ML20010E655
| ML20010E655 | |
| Person / Time | |
|---|---|
| Site: | Yankee Rowe |
| Issue date: | 08/31/1981 |
| From: | Kay J YANKEE ATOMIC ELECTRIC CO. |
| To: | Eisenhut D Office of Nuclear Reactor Regulation |
| References | |
| RTR-NUREG-0737, RTR-NUREG-737, TASK-3.D.3.4, TASK-TM FYR-81-127, NUDOCS 8109080107 | |
| Download: ML20010E655 (4) | |
Text
~
r
\\
t tANKEE ATOMIC ELECTRIC COMPANY
~.
h9@/2.1 1
.m y hq 1671 Worcester Rood, framingham, Massachusetts 01701 yyg Q 1g7
. YANKEE
$.mF August 31, 1981 g
(t_ b,,
~
J.,
9:
SEi 0 41981 *-
United States Nuclear Regulatory Commission y,3, m 3,,,%
g co"2m' Washington, D. C.
20555 c
A Q Ih ;'E Attention:
Mr. Darrell G. Eisenhut, Director s/
Division of Licensing
References:
(a) License No. DPR-3 (Docket No. 50-29)
(b) YAEC Letter to USNRC, dated January 1,1981 (FYR 81-6).
(c) Letter from W. R. Stratton, A. P. Milanauskas, and D. O. Campbell to USNRC John Ahearne on the subject of iodine release during nuclear reactor accidents, dated August 14, 1980.
(d) Paper presented November 20, 1980 by M. Levenson and F. Rahe to the American Nuclear Society International Conference in Washington, D. C. entitled, " Realistic Estimates of the Consequences of Nuc' ear Accidents".
(e) NRC report entitled, " Technical Basis for Estimating Fission Product Behavior During LWR Accidents,"
NUREG-0772, June 1981.
Subject:
Proposed Control Room Emergency Filtration System for the Yankee Nuclear Power Station
Dear Sir:
In response to TMI Action Plan Item III.D.. 4, Control Room Habitability, an evaluation was performed to consider both radiological and chemical releases. The result of this evaluation (Reference (b)), indicated that a control room emergency filtration system would be required to maintain doses within the guidelines specified in NUREG-0737. The purpose of this letter is l
i to describe the filtration system we intend to install at the Yankee plant and to discuss the assumptions used in the design.
Normal ventilation for the control room is provided through a unit ventilator located in the Turbine Building. As shown on Figure 1, the emergency filtration system will be independent of the normal system and will l
contain redundant 3000 cfm filter trains. This system will recirculate air l
from the Control Room through charcoal and HEPA filters designed to reduce the thyroid dose to control room operators to meet General Design Criteria 19 assuming a conservative halogen source term. Operation of the emergency i
I filtration system will require manual action in the Control Room to first secure the normal ventilation system and then initiate the filter trains.
Indications and alarms in the Control Room will provide the necessary guidaice I
for the operator actions. Emergency power for the fans will be provided from the emergency diesel generators.
gf[
.r
//
8109080107 810831 PDR ADOCK 05000029 P
United States Nuclear Regulatory Commission August 31, 1981 Attention:
Mr. Darrell G. Eisenhut, Director Page 2 i
The assumptions used to evaluate the proposed filter system are listed below:
1.
A source term of 50 % of the iodine core inventory released to the containment atmosphere with subsequent natural removal with an effective removal constant of 2.5 hr-1 (Time for concentration to reduce by ane-half is equal to 17 minutes.). This mechanism continues until the airborne halogen concentration in containment has reduced by a factor of approximately 1000.
2.
It is assumed that a time period of 20 minutes exists between the initisl indications of system parameters which could lead to a degraded core condition and the actual release of 100 % of the noble gases and 50 % of the halogens to the containment atmosphere.
3.
The containment leak rate is 0.2 % per day for the first 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> and 0.1 % per day thereaf ter.
4.
The infiltration rate into the control room is assumed to be 34 cfm for the duration of the accident (0.06 volume changes per hour).
5.
Filter system initiation will occur simultaneous with the development of the full source term in the containment atmosphere.
6.
Filter efficiency is assumed to be 95 % for all iodine species.
7.
The filtered recirculation flow rate will be 3,000 cfm.
8.
Occupancy factors for a single person in the control room:
100 %
from 0 - 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />, 60 % from 24 - 96 hours0.00111 days <br />0.0267 hours <br />1.587302e-4 weeks <br />3.6528e-5 months <br /> and 40 % from 96-720 hours.
The basis for our assumed source term comes from a review of the literature on iodine release and its subsequent behavior.
It is evident from the information contained in this literacure that the probability of 50% of the core inventory becoming airborne in the containment is extremely low.
l However, if we assume, for the sake of conservatism, that 50% of the core inventory is airborne in the containment, the further assumption that only one half of this quantity would plate out for the duration of the accident (as is i
recommended by the NRC criteria in SRP 15.6.5), ignores a wealth of empirical evidence from past accidents and experimental tests. Data from these accidents and tests show that physical processes, such as coalescence, agglomeration, settling and plateout, are very efficient in reducing the release of radioactivity and confining it close to its source (Reference (d) a tt a ch ed ). Therefore, based on this Information, we have assumed that af ter the 50% iodine core inventory is released, the natural removal processes described above continue to reduce the airborne iodine concentration for several hours following release from the core.
The conservatism of the NRC source term was recently brought to the attention of the Commission.
In a letter to then Chairman Ahearne (Reference (c) attached), it was pointed out that existing NRC models grossly overestimate iodine volatility for accidents involving substantial amounts of water. This is based on the evidence that iodide is released from the fuel as cesium iodine, rather than as elemental iodine or organic iodide.
United States Nuclear Regulatory-Commission August 31, 1981 Attention:
Mr. Darrell G. Eisenhut, Director Page 3 In response to the question's raised by the abnve references, the NRC prepared a report the objective of which was to provide the best technical information currently available for estimating the release of radioactive material during reactor accidents (Reference (e)). This report concludes that nearly all of the fission product iodine should be in the form of cesium iodide and states that for severely degraded core accidents in which l
containment integrity is maintained, the environmental release of iodine would be low.
t Designing an emergency ventilation and filtration system to a more realis tic source term for iodine is not only a sound use of resources but in this case it also has the advantage of minimizing whole body doses.
If the NRC source term was adopted, a filtered pressurization system would be required to limit unfiltered inleakage into the control room. This scheme would have the distinct disadvantage of increasing whole body doses by drawing unfilterable noble gases into the control room.
In light of the increasing evidence of the overconservatism in current assumptions regarding airborne halogens, it is prudent to adopt more realistic assumptions, especially where is is shown to result in a decrease in whole body dose.
In the case of the Yankee control room, if a filtered pressurization system were included in a design, the airborne whole body dose to control room operators would be increased by a factor of 2.
t In conclusion, it is our position that using the assumptions described above, the Yankee Rowe Control Room can meet General Design Criteria 19 while minimizing the whole body dose from airborne radionuclides.
Schedule for Installation l
Detailed design and engineering on the proposed system will begin with NRC approval of the canceptual design. We expect that the modifications will be installed by January 1,1983. We wish to point out, however,.that this schedule is dependent on a timely review by your staff.
If NRC approval is delayed beyond the next two months, we expect that procurement and delivery of equipment necessary for this modification will jeopardize system installation during the 1982 fall refueling outage.
i Furthermore, we wish to point out that the SEP integrated assessment, j
also to be completed in 1982, potentially could have an impact on the I
installation of this modification.
If an impact is realized, we will notify I
you immediately to identify the difficulty.
If you have any questions, or desire additional information, please contact us.
l Very truly yours, YANKEE ATOMIC ELECTRIC COMPANY
$ duo / 0 YWah J. A. Kay Senior Engineer - Licensing JAK/jgh Attachments
i i
i yigg
. -a_1=-.
i: L_
-ai'-
6 g-44' tc urut_7 oey iq, wut N ""*
- cegg.
_ L so.c3 r3 =
b
"""G'
.t.
a.e,-y
==
\\
~
,b;sts- ~
~
N w um exwn m.v.,
V womm V.
=$
'*""*' "*a"
$:/-"*c'"
JanLTN www:
x uf i
\\
a\\
It 4
w.
Sococh g,
g l
C]
N V.
]
O
- h'=r
+
%s
+
~ ~ -
u crr,cc oru g,
,f
,I
/
_O k w [ - b.i,,gs O*-Y %,d1 t [ "T y _.]"
. c/
z e
i-
- ~,g._.
