vaporization of water droplets caused by a flame propagating past the droplets.
A pictorial tepresentation of the model is shown in Figure 3.
Vaporization begins when the flame impinges on the droplet, and a steam film composed of water vapor, air, and combustion products forms around the droplet, dif-fusing outward.
Because of the thermal expansion of the burned gases, a relative velocity will develop between the droplets and the surround-ing burned gahes.
The relative motion of the gases will distort the surrounding steam film and increase the rate of heat transfer to the droplets and hence increase the ' rate of vaporization.
The steam film surrounding the droplet grows with time and finally mixes with the 75
propagatin/g flame diffusion film
\\
^
\\
/*.
=
c.
c C"
C*
unreacted combustible with complete burning, g:;
e/ e dispersed water mixing and vaporization a cm c
C#
n TURBULENT Co
.l.
c3 MI)(ING e *d;' e o
\\
O m
o c _,, o oi. e
+ gas expansion Figure 3.
Monodispersed Water Droplet Vaporization Model e
combustion products as the droplet is ultimately consumed.
The vaporiza-tion time for the droplet is defined as the time between flame impinge-ment and complete vaporization of the droplet.
Details of the mathematics of this vaporization model can be found
~
in Ref. 5.
Shown in Figure 4 are the vaporization times of droplets with initial radii between 1 and 104 um for several values of initial hydrogen mole fraction.
Note that droplets of initial radius less than 200 um will vaporize in less than 1 second.
A 200 um radius droplet is a fairly coarse droplet and is easily obtained by many different types of spray nozzles.
We will examine the possibility of the fog vaporizing in the thin flame front.
The thickness of a hydrogen laminar flame front is of the order of 1 mm.(2)
For a flame speed of 2 m/s, the residence time is of the order of 0.5 x 10-3 seconds.
From Figure 4 we see that droplets of initial radius less than about 4 um will entirely vaporize in the flame zone.
Another critical aspect of flame propagation within a cloud of water droplets is the possibility of quenching; i.e., droplets acting as heat sinks and extracting the thermal energy generated by the propa-gating flame.
As the droplets become closely packed, the area available for energy loss increases. A critical spacing of the droplet field ex-ists such that a large fraction of the heat released is absorbed, thus preventing flame propagation.
This spacing is called the " quenching distance".
This parameter is experimentally determined by propagating flames in tubes, and for these geometries the quenching distance is de-termined by:
dq = (4Vc/STlcritical where Vc = combustion gas volume and ST = heat transfer surface area.
For a cloud of spherical water droplets of a given water content (by volume), this volume-to-surface ratio is:
77
s k
i H - Air mixtures.05% H 0 2 re
/
1
% relative humidity
( x ) =0.17 H2 St E
e (x%)
initial H fraction (O.14 2
(0. 108) 3
~
g es (0. 075) s i
g flame zone vap.
based on (0. 057)
S 100.cmlsec
}
o I
o'
/
tb
' i b '
ib-'
ib' '
i b ' '
ib' ' " ib ' " ik ' " 'to'
. Vaporization Time (SEC)
Figure 4.
Vaporization Time vs. Droplet Radius at Various Combustion Stoichiometries
4Vc. jl R(1 - n)
ST 3
8 where R = mean droplet radius and n =.olume fraction of water. When this ratio becomes of the same order as d, the heat transfer effects q
due to droplet spacing approach a critical limit.
A critical droplet radius is reached when:
nd R =1 9
e 4 (1 - n)
Droplet arrays which are composed of droplets with radii greater than Rc will allow free propagation of the hydrogen flame.
Figure 5 is a plot of the critical droplet radius for various water contents of the fog and two values of combustion stoichiometry.
(Data for quenching distance were taken f rom Lewis and Von Elbe. (6) }
If the fog is made such that the great majority of droplets are larger than 5 um in radius, then we do not expect the' droplet vaporiza-tion and droplet spacing to significantly af fect the flame structure and -
hence influence the flammability limits.
Fogging could therefore be used in conjunction with deliberate ignition.
The predictions of the vaporization model have been checked'against available experimental data (7) for isolated water dropJets evaporating in heated air.
Figure 6 shows this comparison and our predictions of vaporization time agree to within a few percent.
(The long vaporization times seen in this figure.are due to the large droplet size and low am-g bient gas temperature used in the experiments.)
The model predictions also agree with an experimental study which examined hydrogen flame propagation in a vessel with water spray addi-
-i tion.(1)
Using a mean droplet radius of 250 pm (as was employed in the experimental work) our model would predict a vaporization time of 0.5
-sec, well within the time at which the pressure rise was observed (this corresponds to ~ the time required for the flame to propegate across the 79
c mbustion +
.-.-._4.. - _ -...
a
- 4. ti}: Air._co y 4.-w.
r 4 A +7
.,,A.
+
_L A_.
4_
f..
L-7w
+aw
.I
.a d.b
.d t
- g.,
. p
...p
.y y.-
-g.
7.._
..y
- t,,
1-.A j i -
- , n-.t_...-_. _-_..r.... --- @..re ;-+--- _ _- __
.._-_._4 7
- - ~ ~~
~~
' ~: Tj
-~.-.
-g - -. J-DroN1'e't !foigsNith meani'dia.m eters
. _-. AlfoVe{thes'e-curvos allow _.,
j
.-..-f,re e. pr, agati_o_n_ of,__f_i_ cme-t Lh-b -~
-.a
- q- -
,r
- _ 7_
yU+5
__ y u_u_
. tw _2 u.
i ;.,
3: ;
g_._
i i
_y
.w
_..m. ole.,f raction(- o f_.-~!!
.=_10g_u_- 2._t
_4 -
o
.-..o i !m "2
~
_-Z---~"
-ff--0-4
- c. -._
- _ _.. ___( d =_ -1. k_as'_-T '_'~~ l i,.. _ _..
++'
i 4
.. q
..) A 4_-_-
_o__-
-.__j_.__
_. __q __ _
~
. - _ _ +
__-a s A '> '
~
__..p--_.__. _ J _ __ __. _
. ;i. _ ___
.a____.
g_
i 1-__
I 54
]
I y
F1 7
_j_._.
a..
fi ___
1 i
O2
}_._
q.___ j 3
..p_.
s_,.
..z - -
/p.,
-Q i
.. a.
. -_. - t._
X - =_. 3 0 _
' ~ _ _ _
y
_f._..-_
--y'/.
/
Td =~t 64 tm W
- ? o,1
-+---.
. _.u __
.. _ _ y, ;,,e._
._.-p-.,-
i..-
._ m ;_,__;.
3
- r i
__-.__4_ - _-
..y _/s g_4_
i j
_ p._..Z
-[0 ~01FT -0, 02Z Z~0 0 3 ~_ -g 0. 0 Cg-_T~ 0 0 5.--~
I s
7,. -.E R.
.Jr..
CO, N. TENT-%-(by -volumd WAT
_.i Figure 5.
The Effect of Droplet Spacing on Flame Quenching 80
a Water - Air 500mm Radius Abdel predicticns
.I m Experimental data cf r
isolated dreptets of water in
$M
- heated air REF:' Evaporation of Single Droplet at Elevated Pressures and Temperatures".Hiroyasu and Kadeta, Tran, Jip. So.M.E.,40_,(1974i 200,.
300.
M 500.
Amident temperature 'C Figure 6.
Comparison Between Theoretical vaporization Time and Experimental Data on Isolated Droplets (Ref. 7) 16 5 H2 40.
lE;;; 30.
O gpm of Hp a.
E
?'i a
Gn I \\.
\\
a.
w 73 gpm of H 2
m e
.. s = is a i. = no time (SEQ
- Vaporization time for 500pm dia, water ditplets itEF: ' Flame and Detonation initiation and Propagation in Various Hylrogen-air mixtures with and without water spray", L W.Carlson, R. Knight,and i Henrie,Rockwell intern. report Al-73-29,1971 a.
Figure 7.
Vaporization Prediction ft 500 pm Diameter Droplet in Hydrogen: Air Combustion Environment (Ref. 1) 81
1 experimental vessel (16-inch diame ter).
Figure 7 shows the pressure vs. time of a test at 16% initial hydrogen concentration and the model prediction of vaporization time is the barred interval on the time axis.
PRODUCTION OF WATER FOGS Based on the thermodynamic calculations, a water volume fraction approaching 0.05% may be needed to sufficiently reduce the pressure and temperature following combustion.
For a 3.4 x 104 m3 containment vol-ume, a water content of 0.05% corresponds to 4500 gals of water placed j
in suspension.
Since a large quantity of droplets with radius less than 200 um may be required, a high output aerosol generation device must be used.
Fine mists can be generated in large quantities using nozzles or mechanical atomizat' on devices.
Typical outputs from these systems are i
j usually less than 10 gals / min (6) and thus many nozzles will be needed.
Conventional nozzle systems produce polydispersed sprays (varying drop-let sizes) with a log-normal distribution of droplet sizes centered about a geometric mean radius of 50 pm.
Sprays with smaller droplets (10 um mean radius) can be produced using specialized nozzles with swirl and air blasting.(8)
Fogs with very small droplets, less than 1 pm radius, are difficult to generate in large quantites.
