NUREG-0900, Informs Commission of Status of Staff Reassessment of Source Terms & Related Severe Accident Research Activities,For Use in Implementing Commission Severe Accident Policy Statement. NEA Mar 1986 Rept, Nuclear Reactor...Source Terms, Encl
| ML20210J678 | |
| Person / Time | |
|---|---|
| Issue date: | 03/24/1986 |
| From: | Stello V NRC OFFICE OF THE EXECUTIVE DIRECTOR FOR OPERATIONS (EDO) |
| To: | |
| References | |
| RTR-NUREG-0900, RTR-NUREG-0956, RTR-NUREG-1150, RTR-NUREG-900, RTR-NUREG-956, TASK-PII, TASK-SE SECY-86-096, SECY-86-96, NUDOCS 8604030530 | |
| Download: ML20210J678 (59) | |
Text
- _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
o .
>D Ctc o
^
E o a ,;
\...../
POLICY ISSUE March 24, 1986 (In OrmatiOn) SECY-86-96 For: The Commissioners From: Victor Stello, Jr., Acting Executive Director for Operations
Subject:
STATUS REPORT ON SEVERE ACCIDENT SOURCE TERM REASSESSMENT
Purpose:
To inform the Commission of the status of the staff's reassessment of source terms and related severe accident research activities, particularly risk assessment.
Category: This status report is for information purposes only.
Discussion: The staff briefed the Comission on source-term related activities on November 4, 1985. Since that briefing the Severe Accident Research Program (SARP) has focused en three major documents discussed below:
(1) NUREG-0956 (Final), Reassessment of the Technical Bases for Estimating Source Terms, (2) Draft NUREG-1150, Nuclear Power Plant Risks and Regulatory Applications, and (3) Revision of NUREG-0900, Nuclear Power Plant Severe Accident Research Plan.
The information developed under SARP is being used by NRR for the implementation of the Comission's Severe Accident Policy Statement. The staff plan for implementation was transmitted to the Comission in SECY-86-76 on February 28, 1986.
Contacts: M. Silberberg, RES Telephone: 42-74737 M. Ernst, RES Telephone: 44-37923 0604030530 860324 PDR SECY 06-096 PDR l
2 Status of NUREG-0956:
The public comment period for the draft report ended on January 7,1986, and 66 letters containing about 700 pages of comments have been received to date. Many changes will be made in NUREG-0956 as a result of the comments, but the general findings will not be changed significantly.
. Sections on risk and containment performance, which were not the main thrust of the source term work, are being deleted and information on these topics is being deferred to NUREG-1150.
The code descriptions and numerical results discussed in Chapters 3 and 4 are being upgraded to correspond with the improved set of codes called the Source Term Code Package.
These inproved codes have been subjected to a quality assurance program and the improvements resolve a number of the criticisms raised on the earlier, EMI-2104 version of the codes upon which NUREG-0956 was originally based.
Many comments were received about the limited validation of the source term codes. Existing validation will be displayed more adequately in NUREG-0956, and work is in progress at several of the laboratories to improve code validation prior to publication of the final report. However, the major cause of limited validation for the NRC codes and, in fact, for all source term codes is the limited amount of applicable experimental data available at the time when the BMI-2104 codes were being developed. In several cases, it takes 1-2 years after a test series to fully characterize the data and use it in code validation. The final version of NUREG-0956 will. describe revised research plans for validation of all of the NRC's severe accident computer codes using the augmented data base.from past and current tests.
Many comments were also received about the need for quantifi-cation of uncertainties. It will be pointed out that quanti-
' tative source term uncertainties are being determined for the risk rebaselining study, which is the first application of the Source Term Code Package and that those results will be presented in NUREG-1150. A new quantitative uncertainty treatment and that study will be described in NUREG-0956.
A number of comments expressed concerns about peer review.
The review of NRC's source term work to date has been more open, diverse, and' extensive than that for any other similar methodology, and the staff intends to continue to make source term technology information readily available. By the time NUREG-0956 is issued in final form, all NRC codes, manuals, and reference documents will be published and available in final form.
t #
g 1
In a letter from Chairman Ward to Chairman Palladino, the ACRS stated "that the codes, in their present form, should not be given much weight in making decisions." The staff met with the ACRS Class-9 Subcommittee chairman to get clarification _ of this comment. It is our understanding that the ACRS comment was based largely on (a) the absence of methods for quantitative uncertainty analysis and (b) dissatisfaction with containment performance evaluation as described in Draft NUREG-0956. Both of these subjects are being addressed in the risk rebaselining study, which is the first application of the new source term codes; they will be discussed in the final version of NUREG-0956 and fully documented in NUREG-1150. We will continue our practice of frequent discussions with the ACRS, including a review of the final version of NUREG-0956, and review of NUREG-1150 as it develops.
Finally, the subject of iodine chemistry is still under investigation. Although the analyses discussed in NUREG-0956 treat iodine transport only as cesium iodide in particulate form, draft NUREG-0956 (p. 3-33) cautioned that some elemental iodine may be present in the reactor environment and mentioned several investigations into this possibility. Recent results from scoping experiments at Sandia National Laboratory indicate that Csl might dissociate in a radiation field. Additional experiments are being run at Sandia under NRC sponsorship and at Whiteshell in Canada under EPRI sponsorship to assess this effect and to identify the dissociation mechanism and likely forms of transported iodine under prototypic accident conditions.
In the meantime, a parametric sensitivity study is being performed to determine the impact of different fractions of volatile and particulate iodine on calculations of source terms.
Results of the sensitivity study will be reported in the final version of NUREG-0956. While a change in the chemical fann assumed for iodine would make a difference in the iodine release, there are several factors that are likely to mitigate the impact: (a) CsI is itself rather volatile and current analyses often predict significant Csl releases from containment; (b) releases for some important sequences are dominated by other radionuclides such as strontium, tellurium, and neptunium; and (c) some volatile forms of iodine are more reactive than Csl and even if present would be retained on surfaces, in water pools, and by sprays.
Publication of the final report, NUREG-0956, is now scheduled for July 1986.
4 Status of NUREG-1150:
NUREG-1150, Nuclear Power Plant Risks and Regulatory Applications, is being prepared to provide a greater under-standing of the risks and the uncertainties associated with severe accidents at nuclear power plants. The report is based on the assessment of the risks resulting from internally initiated accident sequences at six reference plants. The reference plants have been chosen to represent a spectrum of plant and containment designs. Thus, NUREG-1150 will provide a summary of the current state of knowledge regarding reactor risk and it will provide insights regarding the use of this~
risk-based information in the regulatory structure.
This report is, in a sense, building on two prior reports, NUREG-0956, Reassessment of the Technical Bases for Estimating.
Source Terms, and NUREG-1050, the PRA Reference Document.
NUREG-1050 discussed the state of the art of probabilistic
~
risk assessment (PRA), provided summary information from existing PRAs, and provided general insights on the use of PRA methods and results in regulation. NUREG-1150 is taking the next step by providing PRA information on six reference plants, rebaselining any previous analyses of these plants to reflect the changes in design and operation of the plant which may have occurred since the earlier studies were published.
Utilizing this information to generate improved estirates of core melt frequency, and combining this with improved knowledge of severe accident phenomenology as discussed in NUREG-0956, NUREG-1150 will contain an assessment of the nature of containment performance and, ultimately, of the likelihood of the release of radioactive material to the environment.
NUREG-1150 also will identify plant-specific insights regarding the dominant contributors to core melt frequency and risk, the uncertainties associated with these estimates, and the principal factors influencing the uncertainties; the report will also provide recommendations on the use of such .infor-mation in making regulatory decisions.
Sensitivity and uncertainty analyses are key elements of the analytical work to be discussed in NUREG-1150. The analyses are being conducted to: (1) provide a perspective on modeling assumptions that are important; and (2) to provide an envelope in which the actual value of results may reasonably be expected to be found. Because of the
5 subjectivity associated with estimates of possible parameter values, this " envelope" will not be expressed in terms of formal statistical bounds. Where possible, analysis of actual data will be used in defining reasonable bounds; in other cases, where data are lacking, expert judgment will be used to estimate possible ranges of parameters. Because of this, the final uncertainty results should be regarded as reasonable estimates, not precise statistical parameters.
The analytical support for NUREG-1150 is progressing well.
The rebaselining of accident sequence and core melt frequencies is nearing completion. Preliminary estimates are now available on five of the six reference plants. These analyses are now undergoing internal review by laboratory and NRC teams for technical accuracy. All planned Source Term Code Package runs have been completed on five of the reference plants and all analyses for the sixth plant should be complete by late April. Additional runs could be desirable, however, if the analyses identify new accident sequences that contribute significantly to risk. We do not anticipate that this would cause any major perturbation in the schedule.
Detailed containment event trees have been developed for two plants, and containment work on all reference plants should be completed with draft reports available for internal review by early June. The determination of a best-estimate risk level with a detailed uncertainty analysis is nearing completion for the first plant. We expect the first draft report for internal review by late April, with all submitted by late July.
Final reports will be available on five of the reference plants by early August, with the report on the last plant, LaSalle, available in mid-September. This information will form the basis for the NUREG-1150 discussions of the factors influencing risk and risk uncertainty at the six reference.
plants. The rebaselining analyses, when combined with earlier PRA insights, will also help identify those additional factors which could alter the conclusions appreciably if they were present at another plant. Such information will influence the ability to extend the results of the reference plant analyses to other plants of thE same general type.
NUREG-1150 will conclude with an assessment of the methods employed and information gained and it will provide insights on the use of such methods and information in regulatory evaluations and decisionmaking.
6 The origiral schedule called for publication of Draft NUREG-1150 in mid-August of 1986. However, as anticipated, the extent and complexity of the job has resulted in substantial slippages in the work contributing to NUREG-1150.
Nevertheless, we plan at present to have the draft available in September 1986.
Status of NUREG-0900 (Rev. 1):
The NRC severe accident research program (SARP) was planned after the accident at Three Mile Island Unit 2 power station in March 1979. Major research experiments did not actually begin until late FY 1982 since lead times for such comolex efforts are considerable. The program plan was documented and published in January 1983 as NUREG-0900, " Nuclear Power Plant Severe Accident Research Plan." The report listed thirteen separate elements of research to be completed under the plan. Although most of the planned work has been completed and much of it has been redirected due to changes in scope of work and reductions in the budget, the original document still serves a purpose as a general guideline.
However, in order to provide an updated description of the additional research to be accomplished over fiscal years 1986 and 1987, and projections of possible research topics in subsequent years, the original document has been revised to be issued as NUREG-0900 (Rev. 1). In this new document, the original thirteen research elements have been combined into three program elements (Accident Sequence & Reliability Research; Source Term & Containment Loads Research; and Containment Behavior Research). Each of these major elements is further divided into two sub-areas resulting in a total of six sub-topical areas of research discussed individually in the revised report. In addition, a new section has been added which describes all the severe accident codes being developed,-their relation to PRA studies, their relation to benchmarking less-detailed risk codes, and their exeri-mental validation schedule and scope.
Although NUREG-0900 (Pev.1) describes the information sources available to NRR to implement the Commission's Severe Accident Policy Statement and the details of the research supporting those sources, considerable information is already available from the NUREG-0956 and NUREG-1150 efforts described above, which originated from, and are supported by, the results of the original program defined in NUREG-0900.
7 Final RES and NRR staff comments on'the draft version of NUREG-0900 (Rev. 1) have been incorporated and the final document is scheduled to be issued in late March 1986. The document will reflect the changes mandated by the current OMB FY 1987 budget mark for the Office of Research. Most-of the source term research milestones listed in the February 18, 1986 memorandum from V. Stello to Chairman Palladino are also reflected _in NUREG-0900'(Rev. 1).
~ Victor- Stello, Jr.
~
Acting Executive Director for Operations DISTRIBUTION:
Cormnissioners OGC OPE '
OI i OCA-OIA OPA s ACRS EDO SECY
1 l
e W - e '. K OW f *<g W
% or -
e.. .
g y N,. ' ., ,r'if-# ' N -
g . ., . j.- e g.
c -
t N 3
~
4
l- P :
4.a$.$C__ @ P 'C4 #~.y' -m e--
(, - . ?.y(&&,a;;* %g_%.s%gg$@s{t k
V ,* Q,s,%,,',m{CC
- s. h. "* Y _ J L.
.,N<.- . n~
a r
- g. -y, < . -s.'*. r s
e
-i . - - -
...r'
.y - . v. ,
- .. % ~-el-arte.M, M , *-
N -
- Q <.ce" . - ~.s, .~ %" f. ,,1_3* ,c. v .- 4* . -
^ . *'._ .
d
, ~
~' -
li re f'ut.
{n ..c.v< ' s, x3.% { k ,' 2m'{
g[. _
y gp ' h~, -ys'
]48'am.s ...a .g
--. m. , m . . . . .
. r- e
~
%*. +---
w .q;. .r%. Gur g g.
- - >-a,n ee e= %* y m= - -
5 $2.ywn-M & h a g 2 W a j $.1( W d g...
.e itJ g_;.'- _
n' .ammw.w. ses m:7,. n . . .y;. ,
i W av8=,MR)af*.%h
- y. 9: .Acm.p:g;.,@%.
M.7 j i* .'4. .Ms -T~',ar * $ ? p :EfSf. _.
- 2. '
} [W .. g ' g$ f
-: .t. stt . _rt ang.g g g y..
.. p aEx erts4 nwh 3, C 3,.a.s ,.w mag App gg~.r- 1 E m 3 pm
. . w ~.. . . .- ~
- n.
. , . n . ..
u m ~ m- e n.- - +
A4d a . y+- M M.;
2 3.
- h. . .n -
e g
- r.qym. . y, %c y , y,p _
_ d m,
~
- l 'YQ .
4 4. ' kk W3 h c ws y::+* : .- ' fM . . % h,1. - ;. . k;:En f ; % s, : .:
, ' g , ,;;w -.;
y% . ?$? w$;
. . , ~
1g ~-
'* R f. -v.m y'f;:,kl:&NS h-Y h h. A
.ww; s.wc..sw f .;we.m'h*.w^
y '? .Y
.ww ppp 4y,gn _
g?WQW
, R N & f f M Y~~'. .AWbk &2 h W l$OO%COEN> % F . M Q M V F w.fY:i; 7:& n W. ..
- yww l YNM*WMNbkNhYdN NMNh W 4; g c 4,y gg,4.m..w Wi2W 2tW v > dg.$.u x,s.g-4%p.,p.. P. .
sg -
.,e ,.
p m. . :. g .
y ..:
n_, ,v . . -
,g6,,r ej:. ~
- j. .
fGvgf i i \ -
pg g k.: g;':
(w m.
wwr
&}W%wsm%
4.*.9 .;. + ,7 t
gt{ ; . - 'yf7%m.g.
.. p-sgm 15W7 4. .. M<..v . 4,, n M..-
e.
(*;'y,!. p m
Q ww.
g,yjK ' ,, f
..+.
W A. ,. .. .. 7 gk.w.w.gun e -i n .ggg -
-w g.e.g@g>.e.n,e;. ess
.r. .:- g. . .T YM t J%[. . ( -p;$. ./ L NQ. 'Jp.. N , f k: ' - -.NQy.Q tgs .
- $w%< . . : , c v ,.4
.3.. . . s, ,
,' , a.;;.ghy ' s e- - f ,., g. f. Wc - g s
- :, g
4 w~. r.