9._ m m. _ = mw_%
w @s 1..A
)
wcou.
u' l 4 *t f
+
./~N" b
EFrLUENT}
+
f f3 m
3
/ untor.xi,=- yp[ Q % i
/ **4"4 "T
j - -~ - - -+, f 4+-f-- -
HALLWAY xa veur etANnto
,e,,%
um,$$
Y'E'M 0576 len=Jt:'."J.~'
i; i
PROPOSED EMERGEMCY FILTRATLord
_ SYSTEr4 (T/PicAL.)
FIGURE 1
l l
l t
i m
t t
i Q )<
d if6L 68 W 3 M August 14, 181D Chairma: John Ahearne U.S. M 'a=
Regula:ny W 2sion 1717 H S::ee:
trashing:en, D.C. 20555
Dear Chairmas Ahaarna:
- 42 vish -a b=1=g.o your a::en:1on a ma:: : da: may be a ve:7 i=portan: deval-opmen: in mactor saf a:7 a=alysis.
We believe da: suffician: evidenca has accumulated to show tha: the behavior of iodina duri=g nuclear ranc:c: acciden:s is not cor:se:17 described by exis iss NRC models and Regula:ory Gnides.
Iodina volatili:y is grossly overas ima:ad by desa modals for ac=iden:s i= which sub-sm-'ai amounts of unter are present, and escape of iodisa o the enviremmen:
will be ex::amm17 small (as 1: uns at "hree Mila Island) as long as raasonable contai= ment istag 1:7 is also maintained.
As a consequenca, de risk *a tha ganaral public presented by iodine is lower da: as:ima:ad, perhaps by orders of mag =1:nda.
Our conca:n vi:h :.6is issue orig 1=a:ad vi:h our involveman tu the several "cc5 ' cal Staff Analysa: for the ? ssident's Commission cu da Ac=1da== a: hree Mils Island.
l'ha mechanism for he behavior of id'n= ca: we propose hara was iartved from dose analyses, f ca fu=: hor M"=: ion of czperimen a1 and theore:1 cal studies i=volving :he A-=d *:ry of iod1=a and casi::a fission ;;o-ducts in light usta reactor fuel and sys:a=s, and from the observed behavior of iodina subsequen: to fuel failures duri=g ac=1 den s and incidants at other reac-to si:ss.
We believe that the expla=ation prasen:ad bara nAll change ha pra-concepts of da bazards i=vcived dcring and subsequen: :o reac:or acciden:s son:
and, therefora, will wouira a cri:1 cal ra==d-m:1on of bow these hazards and risks are calcula:ad, and da critaria o which engineered safeguards are designed and installad.
Although de ':hree Mila Island,(':MI) reac:or cora i=ve::orias of zanon-133 and todise-131 vers cost. arabia, be:veen 1.4 and 13 21"'on c=rias of sanon escaped c de envir---- du:1=g the acciden:, while only 13.o 18 c=rias of iod1=a s' = 17 ascaped!
This 3:sa: dispari:7 was iden ified as a mm:::: cf crucial 1 scar:ance sa:17 i= ca i=ves iga:1on by.he ? asiden:'s W' =sion, a=d a=
cffor. was anda :s find da expla=ation.
I: uns clar: -hat we c=uld =c y' e d-7 cndars and.he accida=: u= 11 :.his diserapancy (a bu::c: ed 100 o 10u) uns ex=laised sa:1sfactorily.
Fur har, 1: uns recog ' ad.ha: de physical a=d cha=ical cond1_1ons duri=g da accide== a: v may =c: have i:ee u=ique.
(We no:a da:, gn=arally, adicied1=e is de con::c"'=g fissio: p cdue: specias vi:h respect o si:a sais:7 analysis as wall as :he desig: and cpara ic: cf
_ __ __ gai=, engd.saera[ _ sad afggds_. L.___ _
- - - - - - - - - ~
c
\\
I Chairman J. Ahaa==a Augus: 14, 1 910 2a azplana.-ion for the very low escape of 4~8d*=
-ha: developed during
'.a investigation by the President's Comission was that, as the tamparature of.he cara increased, iodina diffused out of the fual rods d. ugh te failed cladd1=:
and vaporized.
"As indica escap1=g, if no: alzaady is the iodida for=, than anecantared a chanicany reducing environmen: which convt :ad i: to iodida.
Si iodida subsequently want into solution as iodida icu when 1: contac:ad wa:ar.
Ir as recognized tha: addi.1onal azperi=antal work :_.. naaded :s provida a quan:1:a:19e description of de iodise behavior.
Never:halass, this armim:1c:
accoun:r.1 for the :m:.ch smanar escape of iodina tha: uns observed at 21 com-pared to de amount predic:ad to escape if aleras a1 iodida had baan n aman:, a is assumed is.he Regula:ory Guidas.
l W baliava tha: this description can be s::ang:haned and anda mora dadd d:ive.
l A1: hough de presen: data are not absointaly conclusiva, we believa da: iodise amarged f =m :sa fuel as casium iodida, al:aady reduced to iodida.
"he reactor systan enviremmen: than sustained this chemical stata.
Fur:harmora, 1: vould have conver:ad c:har iodise specias, should hay have baan prasent,
.c iodida.
Casiu= iodida would be azpac:ad to condense or plata-on:
when 1:
mached mm:a; su= faces at :supa:acuras a: or below 400 :o 500*C, and i: vould *d-=1'_y actar isto solutico as iodida ion as soon as m:ar or condensing staan was encocu-
- arad.
Sa we.icns cf iodina species is ustar, and da fae: tha: iodida ion is *."a domi:ms: Species, ensura tha: iodina vola:"'4:7 win be very sma n (compared o that i=pli'ad by the Ragnia:ory Guidas, for azampla).
A ranc:1on causist oxidation of iodida would be macassary o
d-assa he vola.111:7 of iodise.
Additional azparimental work is required to provida a quantizativa description of indise behavior, bu this quali.ative picture is consis an: vi:h the small escape of iodise observed is a number of i=cidents whos untar was pre-I sant, such as at SI.
l l
S1s maah= den is supportad by the fenoviss observations, as van as by measurama=:s ::mde at N:
1 1,
Iodina and casium ars==7 ammad congruan:17 from ?EFA Isakars d= ring power
- ansias:s (the d ~' d - = spiking phenomenen).
2.
2a:modynamic calmi ed ons pe= formed at several sitas i=dicata that Cs! is
- he stable form of iodise is GR. fuel.
yur.har, the fission yield of cssiu=
1s larger :ha:
ha: of iodina, and esadum is always presas: is gras: (about renfold' az=sso on iodina.
2.
I.radia:ad fual has baan caused.o fai! i= arperimen:s perfor=ad under si::mc la:ad accida== ccedi-1ons, and :he iodisa ralaased is recovered predo=1-san:17 as Cs! =a:ha: -han as ::ala
' =- I.
2 O
n
i o
O i
)
Chairman J. Ahmarna Augus: :4, 19E 4.
2a chemis=y of iodise is such tha:, if water is ac=assible, iod1=a vill intaract vi.h the varar so that 1:s concannation is the gas phase v' ' ha meh seallar than its concen::stion is tin. n:ar.
l 5.
In other incidents dat hava led to the destruc 1on of fuel is varar systems ;
l
(% Spar:-1, Snapt:an-3, SL-1, 3C1, CB1, ad Plt"1), va unders and da:
a mch smaller amount of iodine escaped fron the systans dan would be pro-l jacted by he exis ing modala.
Data arm hard to come by for =any of hase ac=1dants and a=parima=:s, and our 1sves:1gation is cone'-"dag.
I= marked contrast, a large frac:1on (20,000 cu=1as) of he iod1=a escaped :o de environment dur1=g de Windscale accident, which occurrad unda/ czidd-1=g condi:1ons and in tha absence at va:ar.
The significance of this mecha=is" for iod1=a escape and ::ansper: ca= bardly be ovaremphasized.
We asser da: :L anampac:adly lov ral ssa of =adiciodise 1=
a the ':MI-2 accidas: is now understood and can be ganaralized :s echar postnia:ad acciden:s and to ochar desig=s of wata reac: ors.
We baliava da: a acciden:
j involving het f=al and a vatar or steam-watar environmen: W11 have the same contraM'M chemical cond1:1ons as did the '!MI-2 cora and primary syscam.
La iodina vill ama.ga as Cs7 (and possibly some other iodidas) and an a= into the l
sein:1on as soon as var staan or va:ar is encountarad.
I: vill persis: in solu-i f.1on as non-volatila iodida ion as long as oxidd-d7 condi:1ons do =c: prevail.
i I
Al. hough wa feel da: da evidence is suffician:17 s eng en jus-d'y ^'
- 1st:ar, 1: is impor an: to qualify our posi:1on.
Iodina chenis=y is vary comple=, and daf1=1:1ve azparimental and analytical studies of iodina behavior duri=g and following loss-of-coolan: accidents ara 2acking.