Since nozzles have small discharge areas, large pressure drops across the nozzle orifices can be anticipated.
For sufficient atomiza-tion, typical pressure drops through pressure nozzles are 100 psi.(1)
The pressure drop across "N" orifices is calculated using the following relationship:(2) 2 p=1p T
T C AN 2
D where i
)
82
f P = pressure drop across the nozzle system p = liquid water density h = total output of water (volume flow rate)
T A = nozzle discharge area of a single orifice CD = discharge coef ficient = 0.8(3)
Consider a total water flow of 104 gal / min input through 5000 orifices of 0.1-inch diameter; the calculated pressure drop is 70 psi.
(If 1000 orifices were used in this calculation, the pressure drop would be
~ 2000 psi.)
Therefore, we conclude that, for pressure atomizing nozzles, a large number (>5000) of orifices will be required.
Probably a superior atomization scheme is the use of air blast noz-zles.
Water can exit through reasonably large nozzles (1/4 inch diam-eter or larger) with very low pressure drop and with high resistance to clogging.
The atomization is accomplished by an impinging high-speed compressed-air jet.
FOG MAINTENANCE AND STABILITY Although high volume production of small water drops is possible, stable, high-density fogs may be difficult to maintain if there are very large settling losses.
Estimates, made under the assumption that drops do not interact and coalesce, yield losa rates f ar below those given by a more complete formulation of the problem.
The range of these losses, when drop interactions are and are not considered, extends from moderate values (<100 gpm) to values so large ( ~ 100,000 gpm) that fogs of accept-able density cannot be attained given reasonable injection rates.
- Thus, the mechanisms affecting losses from containment must be examined more I
i closely to determine the feasibility of maintaining a high density fog.
Available nozzles do not produce monodispersed (single size) sprays; rather, they produce polydispersed sprays covering some range of drop sizes.
As the constituents of a polydispersed fog settle, larger drops, with higher terminal velocities, will overtake, collide, and coalesce 83
with smaller ones, having relatively lower terminal velocities.
This is-called gravitational coagulation (12,13).
The result is to increase the mean drop size and therefore to increase the rate of water loss.
Because the effect is so significant, coagulation plays a critical role in fog maintenance and stability.
Most calculations of aerosol coagulation are based on some form of the " geometric sweep" concept (14) of collisions; i.e.,
if any portion of a drop lies within the volume swept out by another drop as it falls some d is tance, then the two will collide.
This method gives too high a rate of collision and so the swept volume is modified by a collision ef fi-ciency,' E, to account for close-proximity hydrodynamic interactions of c
the two drops.
A commonly used collision ef ficiency is that given by Lindauer.(15)
An efficiency of zero (Ee = 0) yields no collisions and hence no coagulation, while a value of unity (Ee = 1) returns the geometric sweep collision rate.
The actual value must lie between these extremes.
Figure 8 shows losses from containment for two values of the colli-sion efficiency:
zero and that given by Lindauer.
Calculations are based on a widely accepted formulation of gravitational coagulation (16,17,18) and drop settling.(19)
The containment volume, V, is initially empty c
and the uniform injection rate is constant at 310 kg/s (5000 gpm).
Note that the ef fects of coagulation are very pronounced.
At less than 60 seconds, the loss rates when coagulation is considered (Ee=
'Lindauer') are roughly two orders of magnitude greater than when they are not (Ee = 0).
The dramatic consequence of this difference is shown in the instantaneous fog density, Figure 9.
When coagulation of drops is considered, the water content peaks at slightly over 2.0 x 10-4 g/cm3 (0.02% by volume) in less than 30 seconds; this is the highest attainable density for the stated conditions and is only marginal for a successful mitigation design.
After this time, the density oscillates regularly with some decay in amplitude.
The long-l l
84 L
10' ;
Injection Rate -
4 J
4 S 10 E = 'Lindauer*
M LsJ tn (A
O; 10 -
E
=0 c
10' 0.0 12.0 24.0 ss.o 4s.o so.o TIME (S)
Mean Drop Radius = 10 um Log-Standard Deviation = 1.2 Injection Rate = 310 kg/s (5000 gem)
Volume of Containment = 3.4x104 m5 (1.2x106 ft3)
Height of Containment = 40 m (130 ft)
Figure 8.
Settling Losses for Steady Fog Generation - Comparative Effect of Collision Efficiencies I
l l
85
Og - 5.0
^n 3
4.0 -
UNo v
3.0 -
g E
=0 c
Zw HZO 2.0-O Q:
W W
1.0 -
Ec" O.0 0.0 12.0 24.0 36.0 48.0 60.0 TIME (S)
Mean Drop Radius = 10 um Log-Standard Deviation = 1.2 Injection Rate = 310 kg/s (5000 g2m)
Volume of Containment = 3.4x104m5 (1.2x106 ft) 3 Height of Containment = 40 m (130 ft) 1.0x10-4 g/cm3 = 0.01%by volume Figure 9.
Water Content for Steady Fog Generation - Comparative Effect of Collision Efficiencies 86
time behavior is shown in Figure 10.
In contrast, when coagulation is not considered the density continues to grow almost linearly for several hours.
The asymptotic water content in this case is about 2.5 x 10-2 gf cm3 (2.5% by volume) attained at around 5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />.
Note that this value is over 100 times that of tha maximum mentioned before.
As a means of decreasing losses, we have examined the effect of the mean drop size being injected.
With no coagulation, halving the mean would result in approximately a four-fold decrease in the settling losses and a corresponding four-fold increase in the steady-state fog density.
As shown in Figure 11, however, when coagulation is considered, the mean drop size being injected has almost no effect on the water content; be-yond an initial transient, the quasi-steady density is nearly the same for mean drop sizes in the range 10 to 50 um - again, coagulation is clearly dominating the dynamic behavior.
Similar comments apply to the ef fect of spread or log-standard deviation of the drop sizes being injec-ted.
Figure 12 shows that a wide range of log-standard deviation, 1.2 to 1.4, does not significantly alter the quasi-steady fog density.
These somewhat surprising results indicate that generating very fine drops of nearly uniform size will neither significantly reduce losses from con-tainment nor increase the fog density.
Finally, the role of the injection rate has been examined for rates in the range 62 to 620 kg/s (1000 to 10,000 gpm); some results are given f
in Figure 13.
Considering the value of the water content at the first maximum to be a characteristic measure of the fog density, the density appears to grow as the injection rate to the one-half power.
This again may be contrasted with the case when coagulation is not considered, in i*
l which the water content would grow linearly with the injection rate.
l l
Thus, merely increasing the injection rate is not an ef fective way to achieve significant increases in the fog density.
l l
In summary, we find that the maintenance of high fog densities re-quired to substantially mitigate the effects of hydrogen combustion may I
i 87 i
L
Io 7 - 5.0
^mg 4.0 -
ONo v
3.0 -
Zw F-ZO 2.0 -
O EW 10 -
0.0 0.0.
30.0 60.0 90.0 12 0.0 15 0.0 TIME (S) l l
Mean Drop Radius = 10 um Log-Standard Deviation = 1.2 j
Injection Rate = 310 kg/s (5000 gem) l Volume of Containment = 3.4x104m3 (1.2x106 ft3}
Height of Containment = 40 m (130 ft)
Collision Efficiency (Ec) = 'Lindauer' l.0x10-4 g/cm3 = 0.01% by volume Figure 10.
Long Time Water Content for Steady Fog Generation i
88 l
l - _
O 7 - 5.0 m
"3 4.0 -
O\\O Mean Drop Radius (pm) 3.0 -
W Zw b-Z 10 0
2.0 -
O 20 gw
>=<
1.0 -
m 50 }
3 30 0.0 0.0 12.0 24.0 36.0 48.0 60.0 TIME (S)
Log-Standard Deviation = 1.2 Injection Rate = 310 kg/s (5000 gem)
Volume of Containment = 3.4x104 m3 (1.2x106 ft3)
Height of Containment = 40 m (130 ft)
'Lindauer' Collision Efficiency (Ec)
=
1.0x10-4 g/cm3 = 0.01% by volume Figure 11.
Water Content for Steady Fog Generation - Variation with Mean Radius of Injected Drops f
l 1
89
- ~
Io 7 - 5.0 m
3 4.0-O\\$
Log-Standard 3.0 -
Deviation Z
w ZO 2.0 -
1.2 U
l.3 Ek:w 1.4
<t 1.0 -
0.0 0.0 12.0 24.0 36.0 48.0 60.0 TIME (S)
Mean Drop Radius = 10 um Injection Rate = 310 kg/s (5000 gpm)
Volume of containment = 3.4x104 mJ (1.2x106 ft3)
Height of Containment = 40 m (130 ft)
Collision Efficiency (Ec) = 'Lindauer' l.0x10-4 g/cm3 = 0.01% by volume Figure 12.