-vk
.!l*, .. . > y, .4'. .s f>%.
f* ..[s?*a .. .
~
n ' ,Q una --
l' Q h g Ys-y >.
N '*'; , _ .. .
~ y . . ..
?
-y , y* - 9..K '9t
.f
,. y,y
, + ?J '
- ,~h_; * ' . , _ . ~ky:"* ., y _ ' .'-
n.g. . m , * ?y. .%.**. .y i- - .;.,. .,>y+. , ; -
v
.m 4 .
. . b.p-. r. % y. . m- n ? -> ,.q, g y.5%.) ".v 4ej .
( 0, ?
- v. ], .l g .g f,, -- #'g M'!T ,.. ~ - . ;
}
p .
- I .- :q1 . . +4.4 . G. A. ." ~. . -. .z.. 4.e. n.- .u *
< 'Y*'
':--g.
g .~: . .M ..,'::w : s , g.v. ,,.g
- s. 3 ... / , .,
4 .
W,
.q.- ,
4u ..
f fL '\. .
.;J... ..:.
N :... . . _ . . %*h 5 f,. _ .>5N O 'T.f.ap:,$ch d M Q &. &g[-
- I' .
l l
-a- ..
4 9
NUCLEAR-REACTOR ACCIDENT Q l#wh[D4
- t ! Vdyp t(pM.PhgQ*M !
',my? O Report by an NEA Group of Experts March 1986 l
i 1
NUCLEAR ENERGY AGENCY ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT 1 '
1 1
! I
\
Pursuant to article 1 of the Consention signed in Paris on 14th December,1960,and which came into force on 30th September.1961, the Organisation for Econcmie Cowperation and Development (OECD) shall promote policies designed:
- to achieve the highest sustainable economic growth and employment and a nsing standard of living in Member countries, while maintaining fmancial stabdity, ud thus to contnbute to the developrnent of the world economy;
- to contribute to sound economie expansion in Member as well as non. member countries in the process of economic dese!cpment; and
- to contribute to the expansion of world trade on a multilateral, non-discrirninatcry b pis in accordance with international obligations.
The Signatories of the Convention en the OECD are Austria, Belgium. Canada. Denmark.
France, the Federal Republic of Germ.' . G reece, feeland. Ireland, Italy, Luxembourg, the Netherlands. Norway, Portugal, Spain eden, Switzerland, Turkey, the United Kingdom and the United States. The following countries acceded subsequently to this Convention (the dates are thee on which the instruments of accession were deposited): J4 pan (23th Arnt, 1964). Finland (28th January,1%9). . Australia (7th June,1971) and New Zealand (29th May,1973).
The Socialist Federal Republic of Yugoslavia takes part in certain work of the OECD (agreement of 28th October,1961).
The OECD Nuclear Energy Agency (NEA) was established on 20th April.1972, replacing OECD's European Nuclear Energ Agency (ENEA) on the adhesion of Japan as a full Siember.
NEA now groups all the European Siember countries of OECD and Australia. Canada.
Japan. and the Umted States. The Commission ofIhe European Communities takes part in the work of the Agency.
Theprimary objectises ofNEA are to promote co-operation between its Sfember gos ernments on the safety and regulatory aspects ofnuclear deselopment and on assessing thefuture role of nuclear enery as a contributor to economic progress.
This is achieved by:
- encouraging harmonisation of governments' regulatory policies and practices in the nuclearfeld. with particular reference to the safety of nuclear installations. protection of man against ionising radiation and preservation of the environment, radioactive waste management, and nuclea third party liability and insurance;
- keeping under review the technical and economic characteristics of nuclear power growtn and ofthe nuclearfuel cycle, and assessing demand and supplyfor the diferent phases of the nuclearfuel cycle and the potentialfuture contribution ofnuclear power to overall energ demand;
- developing exchanges of scientifc and technical information on nuclear entry, particularly through participation in common services:
- setting up international research and development programmes and undertakings jointly organised and operated by OECD countries.
In these and related tasks. NEA works in close collaboration with the International Atomic Energ Agency in Vienna, with which it has concluded a Co-operation Agreement, as wellas with other international organisations in the nuclearfeld.
hensee en franres ans as sense:
E A OR h L&Al Rappare esa Csemos esapena es FAeN O OECD,1936 Appi' cation for permission to reproduce or translate all or part of this publication should be made to:
Head of Publications Service. OECD
- 2. rue Andr6 Pascal,75775 PARIS CEDEX 16, France.
- w Q
The dominant issue in nuclear' safety over the last few years has been the source term. This refers to the quantity of radicactive catertal .shtCh might be released in a nuclear accident, its physical and chP91c31 form, and the other characteristics needed to completely specify its dispersion tn the environrent.
The importance of the source term issue was actentuated by the acci-dent that occurred in the Three-Mile Island-2 pl3nt in March 1979. This accident did not produce the fission product icakage expected frcm classical regulatory assumptions for accident source ter9s, or as ewpected frum the more quantitative source terms of the 1975 U.S. Reactor Safety Study (WASH-1400). The inert gas fission products behased about as espected, but all the other radionuclides ware released in surprisingly los amcunts, in particular iodine 131. While about siv million curies of xenon escaped to the environment, the todine leakage was only about 18 curies, less than one part in ten million of the iodine in the core. According to classical assump-tions, fission product iodine should have been volatile at the high temper 3-tures of a degraded core and c50ulJ have appeared as a gas in about the same amounts as the inert gases.
It was suggested in 1980 that the explanation for the unexpectedly low airborne release of iodine during the TMI-2 accident was that the re-ducing conditions in the vessel of a water reactor. during a-core-damaging accident would transform iodine into non-volatile iodide. Iodine would there-fore not appear as a gas. This view triggered a worldwide programme of ex-periments and analytical studies on fission product behaviour, for various reactor types and accident sequences. On the basis of these research efforts and analyses, it is now believed that for water reactor accidents resulting in core damage, iodine would appear as caesium iodide and the excess caesium (there is about ten times as much caesium as iodine in the fission products) as caesium hydroxide. Both compounds are highly soluble in water and are much less solatile than elemental iodine nr caesium. Moreover, they can form aerosols that are subject to depletion processes in the reactor primary cir-cuit', as well as in the containment building. Therefore, todine and caesium source terms may be much reduced from the WASH-1400 estimates.
If this assessment proves correct, it could have major implications, in some cases, for such regulatory issues as plant siting and emergency response planning, since in the event of a major accident, the potential radiation exposure of the neighbouring population would be much lower than previously assumed.
However, a number of questions remain, and the previous conclusions must be treated with caution. In order to arrive at a realistic estimation of accident source terms, the NEA Committee on the Safety of Nuclear Instal-lations (CSNI) het set up a number of activities in the areas of source term assessment, reactor containment atmosphere control systems, and acci-dent consequence analysis.
4 9
~ ~
Cne of the most stqnificent activit tes carried out in 1985 .ns an interconparlton of the s arious *nucce term studies published so far m.J ,n analysts of the results of recont ..crk per ormed f
in ()ECO " ember auot r ies.
The intercomparison covered infor'ation from C3nid3, f~ror, =,!b' Federal Republic of Carmany, Italy, Japan, Leden, the United Xa ,. ,, e j ' t .e United States (Nuclear Regulatory Commission and Battelle C;i ,rbus L-ir ra-tories, herican Nuclear C ;tet y , American Physical Sec tety, Indu;trv-Degraded Core Rulemaking S cogram, hre 5 '.lebster Crijineer :nq Corpret t ,,
New York Power Authority; Electr c P, er Peue:irch Institute). IS 2 .nter-corp 3rison pointed cut areas hete considerable prngress h3S DeCn 'Jde 3 W.;e WASH-1400, areas wl-re nource term information is censidered uut f te tent ,
barrters to wplying eurce Leta information to different itqht .;ater re c-Lor pl3nts and generic ta;ues ,pplicable to most pl3nts, uncert a tnt u s 1 il areas of disagree .ent between ,ar taus saurce tetm stuettes, and ro.; c ~ -l a -
t t r1 fer & ,t.nq .tth the un.:ettatottes.
This report is a sur.ary m of the findtrg3 of the interc rparis;n. "7te detalleit Infor~ation supperting the su a,ary report will be published in '! e form of a Tethatcal Annes falle.4ing peer review.
4 w.
W
e TABLE OF CONTENTS I. EXECl*flVE SUM.9ARY
- 1. Introduction ................................................... 7
- 2. Main Developments Since WASH 1400 .............................. 8
. 3. Areas where Source Term [nfntmation is Suf fic ient . . . . . . . . . . . . . . 9 4 Barriers to Applying Source Term Information to Different LWR Plants and Generic issues Applicable to Most Plants ........ 10
- 5. Uncertainties and Areas of Disagreement between Studies . .. . . ... 11
- 6. Recommendat ions for Dealing with Uncertainties . . . . . . . . . . . . . . . . . 12 II. TECHNICAL
SUMMARY
, 1. Sufficiency of Information and Progress Since WASH-1400 . .... ... .14
- 2. Areas o f Disagreement Between Studies . . . . . . . . . . . . . . . . . . . . . . . . . . 34
- 3. Current Uncertainties in Source Term Information . . . . . . . . . . . . . . . 39
, 4. Implications of Uncertainties in Source Term Estimates ......... 44
- 5. Bibliography ................................................... 48
. List o f Task Force Members and Other Contributors . . . . . . . . . . . . . . . . . . . . . . . . 49 c
e-E 4
l
[ p
- ~ - -- - , , - . ,< , , ~ , e- r - - , - - , . ,
I EXECUTIVE
SUMMARY
- 1. IflTRODUCTION 1.1 At its Nove-ber 199'+ meettrg-the CSNI decided, following a prepocal made initially by its Sub-Cc:mittee on Licensing, to set up a Special task rocce on
- Source Terms, whcce *atn task sould be to intercompare the sarious scurce teen studies published in late 19di'early 1985 with a view to :
i) . identifying ateas where scurce term infor, mat 10n appears to be sufftetent ;
ii) identifying areas where current studies disagree. and why ;
iii) identifying generic source term issues that could be applied to most LWRs ;
iv) identifying barriers to applying source term information to dif-ferent LWR plants.
The Task Force was also asked to draw the attention of the Sub-Committee to areas where, the new technical understanding resulting from the review of current source term studies may have direct and broad implications for nuclear safety and regulatory issues.
1.2 The following source term studies were considered by the Task force :
- the Report of the American Nuclear Society.Special Committee on Source Terms ;
- the Report of the American Physical Society Study Group on Radio-nuclida Release from Severe Accidents at Nuclear Power Plants ;
- the technical Summary Report of the Industry Degraded Core Rule-making Program Study on Nuclear Power Plant Response to Severe Accidents ;
- work performed by and for the USNRC, including BMI-2104, Radio-nuclide Release Under Speci!1c LWR Accident Conditions ;
- studies performed by the Stone & Webster Engineering Corporation ;
- studies' performed by the New York Power Authority ;
- studies performed by the Electric Power Research Institute ;
i l
y e p
, , - - - --, , . . ~ , - - - . , -n-, , , . , , - -
~
- work performed in Canada, France, the Feder31 Pepublic of Cerman,, ~
Italy, Japan, Sweden, and the United Kingdom.
1.3 l' embers of the Task' Force utsh to express their gratitude to the var 10us organisations. concerned, for their uupport throughout the study and for the large ascunt of documentation made available.
.' 1.4 Finally , the phrase Scurre Terms has been useo throughout the . l tt.o id .ge
. with a sariety of meanings. In the past, for exaeple, the amoant of *ater tal released fecm the primary circuit ta the cuntaineent and the fracti nal tet,: 4 of the core inventory to the ensironment have both been called the ;ource to' ,.
In_this Report the Task Force has used the definiticn adopted by t i e CSNI Ceo. .-
Group of Experts on Sesere acciden's. For completeness, the Senior Gruuc F,s defined the ?curce Tera as the quantity of radioactise material .. filch T.t9ht P.c released in a nuclear acciJent : its phy; teal and Chemic31 form and tPO at'N r quantitles needed to completely specify its dispersion In the ern to v. o l. e j.
energy in the plume, height of release, duration of release, etc.). Uf coucae.
the pecbability of the varicus accident scenarios must be considered in paralict with the source term.
1.5 The following sections of this Summary present a brief overb tra of the Task Force's main conclusions regarding the sufficiency of current scutce term information and the progress made since WASH-1400, the areas of disagree-ent between the various source term studies, and current uncertainties in Scurce Term information.
- 2. MAIN DEVELOPMENTS SINCE WASH-1400 2.1 Considerable progress has been made since WASH-1400. In particulir. -
problem identification and definition have been accomplished, and source terms can now be calculated in large measure on a mechanistic basis.
2.2 The new information indicates that source terms were overestimated in the past for mest accidents. However, the Task Force has noted that for some types of plants and some sequences, the calculated source terms for some species are presently higher ; further work may change the situation.
2.3 Source Term technology is more complex than the treatment in WASH-1400, and generalisation of source-term values to different types of plants is not possible. Most of the tools for applying the new source term information to individual plants exist and are being further developed.
2.4 Important factors lea' ding to reductions in the release of radioactivity relate to. plant performance. For example, containments can withstand pressures which are significantly higher than their design values and will fall, if at all, at longer times.
2.5 Operator actions have been developed as a first priority to. prevent core-melt, and as a second priority, to mitigate its consequences. Relevant operating procedures are dependent on national practices in this area.
- 8-e mew
l 3. -AREAS WHERE SOURCE TERM INFORMATION IS SUFFICIENT (with qualifications)
For tee purposes of this Report, the def tnit ton of sufficiancy dich has bee, taken is that no substantial R&D vork addit tonal to that in progress or planned is judged neceusary for early recolut ten of the issue- in quest ion. This j will be conditional not only on the magnitude of t he associated ancertaint :es, but also on the likely impact of. these uncertainties en final source term c it t-mates. The magnitude of uncertainty that can be tolerated is expected to depn.d un the particular application.
3.1 The insentory of radioactivity is ame tated with relatively Vall uncor-tainties, and informat ton is considered auf ficiont.
l 3.2 t!ast important in-sessel thermal-hydraulic phenomena are thought ta base j been recognised, also the need for close roupling of thermal-hydraulic codes with codes for evaluating fission product deposition, self-heating and re-i I volatilisation. The current methodology is probacly adequate to allow ptedtc-tion of RCS fission product retention in the early (i.e. pre-heat-up) phase of accident sequences (i.e. large and interfacing type LCCAs) in which natural circulation in the core and RCS plays little part. Elsewhere, further develco-eent of the methodology and reduction of uncertainties are underway, and util give improsed predictions.
3.3 Existing ex-vessel codes are thought to be adequate to predict long-term l overall thermal-hydraulics as they impact on containment loading. Phenomena descriptions relevant to short-term thermal-hydraulic behaviour.are less well developed, especially with respect to the effect of steam spikes, direct heating, and hydrogen burning. There is a consensus among experts that in-vessel steam explosions of suffleient magnitude to fail both the RPV and the contain-l ment are very- unlikely, .but there is still some disergence of views concerning the quantification of the probability. Further information is being developed on the survivability of some safety-related equipment in the contajnment. Also, l
' further thermal-hydraulic work is underway with regard to fission product trans-port in tne containment, notably in relation to departures from good mixing, turbulence levels, intercompartmental flows, local decay heat effects, and a combustion ef fects.