Nonach=1==s, 1: is clear tha: de behavior projected f ca the exis.isg Eagulatory Guides is vreng.
The en= ant 3RC assumption, that alael iodina is da chanical fotz of the
'adiciodina released, is :ngarded as a conservatism, but in his casa da assasp:1on of a wrong chemical form ms: ba =sgarded as an a=or which has com-pounding aff ac a.
If, after dna consida:2: ion, the RRC is satisfied dat our desc=17:1on of iodina behavior is valid, sa recommand that an urgan study and assessman: be mada of all availabla i= forma:1on, and, appropriata ac: ions be under.aken.
With due sspec: va point out four consequanens should our posi:1on be co=ne::
1.
2a freq m -17 quoted fission produe: ascape assusp:1ons (f =m ""D-14 54 1:
1962 :o.he =cre : scan: 3agula:ory Guidas 1.3 and 1.4, and da Reactor Safaty S:sdy, WA$d-1400) should be ra=md-ad.
"he presen: assu=p icus t
grossly overs a:a todina ralaase f =m a mac:o si:a is ma=y :ypes of loss-of-coolas: accident, anc saf any cri: aria based en ca.3a ass %:1cus should be raavslum:ad.
1
-m
--we-e-emv-e w-m--e.ws
-e-e--,,.sww-,,.,--www-,-c.-cc,,-w----w----r--
e,--mem + rw-,
Chai ma= J. Ahem se August 14, 19M 2.
"Ae dispersal oi xdiciodise is the biosphere my no 1=nger dM-=:a and con==ol considera:1on of ac=1da=:s and the design of safa:7 sys:ams.
3.
Many, if not most, =~ ddar: sequences nats: be reezamined is da :il.
Sa iodisa :isk to the ganaral public may, in fae:, be lower than previously estimated, posz1h17 by orders of mag =1:nde.
La impac: of a radue:1on of iodina dak on d e requirements for evacua:1on is particularly i=po ~.as:
1:
this d.me.
4.
The **'=ared sateguards designed for indise control should be :neza=1:ea to assura effectiveness and optist:a.ic for-.he ac:ual iodice behavior rather ^as the behavior on= en:17 assumed.
Finally, we ras 11:a tha: a major :svision of N2C assump:1cus rala:17e to acci-dan: anslyses, dose esiculations, and desig= of safeguards should not.aka place wi:hout an adequata base of tachnology from both experi: nan: and thacry, and especially until he Commission itself is convinnad ha: 1: is appropriata :o accept a :rvised physical and chemical dmscription of iod1=a ::ansport from fuel l
- o tha environmen:.
On the other hand, -Jaa impac: of wrong assumptions is so serious hat an 1stansive effer: should be md' a.o establish.he fachs.
We ara randy to offer omre detailed infor.stion or ts : hor assistanca should.he
?Dtc :nquas it.
We will be plassed to brief ha NRC staff or a=y review md:-
taas you may appois:.
Siscarely, 1
/
. Jw.~I?A W. L 5tratton los Alamos Scientific Labe'ratory k.Y
- e
- i.,.
,..has Qak Ridge National Laboratory f b b,- ff D. O. Caz= bell Oak Ridge Na.ior.a1 labora: cry cc:
G. W.
d"gham, DOI-WASE D. L Kerr, *E E. Postma, CEr
i Sffn&JcE(d.)
l REALISTIC ESTIMATES OF THE CONSEQUENCES OF NUCLEAR ACCIDENTS by M.
Levenson and F.
Rahn The Electric Pcwer Research Institute Palo Alto, California 94303 ELECTRIC POWER RESEARCH INSTITUTE Scved er, 1930 r
bu f 0l 8]hdO'c0hbYf0
REALISTIC ESTIMATES OF THE CONSEQUENCES OF NUCLEAR ACCIDENTS by M. Levenson and F. Rahn EPRI P R E A M B 1. E The safety of nuclear power plants has been defended - and attacked - on the basis of how likely it is tha.t a major release of radioactivity will occur.
Nuclear advocates say once every million reactor years at most; pecole opposed to nuclear power say it can happen at any time and will happen fairly often.
This dialogue has revolved around the probability and neither side has bounded the size of the public risk from the worst release that could really happen.
The lethal content of a physical system is not a measure of its risk.
For example, a swimming pool contains enough water to fill the lungs and thereby drown about 100,000 people, but no one ansiders this a true measure of the hazard of swimming pools.
Similarly, the air in any small office, injected 50 cc at a time into people's veins, is capable of killing ove-500,000 people -
but that air represents no real hazard.
The same is true of the radioactivity in a nuclear power plant - wioely dispersed it could cause a catastroche, but no such dispersal mechanism exists, accident or not.
Every historical reactor accident, every nuclear weapons accident, as well as many experiments demon-strate that the dispersal mechanisms act to limit large releases of radio-activity.
This is why an accident causing widespread and serious health effects to the public will not happen.
Simply stated, the ultimata safety of a nuclear power plant does not depend on the engineering features of the plant. These features determine the plant reliability and frequency of failures and accidents.
However it is natural processes (chemical reactions, aerosol settling, effects of moisture, etc.)
that prevent a public catastrophe from occurring.
This simole fact is often lost sight of in discussions on the safety of nuclear power plants.
Now, in the aftermatn of TMI, people are perhaps more open to asking the questions: Why weren't the public health effects greater? Was it but for the grace of God? No! but it was due to the grace of Nature.
Engineered i
barriers, after all, are always subject to failure. Not so with natural l
phencmena. Our experience has shown natural phenomena to be very effective I
in containing radioactivity.
These same natural barriers will also act in future accidents.
The inherent safety of nuclear reactors rests on nese l
.f oemonstrable phenomena - not on theoretical arguments or hypothetical scenarios. Whether an accident does or does not occur depends on our skill, although some like to think of it in terms of luck or probability. But the consequence of such aa accident is not a question of skill, or luck, or probability - natural processes will limit the dispersal of significant radioactivity to the near vicinity of the accident. As a result, a public catastrophe will not occur.
I l
I l
l l
REALISTIC ESTIMATES OF THE CONSEQUENCES CF NUCLEAR ACCIDENTS by M. Levenson and F. Rahn The Electric Power Research Institute Palo Alto, California 94303
l PP.E? ACE
'ha authors wish to acknouledge che mny people, :00 numerous to name, who have revieced early drafts of this paper and who have mde mny valuchte It became clear during the development of suggestions cauard its i.mrovement.
the ideas contained herein that sevent researchers in this countrj and othere have been :hinking along similar ;n:hs.
ne ace
- dan: a Gree Mile Estand II posed the question as to uhy so tic:le iodine and particuta:e Nuclear Uni:
mienr escaped :he plan: rotative to the gaseous releases.
De fairly obvious conclusion uas that natunt processes vere acting more effician:ty than the modeling predicted. nis paper simply c::empts to reinf.ece that conclusion, and to bring a new perspective on the interpretation of some neu, but also much old, empirical data. As tong as the in erpretaticns of such data vere emergency response and other criver~~a, there pas no motivation no: used :o se:
But the recent emphasis on evacuation and to a :ensively reavalua:e the da:a.
siting policy and C%us 9 accidents mkas reatis:ic reevaluation of the conse-quences of nuatear accidents i.yor:an.
l l
l
I.
INTRODUCTION Radiation exposure estimates form the basis for emergency response planning in the event of an accident at a nuclear reactor. A reexamination of the current estimates show that they may be high by a factor of ten or more.
If this is so, public concerns about nuclear safety may be exaggerated and our strategy for dealing with such an accident may be incorrectly biased, particularly in the case of evacuation policy.
Fo the reactor accidents and the resulting releases of radioactivity that could actually occur, for instance, mass evacuation does not appear to be the safest strategy.
Sheltering (sometimes with the evacuation of the few individuals at close-in locations) appears to be superior, in that it may result in a lower overall risk to the general population.
In a reactor accident, the principal concern is that the engineered safety features will fail resulting in a large release of radioactivity.
The radio-active fission products in the core will then be redistributed by various natural processes (chemical reactions, aerosol behavior, condensation, effects of moisture, etc.). The failure of each engineered barrier to function properly, however, still oces not mean that a significant amount of radioac-tivity will escape. Experiments and experience demonstrate quite the oppo-site.
This raises the question of why current estimates are so high and how much radioactivity could really escape.
The risk to the public from a nuclear emergency is based on three quantities:
(A)
The orobability of some sequence of undesirable events occurring (3)
The consecuences that would follow if these undesirable events occur l
(C)
The action taken to mitigate the accident Considerable work has been done on developing a probabilistic methodology for evaluating part ( A). A good example of this technique is that used in WASM-la00, the Reactor Safety Study (D. We believe the probabilistic models have oesn developed to the coinc where their usefulness is not limited by 1
their technique, but by the validity of the data used in evaluating part (B),
the consequences.
