ML20004C309
| ML20004C309 | |
| Person / Time | |
|---|---|
| Site: | Mcguire, McGuire  |
| Issue date: | 02/28/1981 |
| From: | Hammond R, Hubbard H, Zivi S R&D ASSOCIATES |
| To: | |
| Shared Package | |
| ML20004C307 | List: |
| References | |
| RDA-TR-178700, RDA-TR-178700-0, RDA-TR-178700-002, RDA-TR-178700-2, NUDOCS 8106020579 | |
| Download: ML20004C309 (27) | |
Text
{{#Wiki_filter:_ ._
\
l ENCLOSURE 3 O
. RO A-TR-178700-002 I
' l SOME HYDROGEN CONTROL CONSIDERATIONS '
I FOR ICE-CONDENSER NUCLEAR PLANTS 4, li
+ FEBRUARY 1981 l
T a,
'I SY:
4 H. W. HUBB ARD R. P. HAMMOND
-q S. M. ZlVI J
Sponsored By: 1 LAWRENCE LIVERMORE LABORATORY . 4 P.O. Sex 808 Livermore, CA 94550 i J
-m A
7 J 3
.e
] ;p .- ! R & D ASSOCIATES 4 Post Office Sox 9695 l , j,yktg Marina del Rey,
- California 90291 4640 ADMIR ALTY WAY
- M A RIN A DEL REY
- TELEPHONE: (213) 322 1715 7
-M
\ . i t 8106020 5 79
1 I c
~
TA3LE OF CONTENTS
~
section Page 4 y I INTRODUCTION 2 II ON CONTROLLED IGNITION OF ACCUMULATED l HYDROGEN IN A STEAM-RICE ATMOSPHERE 3
- 1. Analysis and Interpretation of LLNL
~7 Hydrogen Igniter Tests 3 A Discussien of a Possible Reliability 2.
Problem for Igniters . 8 III POSSI3LE ALTERNATIVES TO THE USE OF
~" IGNITERS 12
- 1. Chemical Removal of Oxygen 12
- 2. Inerting by Dilution 17
_' al 3. Segregation of Oxygen by Foam 18 M APPENDIX A. CALCULA'.? ION METHODS FOR TABLE 1 22
-m -*, REFERENCES 26 m
2
-4 ~7 -A 7
M M 1 . e
- f _._.
I. INTRCDUCTICN During the accident at Three-Mile Island in 1979, signifi-cant quantities of hydrogen gas were produced by the reaction of the =ircenium cladding with hot steam. The Nuclear Regula-a tory Ccrmission (NRC) has been rightly concerned that reactors carrently operating should be safe if hydrogen generated dur-7 4 ing an accident were the same amount that actually was gener-ated at Three-Mile Island. The clants of most concern have t 4 been the ice-condenser plants because the volt =e of their containment is considerably smaller than standard plants. The m
+
proposal of the Tennessee Valley Authority (TVA) for coping with this problem in its Sequoyah ice-condenser plant was to 7 put in place gicw-plug igniters so that any hydrogen that is 4 evolved during an accident could be burnt before accumulating into a dangercusly large = ass. Since it was desired to install t i these igniters in the Sequoyah and other plants as quickly as possible, the NRC asked the Lawrence Livermore National Lab-
)
a oratory (LLNL) to carry out sc=e experiments on these igniters
, to delineate the region of their applicability.
4 The present report is an outgrowth of a subcontract from
, LLNL to R & D Associates (FDA) for assistance in part of this J
4 work. Section II deals with analysis and discussion of the data generated by LLNL and some possible difficulties and i 3 4 problems that may be encountered in application. Section III l discusses alternative means of coping with the hydrogen prob-
; lem and the further steps required for their assessment.
4 . 7 l I 1 .< 1 1 7 l l l l 1 - - - - - -
~. .
II. ON CCNTROLLED IGMITION OF ACCUMULATED HYDROGEN IN A STEAM-RICH ATMOSPHIRI
- 1. ANALYSIS h.7D INTERPRETATION OF LLNL HYDROGEN IGNITER TESTS
. The LLNL has tested the performance of gicw-plug igniters i in mixtures of hydrogen, air, and steam (Refs. 1 and 2). RDA has analyzed scme of the test data to locate the initial state-7 ,1 points on a ternary mixture chart and to relate the observed igniter efficacy to the state-point. An anomaly was reported '
1 in References 1 and 2 in which ignition failed to occur in a 4 mixture of marginal ignitability because it contained a sub-7 stantial quantity of steam, but which entered the assumed
'3 region of detonability as steam condensed. We studied this 7 ancmaly and its import to the general question of controlled 4
ignitien of hydrogen in a nuclear reactor containment building. In tes:s where steam was mixed with hydrogen and air, the precise location of the state-point in the traditional ternary r mixture chart was uncertain, .primarily because temperature l4 gradients in the test vessel made difficult the spe'cification
, of the steam quantity. At the suggestion of LLNL, we calcu-
, lated an effective average temperature, t,, at which the sen of the vapor pressure of ^ saturated steam plus the partial f pressure of the hydrogen and air would equal the experimentally 4
observed total pressure. Some details are given in Appendix A. [ The assumption that the steam in the mixture is effectively saturated (rather than being superheated) is believed to be
? justified because steam is supplied to the insulated vessel l6 at the rate requ__ed to sustain the desired total pressure, l
y while liquid water accumulates at the bottom of the vessel. f
. No heat is added, other than that provided by the steam. The steam is provided in pulses through nn on-off solenoid valve, frem a boiler at about 6 atm total pressure. Temperature l
i . variations (spatial) of as much as 10*C in the test vessel are l l t 3 . l
- 7
reported. Table i shows the-calculated effective mixture t . temperature for the eight initial tes: conditions considered, i l and for the final state of the minture in Test 43 where no o ignition occurred.