3.4 There is sufficient information to demonstrate that some containments are stronger than previously thought. However, further information is needed on sources of leakage prior to failure and on the failure mode,. i.e., to determine to what extent leak before break would occur for dif ferent plant designs.
3.5 The available information on the total amount of in-wessel fission product. release from a mciten core is largely sufficient for the volatile species (Kr, Xe, I, Cs, Te) when the data are used in accident analysis. Further data are needed on the identification of released c"'ical species and the ki-netics of volatile fission product release from futi prior to complete melting.
With regard to the fission products of intermediate alatility'(Ba, Se, Mo, Rh, l Pd, Tc, Ru), there are differences and gaps in the kinetic and species identi-l fication data. A complicating factor in some accident scenarios is the conver-l sion of intermediate volatility species to more volatile species. A complicating factor in applying kinetic data for the volatile and intermediate volatility fission products to accident scenario calculations is that the actual process
[ -
y i
Q-
' of core degradation will involse se cuhat random diffusion path lengths the. g h the f'et u rods to the cludoing breaks for the fisstun products. Cspcti ental release dati for the non-.olatile fission products (i.e. , t hn race e irth i' 3r. I limited,.but it is believed that these species will be largely ret ain d in u o melt during the in-vescal phase of an accident, and the up;ct af '9 tn to.A .if data is unlikely to be important ~in relationship to the f toal :,urce ' rms.
1.6 Acrosol behat iour to the cent sp- ent atmospt. ore is cne of the Nttar deselcped areas in source term methodology, and is .031! mpputted by e g,eri-1 v ntal data. There are seme remaining is ,oes, su h 3s turhulent '.yjh 5 - i ti'n and aerosol shape factors. Alan, Lt is recri nded that f uture v.w:e totm os alvations be undert aken using codes incorporating ccupied t hct ,l-h A..u' ;)
acrosol !,chas tour, and ha. triq , ~ultt-cerpartment ad alti-cu punent epW i '
3.7 Infutmt ton un the long-term tchav tour of iodine ia cntatt. ent .. l!
develeped. Additional infatmatten un radiolyqis ef fects, impurit ies and 'ury, ' .
tedtde predoetton is being deseloped.
- 4. BARRIERS TO APPLYING SOURCE TERM INFORMATION TO DIFFERENT LWR PLAN'TS AND GENERIC ISSUES APPLICABLE TO MOST PLANTS The methods for source term evaluation are largely generic and applicable to most LWRs. However, generic applicability of quantitative results to all plants is limited by such factors as the power level of the plant, geometries (such as containment and auxiliary butiding layouts), and other specific plant features' (such as dif fering valve. characteristics and type of concrete used in
' the containment). Nevertheless, some imoortant generic qualitative conclusions can be draan :
4.1 There is general consensus that LWR containments are stronger than 4
previously thought in WASH-1400, thus reducing the likelihood of early contain-ment failure for some containment types.
~
4.2 Radioactive core inventories are principally a function of reactor pcwer.
and the fuel cycle, and independent of reactor type.
4.3 The release rates of fission products from the fuel are, for the most
- part, generic to different LWR reactors. However, if there are significant
! differences in the type and amounts of non-radioactise materials in the core, it may be necessary to carry out plant-specific modellirq to describe the sub-sequent behaviour of fission products. Examples of such differences are the amounts of zirconium, the composition of control rods (e.g., Ag-In-Cd, 8 4C, and Hf are all used in LWRs), and the presence of boric acid in PWRs.
4.4 Under the reducing conditions, likely to pertain in-sessel, the dominar t forms of the volatile fission products immediately following release from the fuel are Csi, Cs0H and Te in all types of LWRs. The effect of boric acid (which is present directly in PWRs or can be produced from 84C controi rods exposed to steam in BWRs and some designs of PWR) to modify chemical forms of Csl and CSCH in the core region or elsewhere in the primary circuit, is under investigation.
7
- 5. UNCERTAINTIES AND AREAS OF DISAGREEMENT BETWEEN STUDIES
~
5.1 The various source term studies under. review had disergent siews as to the implications of. the present source term predictions. The APS felt that ,
although .the source terns for a nw<ber of caso9 were calculatad to be looer than WASH-1400, it was not pussible to mnke the generalisation that source tert 9 are lower in all cases. This agrees with the results of the UNI-21CA study. The [DCOR.results were generally Ic..er and the ANS resiew concloded that Ocurce terms would be lower with only i ; *12 number of ex epticas.
- ~
5.2 The Task Force's review of the various schrce tern studies indic ites that the. uncertainties in predicted source term valtes depend on the specific
). plant and sequences, particularly the time of centatnment fat ture, and on the r:pecific set of cocco used for the dcte atr ation.
5.3 Dif ferent code sets currently result in dif ferent souice term predic-tiens for identical plants and acquences. ThcGo disagreaments can only be reconciled in terms of the uncertaint tes. These uncertainties arise due to differences in the following areas :
- the mathematical models representing the physical phencmena governing the processes ;
the correlation and specification of the physical properties of the various materials of interest ;
- the ina'dvertent omission of possibly important phenomena ;
- the boundary and initial corditions related to the specifications of the accident sequences and the plant geometry ;
- the numerical approximations involved to the application and solutions of the various mathematical representations of the phenomena.
Within the above categories, some of the items that have been identified among the_various studies that have contributed to the differences in the predicted source terms are listed in the following table :
SOME AREAS THAT HAVE BEEN IDENTIFIED AS CONTRIBUTI'.G TO DIFFERENCES IN CURRENT SOURCE TERf1 PREDICTIONS
- 1. Differences in the modelling and predictions of the RCS thermal-hydraulics, including core melt progression, hydrogen genera-tion, and natural recirculation.
- 2. Control rod behaviour and aerosolisation.
- 3. Differences in the models for aerosol behaviour in the RCS and in the containment.
- 4. Retention features included in RCS fission product transport.
- 5. Details of treatment of revaporisation from RCS surfaces.
1 . . _ . . . . . . _ . - . . . .
o.
6". Aasumptions on fuel dispersal on ejection from t.be m:5 /quan-ttty of fuel and geometry of corc/ concrete interactions).
7 Inclusion (or .not) of airbotme water dropleta et.e to tee int-tial bleadcun of some sequences.
- 8. Many differences in the ecdelling details for core / concrete internetton, thermal-h dtaultes and fission product,'ccra al i
release.
9 The assuq tiens an t.he amount of by-pass around suppression pool 9.
- 10. Ei fecti.eness of suppreastun puols and ice cundensces for rem.osal of aerosols.
- 11. The degree of subcorpartmentalisation of containment and contalment butidings.
- 12. Contatnment chermal hydraultes related to the timing and extent of steam condensation, with respect to aerosol sources and to floa among compartTents.
- 13. Assumptions on the locatiens of breaks in the RCS and con-nected piping for s0me sequences, and the local envtronment around the break.
14 Mode and timing of containment failure.
- 6. RECOMMENDATIONS FOR DEALING WITH UNCERTAINTIES 6.1 The lask Force observes that experimental and analytical results for source term prediction will always be accompanied by some level of uncertainty.
Ideally, a distribution of the probability of occurrence of a particular source term for a given sequence is desirable and would be of most benefit for requia-tory applications. However, it is not likely that such distributions will be forthcoming on a timely ~ basis. In the meantime, it'is recommended that the as-
.sessments be based on best-estimate (neither intentionally biased high nor low).
values taking into consideration relevant experimental and analytical results for uncertainty bounds and, if necessary and appropriate, expert judgement. It should be decided by the different countries how the best estimate values and related uncertainties are to be used for any particular implementation.
6.2 The Task Ferce has recognised that the significance of many uncertaintiet varies with the timing and mode of containment failure. Thus, the number of un-certainties needing evaluation can be reduced by categorising them in terms of
~
containment failure modes.1he following is a possible categorisation of some current uncertainties.
y
o Early Cuntainment failure :
~ - RCS te,tentton/ch.ntut.ry ;
-- Ru exidation on.blyh pecucuro ejection and/or steam esplostano ;
core / concrete.interacliun :
- water decplet characteristics rfortnq initial bloudewn.
Late Containment failure-:-
- core 4oncrete interacticn ; -
some anpm:ts of en. sot hc53s tour in cont.un ent :
- revaporisation fec, RCS _:
- suppression Pool Cr4 :
^
- ice enndenaer ;.er formance.
Containment Bypass :
- performance of aus tliary hu1 Lif tngs ;
- RCS retenticn ;
- SGTS per formance ttk t , lik![, Gh'a s ) .
Leakage, Pre-Existing Openings, and Failure to [solate :
- characteristics of leakage pathway ;
- location of leak ;
- same uncertainties as for early and late containment. failure.
6.3 There is a time, depending on the particular sequence and reactor con-sidered, beyond which aerosol removal processes will reduce the source term to an extent that the uncertainties will be unimportant.
6.4 Since the Task force concludes t hat these uncertainties should not delay implementation of the new source term information, then the research needed to reduce uncertainties should be'part of an iterative process. A comparison on the application criteria (selected by regulatory agencies) with the source term uncertainty distributions will determine the need for additional research or code improvements.
l l
i I. I i
t i
I 7
R
~
II T E C H N.I C A L
SUMMARY
The !:l: ~ win; T+:hn ic n1 ^ - ~ , c] is b nei On :!.c 1[.;1n e :. ' . :r ~ ' .
- ?:hnical Annexe3 wh a:-h x1 L: te v * :nb:c, 1!:v: :stnful ; car-ro ..m :1 < ~~
a:nths ; ic Jia: .:ntain3 n::: a;;:n 1: s. . ai in pier. icy 14511.1.1 ^: :b:-
Tan Force.
The +:hnical 3cracy :s les:::tei in:o thcce sa::icns >
ll Suffictency of En!: rsci:n snd Progress Since WA3H-h*M,
- 2) Aceis ! DisaJteeren: 3etween the 5:udies, and
]) Cu:cen: Unce::1Ln:les in 3:urce Tern Informacion.
- 1. SUFFICIENCY OF INFORMATION AND PROGRESS SINCE WASH-1400
1.1 INTRODUCTION
In this section on the source term, we will address the questions of suf ficiency of information both in the context of :
- 1) the applicability of the current methodology to the series of phenomena involved in source term estimation, and
- 2) the plant specificity aspects of the current methodology.
Additionally, comments will be offered on the progress made since WASH-1400.
The phrase source term has been used with a variety of meanings. In tne past, for example, the amount of material released from the primary circuit to the containment and the fractional release of the core inventory to the environ ment have both been called the source term. For completeness, we here define che source term more generally as the quantity of radioactive material which
.might be released in a nuclear accident : its physical and chemical form and the 'other quantities needed to completely specify its dispersion in the envitar ment (e.g., energy in the plume, height of release, duration of release, etc.).
In considering the sufficiency of information for various steps in the evaluation of degraded core accidents, it becomes clear that this is a subjec-
- tive matter. It is related to the question of how good is good enough. The answer to this question can depend on the country addressing the question. For
. g
example, the Ceucan PNS programme ettes the fact that their PhR r?3ctors base ccrparatt.ely strong containments c e pared to the .~..R tce egndt'ncer plants.
Therefore, they reason that they need not be as concerond about the thermal-hydraulic phenomena in the primary circuit 3ince f tSJton pr%ct. retent iun to the primary, whatever it may be, is not consideted necessary to rub 2 ate 3 low cource ters.
To say that an evaluatiun of the suf ficiency of- infot ?at ion 10 a 9ohjac-tise matter und can sary frem gruup to grep to Sm.cser, not an epitc4 tion that the matter is unteportant. The solutton to this proble, lies in the ,,e of the best Pult idiscipline high-Ics e t technical judgNent.
- [n r. m : pot' t , the 7::emil it,qr m t hat has twen a,de ace d%H-l:)O iO . cry grittiyD.g. . . ,b it . . . , , . >l , i'l ' n c bcen 1 .p! 'ed y .al wd, but the perh'ps more dtf ficult 9teps of preblem tdentificat t :n and defintt +yn hwe vparent ly been latgely acevnpli .h"d. 'fe can to mmt case; caletelmt e uaurce terms.
The follovsing Sections .1111 consider specific areas in the dotetmtoat tun of source terms from the v iew points of :
a) accompitsnments since MASH-14C9, and b) the general applicability to sarious types of plants.
1.2 OPERATOR ACil0NS i
! -The actions taken by the plant operators may strongly influence the l course of a hypothetical accident and thus lead to a di'ferent accident scenario j and change the source term. The operator actions are guided by the procedure
, guides furnished to them and are country-specific. The control room operators l may' commit diagnostic or manual errors or, conversely, correct an abnormal plant I situation. In case of a malfunction of safety systems, once an accident has been
! initiated, personnel may succeed in restoring the appropriate system functions l (repair).
! It seems, on the basis of the conclusions drawn from recent source term studies and from operating experience, that recovery and repair actions by the I. operating crew and by external backup personnel have not yet received adequate recognition in the analysis of accident sequences.
l The technical annex to this report which is undergotng peer-review 1- contains an excellent description of the operator intervention situation. In a survey on a country-by-country basis, the approaches used by France and the ,
United States are specifically considered. In the case of the United States, these approaches are also considered on the basis of reactor type.
l 1-
. Much progress has been made in this area since WASH-1400. In particular, l TMI-2 emphasized the reality of degraded core accidents. It can be espected l
, that the methodology of estimating the source terms will be used in evaluating various operator intervention strategies in the future. Because this matter relates to procedures specific to reactor types and some accident scenarios, if the operators are following their instructions, discussions cannot be 1
?
-- _- =, . -
, y
- .- . . _ . . ~ . . .- , . . - .. .. . .
generalized.and therefore the render is urged to consult the matertal presentea in the annex to this~ report ,. hen it becomes available.
In summary :
. Progress Since WASH-1400 t
- . Operating procedures, should a severe accident occur, are currently
- ' at various stages of development in many countries. Significant progress has been made in their development since WASit-1400.
. Cencral Applicability to Various Plants 4 The various ptocedures for operator intervention are plant-upccific and depend upon national policies.
~
1.3 CONTAINMENT PERF0HMANCE Containments are of special significance since they present the release of fission products even if the primary coolant system has lost ~tts_integetty.
[his, of course, sets aside .for the purposes of this discussion the posstbil:t f of containment by-pass scenarios (i.e., the V sequence 1. An sddittenal matter of considerable importance is that cperating esperience indicates that the possibility of untntentional openings cannot be neglected. Efforts must be j
directed toward-reducing the probability of such inadvertent breaches. The containments are designed to carry the design basis accident pressure and tem-perature loads and remain essentially leak-tight, i.e., exhibit an acceptably ,
low leak-rate at design pressures and temperatures as prescribed in the Design Basis Specifications. The acceptable design basis leak rate is in general at 4 design pressure and temperature. These are :
for PWR : 0.1 per cent per day of the contained volume for BWR : 1 per cent per day of the contained volume.
- Design Basis Accidents such as large Loss of Coolant Accidents result 4
in pressures, temperatures and leak rates, that are accommodated within the specified Design Basis Containment performance capabilities.
