ML20028F937
| ML20028F937 | |
| Person / Time | |
|---|---|
| Issue date: | 03/11/1982 |
| From: | Sherry R NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Pasedag W NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| Shared Package | |
| ML20027A699 | List: |
| References | |
| FOIA-82-530 NUDOCS 8302070181 | |
| Download: ML20028F937 (1) | |
Text
A(
W. Pasedag, Section Leader Accident Evaluation Branch FRON:
R. Sherry Fuel Behavior Branch
SUBJECT:
REVIEW 0F ORNL DRAFT INTERI!! SDURCE TERM REPORT
References:
- 1) Albrecht H. and Wild H., " Investigation of Fission Product Release by Annealing and Melting of LWR Fuel Pins in Air and Steam," Paper presented at the Topical :
Meeting on Reactor Safety Aspects of Fuel Behavior, August 2-6, 1981, Sun Valley, Idaho.
t
- 2) SAND 74-0382, " Core Melt Experinental Review," March 1977
- 3) NUREG/CR-1410. " Report of the Zion / Indian Point Study,"
Volone 1. August 1980
- 4) HUREG-0850, " Preliminary Assessment of Core Melt Accidents at the Zion and Indian Point Nuclear Power Plants and Strategies for Hitigating their Effects," Volume 1
' November 1981 sy.
I have completed my. review of the ORNL report.
I think the results of this study are interesting from the standpoint that independent scientists using the sane data base can obtain widely different estimates for accident source tems. This study also shows that it is not intuitively obvious that source tems have been grossly overestinated by pst-studies as has been claimed by some. What these studies do show (and that I fimly believe) is that source tem predictions have large uncertainties.
I believe that the ORNL predic tions are within the range of uncertainty (as are the EPRI predictions).
[
However, I don't believe that the quantitative ORNL results are the Lest estimate values.
i I believe that the ORNL source tem esticates are high. The principal reasons they are overestimates are:
1)
Fission product release during the phase of the accident when the molten core is attacking the lower reactor vessel are being over pre-dicted. The reasons these releases are being overestimated are that release rates from small scale tests (approximately 100g) are being used j
l for very large volumes (approximately 10-100 ton) of naterials. The j
widely different S/V ratios and the enhanced inportance of mass trans-l port resistance within the relt is being neglected.
~
N
~J w.uth..................................................
8,3,0,2070,181 821221 e01 out>
SHOLLY82-530 PDR
.R.q.......
7.................e...............
H. Pas-dag E 1 1 1932 Cote that Albrecht cciculates wch icwer releases for this phase of the accident Msed on his SASCM data and on an assured terperature of 2400*C (Ref.1).
Another reason that the releases are high r.ay be because the r.elt temperatures during this phase of the accident are being overpredicted.
I don't know what temperatures tcere calculated, however.
(I think this type of inforation should be included in an appendix to the report.)
2)
Fission product release during the selt/ concrete interaction phase of the accident are higher than what I would expect. Palt temperatures should be, in general, much lower than for the previous in-vessel phase and this should limit the releases.
The effective surface area of the celt is much larger, hoWever, because of the large volumes of gas which are passing through the relt. The gas flow and the mre chemically reactive environment could significantly enhance release on the other r: T. hand.
The values contained in the ORNL report are based on the inforcation presented by Haaland (Ref. 2).
The values given on Table B.4 (1st
- ~
8 coluen) are the range (s) estimated by Haaland. The lower value being that calculated by Parsley and Fontana (for the Reactor Safety Study) and the upper value being calculated by Haaland assuming a factor of 1^ i s
greater gas release from melt / concrete interactions than Parsley and Fontana assumed. The fission product release value assumed as a best estimate by ORNL is typically within the range (between Parsley and Fontana, and Haaland), but is cuch closer to the higher (Haaland) esti-mates.
I believe these (ORNL) estimates are mch to high.
First of
~
1 gas generation value assumed by Parsley and Fontana all, the got3) agrees quite closely with recent esgwjtes mde using the (1.25(10) ft advanced (WECHSL) core / melt concrete code (1.2(10) ft ) (Ref. 3).
Secondly, the temperature (3000*K) assumed by Parsley and Fontana is ruch higher than the peak polten core teoperature estimates I have seen for this phase of the accident.
For example, MRCH calculations per-fomed for HUREG-0850 (Ref. 4) predict initial melt temperature at the time of vessel failure of frw 2150*C to 2450*C.
Since fission product (and aerosol) release rates are a very strong function of tersperature, I would expect even the estinates of Parsley and Fontana to be high.