.] At the calculated effective temperature, t,, the specific volume of saturated steam was then obtained f cm a curve-fit to the " steam tables," and steam of this density was assumed -v to fill the entire vessel volume of 300 1. This determined the number of moles of steam, N , in the vessel . The number s
cf moles of gas, N , was determined frcm the initial pressure g 1
-1 and temperature, p and ty, shown in Table 1 as obtained in -d Reference 2. Reference 1 cites the hydrogen-air ratio tar which the number of moles of hydrogen, N.d, was calculated.
7) 4 The mole fractions NH /(Ng +Ns ) and N s /(N g 'N3 ) thus calculated
-1 are tabulated in Table 1 as F,.h and ?s respectively. Mole
_j fractions corresponding to the various initial conditions are plotted in. Figure 1. The state-points in Table 1 and Figure 1
~l essentially confirm the original estimates in Reference 1.
a The ratio of partial pressure of steam (at temcerature t e) to -
~} total pressure is also given in Table 1, and is found to be ~"
essentially the same as the mole fraction of steam in the
" mixture, r f -A References 1 and 2 provide gas sample analyses prior to ~7 and subsecuent to ignition. The fractions of hydrogen con- ~* sumed shown in Table 1 (and the percent ccmbustien in Figure 1) -1 were calculated frem the reduction in the number of moles of I hydrogen, preignition to postburn, as outlined in Appendix A.
_9 It can be seen in Figure 1 that except for Tests 34 and 43, d the extent of ecmbustion varies according to the location of the initial state relative to the ILnits of ecmbustion pre-n _ j viously postulated by shapiro and Moffette. In general, the farther inside the limit boundary, the greater the extent,of _4 In Tes: 34, the ~
; ccmbustion. Tests 34 and 43 are exceptions.
_a
-4 L 4 . ,d e
4 e T -- - -v--e .-.-+f-'-w --9---mm,=w- egy -cww ,gw- - - y9-t~ v,--- y +w-r, y p ,,-y,w-y-y y- '+-*MT-g-- '=w-a -g -w-
[ ,l ' ' i l 4 2 01 2 0 0 7 0 0 7 5 7 4 8 7 s 8 3 4 4 2 1 4 7 2 6 2 4 0 0 0 1 2 0
% 1 8
1 2 2 0 0 0 2
. 't
. t l 0 0 1 2 3 1 6 u 4 2 0 6 5 1 5 6 S.
T 3 3 6 . 1 3 1 X i 4 1 0 3 3 2 0 0 0 0 % I f MO 2 9 5 6 0 2 0 6 6 5 I 0 3 9 5 4 1 4 7 t T 7 2 6 3 . 0 . i 4 0 HI 3 1 1 2 0 0 0 0 0 0 CS 5
. OO 9 5 9 9 0 9 6 8 t
i I' 9 1 1 5 0 3 1 3 0 9 DM 3 8 2 4 8 . 0 2 1 9 0 4 YO 0 1 1 0 0 0 0 e l i C r
- 4 4 5
! t D 8 2 5 1 8 8 8 8 2 u l
8 9 7 5 3 0 3 0 4 t l l t 3 1 3 9 x AA 1 1 1 0 0 0 0 0 i 0 1 m MS 3 c AE 8 0 4 0 9 i J.. El t 7 7 0 1 1 3 2 1 1 3 1 9 r Tu 2 9 8 2 1 Z . x t 4 ST 2 1 1 8 0 0 0 6 0 e A 0 1 m l l l 2 i o iTE l I' 7 2 6 6 0 0 0 3 4 h c 3 6 7 J i l M 6 3 3 5 1 0 9 3 1 3 0 5 i o E 2 1 . 1 7 0 0 0 0 t
- ST 0 1 0 S
_ T 7 " t
. SE 2 8 2 0 8 0 8 9 6 6 ER 5 9 7 6
1 4 4 3 0 2 0 4 J TU 3 1 1 8 0 0 0 0 0 _ T 0 1 ._ LX l t I f. I M # t, L ~ . E t RV s s 4 OI e n ) a T g
. FT o F e
C i ( r y J SE t i u r l t F n ) e s d OF g e r s 1 I E ) i g u e l
)
T e u t r a I D g e a a p u F L. DE u r g r d ( l T a o e p l a g i s d r g f e - O A. e n m t n e e J CI ) b i e t o i n r mu r u U n F / t n s - L. D n i * , b / e r u t a n EIC. o n e i n
/
b ( o l ( r r b r o l l A i t I e i u u 1 2 c e p n t i uC c ( r t e t s d
- n. S e u o c r x s n 1
n 2 m o c AD j e t t i e u i n e a o 1l e i a r . A EI n r a j s m :s e g p r n i f t t c f t i u r a n s o o e a l iA s s p e r i e e r v o a r m n m o i t c v r ne m a n e i f 4 a e m e m p i i e d a y e t r o t t g
. . e r e p t l
o a e t t c c s h t t i n f i a g m o J 1 t s m t l a ef a s g t t e a r s s s t r , , i e c s f e d y e a a r o f f 3 g g , l a f t E. r g g i r t e 'F F j r o r l a e s h
- h. i l
o a e e P e mf n , , , l
, ,/ t , , l / t f e 3 g A g I 2 f t eo S f , , i J
T !