4 Most degraded core and core-melt severe accidents will also be accom-modated within the pressure, temperature and leak rate performance capabilitie9 of the containments. Some severe accidents will result in pressures, tempera-tures, and leak rates exceeding the Design Basis Specifications. In these situ-ations it is espected that generally the leak rate will gradually increase witt pressure and temperature. Containments designed to carry a specific pressure load will be able to maintain integrity up to 2 times and sometimes more than i the design pressure, which is the ultimate load carrying capability.
A few severe accident sequences can conceivably, in the absence of leakage, result in pressures exceeding the ultimate capability and over-pressurization failure will result.
)
7
I .
ed The annes to this report will present more detailed coerent0. T.-so impor-tant points to be found in the more lengthy discussion are :
1.-The potential for containment leakage through penett3 tion 4 prior to reaching currently reported containment capability pr'9urc should be conaidered in sesere acetdent risk estim.ites : "d
- 2. Failure of nonectallic seals for cuntainment penetrations 'pr tmartly
~
pressure unsealing equiprent. hah hes, personnel airlocks. Jry-eli neads and purge valses} 'are potential sources of enotainment !caiage.
In summary :
. Progress Since WASH-1400 Since WASit-1400 was published, more thorough analyses have indicated that containments can be expected to withstand higher pressures than were previously assumed.
g . General Applicability to All Plants y Containment response is specific to a given generic containment design.
1.4 FISSION PRODUCT INVENTORY The first information that is needed for the calculation of source terms n is a detailed inventory of the radioacti%e isotopes in the core. A number of computer programs, such as ORIGEN-2 and FISPIN, are available to calculate these for various core operating histories.
Fission product. inventory predictions of ORICEN-2 for example have been compared with the inventories of actual fuel pins taken from operating reactors with well characterized operating histories. Differences between ORICEN-2 cal-culations and experimental results are usually within 10 per cent for U, Pu, Am and Cm. Agreements noted were considered acceptable for analyses of severe accidents. Comparisons of decay heat with the ANS standard showed good agree-ment over a range of time from about one minute to several months. The FISPIN program has also been validated against decay heat measurements from which it was concluded that the predictions are within 8 per cent of experiment.
3s The prediction of the core inventory of radioisotopes 1s not associated 2 with a great deal of uncertainty. Although not all computer programs predict th the same numerical value for each radioisotope, it is the least uncertain por-tion of reactor accident analyses and is generally regarded to be sufficiently accurate for the source term determination.
, - . _ t
g L .;
y In summary :
. Progress Since WASil-1400 inventory matters were also well undersinod at the Lim of WASil-14noc
. General Appilcability to All Plants Applicable, 1.5 FISSION PRODUCI RELFASE The attention of the reader udould be directed to the arnes oft!e report (tu be published later) which contains a thorcugh discu70 ten uf the fission product ehe'nistry in the fuel _and 'in the near vietnity of the ftiel after release.
The chemical states of the fission products in the fuel can to char gej as the core heats up and finally melts in the course of an accident. Tha ec1t important factors in this regard are :
a) the composition of the steam hydrogen mixture through the care, b) reactions of fisston products with the zircaloy cladding matertal, c) possible reactions with boric acid soluble neutron absorber or egC control rod materials, and d) reactions with other control rod materials'(Cd, Aq, In),. steel constituents, etc.
The ratio of st'eam (an oxidant) to hydrogen (a reductant) sets the effective -
oxidation potential in the course of a meltdown accident which in turn can influence the chemical form of elements released and the degree of zirconium .
, oxidation.
.It is generally agreed that the dominant chemical species of iodine released from a degraded core is CsI*. When the caesium uranate in the fuel is exposed to steam upon fuel cladding failure, Cs0H would be formed and would be the principal caesium compound escaping from the core. The detailed knowledge of tellurium chemistry under the conditions expected in reactor accidents is limited. It is currently expected that released tellurium will mainly occur as elementary tellurium'or hydrogen telluride. Under highly oxidi2ing conditions it was found experimentally that the chemical form of the tellurium, which was produced in the SASCHA-facility and had condensed on aerosol particles was tellurium dioxide. There is still some uncertainty about the chemical form of the released ruthenium whicn depends very much on the amount of oxygen present.'
- It'is also possible to react Ba0 and Sr0 with steam to give relatively volatil hydroxides. In c'ertain cases this position may change if the Cs! has an op:or' tunity to react with boric acid.
- 18 -
+ + * * *
- ep me , me.e . ,
e . . . .
As the exposure of fuel particles to air losygen) shculd not occur inside the *
,essel esen in the case of 'a c. mall steam esplosion, the potential for ruthenium astdatten during in-sessel molting shuuld be quite small. A signtficant release uf barium and strnnttum is obsersed tf there is little or no oxidation of the eircaloy cladding so that the hig'ily reactise zirconium is able to redore Ca0 and Sc0 by converting them into the nare sotattle elemental form. It tu also possible to react Ba0 and 500 wtth steam to gtse relattsely .olattle hydrosides.
It should be pointed out that the core is not at a untform teiperature and releaces of differrot species are not isolated in time but may in fact occur s imu l t ancona ly in different regions of the cure. This may potenttally affect chemical speciat ion if the different apet tes are all airborne at the sa o t t+e.
The dtt fotent ~odels and codes for esttmating the release of fission products and structural ~Perials from the caro during heat-up and eciting are referenced in the ain body of the repart. Far solattle fission products, Goth as Kr. Xe,1, and Cs the practical utility of these.dtf ferent modela la cid ar low in those accidents phere core eelting iu involsed, since all estimates aqrce that nearly all of thesc ele *ents are emitted prior to core slumping from those parts of the core which base. reached melting te*perature. for dwell times of 10 mtnotes for the core fractiun at ! 700 oC tt is estimated that more than 90 per cent of the volatile fission product inventory of this core fraction as released. The uncertainty in estimating the release of~ these cicrents in core melting accidents is more associated with the uncertainty in evaluating the temperature distribution in the core than with the uncertainty in estimating the release rates themselves.
In those cases, however, in which fuel overheating, but not melting, would take place, the choice of the escape model can be important because the differences in escape predictions between models tend to be greater at these lower temperatures.
For the less volatile fission products, such as Te, Se, Ba, Sr, Mo, Rh, Pd, fe, Ru, there'are significant differences in the data from various sources.
For Te, differences in volattlity shown in different tests have been explained by a reaction of Te with metallic zircaloy to form nonvolatile tellurides i the more the zircoloy becomes oxidized during the test the greater the relesse of Te. There may also be other te complications. There is also evidence that Mo may volatilize, perhaps in combination with Cs.
The nonvolatile fission and activation products are beltesed to be nearly totally retained in the melt, at least till the time of vessel failure. The experimental release data for these elements are rather limited, so that the determination of escape fractions is relatively uncertain but these values are all quite low.
It should be noted that the uncertainty in estimating release of the less volatile and nonvolatile fission products depends to a great deal on the uncertainty in evaluating the temperature distribution in the core.
In addition to fission products, other core materials present in cladding, ils structural components, or control rods can be vaporized in an overheated core 9- and contribute to the aerosol mass in the reactor coolant system and the con-tainment. The vaporization rates of the most important non-fission products
. - .. y
o Like fe, Cd, Ag, Zr, In, are known to a sufficient degree. It should be appre-ctated that silver and indium, melt and flow to cooler regions of the core where they will very likely pool or resolidify. The control rod materials wtll be vaporized and' escape as an aerosol as soon as the temperatures beccme high enough. Thus, relocation of these matarials can cause a delay in tire at which these materials may contribute to the aerosols in the accident coquence. There-fare, in-sessel volatility and nonfiusionable core maternal behav tour are de-pendent on both the accident sequences and the control rod compostttons v.htch
- are different for PWRs and BWRs. These releases are irrottant because they will determine the aerosol behaviour.
In summary :
. Progress Since WASU-1400 Significant new experimental data have been obtained. Codes pre-dicting fission product release have been developed. There is an increased understanding of fission product species.
. General Applicability to All Plants Most data are applicable to all plants with' exceptions of behaviour of core material which can vary from plant to plant.
1.6 IN-VESSEL AND PRIMARY C00LANf SYSfCM fHERMAL-HYDRAULICS The in-vessel reactor coolant system thermal-hydraulics has been care-fully considered by the Task Force. Although there are aspects which need fur.
ther investigation, the degree of understanding is much greater now than at the time WASH-1400 was published. Indeed the thermal-hydraulics after core degrada-tion had started was generally ignored at that earlier time. The aspect of coupling of thermal-hydraulic codes with fission product heating, revolat11123-tion, aerosol transport and chemical reactions of fission products with struc-tures is treated in the next section of this report.
In general, conservative calculations can be performed and are often appropriate. However, conservative calculations may be conservative only with regard to a specific question. Detailed knowledge may be necessary to do calcu-lations which support a reduction in the source term. In this situation, because some of these important phenomena such as the mechanics of major core degrada-tion in a large core, cannot be studied experimentally, engineering judgment is necessary. Insofar as this situation is judged to be unsatisfactory, the simu-lation of the behaviour of large cores may be said to be inadequate. However, it may be too difficult (expensive) to completely fill th19 gap. It should be pointed out that core degradation and core relt may show some random beha<iour.
Another problem relating to the core thermal-hydraulic methodology is that different amounts of hydrogen are predicted to be generated by various codes. This is thought to be mainly due to different assumptions in the models of core collapse. Therefore again, the lack of knowledge of overall core be-haviour, especially blockage formation and internal flow within the core causes some lack of confidence in the results.
- .. . - .. p
Na(ural convection (mainly for PWRs) has only within the last year or two been recognized as an important mechanicm for energy distribution, p1rticu-larly betweeit the' times of core uncesery and large scale care degt a ht ten. This phenomenon leads to a d.stribution of energy throughout the pri ary c a! ant system and may change the predtcttons of the course of'the acctient la m nq,
- for example, to the prediction of a system failure at a location n " .,* r ' h in the bottom head. This phenomenon is incompletely -odelled in many cede; and the esperimental data are as yet incompleta.
I Another phenomencn which must be *entioned is that a ste m r.ph,3 ten m w occur if .nolten fuel ts poured into a cold fluid result trq in a f 3at mergy transfer from fuel to cold fluid .thich may threaten RPV toh*gr tty. tt n jerer-ally understood that the most tmportant ingredunts in the ancw ent of sto nm esplosions are :
- 1) the a cunt of melt part telpating in the raptd wit ' v iter inurxt tcn.
- 2) the ratto of energy ohich is transferred into the ater, rd
- 3) the energy needed to fati the RPV.
The physical phenomena related to steam e= plosions have teen studted in dif ferent espertmental facilities by different organtsattons with different matettals.
Also, suostantial theoretical work has been done to address the way the three major contributors mentioned above centribute and/or affect the assess ent of steam explosion. Based on the esperimental and theoretical work tt has been ecncluded that the participating mass is considerably lower than necessary to threaten RPV integrity. It is hcweser, generally agreed that small steam esplo-sions may occur.
o Also recently an expert grcup (SERG) has agreed with the broad consensus
- that steam explosions of sufficient energetics which could fall the containment have a very low probaotlity. .
A final point cencerning generality of the thermal-hydraulic codes should te nade. This ts that, at least in the United States, codes such as MARCH were developed with the PWR primarily in mind and the codes were applied to Bars with sete compromises. This situation needs to be retedied.
se In sununary :
(summary of related matertal also in next section) s
. Progress Since WASE-1400 hach progress in identification and understanding of major physical phenosuina and aystom behaviour has been made. The esperimental deta base tus been increased. Much progress has been made in code devel-opeent to simulate the realistic behaviour.
Identification and understanding of mejor local phenomena has been accomplished along with modelling of conservative core heat-up, melt s formation and RPV failure. The random behaviour of highly degraded cores should be recognized.
y
. Ccneral Applicabilite) to All Plants Must of the physical phenomena can be applied to PWRs and UWRs. Coco and nystera behaviour is plant-specific, especially for interna.l ilnus and RPV or RCS failure.
. 1.7 REACTOR CHOLANT SYSTEM (RCS) illERMAL-HYDR AUL ICS REL A T ING l0 F[5SluN PRUDUCi l R ANSI'UR I /CUHL UElf AV IUllR/RCS I A ILultt.
Thi, nection of the w raty includes the thermal-hydraulic epects of -
the accident Ocen:irius .herein the fisalen products may be releud and trans-ported. Rc* ulat titet tua of enti ter d@asited fissiun prcducts is an r eor t .ot phenomenon. The t'wmi! hydemite p:.rs eters of intere3L, ces a funct ten of space and time, durfry the o'urse of the accident are :
- 1) te+perature.
- 2) pressure,
- 3) gas compositions.
- 4) gas ficw rates,
- 5) temperature gradients,
- 6) the rates of degradation of reactor insulation, and
- 7) the presence of appreciable quantities of liquid water in, for example, loop seals or hung-up in the pressurizer and not predicted by current codes.
Certain phases of degr3ded core accidents will involve the interaction of the heating of the RCS structures with the thermal-hydraulics of the RCS due to the deposited fission products. Ccupled codes are now under development.
A number .of source term estimates have now been made on the same or similar plants using dif ferent codes. In terms. of the prediction of thermal-hydraulic conditions (system temperatures, flows, hydrogen production) appre-ciable dif ferences have been observed. These dif ferences may imply different rates of fission product deposition and volattlization. One might be tempted to urge resolution of the differences observed. However, this may not at this time be a profitable exercise since none of the codes used takes complete account of the coupled phenomena and their interactions during the in-vessel progression of the accident. It might be well to wait until these codes incor-potate the newer concepts before undertaking comparisons. Information closely related to the topic under discussion is also presented in the neit section.
In susunary :
. Progress Since WASE-1400 There is progress in the sense that the necessity for close coupling (probably at each time step) of the thermal-hydraulic codes with fission product deposition, self-heating and revolatilization phe-nomene has been recognized. Most phenomena are thought to have been recognized.
O
- 9
. General Applicability to All Plants the information base developed for in-vessel accident progression for PWR and DWR reference plants has. general applicability.
It is useful to focus the predictive methods on PWRs and nWRs separately so that it is possible to represent the urpierate char-acter of their core and system designs in great . detail.
1.8 RCS FISSIUN PRODUCT TRANSPORI AND CHEMISTRY The analysis of fissivo product transport through 'and cha*1 cal toler-action in the RCS l' law circutis during 3escre accidents has belatedly recei.ef attention during the past decade. The fccus has been on thelaore solat tle fis-sico products (icdine, caestum and t ellurium) and core structural mat.arti41s (silver, indium, cad.nium) aad some of. the volat'le elements in etrealoy and stainleuc steel.
1.8.1 Chemical Species and Reactions It is now thought that the major species released from the degraded fuel will be Cst, Cs0H and Ie and of course the noble gases. Various potential chem-ical reactions within the RCS may change these forms, but cost of the modelling codes matntain these forms during fission product transport through the RCS.
Analyses using a version of TRAP-f1ELT tend to follow the separate species but the IDCOR analysis, which uses an empirical aerosol correlation does not main.
tain the chemical ccepound identity during the' initial material transport. The .
only study in which chemical reactions between the vapour soecies and stainless steel surfaces in the RCS are specifically calculated is the Bill-2104 work.
Simple physical condensation is used for Cs!, weak chemisorption for Cs0H and strong chemiabsorption for Te. The other studies simply model vapour deposition of Cs!, Cs0H and Te as physical condensation and evaporation processes.