[
3)
I don't know what the bases was for the leach estimates given on Tables B.13, 8.14, and B.15.
It is claimed that these values are the fractional losses 10 days after water contacts the failed fuel rods. These numbers, however, are very much larger than Katayana's leach data quoted in Table B.2.
Whats going on - did someone forget that the values in Table B.2 are a percentage and not a fractional value as used in Tables B.13, B.14, and B.157 It certainly looks that way.
.. ~..
.c
-y
.........-.,,.nu-an n
r-i r i ai nrennn e-n n v
1 I,'AR 11 yggy
~
W. Pasedag 3_
)
In Lddition to tha cbove co..,2nts. I believe the irportt.nce of prirtry system retention has been underestimated by ORNL as wall as the effects of delayed j
containeent failure on source term reductions.
Enclosed are my detailed corrents on the report.
R. R. Sherry Fuel Behavior Branch Division of Accident Evaluation
Enclosure:
As stated DISTRIBUTION
..SUBJ._. - ~ ~ - ~ ~ ~ - -
CI RC CHRON BRANCH RF p.
RRSherry RF lt RRSherry MSilberberg i
t i
6 i.
+
l s
b) 0
..F.3. 6..k.I.).Ah.c..
....FB..B.:. /?
on cr>
- euwm) RRSher y,/..pr7.MS.i l b.e.r...e..r.
3./../(/. 8. 2.....,.. 3[. /../.8. 2.......
o*">
.. c. c.,,.., n. n o mne com nec.ic,As ovenon rnpv
Detailed Comments P-10, last paragraph, 4th line
" chemical and physical processes and engineered _
safety features".
P-ll, last paragraph, 7th line - Volatilization? (vaporization? i.e., core melt / concrete release).
P-14, Section 3.4.1.3 - Where are the total energy release rates given - did I miss them some where in document?
P-17 Table 3.1 - A companion table should be shown which shows the research currently underway to improve the techn'ml bases.
P-17 Table 3.1
" Transport through containment" "first principles code."
I believe most required data'is available.
In fact, most of the mechanisms are included in one or more of the existing containment codes.
.cMppendix A cr_r - _:....
P A-2, 3rd sentence from bottom
" entirely dry environment" must be' defined j.I (e.g., superheated steam and hydrogen environment).
P A-4, first paragraph, 7th sentence - I don't believe that a majority (or even a large minority) of the fuel rods will fail for a large LOCA (with j
successful ECCS operation) if one look? st it realistically.
P A-5, 5th sentence from bottom
" bypass the drywell (not containment)".
P A-7, 9th sentence "However, if_ water were present in the pressurizer o_r_
pressurizer quench tank".
P A-7, 3rd sentence from bottom "the containment and containment safety features".
Appendix B P B-2, Item 2)
" Diffusion from fuel matrix prior _t_o; melting".
P B-3, 2nd paragraph - Rapid release of volatile radionuclides occurs at the time of cladding rupture and then more slowly as volatile species evolve from their condensed phase (to try to reestablish equilibrium, e.g., condensed phases of cesium uranate and cesium iodide release gas phase constitutes Cs
- and CsI, respectively) which are subsequently released.
(I don't buy this additional release mechanisms of shallowly imbedded gas.)
P B-3, 2nd line from bottom - Delete "at temperatures above the melting point of U0," (approximately 2800 C) and insert just "from molten fuel".
The temperature at which the UO fuel becomes liquid in a fuel rod is predicted to be much less then stoici$ metric UO melting point because of mixing with lower melting point components (e.g.,2 cladding).
.- P B-5, 8th line - (mainly H, 00, C9, and steam).
2 2
P B-5, last line
" thick crusts of concrete decomposition materials Islags) over".
P B-6, 2nd paragraph - same as previous comment on shallowly imbedded gas.
P B-6, 2nd paragraph, 6th line - Replace " plenum" with " Interconnected voidage (i.e., plenum, fuel cladding Sapt cracks, pores, etc. )".
P B-7, line 6 - Replace "do" with "did".
P B-8, 5th line from bottom
. Rewrite sentence, biggest effect of oxidation
.(steam explosion) release was the predicted increase in the Ruthenium group release fraction (due to formation of volatile oxides).
The correct values are shown on Table B.l.
P B-12, 4th sentence - What is "at best" here supposed to mean (imply).
y
~
P B Note, Leach values on this table are a percentage. Later on Table B.13: s B.14, and B.15 these values seem to be interpreted as fractions, s
P B-20, 2nd paragraph - Note that FASTGRASS models only the pre-melt heatup phase of the release.