When accident consequence estimates lead to actions (such as evacuation of an area) which pose significant safety, health, and economic risks, then these estimates must be consistent with what is actually likely to occur (see Figure 1).
In addition, the risks posed by a nuclear accident and the miti-gating action should be evaulated on the same basis.
If the risks of the mitigating action are treated less conservatively than the accident risks, incorrect conclusions will be reached and faulty emergency strategies may result.
j The Reactor Safety Study attempted to model the important natural phenomena to produce a realistic assessment of the risk of a nuclear accident.
It sue-ceeded to the extent that it is much improved over an earlier Brookhaven study l
(WASH-740) (2) on the same subject.
However, in terms of correctly handling all of the details of the many removal processes which limit the release of radioactivity, it is still quite far from what would actually happen in reactor accidents.
The objective of WASH-1400 was to methodically examine potential accident sequences and obtain estimates of the plant response and public consequences for such sequences.
Emphasis was placed on examining large Loss-of-Coolant Accidents (LOCA).
Limits on time and resources led to simplifying assumptions in the study.
It was hampered by a lack of ability to define with precision the conditions existing during an accident.
The outccme was an efficient but simplified model, that contained conservative assump-tions, in many areas of ccmplex or uncertain phenomena.
As a result, l
WASH-1400 has a tendency to greatly overestimate consequences.
In judging whether a model such as WASH-1400 is adequate, experience with previous reactor accidents, especially those involving complete or partial core melt, and those with an absent or breached containment, should be accorded special attention. Also important are the many large-and snall-scale experiments.
If discrepancies exist, results of the modeling must be i
used with great care.
Some of the important benchmarks against which models t
snould ud ccmpared are given in the next section. !
II.
RADIATION RELEASES FROM DAMAGED REACTORS There have been a number of serious accidents at reactors involving signifi-cant core damage where no significant amounts of radioactive material were releasedtotheenvironment(3).
These accidents occurred at Detroit Edison's Fermi Unit-1, the Experimental Breeder Reactor-I in Idaho (1955), the Sodium Reactor Experiment (SRE) facility in California (1959), the NRX reactor at Chalk River (1952), and the Westinghouse Test Reactor (1960).
There have also been at least three major reactor accidents that resulted in radioactive releases to the environment. These occurred at Windscale, the S* -1 reactor, and at ihree Mile Island; at each, there was major damage to the reactor core. Both the Windscale and SL-1 accidents cc^rred in nonccmmercial reactors.
Neither of these twn reactors had containment buildings.
Neverthe-less, the radiological releases were quite limited.
In all these accidents, the peint of interest is the fractional inventory release; i.e., the amount of radioactivity escaping relative to the radioactivity in the core.
In October 1957 a major fire occurred in the Windscale No. I reactor on England's western coast (1). Windscale was a'i aircooled reactor for the production of plutonium, and was not typical of commercial reactors. The burning of the graphite and uranium core and the lack of a containment system
(
allowed the escape of radioactive fission products frem the reactor's 400-foot stack to the surrounding countryside.
The reactor continued to burn for more than two days.
Substantial amounts of radioactive iodine existed in the core, l
much of which was released from the fuel during the fire. Only a small frac-1 tion, however, ever exited the stack.
The highest radiation level reported off-site was about 4 mR/hr.
This reading was reported at a single location about 1 mile from the reactor. Monitoring of the areas surrounding Windscale, and of locally produced milk, was undertaken.
In certain areas, the consump-tion of milk was temporarily halted as a precautionary measure (1).
On January 3,1961, the SL-1 reactor at the Idaho National Reactor Testing Stationexoeriencedareactivityinsertienaccident(6).
The sudden removal or a c0ntrol rud, under abnormal conditions during maintenance, was the cause.
This sudden reactivity insertion led to a power excursion and exten-sive core melting.
Tnree empicyees nere killed due to injuries sustained fr:m - _ _ _ _ _ _.
mechanical effects of the steam pressure. The SL-1 was a small, natural-circulation,3Mhh boiling water reactor (BWR).
It was a prototype military reactor operated by military personnel.
Its metallic fuel elements were con-structed of highly enriched uranium-aluninun alloy, surrounded by aluninum alloy cladding. Few engineered safety features existed.
In these respects, it differed appreciably from a modern power reactor.
Fuel that melted contained about 19% of the total core fission product inven-tory.
However, in spite of the fact that the sheet metal building which housed the reactor was " drafty" and vented to the atmosphere, less than 0.1%
of the nongaseous inventory actually reached the atmosphere during the first two days following the excursion event.
For instance, environmental sampling results indicated that only about 20 Ci of I-131 had escaped from an initial core inventory of 28,000 Ci (7). Further sampling indicated total releases of only about 0.5 Ci of Cs-137 (core inventory 3100 C1) and about 0.1 Ci of Sr-90 (core inventory 3070 Ci) for the accident (8).
In comparing this accident to what might happen in a connercial nuciar plant, the presence of a containment building, and the multiccmpartment nature of such containment buildings would further decrease the amounts of radioactivity released. Nevertheless, at SL-1, relea3es of fission products, particularly of the volatiles and particulates, were quite small because of the physical and chemical laws governing their behavior, not because of the existence of h
engineered safety features or a containment building.
Recent calcula-tions (10) were done using updated versions of the CCRRAL and CRAC codes to reproduce the radioactive releases from SL-1.
The calculations demonstrate that unless the physical / chemical phencmena connected with the initial rapid dispersal are properly accounted for, the analysis will greatly overestimate the environmental releases.
The recent accident at TMI in March 1979 resulted in the release of about 15 Curies of I131 totheenvironment(R).
This was less than one part in ten million of the iodine in the core.
A much larger quantity of the noble gases Xe and Kr were released (aoproximately 2.5 million Curies or 2% of the noble gas inventory). Negligible amounts of Ba-140 were released (,9_). These noble gases were quickly dissipated.
Radiation levels outside the reactor site were cuite icw, mostly below 1 mR/hr.
____________----A-
There was no failure of the reactor containment building during the accident, and as a result there were no direct releases from the containment.
The releases that did occur were secondary leaks from auxiliary systems.
The amount of material leaking from the containment building was further attenu-ated in the auxiliary building by the operation of plating and fall-out mecha-n'sms prior to escaping to the atmosphere.
III.
RADIATICN RELEASE FROM CONTROLLED EXPERIMENTS In addition to the experiences with reactor accidents already described, other empirical data exist which demonstrate the role of natural phenomena in limit-ing the dispersal of radioactivity.
These data come from experiments investi-gating the various aspects of fission product ditpersion.
A.
Small-Scale Exceriments The first point of departure for any evaluation of the radioactivity released during a major reactor accident concerns the melting and vaporization of the fuel itself. Recent experiments (J2,) on high temperature, high concentration U02 aerosols carried out at Rockwell International have shown the tendency for fuel-like aerosols to exhibit a fall-out behavior characteristics of two relaxation times.
The first operates on a time scale of seconds, during which time more than 90'. of the mass of airborne particles is removed from the air, while the second operates on a time scale of tens of minutes, during which remaining fine particles settle out. Previous experiments were not able to detect this effect because of difficulties in making measurements earlier than a few minutes after the creation of the aerosol and in making an accurate mass balance.
The more recent studies further show that at high concentrations 3
(.07 to 1.09 kg/m ) agglomeration is so rapid (milliseconds) and the resulting particulates so large (100-400 um) that the giant agglomerates (containing a large fraction of the available aerosol mass) will fall out raoidly and will sweep out accitional aerosol mass during their gravitational fall.
Studies at <arlsruhe (M) en core meltdowns require that there be between 1 anc 2.5 tonnes of aerosol to be consistent with release fractions.
The total.-
aerosol would consist mostly of fuel and structural materials which are nonra-dioactive.
This aerosol would be distributed mainly in the pressure vessel or reactor cavity area depending on the scenario chosen.
Such a condition is highly unstable, and aerosols sould be quickly removed from the airborne state by natural processes.
Particulate fission product will then be removed with the much greater amounts of inactive aerosol.
Note that this will occur even if moisture is not present, although the presence of moisture would greatly accelerate the aerosol depletion.
An earlier experiment at Oak Ridge National Laboratory (J4) with UO2 fuel showed that indeed nearly all of the iodine, tellurium, and cesium and more than half of the strontium, zirconium, ruthenium, barium, and cerium are released from the melted fuel. With the exception of the iodine, tellurium and cesium, however, all these fitsion products condense and plate out in the high-temperature region around the fuel.