Chemical reaction pcasibilities reeding further attention are, for example :
- 1) reactions between Sn and in with Te,
- 2) the-importance of boric acid-Cs! reactions,
- 3) the possibility of caesium molybdate volatility,
- 4) the significance of the possible formation of alkaline earth hydro-xides which would influence the transport and deposition behaviour,
- 5) the potential for species changes during revolatilization,
- 6) the lack of consideration of fission product trapping by liquid water in the RCS flow paths, and
- 7) the lack of modelling of the effects of air entry into the RCS late in scenarios when regolatilization rates and species may be affected.
1.8.2 Aerosol Formation and Transport Turning now to aero3ol transport and behaviour there appears to be two areas of agreement areng the various source term studies with respnet to the RCS 3erosols. These are, first, that acrosols of apprecioble cancentration will.
form near the interface b?twe?n the core and the RCS. Secondly, that aerosol agglomeratien processes will be q;ite active during transport such that signifi-
- cant parttele growth will take place.
ffcst of the analyses use a version of TRAP-MELT for calculat tng aer ,291 transport. Variations in results arise fecm representing the particle size range with :
f) a discrete binning'model or
- 2) a continuous (log-normal) function.
The IOCCR method is the exception since it uses an empirical correlation, in the area of aerosol formation and transport, some cases of insuf fi-cient information exist. These are :
i 1) aerosol formation, transport and growth, 4-
- 2) generality of the applicability of the IDCOR aerosol correlation, l and
- 3) influence of the hydroscopic nature (or lack of it) on aerosol behaviour.
1.8.3 Fission Product Deposition Processes With regard to fission product deposition processes their importance lies in the fact that they are instrumental in defining the degree of retention that would occur in the RCS during severe accidents. It is now believed that this deposition in the RCS can be substantial. For example, BMI-2104 calculates 83 per cent caesium retention for caestum in Sequoyah's THLB' and 98 per cent for Zion's THLB'. There are other results in a somewhat lower but still very significant range.
There is substantial agreement regarding the mechanisms and in general the mathematical models by which fission product vapours and aerosol would i
deposit on RCS surfaces. These are desertbed in the text of thi3 report. This is true for the earlier phases of the accident sequences. At later times in accident sequences, however, the results of 10COR and other analyses which model revolatilization by decay self-heating can be quite different from the BMI-2104 TRAP-MELT results which contain no decay heating effects. Lack of agreement in deposition (RCS retention) results between studies can especi be caused by differences in the thermal-hydraulic conditions that are supplied to the fission product transport codes by the respective mass transfer-fluid flow analyses.
. - . _ . y
h~ .
L
- . In ~ summary
- -
! . Progics: Since WASH-1400 t
it is thought that the major more volatile species entering the RCS are Cst, Cs0H and Ic. Progress has been made in problem identifica-
. tion of the importance of 1) fission product deposition and revola-tilization in the RCS and 2) chemical reactions between vapour i
- species such as Sn and In with Te, and 3) potential chemical reac-t tions of Col and Cs0H to produce mnre volatile iodine compounds.
Significant areas of agreement have been achieved concerning
- 1) appreciable formation of acrosol at core /RCS interface,
- 2) acrosol agglomeration process during transport in RCS that lead to significant particle growth, and 3) general agreement as to the modelling of the deposition of aerosols on RCS surfaces.
Many aspects of aerosol behaviour and modelling can be assumed to be relatively well established.
. General Applicability to All Plants The RCS vapour and acrosol deposition models or correlations used in current source terms analyses and those to be developed in the future are generally applicable to various LWR designs.
1.9 CORE / CONCRETE INTERACTIONS AND OTHER EX-VESSEL SOURCES The core / concrete interaction occurs when an accident scenario has progressed to the ~ point that the molten core has penetrated the bottom of the reactor vessel and fallen onto the concrete below.'The material is presumed to react with the concrete. Carbon dioxide and steam are released along with copious amounts of aerosols (at least in the U.S. experiments but not so much in the German Beta experiments). The' attack on the concrete occurs, it is thought, primarily by melting. The molten components of the concrete become incorporated into the corium melt.
There are two modelling approaches at the present time. The USNRC approach involves the codes CORCON-M002 and VANESA while.the IOCOR approach uses MAAP (CNAERO). The USNRC approach is a kinetic approac'h while the IDCOR approach uses an equilibrium model. The modelling of this aspect of the source term calculation is not at all well supported by data.
In addition to the core / concrete reaction there are other ex-vessel sources. These are the following :
- 1. Pressurized ejection of core melt from reactor vessel.
- 2. Melt / water interactions.
- 3. Aqueous chemistry effects (revolatilization from water pools).
- 4. Boiling /flashiqq of water pools.
- 5. Nucleation of druplets from cupersaturated uteam.
- 6. Direct relcase from RCS (re ;idual fuel or from RC5 t ransport i.
- 7. Revaporization from RCS surfaces.
- 8. Pesaporization of fission pecducts from aerosols as a re ult of H 2 botns.
In summcry :
. Progress Since WASH-1100 There has been significant progress on the core / concrete reaction since WASH-1400. There are now experimental dato using thermite and metallic melts. These have helped identify the features of thee reactions and the relevant phenomena have been incorporated into computer codes. for example, progress has been made in the following areas :
. Identification of the general phenomena of importance, The relationships among melt temperatures and gas flows with the generation nf aerosols, The rates of release of H O2 and CO2 as concreto melts,
- The nature and chemical reactions between the bubbles and the melt,
. The importance of chemical vaporization of specific fission products.
. General Appilcability to All Plants The general methodology if properly developed may be applicable to all plants. However, the application will be plant-specific de-pending on containment configuration, type of concrete, etc.
1.10 EX-VESSEL THERMAL-HYDRAULICS This section addresses the thermal-hydraulics and associated phenomena which are important in the containment.
Containments are designed for loads resulting from Loss-of-Coolant Accidents. The thermal-hydraulic calculations are performed with conservative assumptions. Therefore, containments can cope with nuch higher loads than they are designed for when realistic assumptions are used.
The thermal-hydraulic behaviour within containments (pressure and tem-perature history, internal flows) can be divided into short-term and long-tern containment behaviour. Short-term usually characterizes the behaviour within the first hours af ter an accident or release of melt from the RPV. Thermal-hydraulics within the containment is important for fission product retention, i
y
dith regards to larg tctm contatnment thermal-hydraultes, the glnbal values fot-containment pecasure and te perature can be calculRed alth routine-ly used codes if the time of melt / water contact and split of 9'o decay halt between cater and concrete can be predicted wtth sufftetent occuracy. Ibe trans-port of energy through the pressure boundary ts plant-spectric an.t can also be calculated with these codes. Due to condensation ef fects tbc t y p _r atures .,ith-in the contaircent stay usually at the saturatten level. [L ;3 only above the melt that superhtated steam may ettst. ihtG is dif ficult to clicul3te. One to dif ferent concentrat tens of ncncondenc tbles tn different volwres strat tf tc.d tan
- may occur and is also dif ficult to calculate. For pressure calculattens unty condensation on structures is L.Len totu account. ffatertel of -otal atll folles containment torperature .ery c:asely and can be calculated e tth root tnely an-d ccdec. Condensation on ccrcrete structures c:n al:0 be calcul..ted atth cuut ac-ly used codes. Only a few carrelattens ex tst for the prediction or tulk cm 4 n-
\ Gdlion. BeC3Use theGe phunoren3 are LTpqttant for 9curCe term psaluation 'ure research chould be performed. Cnly a few cudes are capable of predicting con-Section flows within a multiccapartment contatnrent. Because those flGad (n-fluence fission product retent ion, code irprovementa and salidation are recom-mended. For accident management actions the effectiseness of sprays and beat eschangers (e.g., of ventilation systems) can be catculated with rcutirely used codes.
With regard to the short term behaviour, the containcent behav tour is mainly influenced by melt,' water interaction in the cavity below the RPV, a possible melt / atmosphere interaction and by hydrogen burns. After RPV bottom forces. The failure, melt is drisen into the cavity by pressure and/or gravity behaviour of the jet, e.g., droplet formation, is uncertain and should be studied further, if this effect is of importance.
The contact of melt with water in the cas tty might result in substantial steam production (steam spike). Due to large participating conteinment volumes and large melt drop sizes the threat to containment integrity is assessed to be very low although structures within the containment miq*.i ce damaged.
The possible rapid ejection of the melt into the reactor cavit t may heat the cavity. Depending on the design of the cavity, the connecting channels to the containment and the sequence, a large part of melt debris may be swept out. If large amounts of melt participate in this process the integrity of the containment may be threatened due to fast temperature and pressure increase, further analytical and experimental studies are in progress.
In summary t
. Progress Since WASE-1400 There has been much progress in identification and understanding of-l f
major physical phenomena and system behaviour. Existing codes for DBAs have also been improved with respect to capability of studying accident management measures.
n There is sufficient information with regard to global long term containment behaviour.
.y
. ' General Appilcability to All Pl.nnts The methodology, when developed, should be generally applicable but the results will be plant-specific.
1,11 FISSION PRODUCT / AEROSOL IHANSPORT IN CON ! A i W1.N T This section is concerned WLth the ?(thodolOgles for predicting ridLO-activity transport by natural preceu;es in the containment and other buildings cf an lwr. The period of cancern for this part of the accident scenar:0 u;;al-ly begins with a discharge of cTJ19Nt into the CO3Clor Containment frON either a break in the RCS or the presaurizer quenth tank. The discharge will include both stea9 and parlLculate aM or armal. Discharges of noble gas fission prod-ucts and core dert ed s materials nido t akes place. Later aerosols may be r;ener-ated by the discharge of the *olten corium material and by the subsequent tr teracticn of the hot core materials with the corerete of the contain ent basemat. In the latter c,se the aercsola generated will include constituents of concrete. At some L tme the contatnment building may suf fer overpressure failure.
The methodology is thoroughly discussed in the annes of this report (to be published). It can be divided into three broad groups. BMI and Stone and Webster used NAUA-4, IDCOR used MAAP and an empirical correlatier,, while EPRI in its analysis of the Surry reactor used the MATADOR code. Within these broad groups of results dif ferences are due to differences in applied boundary con-ditions, such as layout and dimensions of the plant and the thermal-hydraulic conditions.
A number of general conclusions can be drawn regarding fission product and aerosol transport in the containment and other buildings from the available results. When containment breach is delayed, the containment provides a large additional retention of the fission products. However, the current methods do not include resuspension by rapid depressurization of the containment, by hydrogen combustion or steam spikes close in time to containment breach. Fission products may also be released later in the accident sequence due to revaporiza-tion or core-concrete interaction.
In summary :
. Progress Since WASB-1400 The modelling of fir.sion product / aerosol transport in containment has been much improved since the first reactor safety study (RSS),
and placed on a rechanistic basis rather than the empirical basis of the CORRAL code used at that time.
In general terms, the calculation of fission product / aerosol be-haviour in these circumstances is one of the best developed dis-ciplines in the field of source term assessment, and is well supported by experiments. Comparison between the studies under review indicates that there is essentially a consensus as to which are the important prccesses governing the aerosol behaviour, and, 7
with certain exceptions, the appropriate mathematical models and input data representing these processes. The principal need for the future is to improve the relevant thermal-hydraulic talcula-tions. Whereas only fission product retention in the containment proper was allowed for in the RSS, retention in various auxiliary buildings is also allowed for in current evaluations of the cnurce term.
. Canctal Applicability to All Plants A properly designed methodology developed for fission product /
aerosol transport in the containment and other buildim;s will be generally applicable to all plants. This does not imply, however, that the results will not be plant-specific because this does seem to be the case.
1.12 FISSION PHODUCT BEHAVIOUR IN WATER AND t.ONG-TERM EFFECTS The issue of Icng-term release of airborne fission products during an accident is generic to all water-cooled reactors and all accident sequences.
With the exception of aerosols, which can be regenerated by physical phenomena late in an accident, delayed generation and release of airborne radionuclide-t bearing. species is associated with chemical reactions within contain,ent. Since chemical reactions are involved, the chemical conditions within the containment may have a profound effect on the quantity and rate of radionuclide release.
The radionuclide which is expected to dominate the long-term contribu-tion to the source term is iodine. ladine is known to exist in seseral sotatile chemical forms including 12 and organic lodides (e.g., CH3 1), and iodine release from reactors under both operating and accident conditions has been. documented in the past.
The treatment of iodine chemistry is cosered in the annex to the report.
The effects of containment chemistry-and potential long-term release of radio-nuclides are generally not included in current severe accident analyses. The treatment of chemistry in containment in accident analyses has to date been based on three assumptions. These are :
- 2) all Csl is removed from the atmosphere in. containment as an aerosol, and
- 3) all Cs! in contact with water dissociates to form I which remains in solution indefinitely.
These assumptions are not valid in the sense of universal applicability to all fission product indine. The airborne radiolodine measured at TMI-2
' demonstrates this. Hence, there is a need to demonstrate, in a complete acurce term assessment, the degree to which the above assumptions are valid.
It-should be emphasized that source terms calculated using very slow releases may not be usable in codes such as CRAC-2.
- r
In summary :
. Progress Since WASil-1400 The WASH-1400 safety- study predicted, in the absence of firm evi-dence to the contrary, that all of the iodine released from t he core during a reactnr accident wnuld be initially airburne as Iz and CH 3 I (0.7 percent). The long-term release of iodine was uuuumed to be controlled by a competition between gas leakage rates from containment and platcout or spray remnval of indine from the containment atmosphere.
Since 1975, and particularly since IMI-2, the foundations for these assumptions have been removed. There is now experimental as well as theoret.ical evidence that iodine is predominantly relea.ied froa fuel during an accident as Csl and that iodine will be intenduced into containment as part of an aerosol or as I- in solution. The aerosol material which settles or is washed out of the atmosphere will also release I . Thermodynamic calculations and experiments have shown that dilute solutions of I under basic conditions present during a reactor accident w'ill remain as I or form 103-and that the total iodine partition coefficient into solution is very high.
Experiments by several groups have eliminated the uncertainty that HOI may be highly volatile and lead to low iodine partitioning.
Studies of iodine kinetics have elucidated rate data on the con-version of I2 to 10 3 and on iodine reactions with organics and other species. Work is in progress to determine in detail the influence of radiolysis on iodine chemistry and partitioning. The experience at IMI-2 plus a re-evaluation of previous work has led to a lower best estimate for the fraction of iodine converted to organic iodine.
The results of the work of the last ten years.plus a critical examination of previous data on iodine chemistry have also been used to identify the conditions which can minimize long-term iodine release.
. General Applicability to All Plants The information will be generally applicable to all plants.
1.13 SCRUBBING Several pathways to the environment for fission products in risk signif-icant sequences involve passage through water pools such as the BWR suppression pool and the PWR quench tank. It is necessary to describe the removal efficiency obtained by the scrubbing of the aerosol beering fission products. WASH-1400 assumed a decontamination factor of 100 for subcooled pools and only 1 for- steam saturated (boiling) pools. The conditions under which the steam-noncondensible gas mixtures bearing aerosols enter the pools change significantly during the course of a severe accident. Thus, estimates of the decontamination achieved by the pool scrubbing require close coupling to the thermal-hydraulics and the aerosol fission product transport.
euer. e-use + + ee e = = e. - p
Recognition of this fact has led to the deselopment of the \RC code SPARC and the EPRI Code SUFRA, CE has also developed models for the scrubbing process assbclated with the quencher, downcomer and hnrtzontal sent injector configurations for the GWRs. These models are proprietary to CE.