P B-21, 2nd paragraph - Iodine chemistry has been considered important for a long time.
P B-21, 2nd paragraph - Rewrite last sentence - garbled.
P B-30, first paragraph - Preliminary infprmation on control rod material release rates are available from Albrecht (KfK) and from Parker (0RNL).
See Reference 1 from cover letter.
P B-31, Section B.5.2, line 3
" molten core materials, (U0,, contal rods, cladding, and structural materials).
P B-31, last two sentences - Using the empirical release rates derived from small scale experiments for the releases from the molten fuel during the phase of the accident when the fuel is in the lower reactor plenum is prob-able excessively conservative because of large differences in the surface /
~ volume ratios.
The melt release model in the START (BCL) computer code could probably be used to generate better estimates than directly using the (0RNL) empirical release rate correlations.
During this phase of the accident, it at temperature)portant to specify correctly the debris temperature (and ti is also very im In this regard the temperature histories of the debris should be supplied for a few representative sequences.
P B-32, line 8 - The INTER code is largely obsolete.
CORCON-MOD-1 and WECHSL are much more advanced codes and are available for analyzing melt / concrete interactions.
, P B-35, lith line from bottom "WECHSL or CORCON which are an improvement".
P B-36, first line - Aerosols are mainly concrete decomposition products.
P B-36, 2nd paragraph, 2nd line - Add CORCON.
P B-38, 2nd paragraph, 7th line (also P B-39, line 1) - Replace " containment building" with " reactor cavity".
P B-4'., line 14 - Replace " melting temperature of the U0 " with " temperature 9
at which the fuel rods are assumed to melt." Reason - mp of UO approximately 2800*C, mp of fuel _ rods (cladding + UO ) may be as low as approkimately 2
1850*C (mp of cladding).
to U? Where P B-40, paragraph 2, line 4 - What has reduced 50% of the U02
, did this assumption come from?._
P B-42, Section B.6.3, lines 7 and 8 - Replace " cladding melting" with " cladding failure".
s P B-42, Section B.6.3, line 9 - Replace " fuel melting" with " incipient melting of the fuel rods".
P B Note previous comment on Leach values for Tables B.13, B.14, and B.15.
P B What is the peak temperature assumed for the AD " gap release." If temperatures were assumed not to exceed 1200*C, (I doubt they would) then the escape fractior.s (volatilization) are excessively high. The ORNL gap release model (Lorenz, Malinauskas, et.al.) was obviously not used for the Cs and I calculations for this table - why not?
P B-55, Table B.16 - Aerosol releases prior to RV failure seem low.
For example, 280 kgs for event V is made of approximately:
140 kg Cs 12 kg I 23 kg Te 30 kg Mo 35 kg Ba 10 kg Sr 8 kg Zr (FP)
~
2 kg Ru 260 kg Indicating only a minimal amount of structural naterials in the aerosol.
This result is in variance to other studies, e.g., Albrecht, et.al. (Ref.1),
where it calculated that 3500 kgs of aerosols are released prior to RV failure (including 1800 kg of control rod silver, 450 kg of structural steel, and 450 kg of U0 ).
Conversely, some of the estimates for aerosol release during 7
melt concreta interactions seem incredibly high (9100 kg, 23000 kg).
_4 P B-59, 2nd paragraph, line 9 - The fact that neither WECHSL nor INTER currently account for gas evolution from radiatively heated concrete above the surface o_f_ the nolten pool is irrelevant since this gas would not contribute to sparging of fission products.
P B-59, paragraph 2, line 12 - The C0P. CON code has a relatively sophisticated '
model for chemical reactions between the gas passing through the melt and the constituents of the melt.
But I don't see ho' the chemical interactions would effect the total volume of gas released s 'ce the typical reactions are:
H 0g + M M0
+H 2
x 2
CO2+M M0 + C0 x
(M = metal species in melt)
~
~
Appendix C P C-5, Table C Indicates chemical reactions are probably not important -
I disagree.
Indications are chemical reactions could be very important - in particular reactions of species such as I, Cs, and Te with structural materials or control materials (e.g., Ag) resulting in less (or more) volatile compounds.
P C-9, first paragraph, line 2 - This sentence is way to strong.
Quick models, in some detail, aerosol processes in a one node region.
However, in no way can it be considered to be better suited than TRAP for RCS transport calculations.
For example, QUICK does not consider FP vapor forms, or inter-action between vapors and aerosols nor does QUICK consider multi-node geom-etries connected by convective flows (i.e., the RCS). Also, TRAP-MELT now has models for gravitational agglomeration and settling.
