ML20072C136
Text
.u,_____n__..
~
~
"~~ ~~
~"
~~
/
f ne:
k UNITED STATES
[
NUCLEAR REGULATORY COMMISSION WASHINGTON. D. C. 20055 January.16,1992 g
<--E
~
MEMORANDUM FOR:
Commissioner Rogers
.~
FROM:
James M. Taylor
~.
Executive Director for Operations y
j
SUBJECT:
STAFF REVIEW 0F RECENT STATEMENTS ON SAFETY GOALS
)
In a memorandum from G. H. Marcus to James L. Blaha, dated November'27, 1991, l
the staff was asked to review a recent analysis of safety goals by Mr. Steven Sho11y. As requested in that memorandum, the staff has focused.Its review on point A.1 of Mr. Sho11y's analysis and the associated Appendix A.
In this appendix, Mr. Sholly concludes that a 3400 Mwth reactor without a containment t
could have a core damage frequency of roughly 3 x 10" per reactor year and still meet the Commission's quantitative health objectives (QHO).
Mr. Sholly makes other assumptions and assertions in his analysis (e.g.,
l average core damage frequency across industry is 3 x 10-* per reactor year) which the staff has not addressed in this memorandum.
i The staff's explanation and observations are provided in the enclosure and the main points are summarized here.
o His choice of population dose consequences for poorly mitigated core meltdown accidents (and his resulting core damage frequency) is at the i
high end of present consequence estimates, is not based on a 10 mile population distance calculation in accord with the safety goal policy, ahd is not clear regarding the use of mean values and consistent assumptions; o
The Commission's individual early fatality quantitative health objective can be more constraining than the Commission's individual latent cancer fatality objective and, apparently, Mr. Sho11y's related goal, by approximately an order of magnitude or more; and o
Possible subordinate goals, such as the large release guideline, could-be more constraining than either of the quantitative health objectives or Mr. Sho11y's goal. Since the exact definition of this goal has not yet been approved, the staff is unable to address how much more.
constraining such a subordinate goal might be. However, NUREG-1150 information, using one proposed definition of "large," indicates that such a goal could be more constraining by one to several orders of magnitude. The Commission in its June 15, 1990, SRM on safety goal
Contact:
Mark A. Cunningham, RES 492-3965 9408170096 940629 I
PDR COMMS NRCC CORRESPONDENCE PDR
g3 Commissioner Rogers 2
j implementation acknowledged that in the large release guideline an inherent order of magnitude constraint could exist relative to the quantitative health 1 objectives.
It must be noted, however, that existing risk analyses indicate that there is i
considerable margin between plant internal event risks and the individual early fatality objective, and therefore, it is possible that plant designs could in concept, have significantly higher risks (for example, a factor of thirty greater than that shown for the Surry internal events risk in NUREG-1150) and still meet the quantitative health objectives.
It is also important to reiterate here that it is the body of NRC regulations and NRC's regulatory oversight that act to define and maintain acceptable design and operational practices in commercial reactors.
It is these practices that have served to constrain the frequency of core damage accidents and the containment integrity requirements against large releases of radioactive material, not safety goals per se.
If the NUREG-ll50 plant and site-specific risk assessment is assumed as a-representative snapshot of the i
levels of safety achieved by our regulations and oversight, then the quantitative health objectives in the safety goal. policy are being attained with a wide margin.
Is I
a Ex uti e Director for Operations
Enclosure:
As stated cc: The Chairman c4tfiiim'issioner Curtiss Commissioner Remick Commissioner de Planque SECY OGC 1
C i
i Enclosure l
Evaluation of Safety Goal Analysis
-by Mr. S. Sho11Y.__
In his analysis, Mr. Sholly has created a safety goal based on collective population dose within 50 miles of the site. This measure is related to, but not identical to, one of the Commission's two quantitative health objectives, i
the average individual latent cancer fatality risk within 10 miles. He has i
used the " acceptable" risk increment of the Commission's quantitative health j
objectives,,0.1 percent. Using the 50-mile average population around active commercial nuclear power plant sites and the average latent cancer fatality i
death rate in the United States, as well as the BIER V dose response information, Mr. Sholly calculates that his goal is equivalent to approximately 7000 person-rem per year in the 50-mile population.
t i
Mr. Sholly then assumes an estimate of population dose consequences (20 million person-rem, citing NUREG-1150) for a poorly mitigated core meltdown j
accident (i.e., one in which either early containment. failure er containment i
bypass occurs). As Mr. Sholly notes, such a' consequence is at the high end of j
those calculated in NUREG-Il50. However, the source of this consequence j
estimate in NUREG-Il50 is not specified. Also in NUREG-IISO, population dose i
calculations are made using site-specific population data; it is not clear j
which site / population Mr. Sho11y used.
i With both an " acceptable" risk and assumed consequences defined, Mr. Sholly j
then calculates the frequency of an accident that would correspond with this i
risk. This results in his conclusion that such an accident could have a i
frequency as high as 3 X 10 per year. The validity of this value is not
]
clear from the analysis presented since the average pcpulation value used in j
determining the 7000 person-rem number may not be the same as that corresponding to the 20 million person-rem number and the 20 million person -
rem may not be a mean value (which are to be used in safety goal comparisons).
i Since the staff does not normally deal with a safety goal such as that defined 4
i by Mr. Sholly, it is difficult to quickly assess all of its implications.