Water Content for Steady Fog Generation - Variation with Log-Standard Deviation of Injected Drops l
l l
90 l
e C3; - 5.0 em M
lE 4.0 -
Injection Rate LJ
[kg/s (gpm)]
NN O
620 (10,000) 3.0 -
g Zw ZO 2.0 -
0 310 (5000),e
~
1.0 -
62 (1000) 0.0 0.0 12.0 24.0 36.0 48.0 60.0 l
TIME (S)
Mean Drop Radius = 10 um Log-Standard Devia; tion = 1.2 volume of Containment = 3.4x104 m3 (1.2x106 ft3}
Height of Containment = 40 m (130 ft)
'Lindauer' Collision Efficiency (Ec)
=
1.0x10-4 g/cm3 = 0.01% by volume Figure 13.
Water Content for Steady Fog Generation - Variation l
l with Injection Rate l
t l
1 91
be difficult to achieve.
The mechanics of drop collisions, and particu-larly the coalescence of droplets in high density fogs, are critical in determining the losses from the fog.
The mechanics of droplet collisions will require further investigation, both theoretical and experimental.
1 We will also be considering the addition of surfactants to increase sur-face tension and possibly reduce the losses from the fog.
THE EFFECTS OF FOGGING ON DETONATIONS AND SHOCK WAVES Carlson, Knight and Henrie(1) found that the use of a coarse (250 pm radius) liquid droplet spray of volume fraction 0.05% suppressed detona-4 tions for hydrogen: air mixtures up to stoichiometric in a 16-inch-diam-eter horizontal tube.
The droplets act to dissipate energy from detona-tion and shock waves, both by mechanical action of the drop drag (22) and by the thermal action of droplet evaporation in cooling the gas.
The possible beneficial effect of fogs in preventing detonations may be of limited interest in the Sequoyah safety study.
If a homogene-ous detonable mixture burns without detonation, the containment will not withstand the resultant static overpressure, even with fogging at 0.05 vol.%.
For the stronger large, dry PWR containments, the suppression of detonation is more relevant.
However, it may be possible to withstand the dynamic effects of a detonation of a locally rich pocket of gas in Sequoyah.
When the detonation wave leaves the detonable region, it be-comes a shock wave of ever decreasing strength.
The rest of this sec-tion of the report will address the following three questions:
- 1) lhat is the effect of fog on detonation waves?
- 2) What is the effect of fog on transition to detonation?
- 3) What is the effect of fog on shock waves from local t
detonations?
l THE EFFECTS OF FOGS ON DETONATION WAVES Pierce (22) in his model of shock wave propagation through sprays, considers that the droplets evaporate behind the shock front in a 92
distance negligible compared to the shock wave radius of curvature.
If this is the case for either shock waves or detonations, we can mu3el the wave as a discontinuity with the jump conditions across the wave given by the usual Chapman-Jouquet theory (see, e.g., Ref s. 5 or 6).
We have done this for hydrogen: air detonation waves.
The reduction in post-detonation temperature is substantial, as seen in Figure 15.
The reduc-tion in pressure, however, is much smaller, Figure 14.
Consequently, the use of fogging, if it does not prevent a detonation, will not greatly reduce the dynamic loads produced by a detonation.
THE EFFECT OF FOGS ON THE GENERATION AND SUPPRESSION OF DETONATION WAVES In the study of Carlson, Knight and Henrie(1) the water spray was able to suppress a detonation wave propagating into the mixture from an adjacent tube which'had a stoichiometric hydrogen: air mixture without the spray.
This is a more severe requirement than that needed to prevent the transition of a deflagration into a detonation.
Evidently, the dissipa-tion of energy in the detonation wave by the evaporating drops was enough to weaken and dissipate the detonation.
The transition from de51agration to detonation is believed to usu-ally occur as a result of the formatior. of shock waves ahead of the de-flagration and the resultant shock pre-heating of the gas.
The effect of l
drop evaporation cooling the burned gas will reduce the expansion of the burned gas and hence weaken subsequent shock waves.
On the other hand, i
the droplets also introduce turbulence in the flame zone and increase I
the flame speed.
The increase in flame speed would tend to promote detonation, j
(
Obviously there are many unknowns concerning suppression of detona-tions by fogs.
Further analytic and experimental work will be required to clarify the possible benefits of fogs in detonable mixtures.
l l
l t
l 93 l
t-16.0
.cowonvions I
swarnat
~
7 e 288 K
- =
AYa.
15.0
+a sees est.
8,**
~
nuntostr 14.0 w
e o
x5 13.0
- *e, m
w
=
Q-12.0 J
o 11.0 g#+
~
z y 10.O e,
E 9.00
- +
w x
Q.
8.00 a<z w 7.00 l
10.
20.
30.
40.
50.
60.
t INITIAL HYOROGEN CONCENTRATION. VOL. PCT.
Figure 14.
Theoretical Chapman-Jouguet Detonation Pressure l
for a Mixture of Hydrogen: Air: Liquid-Water-Drops I
(VL is the vol.% of droplets) 94
2 I
3.0 a
- cs.
+o C+##
x 2.8 4cf I
2.6 2.4 "4
2.2 14 2.0 1.8 l
1.E 8.,Jg 1.4 9%
w
=
1.2
- 2 o
n
'Og 1.0 g
a E
r
.80 U
.60 10.
20.
30.
40.
50.
60.
INITIAL HYOROGEN CONCENTRATION. VOL. PCT.
l i
~
i Figure 15.
Theoretical Chapman-Jouguet Detonation Temperature for a mixture of Hydrogen: Air: Liquid-Water-Drops (VL is the vol.% of droplets) 95
THE EFFECT OF FOGS ON SHOCK WAVES If a cloud of gases detonates, the detonation wave af ter leaving the detonable region will decay into a shock wave of decreasing strength.
(See the discussion in the " Local Detonations" section.)
Pierce (20)
^
considered a simplified theoretical model of the propagation of a shock wave from a point explosion into an inert gas with liquid droplets.
For the same shock wave speed (Mach number) the effect of the vaporization of che droplets is to decrease the temperature rise behind the shock wave, but to increase the pressure rise.
The vaporization of the droplets causes the Mach number of the expanding shock wave to decrease faster than it would without the droplets.
The results we have obtained seem to show that these two effects nearly cancel.
The pressure behind the shock wave at a given distance from a point explosion is nearly indepen-dent of the presence of the fog.
The Pierce model, assuming complete evaporation of the fog droplets, must f ail when the temperature behind the shock wave is low enough so that the vapor pressure of steam is below the partial pressure of steam expected f rom complete droplet vaporization.
As the shock wave propagates outward and the Mach number drops, the vaporization of the droplets will start to become increasingly less complete until at low enough Mach numbers the droplets will act as nonvaporizing particles.
W 96
2.2 Answers to Questions 1)
What would the design concept of a water fog system be?
The practicability of the water fog concept we have been consider-ing depends on the ability to suspend large masses of water in the con-tainment atmosphere.
If 0.05% liquid volume could be maintained, we expect that the concept would work well.
Aerosol codes, which assume all drop collisions result in drop agglomeration, predict very large losses for droplets in containment-sized volumes.
This would require very large flow rates for the renewal of the fog.
We believe that these loss rates are overly pessimistic; droplet collisions will not always result in agglomeration because of surface tension effects.
The addition of substan'ces to increase surface tension could further reduce drop agglomeration.
(Note that droplet charges could also af fect agglomeration.)
However, until the loss rates can be more accurately estimated or measured, it is premature to design a fogging system.
Alternatives to fogging may be possible which also could suspend large masses of water in the atmosphere. We have begun to look at the use of 3
foams.
Foams have been used in airplane hangers for fire suppression.
It would not be necessary for the foam to be able to stay indefinitely in place, since it could be renewed.
2)
How effective would water fog be in preventing a hydro-gen combustion or detonation threat to containment?
We know that droplets of appropriate size will evaporate soon af ter the flame front passes.
Hence, the adiabatic constant-volume combustion calculations assuming complete droplet evaporation should be reasonably accurate in predicting final pressures.
For 0.05% droplet volume frac-tion, Sequoyah could survive a 16% hydrogen deflagration -
i.e.,
not ex-ceed an estimated 45 psig " failure" pressure.
For lower volume fractions 97
4 of droplets, the maximum hydrogen initial concentration would be lower.
For 100% zirconium oxidation, deliberate ignition combined with fogging would be able to handle the hydrogen only if it came out slowly enough j
for it to be burned over an extended period of time with cooling of the atmosphere'taking place.
If the burning was rapid enough to be nearly adiabatic, the containment would be overpressurized, even with fogging.
There is some exparimental evidence that fogging can prevent'the propagation of detonation waves.
This has to be examined further.
Sim-i ple thermodynamic calculations of the jump conditions across a detona-i tion wave, assuming complete droplet evaporation, indicate that the tem-i perature rise is considerably reduced, ut that the pressure rise is re-1 l
duced much less.
If the-fogging does not prevent a detonation, it will i
i not reduce the dynamic loads produced by the detonation suf ficiently to 1
2 prevent damage.
3)
What water fog density and particle size are required to I
suppress containment threat as a function of H / air / steam 2
ratios likely to result from dominant accident scenarios?
?
t From our studies it is clear that a very wide range af droplet sizes will be satisfactory for evaporation behind flames.
The droplet size
]
will be governed by considerations of loss by settling.