The two par 3 eters ;.hiCh dCFin'f c 'bc overall behas ivor or the 2,i a cleo pool are the aerosol p1rticle si/e and the ,leam mass fraction in the carrice gas. The most significant ob;ers.4 tion that can be made concerntrg the tmt peal data is that the experimental DFs are higher than expected based 09 fat ly Tcdct calculations.
- In summary :
. Progtens Since WAS!!-1400 A much more mechanistic understanding of aerosol removal by scrubbing is now available. Ihc appropriate endeu are in the process of being developed it is also now appreciated that for subcnoled pools the greater the particle niic and the larger the steam mass fraction the larger the DF. Furthermore, it has now been shuwn that the DF in saturated pools is greater than the unity as assumed in WASH-1400.
. General Applicability to All Plants Information generated can be employed in all BWR plants, PWR quench tanks and other scrubbing situations when the injection geometry and modelling work underway is completed.
1.14 ENGINEERED SAFETY FEATURES Light water reactors are equipped with engineered safety features (ESFs)-
to prevent or mitigate accidents. Those ESFs include emergency core cooling systems (ECCS), containment and pressure suppression systems such as contain-ment sprays, ice condensers, suppression pools, and emergency air cleaning systems.
The emergency air cleaning system consists of a HEPA filter train and an activated charcoal filter to reduce fission product release outside the containment. In addition to the emargency air cleaning system within the con-tainment, a standby gas treatment system (SGTS) and an annulus air reci :ulation system are provided in BWRs and PWRs, respectively.
- In addition to the conventional ESFs, the concept of a filtered vented
! containment has been developed and adopted for the Barseb5ck BWR in Sweden and
! to the French PWRs.-These filtered vented systems are considered as ESFs.
i l In this section the containment sprays, ice condensers, emergency air cleaning system and filtered vented containments are the subjects of discussion.
Containment Sprays f ' A fairly large amount of experimental data are available for molecular iodine removal by containment spray. These data are summarized in the text of l
i I
(
i
the report. The removal of noi ular todine has been found to be very effectise as has been the reTosal of acrosol particles. Howeser, the removal of crganic iodides is rather poor as might be expected.
Ice Ccndenset rWRs Little experimental data are asallable to evaluate the efficiency of ice condensors in re~oving aerosols iuring sesere accident conditions. A theo-retical treatment of fission product removal in icebeds has been deselcpad in the ICEDF code. The major physical processes are incorporated in the code but-
- the code sertftcation is not posutble due to lack of experimental data.
Filter 3j3:en Approctable infor ation is asatlable on the operat ton of filters. The possibtlity of the less of filter intege tty due to a prolorged sesere accident needs to be evaluated.
Filtered-Venred Contain.-enc Swedish filtered-vented containment utilizes a gravel bed filled with 1 inch. size quartzite rock grasel while the sand bed is used in the French PWRs.
Esperiments were conducted for both cases to obtain the design basts data.
In summary :
. Progress Since WASB-1400 Models of fission product aerosol removal in ice condenser are being developed and incorporated in the ICEDF code. The mechanistic models were used in BMI-2104 calculations.
Filtered vented containments, have been developed both in Sweden and in France. Supporting experiments were conducted to determine filter efficiency in the filter systems.
Because of more detailed evaluation of ESFs capability of fission product removal under severe accident conditions, more credit is now given to ESFs to remove fission products. In addition, the spray effectiveness of fire sprays is evaluated.
. General Applicability to All Plancs The ,information available and generated in the area of ESF is plant-specific to a particular class of plant.
1.15 HYDROGEN COMBUSTION, INCLUDING EQUIPMENT SURVIVABILITY AND CONTAINHENT INTEGRITY It is believed that early containment failure as a result of hydrogen combustion is unlikely for most containment systems and the effect of combus-tion on the chemical processes may be small since the deflagration burns last only tens of seconds. Thus, re-suspension, revolatilization and chemical change processes in the aerosols, due to the temperatures and pressures generated,
~
should have-only negligible effects on the source tern assuming that the can'-
tainment integrity -is maintained for come time (order of t ens of ninutes) after the hydrogen burn.
However, in cases other than early containment fatlure, forthat ey ,tder-
~
ation is sometimes necessary to determine v.hether or not 'the pressuros efwc ated in deflagrations may threaten the integrity of the various ccntainrents current-ly used for LWRs. The lack of a problem is pretty clear for the ! rge-dry 3 3.;R
. containments. The BWR Mark I and Mack II containments have inett at'~,pheres in the United States. The PWR ice condenser and the BWR Mark III plants in the United States have installed igniters in their containments for deliberate ccm-bustion centrol of hydrogen. For thcce ccntatnments, which'have low design prec-sures, the h; . f... -
t +ad may be burned ' inuously (inctead of discretely 1, as has been shown esperteentally in the latge-ecale hydrogen combustion tests at Nevada. In the continuous-burn mode of hydrogen combustion, the pressures generated are very low and containment totegrity is not chall'nged. I!.e d tffu-sion flame type burns may, however, produce locally high te*peratures. Which may be of concern for survivability of scme aafety-related equipment whose proper operation may be mandatory for arresting the progress of postulated accidents. These concerns have resulted in the construction in the United States of the 1/4 scale Mark III containnent facility in which temperatures and heat fluves during prototypical igniter-induced hydrcgen burns are being mea-sured.
Containment challenge from hydrogen combustion in the TMLB' scenario for the ice condenser plants (when station blackout is effective for a certain period and then is restored and igniters become operational) has been a matter of discussion between the NRC and IDCOR. The ICCOR MAAP code predicts continu-ous hydrogen burns in the containment cavity after the corium melt is ejected from the vessel. It also predicts that the buoyancy forces transport a continu-aus stream of hydrogen to the cavity from other parts of the containment. The NRC staff has not agreed with this assessment ; however, proper resolution of the combustion history of the hydrogen generation in 'this scenario can best be achieved after the NRC codes have modelled natural convection flow fields in the containment.
~
The question of the importance of hydrogen combustion phenomena is a complicated one. There are many associated phenomena whose importance is diffi-cult to evaluate (i.e., influence of obstructions on the combustion process and chemical reactions of fission products influenced by combustion). The author of the technical annex on hydrogen combustion (to be published later) has offered the following thoughts :
- 1. Source term estimates for pre-existing openings should include the effect of combustion on the release. In this regard, the ignition time should be considered as a parameter since it will influence the amount of hydrogen burned. Also, the sensitivity of releases to node-size and number of nodes should be determined under combustion condi-tions. The combustion pressure transients for calculating relesses should be conservatively estimated taking into account the influence of flame acceleration mechanisms such as turbulence, venting, con-centration gradients and confinement.
- 2. Source term estimates for pre-existing openings should be updated by ircluding the effect of hydrogen combustion on airborne radioactivity.
The extent of .aporization and osidation of aerosnis, the osidation
~
. of iodides and resuspension of aerosols from surface depostta *.Pould be detetmined from modnis bawd on espectrcnlal data.
- 3. The moWis for hydecgen mixing should be aws ied wre ext. t. ty with relevant esperirental data especially for larr;e t<: p.'rature gradtents and terperature-irpersten condittens.
'4 The diffusion -flame test a results in the quarter-scale LR " ark ((
containment should be used to oscertain the survte sb t i tty of , ifet, -
related equip ent.
S. If cont ino us burning as a diffusion flame is out likel, .n ' he
" irk III and Ice Co.
r denser contatnments in certain accidents ,'e.g.,
station bins outi, it may be ni'cessary to perform ccmbustico tests using travant 'txtures to ascertain whether under the inflo m ' of n ult tple fico.e accelerat ten mechants.es (e.g. , unt ir g and i o o'"'
the peak pressures oblatned can be tolerated.
In summary :
. Progress Since WASif-1400 Experimental programmes have provided burning rate data for laminar conditions, as well as for specific geometries with obstacles. These are useful in ascertaining containment integrity and in determining releases when containments have pre-existing openings. Extensive flammability limtt data have been obtained for hydrogen combustion containment breach.
. General Applicabilitt] to All Plants The basic information obtained from the experimental programmes is generally applicable to all plants. However, the application is very plant-specific.
- 2. AREAS OF DISAGREEMENT BETWEEN THE STUDIES 2.1 PRINCIPLE CONCLUSIONS OF THE STUDIES All the studies which have been reviewed are in agreement that consider-able progress has been made since publication of the Reactor Safety Study in the understanding of severe LWR accidents and the behaviour and release of radionuclides from plants during these accidents. However, there are divergent views as to the implications of these improvements.
Thus, the major finding of the ANS report includes the statement that, "With a small number of exceptions, estimates of source terms associated with severe core damage accidents can be reduced from the estimates in WASH-1400 by more tnan an order of magnitude to several orcers of magni-tude. The noble gas fission products are exceptions."
and more specifically, "For large dry containments, suf ficient information exists to support the calculation of source terms ranging from a small fraction of a per-cent to no more than a few percent of tbc core inventory of important fission product species.
A comparable conclusion is reached for the BWR Mark I!! containment -
bec1use of the suppression pool's htgh availability and effectiveness in nerobbing fission products, and the large size of the containment.
A similar conclusion is reached for PWR tce condenser contair.ments, based on the limited avail 3ble analyses."
and the IDCOR study concludas, "The fission product' source terms - quantities and types of radioactise materials released in the event of severe accidents -' are. likely to be much l'sse than had been calculated in previous studies."
Whereas the APS study, while acknowledging that, "In a number of cases, new calculations indicate that the quantity of radionuclides that could reach the environment is significantly lower
~
than that calculated in the Reactor Safety Study."
nesertheless qualifies this by noting that, "It is impossible to make the sweeping generalization that the calculated source term for any reactor sequence involving any reactor plant would always be a small fraction of the fission product inventory at reactor shutdown."
Particular emphasis is placed on containment isolation failure and con-tainment by-pass sequences by the APS study. It also recommends another itera-tion on the selection of accident sequences in order to provide greater assurance that those sequences important for risk comparison have been identified, whereas the ANS study concludes that the RSS and several PRA's have included secuences which have identified all the release paths of major importance to the evalua-tion of risk.
Recent work undertaken by ENEA/ DISP in Italy [1,2] can be seen as an attempt to reconcile these divergent views, in this case in order to define reference accident sequences and source terms which might be of value in the regulatory context. As can be seen from the following summary the approach taken is illustrative of the relatively severe constraints which have to be imposed on available source term information for the purpose. In this approach two types of accident sequences are excluded from consideration :
Accidents with a frequency of occurrence below a certain de m2nimis
~
threshold (provisionally set at 0.05 x the core melt frequency, but an absolute value may be possible later), on the grounds that the basis of the reference source terms will then be restricted to phe-nomena where knowledge is nnw sufficiently advanced to offer firm
ground for decision making. Acei'donto; involving contain ent failure caused by steam explosions nr hydtngen burning 3re escluded as a
~' result.
~
Accidents which are highly plant specific, such as the F;;R V req,.ence accident.
Calculations are then perforded to codify existing scurce tet n i tofarm3-tien for'some rep'resentatise and/or risk dominant accident cequerces to altcw for two factors .chich are not taken systematically into accuent at the present time 1 The possibility nf pre-ex tsting npenings in the containment Ipfc'.i-sionally up to ~30 cm2 equiv.ilant area).
The.possiblu intervention of the reactor operating crew to mittgate the effects of the accident, where this is warranted.
It then emerges that the source term for the volatile fission piaducts (I, Cs, Te), obtaired by state-of-the-art methods, is never more than a small fraction of the core inventory at shut down. (It is estimated that most of the inventory of the noble gas fission products (Kr, Xe) can sometimes be released, as is concluded in both the ANS and APS studies.]
The E\EA/ DISP report describes original work which incorporates 3 number of new concepts in some senses contrary to the spirit of the mechanistic methods which have lately been developed for source term evaluation (although the authors note that further confirmatory calculations with mechanistic codes are needed). For example, one of the general conclusions of the Specialist Meeting
. on Nuclear Aerosols in Reactor Safety held in Karlsruhe, 4-6th September,1984,
[3] was that,
"... Current LWR source. term methodology is much more complex than in WASH-1400. This militates against the generalization of source terms to more than one plant. The concept of across the board source terms con-tradicts the most recent research findings ..."
The authore of the ENEA/ DISP report recognize that whether and how conclusions of'the kind it proposes, when subject to further development and confirmation, might be reflected in the regulatory process must be left to individual countries. However, some necessary criteria of a general nature are indicated, Emergency procedures for severe accidents, personnel training, and necessary equipment must be available at the plant ~-.
. Measures must be taken to reduce the frequency of occurrence and/or source terms for accidents involving plant specific features to acceptable levels by operational or design changes.
The real possibility for containment failure by direct heating of the containment atmosphere following pressurized melt ejection from the bottom head of the RPV needs to be assessed by further research.
The outcome may well be plant specific.
e r - ~ - - - - ,
The E\FArDISP tre3t ent is essentially restricted to the noble 413 and solatile fission products and th" need to insast igate furthat the pos-sibility of htgher than pres iousi anticipated f releases of. the non-volatiles ducirg corc/ concrete interacttan is acknowicd.jed. ln this the authors reflect the APS utudy, which ident tfies this as one of it s ajor concerns.
It ts niso apparent that the ENEA. DISP treatment is cencerned only with
- setting upper limit source terms (neglecting uncertaintles). '.iheth er
' such an approach is adoquate an its own aill. depend on the bafety ;oa ls chosen by regulatory auo tties and the criterta adopted for cluvre of the source term issue. If further generalitations of source term data are to be made there is a need to deselcp a conststent argunent by .
..hich the source terms for a relat e.cly Small number of accident se-quences below the upper ltatt can be used to repre:ent the hazard fr09 a part icular plant on a particular site with adequate prec!sinn, i.e.,
to re-examine the gatejari. 20::n pr7 cess adcpted in the Reactor Safety Study. Such arguments would need to take account inter alta of the sen-tivity of eff-site concaquences to changes in the cource term. Only -
when these arguments hose been developed would it be possible to judge, for example, whether cite 7ary soutce terms can be defined which are independent of the site or the type and design of the plant, or both.
2.2 SUPPORTINC/ QUALIFYING CONCLUSIONS OF THE STUDIES The following comments relate to the more detailed supporting and quali.
fying findings in the studies under revtew. They include comment on the studies' assessment of the state of the art, the calculational methodologies and data employed, and the results obtained.
2.2.1 Containment Integrity Although quantitative source terms for accident sequences involving early containment. failure are evaluated in some of the studies (notably BMI-2104), there appears in fact to be a general consensus in the studies that Laactor containments are stronger than assumed in the Reactor Safety . Study, and therefore fail, if at all, at later times. This provides the opportunity for almost all of the fission products to be retained in the containment, and is recognized as a principal factor in reducing source terms. Continuation of the research and analysis of phenomena that might present a threat of early large-scale containment failure is recommended.
Several phenomena have been considered as presenting possible threats of this kind, including steam explosions, steam spikes, hydrogen combustion t
and direct heating.
Further experiments and analysis of the phenomenon of pressurized melt l
ejection, which may lead to direct heating, are recommended in the studies.