P C-ll, first paragraph, last sentence - Figure C-1 has nothing really to do with RPV failure times.
All this figure is intended to show is the impor-tance of initial mass concentration and (residence) time on aerosol retention.
behavior.
P C-11, 2nd paragraph, first line - Figure C-1 says nothing about 12 P C-ll, 2nd paragraph, Fth line "one" should be "none".
P C-14, 2nd paragraph, line 4 - To condense o_r chemisorb".
P C-11, 2nd paragraph, line 5 - This also depends on the relative surface areas of the respective surfaces (aerosol vs. structure).
P C-15 and C RE Maximum aerosol concentrations - Table C.3 does not appear to give an accurate portrayal of what the maximum (initial aerosol) mass concentrations would be.
It assumes, I believe, simply X kg of aerosols distributed within some limited volume (subset) of the RCS. A more approp-I l
riate method for determining the maximum initial concentration is to divide
~
the mass release rate M by the volumetric flow rat ~e of steam and hydrogen f
(through the core).
- Hence,
1
. Ci
= M_
(M/sec)_
max 3
j (m /sec)
P C-19. Section C.4.1.4, first paragraph, last line
" unheated water"? How about subcooled or nonboiling as a better word than unheated.
P C-20, last several lines - The assumption that no vapor species retention retention mechanism for many volatile species (e.g., I, Te)y is a major occurs is probably invalid.
First, chemisorbtion conceivabl Second, vapors that condense or chemisorb onto particles may deposit hith the particles in cooler locations in the RCS.
P C-23, Table C.5 - I have major problems with these quantitative results.
Where do the numbers come fro,m? Are they guestimates? For the group A estimates the "best estimate" are always chosen as the maximum value in the range _these are strange distributions.
P C A lot of statements are made in this section which need to be supported.
Why can't the current models (i.e., TRAP-MELT) be used in a predictive mode?-
Why isn't it possible to quantify the uncertainties to any useful degree of
'I resolution? I don't agree with these overly sweeping generalizations.
I do agree that large uncertanities exist in this area however.
Table C.6 - TRAP-MELT does include models for aerosol agglomeration and i
gravitational settling.
Appendix D u.
P D-1, 5th line - Replace " reactor" with " plant".
P D-2, last paragraph, 2nd line - Replace " reduction" with " retention".
P D-3, 2nd paragraph,1st line - Replace'" gases" with " vapors".
P D Delete " larger".
All mechanisms act on particles of all sizes (m hanisms are just more affective for certain particle sizes).
P D-4, last paragraph, D-5, 3rd paragraph, D-6, last paragraph - Several places it is stated that condensation enhances agglomeration.
This may or may not be the case for a given situation.
Condensation increases the mass of an aerosol and hence increases its settling velocity.
Condensation may also cause "long chain-like-agglomerated" aerosols to contract into spherical aerosols (which should also increase their settling velocity).
Increases in settling velocity tend to increase the effects of gravitational agglomera-tion. However, the shape change may significantly reduce the cross section for aerosol interaction and act to reduce agglomeration rates.
P D-6, 2nd paragraph, line 2 - Delete " escaping from the containment".
P D-10, 2nd paragraph, line 4 - MULTI-AEROS (A module in the CONTAIN code) being developed by Sandia is claimed to have a model for steam condensation on aerosols.
y
,.. P D-13, last paragraph - However, the amount of materials that escape from 1
the containment at the time containment fails is strongly dependent on the i
predicted concentration at that time and not on the initial concentration.
{
P D-16, last paragraph - This writeup is confusing.
What the figure shows is that up to a given concentration.(correspondong to 700 kgs) there is an approximately linear relationship between initial aerosol mass concentration (source strength) and leaked mass (note that aerosol removal processes are still acting and may be significant). Above 700 kgs, the effect departs from linearity as increased aerosol concentrations promote more rapid aerosol deposition.
2 P D-18, MARCH III BWRs are nearly as large as some of the "large" PWRs.
P D-21, first paragraph - BWR t%RCH I and II's drywell and wetwell = containment (suppression pool is in wetwell).
BWR MARCH III's drywell is inside containment - suppression pool is also
~
inside containment.
See attached figure.
~
Y P D-21, 2nd paragraph - See previous comment on effect of steam condensation
'i on aerosol behavior.
P D-22, 2nd paragraph - Similar comment as before on MULTI-AEROS.
P D-23, first paragraph - Equal to _orJ1ess than design leak rate occurs for design pressure.