!c p L ll A p t H P A S C F t F F 6
i y . 1 e I .j k f ' :! ; i j, .f3; !l )jl i. ]i ! :I ! !,
100% air S
\
\
* \
g No. 35, 46% combustion
, No.36,54%
\ \\, No. 27, 99%
Shapiro and \ No. 38, 42% 1 Moffette No. 39, 60%
! s
" assumed No. 40, 98%
detenatien"
**** .( '\
No. 43, ~ 0% N \ / .__ - l\ ,<\ No. 34, - 0%
, / v ,
W N
/- /
1 n.. b= ~3~f ~~\7 \ m Shapiro and Mofferte combustion limits 1 1M 100%
-4 steam hydrogen .
MI d Figure 1. Initial state points of LLNL test : nix '9- for tests with stea:2, including the nonburning tra .setory of test no. 43. . a I _d O ,- ,. - - - > - - . . -- - - . , - - , , , - - - - - - - - - - - , - - - , - - - - - - ~ -
igniter was "on" continuously, while steam condensed and the-state-point migrated toward a drier condition. No ecchus-tion was detected until the end of the test when the circu-lating fan was activated. Cembustion at tha' "4"a prcduced a
- 2 srall pressure rise (%1 lb/in ) in the vessel. The gas analysis was sc=ewhat anomalous (Ref. 2 ) ', so that interpretatien of , Test 34 is impaired. In Test 4'3, no ecmbustien occurred, and s the data are sufficient to permit the trajectory to be plotted in Figure 1. The final point of the trajectory is defined in . Table 1 and was determined with the total cressure and the initial composition. ; That ignition did not occur at the initial state-points of Tests 34 and 43 is censistent with the expectation pro-s vifed by the Shapiro and Moffette combustion limit (Ref. 3) in the regien of high steam centent. However, we can be con-i fident of the trajectory of the Tes 43 mixture (since there are no significant ancmalies in the measurements), and this trajectory traversed well inside the state-points of Tests 35 36, 27, 38, 39, and 40. Furthermore, the mixture of Test 43 actually reached the " assumed detonation tone" without ecm- . bustion. Although the cause of the failure to dgnite is still . conjectural, the failure to ignite might be the. most important experimental finding of the te. . program, especially if a con-tainment atmosphere could have similar conditions.
At present, it is our conjecture that a dense condensation fog existed in Tests 34 and 43, but not in the other tests (for reasons unknown). Many consider fog to be an effective suppresser of ignition of hydrogen, and this is being studied as a possible safety measure. Of all the tests, 34 and 43 had the greatest quantity of steam present, and a fog is more i likely. LLNL reports observations of optical obscuration in nenburning experiments subsequent to Test 43 with 50 percent 4
+ ~
s. w m,--- ,. , , . -
- arm concentrations. Also, the method of injecting steam
.s cualitatively consistent with the generation of aerosol nuclei. Whether or not a fog was responsible in. Tests 34 and 43, the conjecture points to a paradox in the simultaneous '? consideration of igniters and fog generators as devices for . 4 4 coping with accumulations of hydrogen in reactor containments. , This paradox is discussed in the next section.
I
- 2. DISCUS $!CN OF'A POSSIBLE RELIA 3ILITY PROBLEM'FOR IGNITERS l In Reference 4, it is recognized that a fog =ay occur in
+ ,
a post-LOCA atmosphere in a containment building. In an ice-
' condenser containment, conditions appear especially conducive 1
i for the generation of fog by bulk condensation of moisture
, in saturated air (i.e., growth of condensation nuclei in the l bulk fluid) as the warm saturated air flows through the ice compartment. If fog does indeed suppress. ignition, there 1
' would appear to be an important set of conditions for which n
igniters in a containment building would not be effective, . and might be countereffective by introducing a spurious hope or distraction.. 7 There is a possibility that while an accident-caused fog 4 might interfere with intentional ignition of hydrogen, the i fog =ight be short-lived, thereby permitting the igniters to t be used when needed. One postulated mechanism for fog removal
? is the contain=ent spray, which could scavenge fog particles.