The part of the technical' annex (to be published later) which is con-cerned with hydrogen combustion, concludes that the effect of this phenomenon has been considered in only; a few studies (IOCOR, BMI-2104, EPRI). It recommends further assessment of the models for mixing and turbulent combustion, and the g *
'66
centinuat tnn of dtffusion-flame testa to asccctain the survivability at Mafety-related-equipment.- Currently the effects of turbulence .cn the burn t ime are bring insestigabed using the .NTS data. The USNRC code HECTR now has a preliminary diffusion-flame model and there is a three dimensional nadel in hMS.
2.2.2 Additional ~ Sites for Retention There also appears to be a general consensus in the studies under re.iew
- that a second principal factor in reducing source terms ta the inclusten of additional sites (suppression pools, ice beds, aux tliary buildings, specifically dedteated filter systems) that trap radionuclides. In the technical annex, the striking difference will be noted between the ef fective decontamination facter in the aus111ary buildings, into which the melt-phase rele'ase occurs in the V ~
sequence acetdent, in CMI-2104 and IDCOR ; but it alco recognizes that dtfrer-ences in design of the reactnes studied (Curry 7nd Zion respectise'y) are prob.1-bly the major enntribut ing factor.
2.2.3 Physical and Chemical Phenomena There are differences in the calculational methods and data used in the various studies to represent the physical and chemical phenomena that af fect source terms, and which impact significantly on the results obtained.
. In the tech 71 cal annex it will be pointed out that BMI-2104 uses MARCH /MERCL and IDCOR uses MAAP. These codes have been compared and large differecces in their predictions observed. Resolution of these differences will have to await a detailed examination of the methods employed.
It should also be noted that these codes are limited in their simplified treatment to relesant in-vessel thermal-hydraulic phenomena during core-melt progression, and their interaction. Effort is under way in the US (NRC, EPRI) ,
to develop detailed mechanistic codes (SCDAP, MELPROC, CORMLT) aimed at improved methods in this area.
. In the technical annex on RCS Fission Product Chemistry and Transport it will be reported that all the studies use a version of TRAP-MELT apart from IDCOR which uses MAAP : the estimates of the effectiveness of deposition during the initial passage of source material through the RCS are general'y consistent. Although there are substantial dif-ferences between TRA.P-MELT and MAAF, the IDCOR results for the early period of the accident also tend to agree because the dominant removal process is gravitational sedimentation, which largely determines the empirical removal rate coefficient used in MAAP.
l At later times in the accident sequencei however, there are differences in the results. The'IDCOR and other analyses (SWEC and NYPA) which include revolatilization by decay self-heating indicate a relatively early release of fission products to the containment. Whereas in BMI-2104, which neglecto decay self-heating, revaporization occurs much later in time. It would appear essen-tial to couple thermal-hydraulics and fission product transport in the RCS in both PWRs and BWRs.
i
, f I
, , With regard to ex-vessel sources there are again Subutant tal dif fer-ences between the UMI-2104 and IDCCR models and their pred?.ct ions.
. With reg'3rd to Fission Product / Aerosol Transport in the Contato cot, apart from the models for diffustephorests and stea,.endeosittop en the particles, there is essentially a consensus bet'.een all the studtes .on the modelling of aerocol growth and remm11 prm.c t wo :n contatnnent. The principal differences are in the numcrtcal o hr es adopted for soluticn uf Lbc cdels.
. With regard to Supprassion Pools BMI-2104 uses the SPARC rc M to calculate pool decentamination factors (DFs) based on IRAP' "Elr calculated 'aercsoi Inadir.gs and particle size dist erbutions. : he,eas IDCCR employs fised DFs.
The.APS report suggesto that the SPARC code and the EPRI code SUPdA predict different DFs in the aerosol scrubbing step. This is of parttcu!Jr importance for predicting the decontantnation in 8WR suppression pools. On further examination it is c19ar that the compartson used different input d.ita (and there was a confusion of radius and dtameter of aerosol particles} for the code calculations. When stmtlar input data are used the disagrec.ent is not particularly signiftcant.
- 3. CURRENT UNCERTAINTIES IN SOURCE TERM INFORMATION J.1 INTRODUCTION The use of complex, detatled computertzed models of the physical prccesses is a rational, defensible means of determining severe accident source terms.
However, there will always be a certain level of imperfection in this procedure.
Models describing the complex processes rely on imprecise knowledge and, some.
times, parametric representations are used for phenomena not mechanistically modelled because of extreme complexity or insufficient knowledge. Uncertainties exist in virtually every phase of the deterministic application.' Some examples of sources of uncertainty are :
The mathematical models representing the physical phenomena governing the processes.
t The correlations for the physical properties of the various materials of interest.
The inadvertent omission of possibly important phenomena.
t
. The boundary conditions related to the specification of the accident sequences and the plant geometry.
The numerical approximations involved in the solution of the various mathematical representations-of the phenomena.
Single-valued source terms, as developed by the various applications of different code systems with specified models and inputs (characteristic of most
-. y
( .
l i
I of the studice under review), are useful. Howeser, such s trgic-valued est imates
(
' should be used with a proper appreciation of the associated orcert lita tes.
Uncertainty / sensitivity analysis methodologies exist erpinying stat tat i_
cal sampling and evaluation techniques that can be used to es1!uate the uncer-tainties in source terms as developed by a particular sot of codes. Oncertatn-ties developed via these techniques can be presented as plabdLtitty distributtons of the key output parameters.
The application of these techniques requires quantification of the ranges over ahich input and code-internal para eters may vary and quantificat ton of the probability distributions of the variat tons within those ranges. Values can _lben be statistically selected from the ranges, <.etghted by the probability distritu-tions. and used with the particular set of codes to determine statistic 31 u'l-certainties in the cutput of each code that are combined and propagated through the entire set to deternine the probability'distribulton of uncertatnty in the source terms as the end proccet of the calculational procedure. Evaluating the uncertainty due to incomplet' . omttled phenomena requires either para *cterizing the models as. they exist so as to encompass some estimated range of uncertainty or directly modifying t le i 3 des *.a include better representations of the incom-plete phenomena. A-danger a t'.at important phencmena may not be tocluded simply because they have been overlooked. One tmportant role envisioned for integral experiments is to help ast.*e that phencmena have not been overlocked and to i
validate that tha predicted uncertaint'ics are consistent with experiment results.'
One of the roles of expert ! views of the nature of this particular one is to l
help assure that such chenc'ena are not overlooked as well as to help identify those elements that are be'leved to be major contributors to the uncertainties.
The objective of this summary, then, is to try to identify the areas of major uncertainty in the source term studies under review and to make judgments as to the possible implications of these uncertainties. It must be realized that a determination of the actual impact of source term uncertainties is difficult to achieve without conducting a full, rigorous uncertainty / sensitivity analysis.
No such conplete analysis is presently available. The QUEST study [4] is the only known attempt to partially apply rigorous uncertainty / sensitivity analyses to quantify the uncertainties in source terna. Although limited in scope, the QUEST study is the best source presently available for identifying and quanti-fying important contributors to source term uncertainties.
3.2
SUMMARY
OF THE QUEST UNCERTAINTY ANALYSIS The QUEST study found.that the uncertainties in source terms are dependent on : time of containment failure, accident sequence, plant type, and plant spe-ci'ications'. The QUEST study found that the ratio of the upper bound to the best est' mate
- release decreased for sequences with increasing release level. The ratic was found to be of the order of 100 for the S D2 sequence (low release sequenu.) and 'o be of the order of 3 for the TML8' sequence (high sequence).
The ratio was of the order of 10-20 for the IC sequence (intermediate release ,
sequence). It ir believed that these values are likely to change as better l
- "Best estimate" in this context means that the calculations were not inten.
tionally biased either on the high or the low side.
1 I
n 1
infor ation becomes available and more complete uncertatnty analyses are made.
For early c'untainment fatlure, the source term uncertainty was found to be domi-nated by in-sessel processes. For late ccntatnrent failure, the scurce teia un-certainties were found to be dominated by ex-sessel processes.
'00EST also -found that all fission pteducts must be considered in chJrac-terizing the source term : iodine and caes tum dcminate for eat ly cent e tre,*nt fat ture, .#ereas refractory fission products and tellurium dominate for later containment fatlure. The suspended aerosol radtoactivity in contatnrant was ubserved to decrease rapidly with time (e.g., by a factor of ten atthin four hours of sessel failure).
3.3 PRIORITIES IN UNCERTAINTY EVALUATION FOR SOURCE TERMS Based on the present rev tew of Soutce Term Studies and the GUEST uncertainty / sensitivity analysis, Lt is clear that containment performance is the controlling factor in source term determination. If the centainment does not fail (either because of its inhereat capabilities or perhaps due to cperator intersention), this would be the dcminant contribution to reduced source terms.
Source term studies, to date, generally have not addressed the likelihoco of containment failure but instead have considered release conditional upon Lt.
The timing of containment failure determines ahtch aspects of source term de-termination will dominate the uncertainties.
Therefore, the major causes of source term uncertainties could be structured under the following categorizing situations :
- 1. Early centainment failure (relative to reactor vessel failure via melt through).
- 2. Late containment failure (relative to reactor vessel failure via melt through).
- 3. Containment by-pass.
- 4. Leakage, pre-existing openings and failure to isolate.
The following categorisation of uncertainties is an example of how such a categorisation scheme can help to focus the evaluation of the impact of un-certainties.
3.3.1 Early Containment Failure The QUEST Study found that source term uncertainty near the time of vessel failure is dominated by in-vessel processes. However, depending on the mode and extent of containment failure, it is believed by this Task Force that ex-vessel processes can also significantly contribute to the uncertainties for
, early containment failure in many sequences. Thus, if early failure is deemed to be of high enough frequency, then quantification of the uncertainty should consider the items covered in Sections 3.3.1.1 through 3.3.1.4 The reasoning leading to these examples can be found in the various reports making up the technical annexes to these summaries (to be published ~later).
~
3.3.1.1 RCS Retention
. Completeness of integration and coupling of all aspects of thermal-hydraulics and fissinn product transport in the HCS.
. Core melt progression models and experimental data base.
. . Mudelling of natural circulation in the reactor sescel and in the PWR loops both before and af ter vessel melt-through fatlure.
In-sessel release of fission products 'particularly Te, ch, 9'. Sr, tto, and Ru' and acrosnt generation from controi rods and Ofri.ctor.'s.
3.3.1.2 Reoxidation on High Pressure Cjection and/or Steam Cxplosions Oxidation of Ru in Centainment At'osphere Resulting From Dispersal uf Fuel Into the Containment Due to Pressurized Ejection From the Pcactor '.'essel or Ex-Vessel Steam Explosions.
3.3.1.3 Core / Concrete Interactions
. Core debris composition and temperature.
ftde and timing of vessel failure (as they affect debris dispersal and melt decay heat).
Presence and effects of a pool of water overlying the debris.
3.3.1.4 Water Oroplet Characteristics During Initial Blowdown Amount and Size of Water Droplets Made Airborne in Containment During the Initial Blowdown (in some sequences).
1 3.3.2 Late Containment failure For late containment failures (order of 5 to 10 h after sessel failure),
CUEST found that source-term uncertainty was dominated primarily.by ex-sessel processes. Particularly important sources of uncertainty 'are, then, believed to be :
3.3.2.1 Core / Concrete Interactions Core debris temperature and gas generation rates from the concrete.
. Composition of the debris bed.
Mechanisms affecting initial. dispersal of the debris.
3.3.2.2 Aerosol Behaviour in Containment
. Aerosol shape factors.
Turbulent agglomeration rates (turbulence level and agglomeration efficiency).
42 -
p, 1
r-i I
. Timing of steam condensation onto containment surf.4ces relati.c to aercsol release from corc/cuncrete interactiers.
. Treatment of the multi-conpurent nat ure of acrocols. I 3.3.2.3 Revaporization from the RCS 3.3.2.4 Suppression Pool Decontamination factors
. Cuantification of irput acrosol stie.
. Amount of hy-pass.
. Possible disti.pt icn of po013 v ia the ef fects of pres sor t t."1 eret ion of fuel from the reactor weasel ';fark III contair-ents).
3.3.2.5 Quantification of Ice Condenser Decontamination factors
. Sequence spcetfic items inctode .
3.3.2.6 Spray Efficiency Removing Aerosols
. Specification of drop size.
. Assurance of spray operation.
3.3.3 Containment By-Pass 3.3.3.1 Performance of Auxiliary Buildings
. Aerosol Behaviour (same as 3.3.2.2 above).
. Thermal-hydraulics (flow rates, steam condensation, etc.).
3.3.3.2 RCS Retention (same as 3.3.1.1. above) 3.3.3.3 SGTS Standby Gas Treatment System (SG'TS) Performance Under Sesere accident Conditions (Mark I and Mark III Containments).
3.3.3.4- Effectiveness of Fire Protection Sprays
. (Refer to 3.3.2.6).
. Actual location of spray system relative to location of aerosols.
e 3.4 LEAKAGE, PRE-EXISTING OPENINGS AND FAILURE TO ISOLATE In add 8 tion to those items discussed in 3.1 and 3.2 the characterization i of the leakage pathway is needed.
I .
rw-Fyrtbecmore some general coments regarding uncert iint Les are appropriate.
A rigorous esaluation of the effects and prnbability of c.tcurrence of steam spikes, fuel / water explostse interactions, pressurtzed ejecticn direct heat ing, and H2 cc-bustion is likely to ahow that the probabtlity of early contatec t fa t tore for large dry PWRs is low - reducing the actual .u tgo t ficarce of c.it e-
<; cry 3.3.1 abose. For large dry PnR3 tf failure is late enough, the natural acrosol removal processes will reduce the source term to an extent that the uncert aint ies v. auld be relatisely uni portant. Howeser, for those failure t ues of up to about 10 h. Lhe releases can be large and dGsuctaled utth large un-certainty.
If early ennt.rincent failure can be asuctated with a hv em;h prota-btlity, then the uncertaint ws of most concern would insolse those asscetated a tth late failure. If contatnnent fa t ture can be shown to be late enough, then natural aerosol remosal procesocs sill reduce the scurce terms to lesels at shtch the nocertatnties are uninportant '.except ing by-pas 3 Gegwncesi . For predicted late containment fat ture acquences, there would be time for cperator action ce mit tgation schemes to be trplemented that could perhaps present cen-tainment . failure.altcgether.
For syutems tn which reliance is placed on engineered safety features, plant specific considerations appear to play a large role and must necessarily .
be tncluded.
- 4. IMPLICATIONS OF UNCERTAINTIES IN SOURCE TERM ESTIMATES Based on the preceding discussion, Table I-provides a list of what we believe to be the present areas of major importance to source term uncertainty.
These are arranged somewhat in priority order.
In other sections of this report and the technical annexes (to be pub -
lished later), the significant progress made since WASH-1400 in source term methodology and the supporting scientific data base is quite prominent. At the same time, the report also delineates important areas where current uncertain-ties are sufficiently large to warrant additional experiment data and/or further improvement in phenomenological modelling, as well as model salidation. It is encouraging to note that in most of these important areas, research, on-going or planned, is in place to address these uncertainties.