For pressures less than design, much less than design leak rate, for pressures in excess of desi.gn,- maybe greater than design leak -
i rate.
P D-25, Section D.4.2.4.1, first paragraph, first line - Why? Because of temperature?
I can visualize situations where H burns occur or where con-7 tainment sprays with stron(y caustic additives are operating.
I think this would qualify as something more than a " region of comparatively low chemical activity." Next sentence - with exception of CH I or species which enter the.
3-liquidpgaseandhydrolyze(e.g.,I2 3
. HOI and I and 10
,-or dissocicte Csl Cs and I, etc).
P D-26, 3rd paragraph, 5th line - Replace " reactor" with plant".
P D-27, 2nd line - The section on 0772 on organic iodine formation mechanisms should be credited to Campbell.
Table D I disagree with the comments on the importance of chemical reactions.
Particularly, when I consider H burns, which could have a major effect on 2
the chemical forms of fps suspended in the containment atmosphere or deposited on surfaces.
Table D PWR Ice Condenser Main steam isolation valve...
N/A N/A (instead of No)
~
Table D Footnote b (see attached figure courtesy of GE).
Table D Footnote d.
Not really true.
Shape factors are allowed (input) which account for nonspherical particle properties.
In any event the "assump-tion" of particle shape had nothing to do with steam condensation effects.
Table D Check your values for containment free volume. Large dry PWRs 6 3 and IP has a volume of approximately 2.6 (10)goxjma'ely 2.f(}0)
For example, Zion has a volume of app are somewga}] bigger.
ft
[7.4(10) m ].
[7.6 (10) m Table D I don't understand what the fractional values indicate for event-Y.
Footnote b indicates they are the fraction released into containment that escapes from the containment.
But for event Y nothing is released,to,the containment.
Table D Footnote b.
These are not numbers. What do you multiply?
-~
Table D-9
" Meltdown column" replace " reactor" with " plant" (or " containment")......
(First and second paragraph.)
Table E.5 and E.6 - Are confusing and hard to understand.
Section E2, first page, first paragraph - Whats the difference between " lack I t,
of understanding" or " lack of knowledge".
en s
l l
1 89'
- - _ - _ _g g gM w m use M gry'u wsk
^' M
-Mk i
utke MK
% u r;+rcs oveu.t b ownu rw w yen p,
ji chgoJM j
MARK 111 CONTAINMENT prywq cchndd I
1 N
f
,,.-=. '- e Q N f,;
REACTOR BUILDING
-r I
- 1. Shield Building g [g
)
N
- 2. Free. Standing Steel Containment
- 3. Polar Crane f(, i
,'ig,
l
'). g.;
,;tl j 'q
[
- 4. Refueling Platform 3
=
R a to e
1 g
- 8. Steam Line j
'h'!a ll,g' j g {
etj mr,7
'.I ih.
f
(
-g{ t j
1
,q
- 9. Shield Wall g
- 10. Feedwater Line j
,.g.
{
5
- 11. Drywell
[. Ip i.p f 3 i r' '
5 7
- 12. Recirculation Loop
- 13. Weir Wall
!I f
i, k, [, d'., f(f 8 l
l ii 1 j'y
- {Q h..
i V
l
- 14. Horizontal Vent
~l s
h 4
(,,
[,,
b
- 15. Suppression Pool i
i-p.
j lJp.q,
.t'(.[.,
'l AUXILIARY BUILDING
(
- .,I
.;g y
- 16. Steam Line Tunnel C,?
3.g l
j
'; o
'b
'?-
M HR S t
[.
ry
.,,, y ;r :
' f.
p i
- ' ;., 4,
FUEL BUILDING
, [,,}{g
,g;,
_.1/
7
' * * ' g;f'r,", l n- $
1
'7 8
- 19. Fuel Transfer Bridge
, g)j,,. :
rc:i p
JQ,
,,; j,
- p <
q
- 20. Fuel Transfer Tube
@j.1 q
,,y
.{,.,
,3 ' \\,,,, f
- 21. Cask Handling Crane
'4
,,,. 4,s
',.,, (
q
'y
' 9..
i
- 22. Fuel Stor' ge Pool a
~,A.
fr, '
(\\
' h I..,
N
- 23. New Fuel Vault I
- 24. Cask Loadi:.g Pool u
h M
- c. W T
- 25. Spent Fuel Shipping Cask
.P i
- 26. Fuel Cask Skid
'i I
'i G E N E R A L '@9) E LE CTRIC
."..o.
i.
c'. * -
- .t