Droplet
. diameters on the order of 100 pm appear to be satisfactory for pressure suppression, but may be too large for limiting less rates.
We have been l
using 0.05% droplet volume fraction as a goal since this was the value used in a previous study.
Our calculations, however, also indicate that 0.05% is reasonable to achieve significant pressure reductions.
~
4)
Assess the problem of maintaining a water fog in the post-accident atmosphere.
i.
The suspension of a large mass of water droplets in the containment atmosphere for an extended time may be the most dif ficult technical 98
problem with fogging.
The results of computer codes (which assume all drop collisions produce agglomeration) indicate very large droplet loss-es, requiring very large water flow rates to renew the fog.
We believe that these loss rates are too high and that surf ace tension ef fects may limit agglomeration.
This issue must be settled before fogging systems can be designed.
5)
Discuss the pros and cons of a water fog system.
Pros 1.
It will reduce the pressure and temperature rise caused by hydrogen combustions.
2.
It will act to condense gteam, reducing containment steam pressure during a LOCA.
3.
It may prevent detonations.
This issue is not yet re-solved.
4.
It could effectively remove fission products and aerosols f rom the atmosphere.
5.
It requires no new penetrations of containment.
It may be simple and inexpensive to install compared to other mitigation schemes.
6.
In combination with deliberate ignition, it will perma-nently remove hydrogen from the post-LOCA environment, while mini-mizing the overpressurization threat to containment.
Cons 1.
It is an active system requiring power to pump water to maintain the fog.
2.
If the loss rate of drops is as high as predicted by aero-i l
sol codes, then the water flow rate required to renew the fog may be f
impractically high.
I 3.
In condensing the steam, it may cause a hydrogen-rich mix-(
ture to go from nonflammable to highly flammable.
1 i
I 99
6)
Will water fog reduce steam concentrations and yield a more combustible or detonable mixture?
Yes.
The fog will condense steam.
If a high concentration of hy-drogen exists which has been inerted by steam, the condensation of that steam will produce a combustible mixture.
7)
What steps are to be taken af ter the water fog is in i
place in a containment containing H / air / steam / fission 2
products? What are the final steps to recovery af ter a water fog has initially prevented or reduced the effects of combustion?
We do not expect the fog to signif'icantly ' alter the flammability-limits.
Therefore,' fogs can be deployed in conjunction with deliberate ignition without interference.
Deliberate ignition would be carried out until the atmosphere was no longer flammable.
The fog is used to reduce the danger of overpressurization to the containment by combustion and/or steam.
When the hydrogen has been removed, the fog can be stopped.
8)
How long will it take to install one?
It is premature to answer this question since the basic practicality of the idea is in question.
If the loss rates are shown to be reasonable, it should not take too long.
It may be possible to modify the existing spray nozzles and have them produce the fogs.
- The following two questions were not part of the original NRC work scope.
They were raised at later discussion sessions -
9)
What effect will the existing water sprays have on fog stability?
If water fogging is adopted, it should serve as a complete replace-ment for the spray system; i.e.,
the fogs will provide all the protection 100
that the sprays did, in addition to greatly enhancing the hydrogen com-bustion mitigation features.
If, for some presently undetermined reason, it was desired to maintain both capabilities with simultaneous operation, the integrated effects can be analyzed.
We have performed a computer calculation using a bi-modal injection to simulate the concurrent operation of a fog and a spray generation sys-tem.
The mean droplet sizes of the fog and spray were taken to be 10 and 500 pm, respectively.
The injection rate of the fog was 310 kg/s (5000 gpm), while that of the spray was 62 kg/s (1000 gpm); a frequency dis-tribution by mass of the injected drops is shown in Figure 16.
Figure 17 gives the fog density or water content, as a function of time.
The addi-tion of a water spray does not produce pny significant change in the fog density - either increase or decrease.
The most notabla result is an apparent stabilizing ef fect of the spray - the limits of oscillation of the overall fog density are significantly reduced.
10)
What ef fect will ice condensers have on the fog?
Simple calculations show that the number of drops suspended in the turbulent flow through a duct decays exponentially with distance X from the inlet:
N=N exp(-( 2U X)/(UKR)]
o s
where U is a characteristic transverse velocity, 0 is the axial veloc-s
(
ity, and R is the duct radius.
The solution presupposes that all drops which contact the duct wall are lost from the system and cannot be re-entrained.
For turbulent flow, the transverse velocity may be taken as i
the RMS eddy velocity, U, which is in general some fraction of the axial velocity - typically on the order of 3%:
Us = U = 0.03 U.
This assumption gives 101
10' :
10'-
^2 O
N O
2 o 10 3 z
W DO w
E 1
L 10 3 10'10,,.........,,........,,
10 10, 10 10 DROP RADIUS (CM)
FOG:
Mean Drop Radius - 10 um Log-Standard Deviation = 1.2 Injection Rate = 310 kg/s (5000'gpm)
SPRAY:
Mear. Drop Radius = 500 um Log-Standard Deviation = 1.3 Injection Rate = 62 kg/s (1000 gpmj ma (1.2x106 ft3) volume of Containment = 3.4x104 Height of Containment = 40 m (130 ft)
Figure 16.
Frequency Distribution by Mass of Generated Fog and Spray 102
'o 7 - 5.0 m
"2 4.0 -
UNO v
3.0 -
Z hJ F-Without Spray 2.0-U 5
/
1.0-t With Spray 0.0 0.0 12.0 24.0 36.0 48.0 60.0 l
TIME (S) l l
FOG:
Mean Drop Radius - 10 um Log-Standard Deviation = 1. 2 Injection Rate = 310 kg/s (5000 gpm)
SPRAY:
Mean Drop Radius = 500 um l
Log-Standard Deviation = 1. 3 l
Injection Rate = 62 kg/s (1000gpmg(1.2x10 3.4x104 ma 6 ft3)
Volume of Containment =
Height of Containment = 40 m (130 ft)
'Lindauer' Collision Efficiency (Ec)
=
Figure 17.
Water Content for Bimodal Fog Gener ation - The Zf fect of Sprays on Fog Stability 1
103
. _ =. -.. _... _.. _........ _. __
4 i
s j
i
)
j N=N exp [-0. 02X/R]
o l
. Appropriate lengths for an ice condenser are X = 15 m (48 ft)
R = 0.10 m ('. 33 f t) giving N = 0.05 N o
That is, virtually all fog / drops entering.the ice condensers will be lost j
from the system.
4 U
0 104
....._,... _ _., _ _ _ _,.. ~ _. _.
REFERENCES FOR WATER FOGGING 1.
L. W.
- Carlson, R. M. Knight, J. O.
Henrie, Flame and Detonation Initiation and Propagation in Various Hydrogen-Air Mixtures, With and Without Water Spray, Atomics International Rpt. AI-73-29 (1973).
2.
I. Glassman, Combustion, Academic Press, p. 58 (1977).
3.
J. Warnatz, " Calculation of the Structure of Laminar Flat-Flames' II:
Flame Velocity and Structure of Fully Propagating Hydrogen-Oxygen'and Hydrogen-Air Flames," Ber Bunsenges Phy Chem 82, 643-649 (1978).
4.
M.
P. Sherman, et al., The Behavior of Hydrogen During Accidents in Light Water Reactors, SAND 80-1495 Sandia National Laboratories, Albuquerque, NM (August 1980).
5.
F. Williams, Combustion Theory, Addison-Wesley Publishing Co.,
pp. 47-56 (1965).
6.
B. Lewis and G.
Von Elbe, Combustion, Flames and Explosions of Gasses, 2nd Ed., Academic Press (1961).
7.
H. Hiroyasu, et al., " Evaporation of a Slagle Droplet at Elevated Pressures and Temperatures (Experimental Study), "Trans. Japan Soc.
Mech. Engrs. 40, 3147 (1974).
8.
R. Dennis, ed., Handbook on Aerosols, Chapter 2, Tech. Inf. Center, USE RDA (1976).
9.
N.
P. Fenger, " Dimensional Analysis of Several Atomizers," Engin -
eering (Dec. 1976).
10.
D.
K.
Hugel and D. H. Huang, Design of Liquid Propellant Rocket Engines, NASA SP-125, p. 128 (1971).
11.
D. T.
Harrje and F. H. Reardon, Liquid Propellant Rocket Combustion Instability, NASA SP-194, p. 47 (1972).
12.
N. A. Fuchs, The Mechanics of Aerosols, MacMillan Co.,
N.Y.,
Chapter 7 (1964).
13.
G.
C.
Lindauer, A. W. Cas tleman, " Behavior of Acrosols Undergoing Brownian Coagulation and Gravitational Settling in Closed Systems,"
Aerosol Science 2, 85-91 (1971).
14.
G.
D. Kinzer and W.
E. Cobb, " Laboratory Measurements and Analysis of the Growth and Collection Efficiency of Cloud Droplets," J. of Meteorology 15, 138-148 (1958).
15.
G. C. Lindauer and A. W. Castleman, Jr., "The Importance of Gravi-tational Coagulation on the Settling of High-Mass Density Aerosols,"
Nuclear Science and Engineering 42, 58-63 (1970).
16.