Recent experimental work at CRNL (J1) shows the formation of Cs! in the fuel prior to release from the matrix. A similar chemical reaction of tellurir with cesium in the fuel is expected.toformCs2e(J62). As a result, during an accident the iodine, T
tellurium, and cesium isotopes are predcminantly in the ionic state and retained by any moisture present.
This is an important phenomena, due to the importance of these isotopes in predicting early and latent fatalities as the result of an accident.
Still other work at CRNL (J8) showed that in partially melted multi-pin fuel experiments, only very minor amounts of particulate activity escaped the immediate furnace liner surrounding the experiment. A most striking reduction in release, compared with the more comonly performed single-pin experiments, cccurred in the multipin release.
This release was lower by a factor of one hundred.
The results s.iowed that the unmelted parts of the fuel and surrounding structure offers a suitable plate-out surface for l
released fission products.
1 In a reactor accident which includes core melting there will be many cooler regions above the core (in the pressure vessel, piping, or pressure vessel compartment).
This condition will be assured by the presence of single-and two-phase water-steam mixtures.
Results (19) from the General Electric Air-craft Nuclear propulsion Cept. ( ANp0) shcw that cesium plates out :n such l
surfaces nnen :ne temcerature is in the range 1000-1800*F, and iodine in :ne l.
~,.-... - -,
range 80-600*F. Other work at BNL @) found that in certain instances, 90%
of the iodine released into air in a reduced state, due to a steam environ-ment, can be collected on surfaces whose temperature is below 120*F. Qualita-I tively identified in still other experiments (R), but not measured, is the absorption of cesium and iodine on the surface of particulates.
In high-concentration aerosols, this phenomenon can take place rapidly.
This ooserva-tion has important implications in considering accidents where large amounts of water may not be present in the immediate vicinity of the core.
In such cases, materials (such as the 500 kg of Ag-In-Cd in the control rods of PWRs) with low melting points may become aerosols coincident with the release of the iodine and tellurium, and thus serve as a blanket of condensing and sorption surfaces for these elements.
Other work conducted at Hanford (g) on high-temperature release of fission products from molten fuel in helium, 2 team, and air atmospheres produced the following result:
radioactivity released in steam was between two and ten
~
times less than that released in air.
This experiment was carried out on i
metal fuel, but the aerosol behavior is directly applicable to the oxide fuel used in commercial LWRs. A second important result was that after the fission oroducts were released from the fuel, the fraction of the released volatiles--
l fadine, tellurium and cesium--deposited in the apparatus was significantly i
higher in a steam atmosphere. Such deposition occurred within a few centi-(
meters of the molten fuel.
In the case of icdine,10". was deposited in dry air, 60% wnen steam was present, roughly a sixfold increase in attenuation.
The effects of steam condensation in removing fission products was next inves-tigated. Approximately 97% of the fodine, 77% of the tellurium and 30% of the cesiun were found in the steam condensate.
It was concluded that condensation of fission-product-laden steam is nearly as effective as high efficiency filters in removing fission products released from the melted fuel.
Other experiments show similar results (13 da).
' edk paths t.1 rough the concrete walls if failure were to occur would be long irregular cracks wnich have rough surfaces so that additional aerosol removal phenomena, sucn as impaction, are coerative and reduce even further the mass of the aerosol transmitted (E). Ex:eriments on aerosols snow that such reatoval pnenomena are very effective and that a major fraction of -he entering..- -..
aerosol mass is retained in the crack. Moreover, moisture will collect in such cracks, serving to further filter the releases.
B.
Large-Scale Containment Tests 3 Con-Six experiments Q6) were performed at BNWL in the 2,286 and 25,500 ft 6
tainment Systems Experiment (CSE) containment vessels in the early 1970s.
The time dependence of iodine, cesium, ruthenium and uranium concentrations were studied.
The experiments were carried out in containment vessels of two sizes, of which the larger was approximately a 1/5 linear-scale model of a PWR reactor containment building.
No engineered safety features were provided.
All fission product retention occurred solely by natural, passive processes.
The natural attanuation processes, in increasing order of importance, were retention in the release apparatus, in-containment removal by surfaces, and removal in leak paths ( E.
This study also found that iodine attaches itself to solid particles and is absorbed by liquid droplets.
The cesium particles which were introduced with the iodine reacted to form cesium iodide.
In spite of the fact that 100%
release was attempted, 28% of the iodine and 67% of the cesiun were retained in the release apparatus and injection line. As soon as the particles were introduced into the steam in the containment building. they acted as conden-sation nuclei to form fog droplets.
Elemental iodine was absorbed into these fog droplets very rapidly until the equilibriun relationship was reached between gas and liquid.
The initial time for 50% removal of the iodine in the gas space was found to be between 9 and 24 minutes; later this " half-life" increased to 20 or so hours. After two hours, iodine decontamination factors ranged from 30 to 1000.
After one day, they ranged frem 100 to 2500.
Cesium behaved in much the same way, although decontamination was less at 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> and much higher at one day.
Most of the cesiun (72-90%) was observed to sett!e on the floor by gravity. About 50% of the iodine and 10% of the cesium was retained by tne paint on the inside surface of the vessel.
(The average LWR has 10 to 20 tons of paint en surfaces within centainment.)
.a-
C.
Exocrimental Reactors Tested to Cestruction
- t various occasions in the past, experimental reactors have been deliberately tested to destruction to verify that large reactivity excursions were self-limiting and would automatically terminate the nuclear reaction. These tests verified that this was indeed the case.
The tests were designed to violently disassemble the core and melt or vaporize part of the reactor fuel. Disper-sion of radioactivity was monitored and provided information on how widespread such dispersal was likely to ca. Three tests of this nature were the 30RAX-I test (3) (1954), the SPERT-I test (g) (1962), and the SNAPTRAN tests (R)
(1963). All these tests were conducted in the Idaho desert.
The cores involved were relatively clean, with low fission product concentrations.
If higher concentrations had been used, other natural processes such as high density aerosol behavior, might have further limited radioactive dispersal.
The BORAX-I experimental apparatus had been used for a highly successful series of tests on reacter transients.
It began to show signs of hard use.
In view of indications that its effective usefulness was near an end, it was decided to run a destructive experiment tc find out what would happen.
One of the effects to be investigated was to see wnat fraction of the fission product inventory in the core would be released to the environs upcn destruction an'a vaporization of the fuel.
The reactor was fitted with special control rods designed for explosive ejection and loaded with excess reactivity.
l
)
l l
The reactor was contained in a tank, which was sunk partly into the ground.
l l
There was no building over the reactor. Motion pictures taken during the test 1
showed that the low-pressure water tank holding the experiment burst and most of its contents were ejected into the air.
Recognizable fuel fragments were tnrown as far as 200 ft, but essentially all the fuel could be accounted for witnin 350 ft. of the reactcr. A wind of 8 mph at ground level (20 mph at 250 ft. altitude) was blowing. Even under these conditions, the phenomeno-l logical mecnanisms limiting dispersal were operative.
l l
The SPERT-I destructive experiment also was conducted in an ocen tank facil-ity.
- t was covered oy a light structure not intended for containment pur-poses.
A large insertion of reactivity was performed en November 5,1962,
-9
under fully documented meteorolcgical conditiens. Approximately 35% of the aluminun alloy core was melted, with all the fuel plates in the core experi-encing melting to sone degree.
The maximum temperature of the fuel exceeded 1200*C. Approximately 20 kg of " spongy" metallic debris ranging in particle size down to below 100 um was recovereo from the reactor tank. An estimated 5
2.4 x 10 curies were released tr, the atmosphere, representing less than 1% of the fission-product inventory in the core.
Iodine was detected onl'r in the reactor water.
The building was reentered four hours after the test.
A radioactive cloud, ranging between 700 and 2000 feet wide, was monitored for a distance of 15 miles, and deposition rates recorded.
The measurement of the dissemination of fission products in the SPERT-I test indicated that the release to the atmosphere was roughly 1% of core inventory.
This was more than an order of magnitude less than that expected from pretest hazard evalua-tions (16%).
The SNAPTRAN-3 destructive test was conducted in May 1963 in an open tank without any covering structure.
Again, a large amount of reactivity was inserted, destroying the core and ejecting half the water out of the tank.
About 500,000 Curies of radiciodine was generated in the burst.
All the iodine was found in the remaining water.
In an earlier test with a dry tank, l
a large iodine release occurred.
The significance of the source-term evaluation experiments described in this section is tnat even though the laboratory and larger scale experiments were designed to give maximum release, they all resulted in smaller source terms than that predicted by the models used currently for licensing reactors.
IV.
POTENTIAL CFF-SITE HAZARDS EV0LVING FROM REACTCR ACCIDENTS A.
A Cuestion of source Term Although analytic studies such as WASH-1400 have their li:r ~ tations, an impor-tant insight derived from them is that only reactor accidents involvi.'g sig-nificant core melting will result in any significant risk to the cuolic(2).