! # For a spray-water volumetric flow rate , and with the spray 9 4 In those optical measurements, the dense fog abruptly cleared as the steam concentration diminished to between 40 percent
~
l and 30 percent, which leaves a mystery as to why ignition failed to occur in Test 43 as the mixture approached its~ final ', state of 27 percent steam. The answer may lie in the possi-
' bility that although the optical length of the fog had increased in the post-43 optical measurements, there was still a suffi-7 .
cient concentration of fog to suppress ignition. 1 a 3 Q e m===- 3_g n , --.e- ,,- --.-----n-, ~~. . - - - - . ~ --,-.v, , , , , - - , - - , - - - - ,n,. ,- w --
4
. fr:plets having a uniform radius R and terminal velocity C, the volume of spray water falling through height H at any instant would be:
9 l V s
=HW U
{ The number of falling dr0plets ND "U * ** # - * "' at which this spray would " sweep through" the contained i, atmosehere would be: 4 V
-t s N =
D 3 47R 3 N
; D = nu=her o f droplets i of radius a I
Y = NO na U I Y = volur.etric displace-T ~ ment rate of falling 1 droplets 1 vs UV s Y = UR2U =- 3 i 3 4 R _4.:R T. 4 Let the relevant containment compartment volume be V . Then 9 e i 4 the fractional rate of " sweeping" of the containment ecmpart-ment volume would be
-.3_W V
Y 4 RV s
; C C 3
If the fog-particle collection efficiency is c, and the particle i concentration is y, the rate of removal of fcg particles is d- Y y cV C c. 9 . 4
cr tv - 3c s Y=YO **P
~ lI RV c .
The fog-clearing effectiveness of the containment spray is indicated by the =agnitude of the decay constant: the larger the constant, the more effective the clearing. The relaxa-tien er e-folding time for fog clearing is: 4KV c T *
- 3 GV sc 5
Consider a total contai==cnt spray-water flow rate of 3 x 10 cm3 7, a droplet radius of 0.1 cm, a terminal velocity of 10 cm/s, 10 3 and a containment cc=part=ent volume of 2 x 10 cm . These
' numerical values are believed approxi=ately valid for the
. Sequcyah containment systems. For this example, and for ti=e i T in seconds,
; (4) (0.1) (2x1010) 1 3 J c
. 3 x 3 x 10 x 10 6
. = _80-=-3 1 ~ 10 9 c c
~
( l Depending on how many exponential factors of reduction are l* recuired to bring the fog concentratien down to an ignitable A relaxation l-1 level, that many : intervals are required. time of a few minutes is precably acceptable, whereas a time of a few hours is probably not. For the particular assump-l ticas-used'in this example, a value cf c less than about ! 3 x 10' would be cause for doubt that the contain=ent spray l could clear sufficient fog prior to intentienal hydrogen s ' 10 . t l
~
I __. l .- . -- - . . . . . . . . -- ._.- __. , _ , _ _, , ,, . , _ . . . , , , . _ _ . . , _ . _ _ , . _ . __
ignition. The value of c depends upon the fog particle si:e and the spray droplet si:e, and magnitudes as small as 10 ~3
-2 and 10 are reported in the literature on aerosol coagulation.
Further work to define the conditions would be necessary before g there could be confidence that the containment sprays could 4 assist igniters by clearing whatever fog that might exist. I Other mechanisms might assist in reducing a fog concen-tration promptly. Particle agglomeration by Brownian motion does not appear very helpful, however, because for particle i 5 -3 concentrations en the order of 10 cm (which would be quite
, dense, optically) the particle half-life would be many hours.
" Plating out" on cold surfaces (e. g. , ice) would be of scme c
help, but a quantitative assessment does not appear easy. j Turbulence in the air-steam mixture would also provide some increase in aggicmeration rate, but would again be difficult
} to assess.
4 At present, we consider the reliability of igniters to be in doubt because of' (1) the experience of Tests 34 and f 43, (2) the above arguments on fog coagulation by contain-
- ment sprays, and (3) an apparent lack of other reliable meth-k cds for prompt reduction of condensation fog.