V. M. Voloshchuk and T.
S. Sedunov, Eds., Hydrodynamics and Thermo-dynamics of Aerosols, John Wiley & Sons, N.Y.
(1973).
l l
I 105 l
l
REFERENCES (cont.)
17.
A. Kovetz and B. Olund, "The Effect of Coalescence and Condensation on Rain Formation in a Cloud of Finite Vertical Extent," J. of the Atmospheric Sciences 26, 1060-1065 (1969).
18.
F. Tolfo, "A Simplified Model of Aerosol Coac".iation," J. of Aerosol Science 8, 9-19 (1977).
19.
R. Gunn and G.
D.
Kinza, "The Terminal Velocity of Fall for Water Droplets in Stagnant Air," J. of Meteorology 6, 243 (1949).
20.
V. Shafrir and M. Neiberger, " Collision Efficiency of Two Spheres Falling in a Viscous Medium," J. Geophysical Res. 68, 4141-4147 (1963).
21.
A.
A. Borisov, B.
E. Gel'fand, S. A. Gubin and S. M. Kogardo, "Ef-feet of Inert Solid Particles on Detonation of a Combustible Mix-ture," Combustion, Explosion and Shock Wavas 11, 909-914,-(in Russian, Nov. -D e c. 1975 ; in English, Dec. 1976T.
22.
T.
H.
Pierce, " Blast-Wave Propagation in a Spray," J. Fluid Mech.
88, 641-657 (1978).
106
3.0 HALON (CHEMICAL) INERTING 3.1 Introduction and Discussion This sectio;. discusses the feasibility of installing.a Halon-addi-tion system as a hydrogen control measure in a nuclear plant containment a
building.
The concept is to chemically inert the containment air, i.e.,
prevent a hydrogen burn, at some time af ter a serious accident begins
~
but before significant quantities of hydrogen are generated.
Fortu-nately, most of the issues concerning a Halon 1301 (CF Br) system in a 3
nuclear containment have already been examined experimentally in a pro-g ram ( 1, 2) conducted for the U.S. Maritime Administration.
Answers to specific NRC questions will be addressed in the sections which follow a
this introduction.
is well known that CF Br extinguishes combustion by means of It 3
chemical, rather than thermal (heat removal), processes.
A physical picture is that an ignition source produces an incipient flame which is then chemically quenched by free radical removal.
However, detailed interpretations of the exact chemical mechanisms remain somewhat controversial.(3,4)
Bromine is generally thought to play the key role in combustion inhibition by CF Br.
For 112:02 flames, three reactions 3
sill directly inhibit combustion:
11 + HBr --- H2 + Br O + HBr -*- OH + Br O!! + HBr -- - H O + Br 2
This direct inhibition results from the replacement of highly reactive radicals ( !!,
O, OH) by less reactive bromine atoms.
Direct inhibition will not completely explain the magnitude of the phenomenon, however, and it is necessary to consider additional reactions that maintain the
" radical pool" at more or less its original overall size.(3)
For H2:0 2 flames, the reactions that contribute to this " regenerative inhibi-tion"(4) are:
107
Br + HO2
-HBr + O2 Br + H 02 2 --* HBr + HO2 Br + Br + M
- Br2 + M HBr + M Br + H + M where M represents any gas-phase species.
One final point should be mentioned before we begin to address spe-cific questions.
At Halon concentrations below that required to inert a combustible gas mixture, the presence of Halon may not be beneficial at all and could be detrimental.
Macek(5) studied the ef fect of additives on detonations of H :02 mixtures.
The presence of 2-4 vol. % Halon 1301 2
actually decreased the minimum ignition energy in such mixtures. Johnson, Furno, and Kuchta(6) observed the combustion-inhibiting effect of Halon 1301 on CH4: air mix'tures.
While 3.6 vol. % of the 1301 rendered the mix-tures nonflammable, 3.4 vol. % of 1301 had no beneficial effect in terms of decreasing the peak combustion pressure.
McHale(1) observed a simi-lar behavior with H2:02:N2:1301 mixtures when he reduced the 1301 con-centration below that required to inert.
Furthermore, if combustion oc-
. curs in the presence of non-inerting quantities af Halon, the Halon will thermally decompose into very corrosive halogens or halogen acids, which could adversely affect the plant and safety systems.
The conclusion to 4
be drawn from these studies is that the Halon conc atration must be maintained above that required to inert at all times and in all loca-tions.
108
e
?
1,
3.2 Answers'to Questions 1)
How much Halon is required to prevent deflagration of i
f.
H;tair: steam mixtures?
Several researchers have-answered this question experimentally for 6
~
Halon 1301.
The concentration of Halon 1301. required to inert H2: air i
mixtures varies with the H2 concentration.' Maximum inerting concentra-4 tions occur for a H2 concentration of about 14-17 vol. % in a mixture of I
H2: air:1301. ' Bajpai and Wagner (731used a vertical Pyrex glass tube of 5.8-cm diameter and 127-cm length to measure H2: air:1301 flammability
~'
limits at 25'C and 760 mm Hg.
Their ignition ^ source was an electrical-i spark (energy unspecified) between needle-point electrodes at the bottom-of the tube.
The criterion for inerting limits was based on' flame propa-j gation (determined by thermocouple response):less than-25% of the tube--
height.
Their result for peak Halon inerting concentrations wasL28 vol.
t i
% Halon 1301 and 32.5 vol. % Halon.1211 (CF ClBr).
2 L
I' McHale(l) examined Halon 1301'inerting limits in a similar labora-tory-scale device and'then confirmed his results with a few tests in in-I termediate-scale (S.6 ft3 volume) and large-scale (1200 ft3 volume) ves-1 sels.
The laboratory-scale device was a vertical Pyrex-tube (5.08-cm-diameter and 122-cm length) with ignition at the lower end by either an electric are or an electric squib (energy not specified for either
-source).
McHale's criterion for the inerting limit was'for flame propa-I gation (apparently determined by thermocouple response) less than 50% of His result for peak Uslon 1301 concentration required l
the tube height.
[
to inert H2: air mixtures at 70*F and 1 atmosphere was 23 vol. % with spark ignition and 24 vol. % with squib ignition (see Figure 1).
I McHale also examined the effects of steam addition, increased pres-sure, and increased temperature (see Table I).
His results for peak N :02 (with Halon concentration required to inert mixtures of H :1301:
2 2
i i
N2=O2) were:
a) the addition of steam to the mixture lowers the l-109 i
l
O 1301
/
\\
\\
ONo Explos on 80 i
OExplosion Propagated
[
O Partial Proyagation
\\/
_\\
/'~;e:g\\gom,,>,,o.
/\\j/
'xJM
/\\/VVNyV\\
/VVVVW/\\
/\\/V\\/VN/\\ \\/\\
/V\\/\\/VNYxf\\( /\\
/\\/N/\\LVYM VN' N H
Aa.r 2
Figure 1.
Explosion Limits, I!2: Air:1301 (From Reference 1) 110
Table I Summary of Experimental Studies of Halon 1301 Inerting 0-Peak Halon Concentration Gas Ignition-to Inert Mixture Conditiosas Source (Vol. %)
Reference H2: air No steam Spark 28
~7 25'C, 1 atm.
H2: air No steam Spark 23 1
21*C, 1 atm.
l II2: air No steam Squib 24 1
21*C, 1 atm.
r II2:02:N2: steam Saturated steam Squib 51-1 N2=O2 4 9 ' C,,
1 atm.
II2:02:N2 No steam Spark 54 1
N2=O2 21*C, 1 atm.
II2:02:N2 No steam Squib 58
-1 N2=O2 21*C, 1 atm.
5 i
H2:02:N2 No steam Squib 58
.1 N2=O2 49'C, 1 atm.
II2:02:N2 No steam Squib 63 1
N2=O2 49 C, 3 atm.
t e
n i
111
required 1301 concentration; b) changes in temperature (70-120*F) have no measurable effect; and c) increased pressure (1-3 atm) raises the re-quired 1301 concentation.
Extrapolation of these data to mixtures of H2:1301: air leads one to the conclusion that the peak Halon concentra-tion required to inert would:
be unaf fected by temperature over the range 70-120*F (21-49'C); decrease by about 1 vol. % with low temper-ature saturated steam present; increase by about 1 vol. % for increased initial pressure up to 3 atmospheres.
The National Fire Protection Association requires 31.4 vol. % Halon 1301 to inert H2: air mixtures (NFPA-12A).
This value includes a safety margin of approximately 10-20%.
Based upon the experimental data dis-cussed above, the 31.4 vol. % requirement should prevent deflagration of any H2: air: steam mixtur'e over the lower range
- of temperatures, pres-sures, and ignition sources to be expected in a containment building following an accident.
The actual quantity of Halon 1301 required to inert the Sequoyah containment can be calculated as follows.
The number of moles of air 6
inside Sequoyah at 120'F (49'C) and 14.7 psia is 1.3 x 10.
If the peak 1301 concentration (31.4 vol. %) is required when the H2 concentration is 17 vol.
then the initial quantity of air represents 51.6 vol. % in the final mixture of H2: air:1301.