However, for simplicity, these models usually assume that any melting of the reactor core will within minutes lead in all cases to a cata-l strophic failure of the reactor pressure vessel and containment building.
This assumption and others listed in Table 1 are not realistic.
But even with these assumptions, the studies indicate that in less than 2% of the instances will the failure of the containment building be an above ground failure.
The other centainment failures considered are due to the core itself penetrating the building by melting through the concrete base mat.
In either event, these analyses predict that the amount of radioactivity escaping the containment building would be quite large.
The near-term dose to the population in these examples is due largely to the radioactive iodine and tellurium released.
The second largest contri.2 tor is the aerosols.
Less significant, making up only a few percent of the total, is the dose due to the noble gases.
Such models may be useful in illuminating the sequences leading to core melt-down and in doing relative risk studies.
The data currently used and the lack of detailed consideration of postmelting physical phencmena, however, give rise to predictions of amounts of radioactivity released to the atmosphere that are invariably nigh.
An example is tha iodine reduction factor estimated in ene Reactor Safety Study accident sequence.
Table 2 shcws such a case (31), which only partly accounts for condensation or solution effects, washout due to dripping water and condensing steam. A total attenuation factor of 1.5 results.
When di f-ferent sets of assumptions (Table 3) for the same accident sequence are used (including some dissolution in the quench tanks but no effect of water and steam in the containment building, or significant aerosol fall-out), the 5
attenuaticn factor increases to between 6 and 10.
This indicates the sensi-tivity of the calculated results to small changes in assumption;.
Inclusion of all relevant phencmena may give even higher attenuation factors.*
When discussing the consequences of reactor accidents some of the important I
pnysical properties of radioisotopes to keep in mind are:
.ere approximately 10g SL-1 and Windscale accidents, the attenuatien factors
'For comparison in tn
, for TMI acout 5 x 100 1
o Stable, dispersible aerosols are difficult to create. Highly con-centrated aerosols coalesce rapidly.
Low density aerosols increase their effective density extremely rapidly in the presence of water vapor, serving as condensation nuclei. The effective size of the particle becomes that of the water droplet @ ).
o Aerosols agglomerate and tend to be trapped when passing through cracks and penetrations whether in pipes, compartment walls, or containmentbuildings(33).
o Agglomerated aerosols formed at high concentration are physically dense, and settle out close to their source.
The original mass of particulates, although it may be large, is not signifiunt, because only a small proportion survives this settling process and remains airborne (34).
o Iodine in its many forms is chemically and physically reactive.
Since nearly all of the s9rface area inside containment is covered with paint, plastic or organic films, iodine retention is high.
In addition, iodine will be adsorbed en the surface of aerosol parti-cles, that themselves are rapidly agglomerating and falling out(3_5).
In either instance, much of the iodine is quickly imobi-5 lized.
o The reactor containment building and the equipment in it present a large amount of surface area for fission product plate-out and adsorption.
The compartmentalization of the building and the ccm-plexity of piping and hardware means tnat any escaping material passes multiple surfaces prior to escape.
This is at best cnly partially accounted for in the modelling (16).
o The moisture conditions in the reactor containment building will cause most of the soluole fission products that become aircorne to go into solution (J7). A core melt accident will always be accompanied by large amounts of steam and water becausa coolant loss frem the primary system is :ne sine qua non of core melting.
" Rain" or "fcg".
will exist in the building even if the centainment spray system is never used.
This is due to the heat capacity of the building and equipment causing condensation and dripping frcm all the surfaces.
Such a condition would wash out large fractions of the various fis-sion product; prior to atmospheric release (25,). As mentioned earlier, moisture further tends to agglomerate aerosols and enhance their density.
o The earth itself acts as a filter and effectively suquesters any escaping fission products in the event of a " melt-through" accident or an " atmospheric release" accident (which, in spite of its name, would likely result frcm a below-grade failure of the containment building in many cases).
If the overpressurization in an accident blew out the penetrations or seals in the reactor containment build-ing, the path for escaping radioactive materials usually would be through other buildings.
This would provide further opportunity for plate-out and fallout of radioactivity.
o The presence of large amounts of water and vapor plus the heat capac-ity of the containment building and debris would be sufficient to imobilize a large fraction of the radioactivity in the event of a postulated massive reactor building failure (38,). The important role 8
of moisture was demonstrated by the SNAPTRAN tests ( E.
As a result of these phencmena, the potential off-site hazard from a nuclear l
l accident is greatiy diminished.
The above phenomena all ac' in the same j
direction to reduce the maonitude of the predicted fission product release and change the character of the release in that iodine and particulates are greatly reduced : elative to the noble gases.
Both changes reduce the conse-quences to tne puolic in terms of acute and latent fatalities and greatly i
diminish the area around the reactor over which a serious threat may exist.
None of these phencmena is dependent on somebody making the right decision, equipment functioning correctly, or power being available. They are always 1
acting.
l l
l. _ -.
The fact tnat the ccmonly used models do not treat in sufficient detail the phenomena that reduce the fission products available for release explains, at least in part, why the models predict consequences from accidents so much greater than any that historically occurred.
B.
A Question of Time If r ealistic consequence scenarios are considered, it becomes apparent that evacuation of very large areas is neither needed nor effective.
The principal threat to the majority of the population is the passage of a dispersing radio-active cloud.
This cloud would contain mostly the noble gases xenon and krypton. Against this threat, sheltering may be the best option in the short tenn (hours and days), and time then exists to detennine what long-term actions (months and years) are required.
There is no acute need for evacua-tion.
Concerning the evolution of an accident, some of the current analyses assume that once any local region of the reactor core, no matter how small, reaches a sufficiently high temperature, melting of the entire core occurs in short order, and there is an inexorable and quick progression to pressure vessel failure, containment failure and major radioactive releases.
In fact, the completion of physical processes for this to occur does not happen instan-taneously, nor is the progression inexorable Q9).
The timing of radioactive release s e..orios is important in the consequence modeling. Even a few minutes between core melting and containment failure would be extremely important.
For example, consider a postulated metal water explosion leading to early penetration of the RCS. Although such an explosion is no longer considered energetic enough to rupture the pressure vessel, let alcne containment (40_), the time between the release of the volatile fission products from the fuel and the drop of the molten core into the plenum of the pressure vessel allows sufficient time for chemical reactions, condensation anenomena anc the effects of rasture to occur. A subsecuent explosion would produce a high density aerosol, initially in contact with water, that would rapidly coalesce and fall-out, not unlike the destructive ex:eriments described in Section IIIC.
A similar case could be made for ;ostulated early ccntainment failure due to a hycragen ex:losicn.
If an accident progresses at a modest rate, the time gained thereby helps in three ways:
the residual decay heat decreases, the energetics of core damage diminishes and the radioactive inventory decays. More importantly, hours elapse before the point is reached where the last engineered barrier between the public and the radioactive fission products, the containment building, might be in danger of being breached.
Recent work in Germany indicate that a failure of the containment building due to overpressurization would require several days to materialize (M.
In the meantime, all depletion phenomena have been functioning to further reduce the source term available for release.
V.
VALUE OF SHELTERING VERSUS EVACUATION If a reactor accident were to occur, those charged with the health and safety of the public would have to decide how to protect the public.
Various factors should influence their decision, including the risks of evacuation, deaths due to traffic accidents and heart attacks, and psychic trauna brought on by the stresses of evacuation, relative to radiation risks.
To model the effects of a given emergency response, detailed sheltering and evacuation models exist which consider the dynamics of radioactive plume dispersal and that of popula-tien movement.
Even with the models and source terms used in the Reactor Safety Study Q2,), the technical basis for widescale evacuation is marginal.
When more realistic source terms for radioactive release are considered, even less justification for such an evacuation exist. For core melt accidents, the off-site doses would probably exceed those specified in EPA's draft Protective Action Guides @3_) only within a very limited area cutside of a reactor site boundry.
Only within this area would it appear that evacuation might be prudent to consider, although not necessarily more effective than sheltering
'n mitigating the whole body dose to the population.
The time before such a threat would evolve is relatively long. However,
, should be recognized that if a threat were to materialize very early in an accident, sheltering would be tne only real Option. Also it should be recognized that while evacuation clans may be prudent to develop the decision to imolement such a plan should te based on actual conditions that exit at the time. _ _
~.
Also important is information, or lack there of, concerning the magnitude of the actual danger.
While calculations that employ " conservative" assumptions are generally believed to increase surety margins, in instances where an evacuation decision is required such a treatment may significantly increase the risk by inadvertently introducing hazards not considered in the calcula-tions.
The concept that evacuation of very large areas is dr,Rable or neces-sary for public safety is probably wrung on both counts.