1 4
'1, 3
4
- i
'1 1
- l. .
e 11 O b
III. POSSISLE ALTERNAT!7ES TO THE USE OF IGM!TERS Centrol of oxygen availability is a positive method of hydrogen control in reactor containment buildings. It has
, received little censideration so far because experience with fully inerted buildings has shown many di::iculties, higher costs, and possibly impaired overall safety frem insufficient s
access for maintenance. However, the possibility exists of inerting the containment only upon demand, after an accident has occurred. Such methods would offer the highest prospects i for effective hydrogen control with minimum cost and incon-venience. , Three different approaches are possible for rapid reduc-tion of ecmbustibility: (1) use up the oxygen chemically, (2) dilute the exygen with another substance, and (3) segre-7 gate or confine the oxygen so that it is not available for I ccmbustion. Described below are exa=ples of each type which
; appear to meet the requirements and which merit fuller inves-2 tigation. '} s'
- 1. CHEMICAL REMOVAL OF OXYGEN
- a. Gas turbine combustor system--Hydrogen control in a l14 reactor contain=ent building through rapid depletion of the oxygen in the building atmosphere is a completely practical
, 1 cperation that can be carried out using commercially avail-l1 , able equipment and simple, automatic controls. The equipment 3 required consists of a small gas-turbine engine generator set, II appropriate ductwork, a supply of fuel, a spray tcwer, a heac 1 exchanger, and two pumps with piping. When started, the sys-1 ( tem would draw the containment atmosphere (which under acci-dent conditions consists of air, radioactive gases, and pos-l9t sibly steam and hydrogen) through the gas-turbine combuster where a fuel such as kerosene or natural gas would be ignited. 1s l The turbine would provide pcwer for circulating the htmosphere l, 12
?
a L l l ----_ l
and driving the pumps. Additional fuel would be burned in an sf:erburner to further depletc the exygen. Any hydrogen
*- in tne inle:. stream would be burned in the flame and assist in depleting the oxygen. The exhaust stream would be cooled rapidly in a spray tcwer and then discharged in the contain-ment so ac to minimize recirculation to the inlet.
The size and ca;acity of this system can be chosen to i suit any likely set of requirements for rate of oxygen removal. The example considered here is capable of reducing the oxygen v concentration by about 75 percent in 30 min in a building the size of the Sequoyah containment. The engine would continue-to run until the inlet gas no longer supports ccmbustion, i whereupon it would stop by itself. During operation, the heat of ecmbustion produced must be removed to the curside of the containment to avoid centributing to the overpressure , that =ay exist from release of primary peactor water and. stored i core heat. This heat must be removed without releasing any radicactive materials. Y 3 The heat quantities involved are substantial: for the e Sequoyah plant, to pass all the containment air through the ' turbine in3 30 min requires a throughput of 2,000,000 ft3 /h, or 550 ft /s. Complete depletion of the oxygen in one pass l would require burning fuel at the rate cf 2.9 lb/s for kero-sene and 2.5 lb/s for natural gas. Assuming on the average that only 75 percent of the oxygen is censu=ed, which seems readily attainable in practice, the fuel consumption is 2.2 ,1b/s for liquid fuel, producing heat at the rate of 49.5 MW, or 1.87 lb/s and 52.2 MW for natural gas. If hyd:cgen is present in the containment atmosphere, the engine controls will sense the higher heat input and reduce fuel flew accord-ingly. Thus the energy developed in consuming the oxygen will re=ain about the same. 13 ' 9 %"'- wg emmis e .
During this time the reactor core may be producing decay heat at the rate of 40 MW cr so, and if hydrogen is formed there is still more heat frcm the zirconium-steam reaction, as well as the energy in the pressurized het water released in
] he escape of the primary reactor fluid. In this illustration it is assumed that the ice condenser, the contain=ent ecoling sprays, and other e=ergency systems are capable of handling
'I these heat releases, but'nct the gas-turbine cc=bustion energy. 7' Therefore, the description that folicws treats the oxygen i depletica system as a separate, independent system with its I cwn heat rejection means. Detailed analysis may show that
- existing systems could be readily modified to accept the addi-tional cooling load.
] Alternatively, id other systems were J u deemed marginal in capacity, the gas-turbine cooling system j cculd be readily enlarged to assist in centrolling centainment '? -cressure at the mest crucial time. s .s The effect of the gas-turbine operation upon the contain-Il
- 1 ment pressure is s=all but in a favorable direction
- eperatien i of the system serves to replace oxygen in the building at=cs-
] phere with CO and H 20. For kerosene fuel the reaction is:
2
*4 CH,*l.5 o3+CO +H 2o while for natural gas it is:
l
- - 2 CH4+
g 202 - Co 2 + 2H 20. Since the water vapor formed is condensed i{ by the heat removal system, the effect of 75 percent combus-a e tion of exygen is to replace 0.2 atn of oxygen with 0.05 ats d
~
of unburned oxygen and O'.1 atm of carbon dioxide, a net reduc-tien of 5 percent in containment pressure. For methane fuel, the reductica in pressure is 7.5 percent. l
- t The presence of carbon dicxide has an important edfect 6 upon the flam= ability limits for hydrogen because it is a l
, much =cre effective diltent than nitrogen alone. Drell and a
3 3elles (Ref. 5) state that no mixture of hydrogen, air and r nitrogen can propagate a flame if it contains less than 14
4.9 percen oxygen, while mixtures of hydrogen, air and car-ben dioxide require 7.5 percent oxygen to propagate. Although it may be possible to adjust the engine to consu=e more than 75 percent of the oxygen, this value seems readily attainable 7 with realistic ducting arrangements and si=ple controls. It also seems to be an adequate value, in view of the carbon
, dioxide produced, to give effective inerting to hydrogen in I
o all concentrations. The gas turbine required for the flow-rate assumed is about 4 feet in diameter and 12 feet long, with additional
- space required for the gearing and generator set. The hot turbine-afterburner exhaust gas is cooled rapidly by passing it into a spray tower which might be 7 feet in diameter and , 30 feet tall, for exa=ple. Water is sprayed down this tower at a rate of about 800 gal / min, collected at the bottom, and l pumped through a heat exchanger and back to the spray no::les.