This requirement means that the final mixture has 8 x 105 moles (264,000 lbs) of Halon 1301 and 4 x 105 moles (1800 lbs) of hydrogen. This quantity of 1301 corresponds to a partial pressure of about 9 psia at 120*F.
In the course of his investigation into the suitability of using Halon 1301 in a' nuclear plant, McHale discovered two mechanisms whereby
- The experimental data do not cover the full range of temperatures and steam concentrations that might be expected in a serious accident.
Neither parameter would be expectd to significantly increase the re-quired Halon inerting concentration, but this should be confirmed ex-perimentally.
i 112
1301 could be removed from the containment gas.(1,2)
The first process is solution of the 1301 in the water in containment, and the second is radiolytic decomposition of the dissolved 1301.
In order to calculate the quantity of 1301 lost by these mechanisms, we need to know the qcan-tity of water inside containment.
The maximum amount of water available 9
to dissolve 1301 is 3.8 x 106 kg (8.4 x 106 lbs) at Sequoyah. ( 8)
The solubility of 1301 in water at 120*F (49'C) is 8 x 10-6 kg 1301/
kg H 0/ psi 1301 gas.
For 9 psi of 1301 in Sequoyah, this is a negligible 2
quantity equal to 270 kg (600 lbs) or 0.2% of the total 1301.
McHale(2) determined that radiolytic decomposition of the 1301: water solution could attain a maximum level of 5.2 x 10-3 moles 1301/ liter water.
Radiolytic decomposition would remove less than 30,00 kg (6500 lbs) or 2% of the total 1301 from the containment atmosphere.
Adding the maximum losses of 1301 to the desired quantity in the gas phase, we determine that 1.2 x 105 kg (271,000 lbs) of Halon 1301 should inert the Sequoyah containment with about a 10% safety margin for low temperatures and steam fractions.
2)
How much Halon is required to prevent detonation of H2*
air: steam mixtures?
The detonation limits of H2 in air are 18 (lower) and 59 (upper) vol.
The lower limit of 18 vol. % H2 requires a Halon inerting con-centration that is slightly less ( ~ 1 vol. %) than the peak Halon concen-trations discussed in the answer to Question 1.
In other words, 23-27 vol. % 1301 should prevent a deflagration in a detonable H2: air mixture.
If follows that a detonation must also be prevented by a Halon concen-tration that is.less than or equal to 27 vol.
Including a 10% safety margin, we conclude that 30 vol. % Halon 1301 should prevent the detona-tion of any H2: air: steam mixture in a post-accident containment environ-ment.
The actual quantity of Halon 1301 needed-t'o prevent detonation in the Sequoyah containment is slightly less than that calculated in answer-ing Question 1.
Based on the discussion in the introduction, it would 113
i not be productive to determine the minimum amount of Halon required to prevent detonation.
Since that minimum would not prevent deflagrations, all the detrimental aspects _of Halon in non-inerting concentrations would ensue.
3)
Will thermal recombiners operating on a II2: air: steam:
i Halon mixture produce halogens or halogen acids in quantities likely to adversely affect stainless steel?
If state-of-the-art hydrogen recombiners could burn the mixture de-scribed above, then they would produce halogens and halogen acids in suf-j ficient quantities to adversely af fect stainless steel and virtually all other materials inside containment (see Question 5).
However, if the Halon concentration in the mixture remains above the inerting level, l
none of the mixture will burn.
Current state-of-the-art hydrogen re-combiners are designed to operate on containment gas with a content of i
2.I9)
Containment gas outside these de-i f 4 vol. %H2 and b 5 vol. %O sign concentrations can be treated by recirculating the burned ef fluent or by adding 02 - but this could lead to the adverse situation described in the first sentence above.
Obviously, Halon-inerted containment gas must be treated in a special way and this will be discussed in the an-swer to Question 4.
1 4)
How should a containment filled with H2: air: steams i
Halon: fission-product mixture be handled after an accident?
As indicated in the response to Question 3, state-of-the-art hydro-gen recombiners cannot handle the mixture described above without caus-t ing significant problems.
There may be a number of methods to safely de-i fuse Halon-iner :ed containment gas.
One of the most obvious is to modify existing recombiners such that they remove the Halon before the contain-ment gas mixture enters the burn region.
Cryogenic traps or chemical l
~~ ~ 114
T i
getters could be installed at the recombiner inlet.
In this way, the hydrogen could be safety burned with oxygen - without the fear of -ther-mally or chemically decomposing the Halon and producing copious amounts i
l of halogens and halogen acids.
l We contacted technical representatives of three firms that manufac-I j.
ture hydrog a recombiners.(10)
These people were briefed as to the prob-lems associated with recombiner performance in a containment filled with a II2: air : steam:Halon: fission-product mixture.
Then we discussed two spe-l cific questions:
1)
Can recombiners be modified to remove Halons before the mixture enters the recombination region?; and 2)
Can recombiners be 2
All of modified to treat mixtures containing as much as 20-30 vol. % H ?
the people contacted indicated a definite "yes" in reply to Question 2, l
but none of them had enough knowledge of Halons to answer Question 1 de-finitively.
Clearly, some engineering development work will be required in order to modify existing recombiners so that they can safely treat a Halon-inerted containment atmosphere.
5)
Are chemical rections between Halon and~the post-accident atmosphere likely to produce halogens or halogen acids in quantities likely to adversely af-fect stainless steel?
I Chemical reactions between Halon and the post-accident containment gas would not be expected to produce significant quantities of halogen L
Tests of the high tempera-products except at elevated temperatures.
ture stability of Halon 1301 have been conducted using sealed tubes. con-taining liquid 1301 and either stainless steel 316 or mild steel.(ll) i When these sealed tubes were maintained at 600*F (316*C) for 25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br />, the metal corrosion penetration rates were 7.6 x 10-3 and 14.4 x 10-3 in-ches/ year for stainless and mild steels, respectively.
Similar-tests (ll) were conducted with a variety of common metals, at two moisture levels (2 and 72 ppm), for a duration of 44 months, and at temperatures between i
I i
115 i.
room temperature and 250*F (121*C).
These tests revealed very Icw cor-rosion rates.
Common elastomers and plastics, sealed with liquid 1301 at room temperature for a period of two weeks, appear to be compatible with Halon 1301, except for silicone rubber, ethyl cellulose, and possi-bly cellulose acetate /butyrate.(ll)
At temperatures above 950*F (510*C) Halon 1301 will thermally de-compose.
In the presence of water, HBr'and HF acids will be formed as well as small quantities of Br2 and carbonyl halides (COF2 and COBr2)*
If the concentration of Halon 1301 were to be below the inerting concen-tration and a burn were to start in a H2: air:Halon mixture, the Halon would participate in exothermic reactions with H2 and O2 to produce halo-gens and halogen acids.
As indicated in the answer to Question 1, Halon 1301 will radiolyti-cally decompose when dissolved in water.
The decomposition products of concern are HF and HBr acids.
McHale(2) found that a solution that was buf fered to a pH of 10.8 would attain a saturat'on pH of about 3 and an unbuffered solution would attain a saturatio-pH of about 2.3 (5.2 x 10-3 molar HBr, with HF ionization suppressed).
These acidic solutions would cause significant stress-corrosion of steel, zirconium, and other materi-als(12,13) if the solutions were not neutralized la a timely manner.
The actual decrease in solution pH follows an exponential function of the product P*R, where P is the Halon 1301 gas pressure in psi and R is the integrated radiation dose absorbed by the solution in Mrad.
For the Sequoyah containment P is about 9 psi.
The solution would be capable of significant corrosion (pH f 4) after a radiation dose of approximately 10 Mrad.
In order to neutralize the pH of the solution it would require the addition of a strong base to the containment water.
The maximum quantity of base would be equal to the maximum HBr concentration, 5.2 x 10-3 mol/ liter.
Using NaOH in the Sequoyah water, the neutralization would require about 800 kg (210 ppm by weight).
116
This problem of Halon decomposition is probably the most difficult technical issue faced by Halon inerting systems for nuclear plant con-tainments.
6)
Are there adverse effects which result from the energy absorption on the addition, expansion, or evaporation of liquid Halon?
The heat of vaporization of Halon 1301 is roughly 3000 cal /mol (35.5 Btu /lb) at 70*F (21*C).
We calculated earlier that 8 x 105 moles of 1301 would be required to inert the Sequoyah plant.
This means that 2.4 x 109 cal of heat must be supplied to evaporate all of the 1301.
The air in containment ( ~ 1.3 x 106 moles) cannot supply this quantity of heat Dy itself, but the solid surfaces exposed to the gas can.
A simple analysis will be developed below and the data recorded by McHale(l) will be discussed.
The surface areas of steel and concrete inside the'Sequoyah con-tainment have been estimated (8) as 3 x 105 ft2 and 1 x 105 ft2, respec-tively.
Using these surface areas and average values of heat capacity 8
and density (see Table II), we compute a thermal capacity of 2.5 x 10 cal /K-cm and 0.6 x 108 cal /K-cm for the steel and concrete surfaces, respectively.
This means that if all steel and concrete surfaces are cooled from 120*F (49'C) to 70*F (21*C) to a depth of 0.3 ca, the heat so removed will be suf ficient to evaporate all of the 1301.