Inadequate recognition is being given to the safety margin provided by shel-tering and controlled air supply - these mean nothing more complicated than staying indoors, closing the doors and windows, and shutting off ventilation fans. The relative merits of evacuation versus sheltering depend greatly en the particulars of a given accident.
Parameters to be considered are severity, site location, meteorological conditions, etc. However, only in a few instances, and only for a few individuals, will evacuation be better than sheltering. Precisa answers to the questions of whether to evacuate partic-ular individuals, when to evacuate them, how far, and in which direction to evacuate them, are site-and accident-specific. But in no case can an analy-sis be considered complete if sheltering calculations have not been included, and the nuclear and non-nuclear risks considered on an equally conservative basis.
As has been outlined above, the primary scurce of exposure to the general population in the near term probably will ccme fr:m noble gas fission products. This is likely to be true even if the containment building suffered a major breach.
Due to the dilution and dispersal characteristics of gaseous fission products, the radiation dose that any off-site lccation receives will l
be small and t: ansient in time.
At Windscale, as at SL-1 and TMI, the radiation from the radioactive plume l
i represented the largest exposure.
Although scme radioiodine vas dispersed over a large area around Windscale, the dose frem it was qui.e small.
The hazard, if any, would have been due to its subsecuent concentratien in hu""
This does not cccur directly, and it was guarded against 3y One temporary dumping of milk : reduced in affected areas.
Aerosol dispersal was l
not a proolem at SL-1 cr 'N.
The E?A draft Protaction Acticn 3dids current y l
l establishes levels of 500 mrem whole body dose and 1500 mR to the thyroid as
" action" threshold doses.
If projections indicate that these levels will be exceeded, then protective action should be considered. Clearly, in each of the historical incidents, much time was available (several weeks in the most serious, the Windscale event) before these dose limits would have been reached.
The combination of dilution dynamics of the noble gases, plus the fact that physical phenomena associated with aerosols and iodine prevent their gross release, assures that time will be available to take whatever further precautionary measures are required.
Equally important is the matter of taking advantage of simple protective measures.
Closing the windows greatly reduces the potential inhalation dose M ).
The concentration of noble gases is not as strongly reduced by such measures, although factors of two or three are likely.
Precise estimates depend on the ventilation rate.
If the ventilation rate were high, however, due to the presence of windy 7teorological conditions, such conditions would also considerably shorten the time of passage of any radioactive cloud that existed and rapidly disperse it.
The shielding ability of structures also offers subtantial protection. Even a simple wood frame house reduces the dose rate from a passing clouc by a factor of two @ ).
A masonry structure may give dose rate reductions up to a factor of 10 on the first floor, 50 or more for a person staying in the basement.
These shielding factors are for gamma sources with mean energies close to 1.% V.
For sources containing primarily ncble gases released a day or two after the accident, the actual shielding offered by such structures is con-siderably greater because of the much lower average energy of the radiation.
These values are also for isolated structures. A town where a third of the area is covered with buildings may provide another factor of three protec-tion @ ).
In fact, the greater the concentration of people, the more protec-tion is afforded by the surrounding buildings, and also--the more difficult is i
avacuation. Evacuation, en the other hand. may actually ex;;ose pecole to increased radiation doses, depending upon meteorological conditions, if the evacuation direction coincides with direction of the radioactive cloud.
In acdition, there is the loss of shielding provided by buildings.
-U-
A recent ttudy of the relative safety of sheltering versus evacuation in the case of a tornado is instructive (47).
The majority of the fatalities, as well as the highest risk of fatality, was incurred by the group evacuating in the face of the danger.
Often they attempted to evacuate across the path of the tornado with tragic results.
Those who stayed behind in the relative security of their own homes fared considerably better.
The effective rate at which evacuations have been carried out in the past is quite slow.
Evacuations carried out because of natural disasters and trans-portation acc: cents have a mean rate of less than 5 miles /hr and a median rate of close to 1 mile /hr (4!).
For a city or mrior population center, the time required to evacuate would be very long, probably several days.
Even with an effective evacuation procedure, it has been observed that 5% of the population will stay behind regardless of the perceived risk.
This last fact was again demonstrated in connection with the attempted evacuation of the area around Mount St. Helens.
Also to be considered is the ease of implementation of sheltering compared to evacuation.
When formulating emergency preparedness plans, the simpler of two otherwise equal alternative strategies is always the better one to adopt, as it has the higher probability of being correctly implemented in a stressful situation.
In this regard, also, sheltering would be by far preferable to evacuation.
VI.
SUMMARY
In estimating the real risk to the public frcm an accident at a nuclear power plant, several quantities are imcortant:
the probability and consequence of the accident itself and the risk resulting frcm any mitigating acticn taken.
The uncertainties of the risk associated with the accioent seem to be demi-nated by the uncertainties of the consequence estimates.
The current proce-dure of using " conservative" assumptions (usually at eacn stage) in :ne calcu-lations produces an estimate of the risk that is likley to be mucn too hign, by an order of magnitude, or more.
-13
~
.. e In and of themselves, conservative estimates as typically made in the licens-ing process may in fact contribute additional risk by overestimating source terms and thus overestimating benefits of activities such as evacuation.
This process, in turn, leads inadvertently to putting major segments of society at greater risk than is necessary by encouraging decisions which have higher risk.
The principal areas of concern focus on the treatment of a nwnber of physical processes.
These processes are always operative and can be counted on to limit the consequences of a reactor accident.
Sufficient credit is not taken for their ability to reduce the release of radioactivity and confine it rela-tively close to its source. Estimates of risk will improve in direct propor-tion to improvements in quantification of these phenomena.
Empirical evidence from many sources shows that these processes are indeed operative and very efficient in reducing the release of radioactivity.
As a result, the policy decisions based on the source term in the event of a major reactor accident must be reassessed.
I l.
Rif trrna.ss esJ Ilotas
- 1. ken tur Sals t y htudy. An Assessment u t 19.
l.D. h berts. Ilsslun-Prede t DePussteoA 35.
For eaample, the oudeling af the entire in a Turbulent Air Stream, Iransactions teatter cocaissnent building mey entall Aa lJunt hists in u.5. foamiercial hslear Pu.er Planh, WAut-14uo (NuulG-75/ ult),
of the American Italear society. @.
three or fewer w oortments using the Ot teber, 1975.
- p. 264, (l%3).
tisputer tuJe ColutA1. WA5H-1400, appen.
A.W. Castleman. Chessical Considerations Jim Vll.
- 2. Ihrus et t.al Pussibilities anJ Conse-20 37 Nuclear Safety Quarterly Repset for in peactur Safety. Iransactions of the quent es o f N jue Ascidents in lange Ameritan Nuclear Society. 6, p. 128, llovember-December,1%8. Snut -lou 9 pp halcar Plants. WA518-740 Mar (h, 19$7.
- 3. See i.J. thumpsun enJ J.G. Secterley, the (1961).
2.23.27 Battelle A rthwest, Rlthland, Technulugy at Malear pleacter Safety. the 21.
R. A. L orena, e t. a l.. f ission Produc t WA. Nrch 1969.
II"*** Ias Highly irradiated L WR fuel.
38.
ft. Cowan, SC.R.64-127. *k asurements Nil Press (1965) Chapter it AcclJents NuNIG/CRWl2. 0.k Rid.je htlanal Lab-from the Gravel Gertie Tests.* {1964).
and Destrative fests.
oratory, february (19H0).
39 Mit14 tion of Small Break LOCA's in 9
- 4. Au.lJent at WinJwale No. I file on Oct.
- 10. 1557 Report to Parliament by the 22.
'R.E.
titillard, et. al., fisslun Product Pressurised hier Reactor Systeens. NSAC-Release frum Overteated 4.ranium. Health
- 2. Nuclear Safety Analysis Center, Pale Prime Minister. Her Majesty's Stattunery i
Of fite, hvember 1957.
Physits. 7. pp l-10. (8%)).
Alto. CA. Nrch,1980 40 E.81. becker, " Steam implosions in tight
- 5. D.V. Sauber. Ph,
..I h asurements ut 23.
W.B. Cottrell, et. al.
U.S. Esperience Attivity in Samples from WinJuele, Atug on the Release and Iransport of fission Water Reactors,' g[u.htt-27 (1980).
- P/R 2607 (4958).
Products within Contalmeent Systems under 41 J.P. Hosemann. private communication.
- 6. htleunit s Week, Volume 2 No. 2 Jan, Slaulated Reactor Accident Conditions.
- 42. WA5H-1400. Appendia VI. Section Il anJ 2nd Geneva Conference on the Peaceful Appendia J.
4 1
- 32. 1 %).
uses of Atumic Inergy, p. ' 85 (1963).