s The heat exchanger is required to pass the heat collected to
}
the exterior of the containment without releasing any radio-active materials, which must be assumed to have contaminated
" the interior of the building. For the assumed capacity a 1
" heat exchanger about 36 inches in diameter and 65 feet long
, would suffice, although the equipment can be arranged in other d proportions to fit the space available, and can be either l
, vertical or horizontal. A second water circuit in the shell side of the exchanger picks up the heat, passes through the ld l containment wall, and is circulated by a second pump to an
; auxiliary cooler, which might be the main cooling tower of the plant. This water circuit does not get contaminated. ,
l l Except for the ductwork needed to control the intake and 4 discharge points of the air, the system is compact in size
, and flexible in arrangement. It is described here as being located within the containment, since it will be expected to become contaminated, and containment penetrations are to be
~
avoided if possible. However, if space inside were tco
- 15 .
O 4 1 . - - - . 1
limited, the system could be located outside with scme shield-ing around it if it were permissible to bring the air ducts (30 inches in diameter) through the containment wall. The water pipes are only 6 to 8 inches in diameter. The above illustration represents the equipment needed to meet the assumed goal of producing an inert at=osphere in the containment within 30 min. The engine-generator set is the most expensive item, estimated at a little under 51 mil-Lion. The additional equipment would cost less than 50.5 mil-lion = ore, exclusive of installation, but the installation costs and the cost of plant shutdown time may tend to dwarf the equipment costs. If the time required for inerting is relaxed to 60 min, for example, all the capacities given above could be halved, and the cost reduced. The apparatus could be more ecmpact and installation might be easier. The gas-turbine inerting system is a highly practical app cach to hydrogen control, once instgiled, but the ques-tion of its fea'sibility may depend upon the extent to which it can be programmed for installation during plant operation or during scheduled shutdowns. At S600,000/ day, only two or three extra shutdown days would generate down-time costs that exceeded the equipment costs.
- b. Recembiners--The ideal way to remove hydrogen and oxygen is, of course, to cause them to react or reccmbine directly and to condense the resulting , steam. This =ethod would treat the hydrogen problem directly and simply. There l
has been extensive industry research on hydrogen recembiners, l with the first unit installed on the Los Alamos Water Boiler reactor in the early 1950s. Much of the work has been done on catalytic recombiners; however, they are subject to poisoning
- in sc=e circumstances, and the preference for ecm=ercial units is for a thermal recembine , in which a heated furnace
- l 16
' ~~~~ p
wall instead of a catalyst causes the reaction.to proceed. Complete with a blower and a cooling spray system, the recem-biner operat.es in much the same way as the jet engine system except that it introduces no additional gases to the contain-
=ent atmosphere.
Unfortunately, the largest commercially available system operates at about 150-ft /3min throughput, so it would require
}
100 h to creat the Sequoyah containment of 1,000,000 ft3 . Alternatively, 50 units could be used to make one pass in 2 h. The air stream is heated by electric heat.:rs to about
,, 1300*F, requiring 5 MW of electric input for the 50 units. ! , If the requirement was for one complete pass in 30 min, as hypothesized for the jet-engine case, 200 recombiners drawing ! 22 :M would be required, probably an impractical goal in both f cost and power sopply. ! To compare the jet engine with the recombiner, one should i
e note that if the requirement is for very rapid inerting of [f the containment, only the jet-engine s'e ems practical. If a sicwer treatment rate is permissible, either could do the
'} Jcb. Hewever, the operation of the jet engine removes oxygen from the containment whether or not there is any hydrogen present at the time, while the recombiner accomplishes nothing l
,}
,4 unless hydrogen is present in the air stream.
>l
.j 2. INERTING SY DILUTION-j An otherwise combustible mixture of air and hydrogen can j be diluted below the flammable limit by adding an inert gas r such as carbon dioxide. However, under conditions where hydrogen formation can be expected, the containment building
, would already be close to its maximum pressure, and the inert
[r gas would tax it even more. If hydrogen should reach 10 per-
;i cent by volume, for example, the building pressure would have to be nearly doubled:to produce a nonfla=mable mixture with I cO 2.