We have learned from the source term studies that the importance of uncertainty in a given phenomenon or process with respect to its impact on source term estimation will depend upon both the specific design of a plant and the accident sequences of interest to that plant. Another dimension which must also be considered is that our ability to cope with the current uncer-tainties, or our need to reduce them further, will depend uoon the specific application of the source term information.
It is observed that experimental and analytical results for source term predictions will always be accompanied with some level of uncertainty. Ideally, a~ distribution of the probability of occurrence of source term values for a given sequence is desirable and would be of most benefit towards making regu-latory decisions. However, it is not likely that such distributions will be f
s f^,
. Table I PRESENT AREAS CF MAJOR IMPORIANCE TO SOURCE TERM tACERTAINTY 1 Items associated with containment loadings and mode and timing of failure.
- 2. Complete integration and coupling of all aspects of thermal-hydraulics and fission product transport.
- 3. Ex-vessel sources related to core / concrete inter-actions, pressurized melt ejection, and w3ter blow-down fragmentation.
4 Core melt progression modelling and experimental valioation.
- 5. Modelling of natural circulation in the reactor vessel and in the PWR loops both before and after vessel melt-through failure.
- 6. Aerosol behaviour related to turbulence effects, multi-compartment and multi-component treatment and to uncertainties in form factors under RCS and con-tainment conditions.
- 7. Retention of fission products in the reactor coolant systems.
- 8. In-vessel fission product release.and aerosol genera-tion from control rods and structures.
- 9. Scrubbing efficiency of engineered safety features.
I e
O
~
fortheoming on a timely basis. In the esantime, 'it is recommended th1t the assessments be based on best estimate (neither intentionally biased high nor low) values taking into consideration relevant esperimental and analytical results for. uncertainty bounds and, 'if necessary and apprcpriate, 'espects' judgments It should be decided by the different ccuntries how the best esti-mate values and related uncertainties are to be used for any of their picticular implementation.
We believe that the presently available analytical tools and data bases are sufficiently adequate to warrant the start of a regulatory implementatico phase that will consider specific applications of the current acurce teen information. Starting the implementation process new wtll alluw the exploration and determination of the eventual likely.applic7tions, and a balls can be dm.el-oped for establishing end-use criteria for defining acceptability. Then, the acceptability /non-acceptability of the present uncertainty levels for vrocific end-use applications can be identified and the research in progress or any additional research needs can be'prioritized. Hence, we see the proposed imple-mentation phase and the research needed to reduce uncertainties as being part of an iterative procedure. An example of this process and the relatlanships of various research elements (such as uncertainty analysis, data base, ethodology',
to implementation and end-use are illustrated in Figure 1. Figure 1 is divided between research and applications according to the dotted horizcntal line. The research elements proceeding concurrently are the development and improvement of the data base, and the methodology, along with uncertainty analysis develep-ment.
The important aspects of the flow chart can be summarized as follcws :
- 1. An Jarly data base and the WASH-1400 methodolugy have been used for source term estimates, but with little uncertainty ana'lysts.
- 2. The appreciation of the problems is now more sophisticated and the 1984 methodology supported by the larger 1984 data base has been used to obtain plant-specific source term estimates.
- 3. The reference plant analyses have included a scoping uncertainty analysis and significant uncertainties have been found to exist.
Improved uncertainty analyses are now under d,evelopment.
- 4. However, the information and methodology available is adequate now for initial use in a regulatory implementation. phase to assess limi-tations and to develop end-use criteria for the on-going research.
Some initial regulatory decisions are possible at this point while others may be limited by uncertainty, awaiting improved research information.
- 5. It is expected that additional data and further improvements in the methodology uncertainty analysis will allow additional use for further implementation and decisions.
- 6. Further iterations of this process will reach closure as the end-use criteria and uncertainties converge for remaining specific applications.
e
.- -~
,~.
- , . o a
.u..
- ~ . o e -
N -
O I -
T A
T
', 0 L
A a
E N
'n E
P s
-
- t; al -
oe - ,.
8 -
A .
S m
IE o
T -
s p
l A - .
T ty .
R e -
E t. .s -
C D t, $,
- , e.
N .s i U ,. c nys - .
n r U l.
L
- t.
H, e 1 J e
- s. n - .
C d d.A * , n R d w
A A t.p - .
E 1 S m E i -
R -
e r ; -" ; -
i a
u R g E i T -
F -
E C -
t s -
u ,.
O -
S ty - i.
s e e s l a lass - . .
E E ts rtey a
W T fe.l t nA a
n -
. A.
E U S .
- n -
PI 5
M m" y
8 g -
9 . o O o -
I e lo T B d A e h
o .
q .
RE ta t. -
R D o t
4 4 -
8 6 9 -
1 91 e -
e S -
)
m4 r -
ia r 1
De i lyt -
r r -
a o 4
(
E f -
i I3eiI IUE4 . a <u.L.4.ilIk
. V4 e k
I l i
O
- 5. BIBLIOGRAPHY
[1] FETRANCELI. C. and Z Aff lRO, C. (ENE A/ DISP - Italy), ! aE A.'iM-201/5 3 (IE A Symposium on Source Term Evaluation for Accident Cond:ttuns, Columbus, Ohio, 29 October-1 Noverrber 1985).
[2] PETRANCELI G. . Workshop on the Reiults of the CEC Study C . ,t r wt3 Scarce Ter., ( 1983-1985) : Grussels, 18-17 September, PM S , -
. ':e App 1Leaticn3 c! :he ?eaulcs of Recent Source ter.m 2t i!.vi c: ~ ;a-latory Issues, E.NEA/ DISP - Italy.
[3] SEN/ SIN (84)S0.
[4] LIPINSKI, R.J. et al, f/ncere stccy in Rafinnuclide 3eb .ise 2.%-r Spec:fic :XR Acc:!cn: C:nditicns, S A'.D 64-0410 ( three .al res).
O
LIST OF TASK FORCE MEMBERS AND OTHER CONfRIOUf0RS Members of the Task Fnrce Cincluding 'e sperts wno draf ted some of the Technical Anne.es) :
. Canada Dr. David F. TCAGERSCN (AECL), Chairman Dr. Wannan N. IENNA'*CriE ( AECL)
Dr. David J. WREN (AEct)
France Mr. hcques CUC0 (CEA)
Federal Republic Dr. Mars-GUnter FR'FCERICHS (O?S) of Cetmany Prof. Dr. E"ro F. ICxEN ICRS), Vice-Chairman Italy Mr. Gianni FEinANCELI : CiEA)
Japan Dr. Venihtsa SODA (JAERI)
Sve~ den Mr. kjell 0. JOHANSSCN (Studsvik Enctgiteknik)
United Kingdom Mr. Frank 486EY (L'KAEA)
Dr. Michael R. HAYNS (LKAEA)
United States Dr. Themas S. KRESS (CRNL)
Dr. Richard CEHLBERG (EPRI)
Dr. Robert L. RITZMAN (EPRI)
Dr. 8. Raj SEHCAL (EPRI)
Mr. Melvin SILBER 8EPG (USNRC)
Dr. Richard C. VOCEL (EPRI)
Other contributions were made by :
j Mr. Robert M. BERNERO (USNRC) i Dr. Brian 80WSHER (AEE Winfrith) l Dr. R. Allan BROWN (Ontario Hydro)
Dr.-Donald M. BRUCE (AERE Harwell) l Dr. James F. COSTELLO (USNRC)
Dr. Mario H. FONTANA (Energex ; IDCOR Report)
Mr. Richmond CARDNER (SWEC)
Dr. John T. LARKINS (USNRC)
Dr. Walter B. LOEWENSTEIN (EPRI)
, Mr. James METCALF (SWEC) l Mr. Giovanni NASCHI (ENEA)
Dr. Alan NICHOLS (AEE Winfrith)
' Mr. Andrew J. PRESSESKY (consultant to ANS)
Dr. Denwood F. ROSS (USNRC) -
Dr. William R. STRATTON (Stratton & Assoc. , Inc. : ANS Report)
Mr. Edward A. WARMAN (SWEC)
OECD Nuclear Energy Agency :
Dr. Jacques ROYEN. Secretary
. . . _ y
SOME OTHER NEA QUELQUES PUBLICAT JNS PUBLICATIONS ON DE L'AEN SUR LA NUCLEAR SAFETY SURETE NUCLEAIRE Internst.onal Comparison Study on Reactor Comparaison internationate sur la med4tisa-Accident Consequence Modehng (Summary tion des cons 4quences des ace,donts de Report to CSNI by an NEA Group of Cuperts. r4acteurs (REsum6 d'un rapport au CSIN par 1984) un Groupe d'esperts de TAEN.1984)
E9 50 US$19.00 F95.00 Nuclear Aerosois en Reactor Safety Les airosols nuctsaires dans la sGrete des (A State 4f-the-Art Report by a Group of reacteurs E mperts.1979) (Rapport sur ritat des connaissances 6tabli par un Groupe d'emperts.1979)
E8 30 US$18.75 F 75.00 Nuclear Aerosols en Reactor Safety - Supple- Les a6rosols nuct&aires dans la s0rets des rnent to the 1979 Report rsecteurs - Suppf4 ment au rapport de (Peport to CSNI by an NEA Group of Emperts. 1979 1984) (Rapport au CSIN par un Groupe d' experts de r AEN.1984)
C20.00 US$40.00 F200.00 Air Cleaning in AC tdent Situations L'Eputation des gat en situations acciden-(Recort to CSNI by an NEA Group of Experts, telles.
1984) (Rapport au CSIN par un Groupe d' experts de r AEN.1984)
C 10.00 US$20.00 F100.00 fn-Core Instrurrentation and Reactor Assess- instrumentation s. 6 valuation de rdtat du ment coeur des riacteurs nucidaires (Proceedings of a Specialists
- Meetsng. (Compte rendu d'une r6 union de sp4cialistes.
Frednkstad. Norway,1984) Frednkstad. Norv&ge.1984)
C17.00 US$34.00 F170.00 Continuous Surveillance of Reactor Coolant Suveillance en continu de rint6gnts du circuit Circuit inteenty de refroidissement des r6acteurs (Proceedmes of a CSNI Specialist Meeting. (Ccmpte rendu d'une r6unien de sp4ciahstes Pans.1985) du CSIN. Pans.1985)
C20.00 US$40.00 F200.00 e=,y_ e e en ey M h- *- ** -eewe * * - e em - g
OECD SALES AGENTS 08POSITAIRES DES PUBUCATIGNS DE L'OCDE
.. ,s.g M.. .as . l . . .g ... (
v13 3 - ..Cgges.scs....,,'-...c.-.,..
. * . ..' ,, a
...a *6...,,....,..
. . . .. .t g im , M '.22
. . va......a....,.
.!P
- . .s: n . .3
.u.t N.s -.v.. * . v.
3
. .,o..,,....,.u,3..
c
- 3. n . u. . . s..~,.........
~ .
-..r.
.,,......c
=, . ,s.-. -. m......u ~a.m. ~~. .
- ....s ...
- n. . -. c. .e... .
u....... ~ . wnm
.: - c. 1 1, m. . .. ,, u n .. n a + ovuu. nu. n
. u * * - -* ' , ~ :~.-~.
. - . ~ . . .m.....c. 3 .*.v ,- -
- x. , . n s . .
. . 2. , ; =. ,n *; 43, ;-
..,e..-....
.e . ..a
....i.,.......
.~..m..,,...,u
_......,.s...,
w.c.- . .
.v...,.... . ..
.s . . om . s~ u . . . w. . . m. , ...
s
.~,s..
.v . r.. . u. i 3, n .. .
.,..,.o- - C. ,......s .
.c .
,s. , o . . . .. ;
s.v,. u . , s
. , . m. <. . . . i n w.
.. oc. . . . . o..
. o . -.
... . .s-..,,,. o,.,~., "s -
mr u o . . ~m
, n 2sv............,,,,,
......-.,....-.o
. _ . . . ~u.
. . . m.oo . . .. . .. . . n
...a
, :, vea=u, 3;,l, ;
g,=,,,,,7,g..- n ns . . ..n ax. ..
m, w . .. n. .
g,=,g;,';pa a ~~ ~..- ~~ .: ::::' n e.T.:.~::" .u .. m co m. ,,.
. . .m.o.. - e-. , , . . . . . , , ,
o..C.. .m .co .. en n .. . . ..
.a. .
.c an .. : .. . . n.-..
....._ ,.. m .~. . m.a. C....,, n
. .n . u .. n w .. m u = m u ,..
. , ,, , % . . 0. . . .u. D.
- c. s.n. = e. n i . = . .o 5 C ..
s.a ..d... m . ..
- u. .n..v4v.n ~ u 2.
,,,,,,,,,.,,,,,,,, n_sm .-.
~ . . _
s
... , n a . n.".n .o .C. .,., n .u.. s. .o.m. ...n ,oC. . . a . . , a.
=. ,,,,,,y..,,,,,,,
o.. C.3 **. .= ...m .e.
a- . H . m C
C. e .. u v .a n
. . e.
C-Co .
...~. .
- m. u.n m.. e. .
C. . ,o c .. n .. .n v n
.c o.+
- O .se 30.. ; J.? .Y. f. 3 F e ..o o 2,C .e
. . m
.. ,on.. ..n s.,-... -..C. .
.. m >= m em oo.
l >
rm... .,
us. C.s un ,.,
.....~...,.n.....,....
m ..- n w n w.e <
n u.a.
,,.,,C,.,.,..,.....,.,,.,.
, . . , , , . . , . 3 ,
.. . .e.
. o., n r. v . .ro.--...c~,,,.m
- .;
- ,,,; .~
- a,.."aa n
......r..u. ,
.. . .n , as.ra v. == .. a n =
.. -. , .. - . . . .3 ,cu
. n. . c.o . ..
t_e.v . .. ,., . . m n , , ,
,,. . a. . *. v. so. o ,
. g. . . 3.e 3 3 = .- F .. Fe.
.*.*.******.**4H.1233...
J g
.c gmu.en ,.mm a' -e .- -. w=ma C..+.
- *co' t i.
w e.,m. oC33.. .., .e n 'o.4e,
.e e.,.
,..n.y.,.
n .e ..=C m .51'o***.WF aaca-f.O p C . . . f. . .-ve J.a .
.? *. ... . O.a.. 9 e .m 0.C.3***.=w. C . r.,0=
CAC.D..w t .s m.e I3 FC J.f.m 90 ..F ?.4
. - C.".
W. t' .9' .3035 72.F* .. 4 * . 0 C 2000. . M.2 9.0 . . C.agg (e FU tA. . . 9 2 G n
- . . ..s. . . , C e,m . e R ?.L F3 73 C. C.s eo. '. 33 23. 33 2. .,D.
,en 3.
L.
O . L.a .. s.
,m vv.00 L.a . .O.u.0 A 4 3 e g . 3. .soca.o
.,.e.3.ba.g ... M 3.
. s p 0 .e. + maut ,.3,... a
....i L es s. n e .w e e>. .
OCO. .i . '.'ICO. . .
- i.i4.,.*f.... C.t.. og D . m. u e. e C.C3 _ 0 i 8 *.es. PS ' '. p.e.s GM.. g.. ., . e . u.
y
6 4
an C4C3 AceCarcars a w a.e,p.. , y,r'S paars Cfott 10 ane 42533 sW M N 'f 3 im SEANCE 44to03ft44=SJ44-ea m 1
I
. _ . . _ _ _ . _ - . . - - - _ . . . . _ _ _ j l