43, h aual of Protective Action GulJet and t
l
- 7. "Repus t on the hclear latlJent at the 5t-1 keetter." USALC heport itM3 19602 24.
G.W. Parker, A Revlev uf fission-ProJuct-Protec t'.ve Ac tions for Nuclear incidents.
(Ja n. 1%!).
Release Researth, Iransactions of the EPA-5?J/l-75 001, 5cgit.1975.
- 8. J.R. Na an and W.P. Geamalli "u.4 steal t h American htlear Society 6, p.120 (1963) 44.
D. f 4ade (td.) hteorology and Atunic Physlos Aspects of the 51-8 Accident."
25.
C.I. Nelson and R.P. Johnson, Aerosal Energy 1968. 150-24190, p. 360. 1968.
88calth Phnits, s,877 (1963).
Leakage Tests. Luna-%6. (1975).
45.
2.G. Burson and A.E. Proflo. Structure
- 9. W.N. Bishop, et. al., *f ission ProJuc e 26 lee for eaample: k.H. Hilliard and L.f.
Shielding frsan Cloud and fallout Gaamma Release inom the fuel folloutng the IMI.
Coleman, Itatural Transport fifects of Ray Sources for Assessing the Conse-2 Accident (1980).
fission Psoduct Behavler in the Con.
stuences of Reactor Accidents. (GG-ils).
- 10. 2.I. Menduta, et.al..*RaJiat ion Releases Ialhaent Systems Esperiment. Binet-1457 1670 (1975),
from the 58-1 AcclJent," paper presented December 1970.
46 M.0. Cohen, private communication at the ANS International Conf. W ahing.
27 R.E. Hilliard and A.K. Puntabt. *Largs 47.
R.I. Glass, et. 41.
Injuries frian the ton. DC, Novembes 16-21 (1980).
Scale fission PruJuct Containment Tests,*
Wichita falls Tornado Science, 207. p.
II. A. Miller, kaJiation Source terms anJ NtDL-5A 2254 (1980).
734 (1930).
Shielding at IMI-2 Transactions ut the is. See group of papers on the SPIRI-l Des-48.
See WA541 1400 Appendia VI. p. Il-7 and neerican b leer klety. 34. (19u0).
truttive fest Results, fransactions of appenag, j, i
- 42. H. A. breut ta, et. a t. hclear les hno-the Amerleen hisar Society 6 pp.137 lugy 46, p. 332. Det ent.er, l979.
I41.(1963).
- 13. H. Suid and W. 5thuett, "the Natus al Ne-29.
W.E. kessler, et. al.. Iransactions of the naval of Pas titulata RaJioactivity is an American Nuclear Society, 2, p. 383 IWW Cuntateunent lauring Cure MeltJu=n (1%4).
Attidents," theemal Reatter Sa fety Neting,30.
1.8. Wall. Applicattun of Neactor Safety Enuaville, IN. April 7 11 (1980).
Study hthodulogy to Emer 3ency Response l
- 14. W.I. Brownsag et. al., Nelease of fission Planning. Cund. of Radiallon Control Pr odu t s Dur ing I n-P Lle Mel t ing o f UO,
Pro 9 tam Alrectors. Hasrisburg. PA.
j Iransattluns of the American Ntlear Ny 4,1975 h iet y. 6, p. 125 (l % 3).
31.
R. kitiman, private communicattua.
- 15. ft. A. R osent, e t. al.. Nunf G CR.0722, leh.
32.
A.E. Postma and t.f. Coleman, filett of (198u).
Continuous Spray Operation on the Re-
- 16. R.S. forsyth, et. al., "Vulatile fission soval of Aerosols and Cases in the Con-Psoduc t Rehavlur in Reet tur f uel bJs taluma:nt System Esperiments; IWHA-1455*.
thaler Ata lJent Cond e t tons." Prot".edings Battelle Pattitc Nortlwest Laburat441es, of Spetlalist Meeting on the Behaveur of Richland, WA, 1970.
Wter Reu tur leal Elements thiJer Au s-35.
H. A. Museut ta, l eakage of Aerosols trous Jent twentit tuns. (Wf D bolear ines gy Contaliument Rutlding, Nu(lear Safety Agena y. brua y. (1976).
Analysts t enter Reler t. (in publicatlun).
- 17. D. C=Llu tut t i and J.f. Sanetki. J.
$4.
G.W. Parker, et. al.. Repor t OR'et -3319 holear NI. in 90 (1978).
Oak Ridga: Hattenal lat.usatory (Aug, 1962)
- 18. f. W. Pas tes ail,u. A. B us ens. l is. son ig, R.g. Hilliard, et, al., ilsslun ProJua.t f
Pse.Ju(t Nelease faua Heltdown ut a Clu,-
Walease trum htelic uranus. Trans. Am.
tre ut Center-HeateJ U0, fuel Plus, Irans.
htlear h iety { a. 17a (14ed).
at t eun of the Anmerican b lear hirt y, b, p. 124 (1963).
4
o.-
o VARIOUS ASSESSMENTS OF PUBLIC HAZARD Public Hazard 1000
)
\\
Reactor Safety Study Skeptic's type analysis 1CO concern 10 This study 1
l l
l 1
100 10,000 1,0CO,C00 Interval Between Occurrences (years) e Most detailed probability analysis such as the Reactor Safety Study (WASH-1400) indicate that a public catas phe mignt occur no more than once in a million reactor years e
Many people fear that th is not correct and that such a catastrophe might occur more often e
This study suggests that natural processes limit both the spread of radioactivity and associated public hazard.
O 4
1 Risk
-Risk F
i i
3-
! l1 2
j Consequence Probability Action i
Piqure L
'I'he size of the overlat) between the circles is a measure of the risk.
If area 1 is much greater than area 2, action to mitigate the consequences of an accident is called for.
If, however, the consequences are small, the risk ret >re-sented by area 3 is smaller than the risk of the mitigating action.
In such a case, no action should be taken.
t
<., s,
TABLE 1 WASH-1400 Assumptions Concerning Fission Product Release to the Environment primary System Assumotions no platecut along transport path for any species in any ECC in-a jection failure sequence no significant iodine soluability in residual water e
Containment Systems Assumotions no deposition along leakage paths to the atmosphere for any e
species in any accident sequence no trapping of any species during water flow through pools e
e limited compartmentalization of the RC3 no retention of any species by auxiliary buildings or structures e
outside containment Release from the Fuel used 100% release for the volatiles (Xe, I, Cs and Te) e I
assumed fuel oxidation very effective in releasing Ru group e
after steam explosion 1
Chemical Forms e assumed iodine would exist in elemental form rather than CsI Aerosol Behavior neglected particulate agglomeration e
e only partially modeled steam condensation effects e neglected particle deposition on dalls Release uoan Containment Ruoture treated as instant percentage loss of airborne contents o
neglected heat capacitj of rubb'e in condensing and trapping e
fission products
i,s
=
TABLE 2 Iodine Attenuation Factors Using WASH-1400 Scenario and Models TMLB'd Sequence o electric power never recovered e
sequence treated like a hot leg break large LOCA e conceptual pathway:
Core
, Upper RCB
= Outside Region RPV
_ Space Event or Value or Attenuation Reason or Process Assumption Factor Cor.ent Full core celt.
Melt release 90%
in vessel High S/V PCS plateout none 1.0 High volatility (I2 & HI) liigh temperature Short residence tin.c RCB plateout some 1.3 Matural depositien (I )
2 Limited time RCB rupture gross 1.16 Instant depres-surization Leak path none 1.0 Huge hole plateout Total Attanuation Factor 1.5 605 release 0 ' 4 hr
<. *o n TABLE 3 Iodine Attenuatten Factors Using Basic WASH-1400 Scenario but Modified TML3'8 Sequence e electric power never recovered e realistic PCS path RCS overpressure failure not catastrophic e
e path for in-vessel release:
Core ___ Upper _ Leg Hot
, Surge Pressurizer Region RPV Line
}
Outside -
eak,
RC3 Quench,
Discharge Paths,
Space,
tank Line Possible Event or Value or Attenuation Cri tical Process As sump tion Factors Ccndition:
Melt release 90%
Melt S/V in vessel PCS plateout Condensation 1 - 10.
Temperatures Residence time Chemical / physical forms Water trapping Dissolution 2 - 100 Water in quench tank or pressurizer Chemical form l
Steam - H2 ratio l
Water temperacure
~1 RC3 plateout Ag = 1 - 2 hr Surface area
",7 At = 1 hr-1 Leak rate l
Plateout in fiany crackt 1 - 100 Leak path gecmatry i
leak path (Length, turns, ecught.ess) i Steam condensation l
Residence time Chemical form Possible Attenuaticn Fr.ctors Lower vahe = 6 em5