T - I I 17 J
-w a *a0 A more efficient diluent is Halon 1301, a well-known fire-l J, extinguishing agent. Halon 1301 is a member of the freon ] family of halogen-substituted mechanes. It is nontoxic and a
inert generally, but in a flame front it reacts efficiently
- with the free radicals which propagate the flame and cause ~'
it to go out. With ordinary hydrocarbon fuels as little as
-- 5 percent by volume will extinguish a flame, while 7.7 percent -- renders a highly fla=mable mixture completely inert. With hydrogen it is not as effective, requiring about 31 percent
- i. for a ecmpletely inert system. The required concentration would be 0.13 lb/ft , or 130,000 lb for the sequoyah contain-3 At $8/lb, the chemical alone would cost over 51 million,
] ment.
with approximately an equal amount for installation. r The cost of a Halen system does not seem high in ccmpari-son with the jet engine or recc=hiner syste=s, for example,
]) but the necessity of adding 30 percent to the containment pressure at a time when it may be near its operating pressure limit gives one scme pause in recommending its use for plants
{} with small containments. For a new reactor, which could be
-' designed with a beefier containment building, further study "' may show it to be one fairly convenient, sure, and Icw-cost q way to inert the containment on demand.
- 3. SEGREGATION OF OXYGEN BY FOAM The use of foam for rapidly inerting a containment build-
((] ing has not been previously considered, although it is well-Foam works in two ways to prevent
] proven for use on fires.
ccmbustion: first, it immobilires the oxygen or air, and second, it provides a persistent form of dispersed water.
-{
The first effect would be most important for containment use.
--. A foam generator uses water detergent and air to produce a 1' heavy, stable foam made up of bubbles about 2 =m in sire and containing 200 to 1000 volt =es of air per volume of liquid Shb e
S m en i o
j f water. The practice is to ccmpletely fill the room or ccm-partment to be protected. The foam-flows freely and rapidly
~l
. .< around, under and behind all obstructions during filling, but
~f once full the air is held stationary and segregated in a semi-rigid mass, so that all convection is suppressed.
a4 i To be effective, foam would have to be in place in the
} containment before much hydrogen was formed. The foam gen-erators work very rapidly, however--present installations already have the capacity to fill the Sequoyah containment in 10 min or less. Careful study and testing would be needed to determine th'e course of events when a stream of steam and hydrogen escapes into a foam-filled containment. It is cer-j- tain that the i= mediate pathway between the source of steam and the ice condenser would be swept clear of foam, and the air in that flow volume would be mixed with the hydrogen.
s A quick calculation indicated that as ,much as 90 percent of 1 the oxygen would still remain immobilized, but as was noted J oarlier in the discussion of gas turbines, removal of only n three-fourths of the air seemed to be adequate for an inert system. By properly chcosing the location of the foam generators
, s. and the connecting openings between compartments, the path-way of the hydrogen could perhaps be controlled and the point of hydrogen accumulation chosen. One possible sequence, which would have to be verified, is that steam blev-down to the ice ~
condenser would create a pathway through the foam, that the hydrogen-steam mixture would follcw during the core-dryout 3 stage, and would reach the ice with probably little additional 1 admixture of air. As the steam cendensed, one would expect , I the hydrogen to form a pocket of highly enriched gas which could slcwly rise through the foam in the upper ccmpartment to lodge on the ceiling of the deme, still without mixing. 1 P 19 _ .--,_.._.,_._._m.--. , - , _ , _ _ _ , . , - . - . - . - _ . _ , , . , . . , , , , . _ . , _ _ _ _ . . _ . , . _ , _ . . ,,_ , , - , .-
The foam mixture developed for fighting large indurtrial fires persists only for a few hours, which is desirable since fairly rapid reentry is needed. Perhaps a more. persistent form can be fcund for containment use, but at best the segre-gation of the hydrogen and air effected by the foam would not be long lasting. Scme active means would be needed to drain off or recembine the hydrogen without danger of burning or explosion. In concept, the simplest of these would be to conduct or pump the hydrogen to a compartment which has been freed frcm ignition sources and which could be isolated for the duration of the accident. It is clear that the use of foam to control containment hydrogen cannot be ascessed without =cre detailed study. In particular, the interaction of containment cooling sprays and foam is not clear. In scme fire-fighting situations, fcam and sprays are used together to produce a heavier, denser foam, yet a heavy jet of water will knock down foam. If some of these questions can be answered, the fcam system has imper-tant qualities:
- 1. The equipment is simple and well-proven. A foam generator consists of a source of water under pressure, a supply of detergent and a metering 1
device, a water turbine driving an air blower, l . a spray no::le, and a wire netting that estab-lishes the size of the bubbles. It is set in operation by turning on the water, which drives the turbine and bicwer, and sprays water and detergent into the air stream ahead of the net-ting.
- 2. It is safe. An accidental or false alarm release of foam would not be dangercus to persons trapped in it (althcugh they could not see to walk) and the fcam causes little er ne wacer damage ccm-pared to a spray system.
20 . 1 i l l l i
I- ,
- 3. It is very quick (N10 min to fill).
- 4. It is commercially available.