If the 1301 is injected into the containment over a sufficiently long time (e.g.,
1000 seconds), the thermal dif fusion rates into steel and con-crete are sufficient to supply the quantity of heat needed (see Table II).
a 6 cal /K or 0.2 The thermal capacity of the containment air is only 7 x 10 x 109 cal for the change in temperature described above.
The net result of these considerations is the conclusion that the cooling capacity of the liquid Halon is very large but that the walls and surfaces inside containment can easily supply the necessary heat 117
h Table II Physical' Properties of Steel and Cancrete -
Inside the Sequoyah Containment 5
Parame ter Value for Steel Value for Concrete Surface area 3.3'x 105 ft2 0.9 x'105 ft2 3.1 x 108 cm2 8.4 x 107 cm2 i
4-Density 487 lb/ft3 157.5 lb/f t3 3
2.52 gm/cm3 7.80 gm/cm Ileat capacity 0.113 Btu /lb *F, 0.238 Btu /lb *F 0.113 cal /gm-K-0.238 cal /gm-K i
25 Btu /hr-ft *F 0.8 Btu /hr-ft
- F' lieat conductivity 0.103 cal /s-cm-K 3.3 x 10-3 cal /s-cm-K i
Thermal diffusion 0.117:cm /s 0.0055 cm2/3 2
coefficient 4
J Diffusion depth
.l.1 cm 10.2 cm in 10 s Diffusion depth
- 3. 4 cn.
0.7 cm in 102s 1
Diffusion depth 10.8 cm 2.3 cm in 103 3 i -
t f
d
^
m j -
118
required for total evaporation.
If the Halon-is injected over a time scale of minutes, then the containment pressure will increase during the injection.
For the example considered above, if the initial pres-sure was 14.7 psia at 120*F, then the final pressure would be about 21.9 psia at 70*F.
3 McHale(l) conducted tests of Halon 1301 injection in a 1200 f t tank.
The tank was a cylindrical vessel with a volume-to-surface ratio of about 2 ft.
McHale injected 1200 moles (400 lbs) of 1301 in about 20 minutes; the tank contained 220 moles of air initially.
The tempera-ture of the gas decreased by only 15-20*C due to the Halon evaporation.
The Sequoyah containmeat has a volume-to-surf ace ratio of about 3 f t' and consequently McHale's results are roughly applicable to Sequoyah if 1301 injection times are comparable.
His data support the contention dis-cussed above that injection of Halon will not cause adverse ef fects (in the form of reducing containment pressure below its initial value).
7)
What would be the design concept for a Halon addition system?
The standard design concept of a Halon addition system is quite simple.(14,15)
Tanks of pressurized liquid Halon are stored outside of, but close to, the hazard area. When detectors indicate a fire dan-ger or fire in progress, the Halon is discharged by manual or automatic controls.
High-flow-rate nozzles discharge the Halon as a liquid spray that rapidly mixes and vaporizes.
One possible design concept of a Halon addition system for a nu-g -
clear containment will be described now.
The Halon would be stored in bulk tanks such as an ISO tank or tank trailer, either of which will a
hold 40,000 lbs of Halon.
A minimum of seven of these would be located outside the containment building.
Each tank would feed a number of noz-zles (say 3 or 4) located inside the containment.
Commercially-available nozzles have discharge rates up to 300 lbs Halon/sec(16) so sizes would 119
be selected based upon the time required to inert the containment.
The discharge system would be actuated either manually or automatically.
Accident scenarios capable of producing significant quantities of hydro-gen would be factored into the automatic actuation system.
Since the Halon system uses high-pressure gas to inject the 1301, an electrically-y passive discharge option (requiring no electrical power; using manually-operated valves) could be designed into the system.
Detectors capable 2 and CF Br would be located throughout the containment of sensing both H 3
building.
Post-accident treatment of the containment gas would require special attention (see Questions 3 and 4) in order to safely remove both H2 and CF Br.
3 8)
What are the pros and cons to' its (Halon) use?
The pros and cons associated with using Halon to inert a nuclear plant containment building are summarized below.
Pros 1.
There is extensive experience with Halon systems includ-ing large-volume (106 ft3) applications.
2.
The ability of Halon to inert H2: air: steam mixtures is proven.
3.
Most questions for nuclear reactor application have been addressed previously in McHale's(1,2) study.
4.
Without decomposition, aH2: air:Halon: steam mixture should be stable for a long time.
5.
The Halon discharge system can be designed to be elec-C trically passive (requiring no electrical power).
i 120
k Cons 1.
Addition of the Halon will increase the gas pressure in d
conta inment (about 9 psi for Sequoyah).
~
2.
Corrosion by the decomposition products of Halons can be significant (especially at elevated temperatures and for long peri-ods of time).
3.
Treatment of the post-accident containment gas will prob-ably require a new or modifimi hydrogen recombiner.
4.
The Halon concentration must remain above the required inert level at all times or the presence of Halon in the mixture can actually be detrimental.
9)
How long would it take to install one (a Halon-addition system)?
In order to get estimates of cost and time to install, we contac-ted two major manufacturers of Halon fire-extinguisher systems:
the Ansul Company (16) and Walter Kidde and Company, Inc.(17)
Both of these firms have experience with Halon protection systems for very large 5 ft) volumes.
In particular, the Ansul Company is presently in-3 (210 stalling Halon systems in oil tanker engine rooms (0.75 x 106 ft3).
Costs for a complete Halo 1 system, including engineering, installation, detectors or recombiners), range from and all hardware (not inclading H2 S.50 - $3.00/ft3 Halon 1301 can be purchased under a GSA contract for
$2.50/lb.
The total cost of a Halon system for Sequoyah would probably be 1.5-4.5 x 106 dollars.
Both firms believe that if they used contain-o ment drawings to prefabricate modular piping components, an all-welded pipe system could be installed inside a containment'in 1-2 weeks (pos-sibly using several daily shifts of men).
Installation of all compon-er.s outside containment would probably require 1-2 months, i
121
REFERENCES FOR HALON 1.
E. T. McHale, Hydrogen Suppression Study and Testing of Halon 1301:
Phases I and II, Atlantic Research Corporation Report No.
ARC 47-5647; Maritime Administration, U.S. Department of Commerce, Contract RT-3900 (Dec. 1976).
y 2.
E. T. McHale, Hydrogen Suppression Study and Testing of Hrlon 1301:
Phase III, Atlantic Research Corporation Report Nr ARC 47-5702; Maritime Administration, U.S. Department of Commerce, Contract T-38169 (Mar. 1978).
3.
G. Dixon-Lewis, Combust. Flame 36, 1 (1979).
4.
L. A. Lovachev and L.
N. Lovachev, Combust. Sci. Tech. 19, 195 (1979).
5.
A. Macek, AIAA Journal 1, 1915-1918 (1963).
6.
A.
L.
Johnson, A.
L.
Furno, and J.,M.
Kuchta, Infrared Spectral Radiances and Explosion Properties of Inhibited Methane-Air Flames, Bu Mines RI 8246 (1977).
7.
S. N. Bajpai and J. P. Wagner, Ind. Eng. Chem., Prod. Res. Dev. 14, 54-59 (1975).
8.
B. W. Burnham (input data for the MARCH computer code), Sandia National Laboratories, private communication.
9.
M.
P. Sherman, et al., The Behavior of Hydrogen During Accidents in Light Water Reactors, S AND 80-14 95 Sandia National Laboratories, Albuquerque, NM (Aug. 1980).
10.
L.
R. Stone (Atomics International, Canoga Park, CA),J. O' Hare (Westinghouse, Pittsburgh, PA),and R.
Borello (Air Products and Chemicals, Inc., Allentown, PA), private communications (Nov. 1980).
~
11.
E. I. du Pont de Nemours and Co., Du Pont Halon 1301 Fire Extin-guishant, Report No. B-29D (1977).
12.
L.
L. Shreir (Ed.), Corrosion Volume 1 Metal / Environment Reactions, Newnes-Butterworths, London (1976).
13.
H. H. Uhlig (Ed. ), The Corrosion Handbook, John Wiley and Sons, Inc., N.Y.
(1948).
14.
J.
E. Echternacht, "Halon Extinguishing Systems Design Criteria,"
Seventy-Fifth NFPA Annual Meeting, San Francisco, CA (May 17-21, 1971).
s 15.
C.
L. Ford, Actual Specifying Engineer, 74-83 (Jan. 1972).
16.
D. Plunkett, Ansul Co., Marinette, WI, private communication (Oct.
1980).
17.
B. L. Warner, Kidde Belleville, Belleville, NJ, private communica-tion (Sept.-Oct., 1980).
122
ATTACHMENT 1:
NRC WORK SCOPE, August 25, 1980 SCOPE FOR 2-3 MONTH EFFORT ON SEQUOYAH 1 E, Investigate using existing literaturc (i.e., no experimental work is required) the effectiveness and practicality of three hydrogen control measures:
1.
Deliberate ignition 2.
Halon addition after accident initiation 4
3.
Water fog j
Objective - Provide an early assessment (prior to December 1, 1980) of the ef ficacy and practicality of these three mitigation schemes for de-l graded (still coolable - not molten) cores.