- 5. Indications are that it would be less costly than any other proposed system except the igniters.
e l I i l I T k . .o d c 4 o . 1> l' 9 r L i, /
?
4 21
,_w=---- r-e --a w-n , , , - - -, , n.- - - - - , - - - - - - - - . . , , - , - , - - , - --,,.-,.,,n,--,,-- --,,- ,s~ -
; s' f
9 APPENDIX A. CALCULATION METHODS FOR TABLE 1 2 .
- 1. DETERMINATION OF EFFECTIVE MIXTURE TEMPERATURE A numerical iteration was performed, varying the effective mixture temperature t, until the total pressure in E'q. (1) agreed with the observed total pressure to within an error of O.004 atm.
e PT"21Y 1 +Pg(T ,) * (1) a Capital symbols are the absolute pressures and tempe.ratures corresponding to py ,ty, and t,.
} Pg(T,) is the function defined by the " steam tables" for saturated steam. An e=pirical curve fit was used for Pg (T,) .
This relation is given in Reference 6.
=
[
- 2. DETERMINATION OF STEAM FUNCTION T
j Let F 3 represent the steam function. The specific volume of saturated steam is v(T,) (empirical curve fit from Ref-erence 6, 1/gm): Vessel volume = V il di Mass of steam = V/v(T,)
~ '
t 1t V N, = Moles of steam = {yg) vTe) - 1 i i
/y V ~1 Ng = Moles of gas = (22.4 ) (( yT273[1 (T, I
i 22 l 0 l ' , 2
r i N - T ' _S = 22.4 1 . l N g (18) v(Te) 273 P 1 3 1 i 5 N T _s. = 4.558 x 10-3 1 N p y(7 ) g 1 8 v , N S NS /N q
'S NS+N g 1 + NS/N g
- 3. EXTENT OF COMBUSTION If the final number of moles of hydrogen in the vessel is SH, the fraction of combustion C is defined as:
N -N N E H H C= =1 -- N N H H Ng is obtainable from Ng , the initial moles of hydrogen and air, since the ratio of hydrogen to air is given in Refer-ence 1. In order to cbtain 5 H frcm the final gas analysis in Reference 1, wherein the fraction of hydrogen in the final i noncondensible gas is given, it is assumed that no gas is lost (or dissolved), and that for every two moles of hydrogen removed by combustion, one mole of oxygen is also removed. Let the initial number of moles of air, hydrogen, oxygen, and nitrogen be: N,, N,N g O' N N l 23 , l
and designate the final number of moles of each by an over-scored symbol. Frem Reference 1, we know N H ' S=_ _ _ Ny+NO H where S is a mass spectrometer determination. Then S - - N H
=
1-S (NN+NO) But 53=NN and 9 O =N O - 0.5(N.-5 d H. ) Then N. d
=
1-S N
,N+N O
. 5(N2 -N3, )
l
*SS' S
~ ~
N 1 = N -
.5N H i
H . 1-S. 1-S .N+NO I l l _ g . . . l N. = - N - 1. 5 N d 1-1.2S. .g H. ( N . . .N , H S c . = - - - 1.s-
- N 1-1.35 - N t .. H -
_ d. . I 4 24 l
- -~, ,% , - - . . , ,-. ,,.%. _ . _i-_ , . . , p .. ---,.---#. , - ---.g.% (,, , - +- .ye , ,c , yg y , -,.
.. . o 4
i i
. . N !
8 C=1-
.l-1.cS-d - 1.5 I N.d .
.s S
. .Na -
C=1- - 0*5 !' l-1.5S N
.H . ,.
a t t h I I l l t
.l i
l 1 . I I i i l I 1 i I F l I ! I I 25 [
L.
, REFERENCES
- 1. Lowry, 3., Preliminary Results:
A Study of Hydrogen Ieniters, Lawrence Livermore Laboratory Report ENN 80-45, November 1980.
- 2. Private cc=munications between Zivi, S. .M. (RDA)' and Lowry, 3.
(LLNL), providing gas temperatures and pressures before injecting 23 December 1980. steam in Tests 35, 36, 27, 38, 39, 40, 43, and 43, ,
- 3. ~ napiro, Z. M., and Moffette,.T. R.,
;ata an "Hydrcgen Flammability ;
WAPD-SC-545, 1957. Application to PWR Loss-of-Coolant Accident," !
- 4. Almenas, K. K., and Marcheiro, J. M. , "The Physical State of j Post Loss-of-Coolant Accident Containment Atmospheres,"
Nuclear Technolocy, Vol. 44, August 1979. I
- 5. Drell, I. L., and Belles, F. E.,
Survey of-Hydrocen Combustion Properties, NACA Report 1383, 1958.
- 6. Keenan, J. H., and Keyes, F. G., Thermodynamic Procerties of i Steam, John Wiley and Sons, 1st ed., p. 14.
I i t i i . i I l ! I I l i I l ! l l l l i 26 i i
._ _ . . - . . _ _ _ . , . . . _}}