Assess whether these mitiga-1 tion schemes will, in the near term, mitigate the effects of a signifi-cant fraction of the accident scenarios that lead to the degraded mode.
Determine if installation of the mitigation system will degrade or im-
~
prove safety.
On Deliberate Ignition (based on TVA plans for a glow plug system in.
Sequoyah):
s
- What ignition strategy should be followed: on continuously; turn on for accident; or turn on at specific times?
i
- For degraded core accident scenarios (short of core melt),
will ignition avoid containment threat?
e
- Are there negative aspects to deliberate ignition vs. exist-i ing potential ignition sources?
- Calculate pressure rise for partial combustion in H / air /
2 steam mixtures and compare this with' literature data to 123
estimate completeness of combustion as a function of H2 concentration (small effort anticipated).
On Halon If Halon is added to containment early in an accident sequence:
- How much Halon is required to prevent deflagration of H /
2 air / steam mixtures?
- How much for detonation?
- Will thermal recombiners operating on a H / air / steam /Halon 2
mixture produce halogens or halogen acids in quantities likely to adversely af fect stainless _ steels?
- How should a containment filled with a H / air / steam /Halon/'
2 fission product mixture be handled after an accident?
- Are chemical reactions between Halon and post-accident at-mosphere likely to pr7 duce halogens or halogen acids in quantities likely to adversely af fect stainless steels?
- Are there adverse effects which result from the energy ab-sorption on the addition, expansion, or evaporation of liquid Halon?
- What would be the design concept for a Halon addition sys-tem?
- What are the pros and cons to its use?
c
- How long would it take to install one?
w On Water Fog
- What would the design concept of a water fog system be?
u ---
12A
I I
i
- Ilow effective would water fog be in preventing a hydrogen combustion or detonation threat to containment?-
- What water fog density and particle size are required to '
suppress containment threat as a function of'H / air /stean 2
,k ratios likely to result from dominant accident scenarios?
j
- Assess the problem of maintaining a water fog.'in post-
~
accident atmosphere.
- Discuss pros and cons.of a water fog system.
e Will water fog reduce steam concentrations and yield a more f
combustible or detonable mixture?,
- What steps are to be taken after the water fog is in place in a contaiment containing H / air / steam / fission products?
2 t
1
-- What are the final steps to recovery af ter water fog has Initially prevented or reduced the ef fects of combustion?
l
- !!ow long would it take to install one?
i,._
I I
l D
- a i
125-17C
f i
i Distribution:
J U. S. Nuclear Regulatory Commission Cathy Anderson (400 copies for R3)
Nuclear Safety Oversight Committee Division of Document Control 1133 15th Street, NW Distribution Services Branch Room 307 7920 Norfolk Avenue Washington,.DC 20555 Bethesda, MD 20014 4
Wang Lau U. S. Nuclear Regulatory Commission TVA Of fice of Nuclear Regulatory Research 400 Commerce, W9C157-CK Washington, DC 20555 Knoxville, TN 37902 Attn R. Bernero, NL5650 M. Cunningham, 3103 B. R. Bowman -(2)
S. Fabic, ll30SS Lawrence Livermore Laboratory D. A. Hoatson ll30SS (5)
L-140 C. N. Kelber, ll30SS P. O. Box 808 R. Minogue, 1130SS Livermore, CA 94550
'M.
L. Picklesimer, ll30SS R.
R.
Sherry, ll30SS Peter Cybulskis i
M. Silberberg, 1130SS Battelle Columbus Laboratories L. S. Tong, ll30SS 505 King Avenue
, Columbus, OH 43201 U. S. Nuclear Regulatory Commission Office of Nuclear Reactor Regulation Nancy B. Willoughby Washington, DC 20555 Bechtel Power Corp.
3 Attn:
V. Benaroya, P-822 15740 Shady Grove Road W.
R.. Butler, P-904 Gaithersburg, MD 20760 H. Denton, P-428 H. N. Lauben, P-1030 G. M.
Fuls J. K. Long, 268 offshore Power System A.
R. Marchese, 268 8000 Arlington Expressway J.
F. Meyer, 268 Box 8000 T. E. Murley, ll30SS Jacksonville, FL 32211 W.
C. Milstead, P-904 i'
K.
I. Parczewski, 516 P. B. Bieniarz D.
F.
Ross, 278 Energy Inc.
Z.
Rosztoczy, P-1030 300 San Mateo, NE i
L.
S. Rubenstein, P-433 Albuquerque, NM 87108 J. Shapaker, P-904 T.
P. Spels, 268 K. Niyogi C. G. Tinkler, P-904 United Engineers 30 South 17th Street Loren B. Thompson (2)
Philadelphia, PA 19101 EPRI 3412 Hillview Avenue J. J. Wilder Palo Alto, CA 94303 TVA 503 Power Building R. Tosctti Chattanooga, TN 37401 Bechtel Power Corporation P. O. Box 3965 R. G.
Zalosh
^
San Francisco, CA 94119 Factory Mutual Research
+
1151 Boston-Providence Turnpike Jack Olhoeft Norwood, MA 02062 Westinghouse Corp.
P.
O. Box 355 H. Tamm/D. Liu (2)
Pittsburgh, PA 15230 AEC, Ltd.
Whiteshell Nuclear Research
~
Robert A. Bari (2)
Establishment Brookhaven Natic'31 Laboratory Pinawa, Manitoba, Upton, NY 11973 CANADA 127
J.
E. Antill Dr. M. V. Banaschik Berkeley Nuclear Laboratory Gesellschaft fur Reaktorsicherheit Berke'.ey GL 139PB (GRS)
Gloucestershire, Postfach 101650 UNITED KINGDOM Glockengasse 2 5000 Koeln 1 A. Morici FEDERAL REPUBLIC OF GERMANY CNEN Nuclit Rome 2165 J. E. Gover
^
ITALY 2166 -
II. M. Barnett 2166 -
E. C. Neidel Eric Ahlstroem/ Wiktor Frid (2) 2510 -
D. II. Anderson I'
Swedish State Power Board 2513 J. E. Kennedy J. Shepherd El-Och vaermeteknik 2513 SWEDEN 2514 -
D. E. Mitchell 2516 -
W. G. Perkins i
Brian Tolley 2516 -
V. M. Loyola A. Narath Directorate of Research, Science 4000 A. W. Snyder and Education 4400 4
D. J. McCloskey Commission of the European Communities 4410 A. S. Benjamin 1049 Brussels 4414 BELGIUM 4414 -
W. B. Murfin
<4420 -
J. V. Walker i
liassmann (2) 4421 -
R. L. Coats Dr. M. Peehs/Dr. K.
Kraftwerk Union 4422 -
D. A. Powers M. J. Clauser Ilammerbacherstrasse 12+14 4424 4
F. Briscoe Postfach 3220 4425 D-8520 Erlangen 2 4440 -
G.
R. Otey D. Jackson FEDERAL REPUBLIC OF GERMANY 4440 4441 -
M. Berman (40)
Dr. J. P. Hosemann (2) 4441 -
A. L. Camp Gesellschcft fur Kernforschung 4441 -
R.
K. Cole, Jr.
r 75 Karlsruhe 4441 -
M. L. Corradini J. C. Cummings Postfach 3640 4441 FEDERAL REPUBLIC OF GERMANY 4441 -
J. F. Muir M. P. Sherman l
4441 Dr. H. Karwat 4441 -
G. G. Weigand.
T. E. Blejwas Technische Universitact Muenchen 4442 A. W. Dennis i
D-8046 Garching 4442 FEDERAL REPUBLIC OF GERMANY 4442 -
W. A. VonRiesemann 4443 -
D. A. Dahlgren Prof. F. Mayinger 4443 -
J. L. Sprung Technische Universitaet Hannover 4444 B. W. Burnham L. D. Buxton 3000 llannover 1 4444 FEDERAL REPUBLIC OF GERMANY 4444 -
R. K. Byers J. L. LaChance 4444 1
Dr. Werner Baukal 4444 S. L. Thompson L.
L. Bonzon Battelle-Institut E. V.
4445 i
Am Roemerhof 35 4445 -
L. O. Cropp 6000 Frankfurt am Main 90 5131 -
W. B. Benedick FEDERAL REPUBLIC OF GERMANY 5500 -
O. E. Jones
]
5510 -
D. B. Hayes E. F. Hicken/H. L. Jahn (2) 5513 M.
R. Baer GRS 5513 -
S. K. Griffiths 8046 Garching 5513 -
D. W. Larson L. W. Davison FEDERAL REPUBLIC OF GERMANY 5521 C. J. Northrup 5812 5846 -
E. K. Beauchamp B'.
F. Murphey 8300 8310 -
D. M. Shuster 128 l
t
8320 -
R.
Rinne 8331 -
R. J. Kee 8340 -
W.
Bauer 8350 -
D. Ilartley M. A. Pound e
8214 3141 -
L. J. Erickson (5) 3151 -
W. L. Garner (3)
For DOE / TIC (unlimited distribution) 3154-3 R. B. Campbell ( 25)
For NRC for distribution for NTIS 129