In
{
particular, some additional effort would be required to analyze the i
relationship between his goal and the average individual latent cancer l
fatality objective, although the two do seem closely related.
I j
In lieu of performing such work, the staff has examined NUREG-1150 data to q
address the question of how much worse a plant could be and still meet the i
Commission's quantitative health objectives. Figure 1, taken from NUREG-1150, i
provides comparisons of the accident risks of five plants with the j
Commission's two quantitative health objectives. As may be seen, Figure 1 i
shows that, for internally initiated accidents, the plants meet the objectives i
with considerable margins. This can be interpreted that these plants yield very small risks from internally initiated accidents. This also means, i
however, that these plants, if otherwise unconstrained, could have j
significantly higher risks and still m et the objectives.
A j
As you are aware, the staff is assts+g the use of (and definition of) a j
subordinate goal relating to the Mgacy of a "large" release of radioactive i
material. An evaluation of such a m i, using one proposed definition of i
i i
= -_
i, -__
.l.
I i
i l'-
Commissioner Rogers 2
"large" was also made in NUREG-1150,. and is shown in Figure 2.
If such a goal l
were accepted, it would be considerably more constraining than the individual i~
early fatality QH0 (by one to several orders of magnitude); the individual latent cancer fatality QHO, and, apparently, Mr. Sho11y's goal. The Commission has previously acknowledged (in its June 15, 1990, SRM on safety j
goal implementation) that a large release goal could be inherently more
{
constraining than the QH0s.
j NUREG-1150 information can also be used to assess how much more frequent a specific accident could be and still meet the quantitative health objectives, 4
j The Surry plant, the risk of which is dominated by centainment bypass accidents with relatively large consequences (the type of accident discussed i
by Mr. Sholly), can be used as an example. The Surry plant has a somewhat 1-lower power rating than the plant cited by Mr. Sho11y. (2441 Mwth vs. 3400 i
Mwth); this difference is not believed to be significant for purposes of this i
analysis. The mean 50-mile population dose from a containment bypass accident i
at Surry is estimated to be roughly one million person-rem. This is j
considerably less than that assumed by Mr. Shelly. For Surry the 10-mile population-is approximately 100,000, which would translate to 240 person-rem i
per year, using an " acceptable" risk increment 0.1 percent.. Using Mr.
j Sho11y's method of calculating a frequency this would result in 2 X 10" per year. However, the staff considers this population dose more appropriate for i
use, since safety goal. assessments are focused on mean values.. This 3
difference in population dose, however, does not have a major impact on the j
staff's conclusions.
I i
1 As Figure 1 indicates, the Surry plant mean risk is roughly a factor of thirty j
below the individual early fatality QHO, and roughly a factor of one thousand below the individual latent cancer fatality QHO. With this margin, and if j
unconstrained by regulations, a containment bypass accident at Surry could j
have a mean frequency cf roughly 1 X 10" per reactor year (vs. the NUREG-1150 estimate of roughly 3 X 10" per reactor year) and just meet the-individual 1
early fatality QHO, while meeting the individual latent cancer fatality QH0 with some margin.
i If unconstrained by regulations sit the individual early fatality objective, a 4
containment bypass accident at Surry could have a mean frequency of roughly j
6 X 10" per reactor year and just meet the individual latent cancer fatality
{
objective using the information in Figure 1.
Since the population dose goal j
defined by Mr. Sholly is more closely related to the individual latent cancer i
fatality quantitative health objective than the individual early fatality 1
objective, this example appears to be more directly comparable to his 3
analysis.
3 In summary, Mr. Sholly's analysis of safety goals may be pessimistic, because:
o His choice of population dose. consequences for poorly mitigated core 1
meltdown accidents (and his resulting core damage frequency) 1:: at the j
high end of present consequence estimates, is not based on a 10 mile population distance calculation in accord with the safety goal policy, and is not clear regarding the use of mean values and consistent assumptions; 3
1 3
a 1
,=
=.:. a - -..
^
I-I Commissioner Rogers 3
l c
The Commission's individual early fatality quantitative health objective can be more constraining than the Commission's individual latent cancer f
j' fatality objective and, apparently, Mr. Sho11y's related goal, by f
approximately an order of magnitude or more; and o
Possible subordinate goals, such as the large release guideline, could be more constraining than either of the quantitative health objectives l
or Mr. Sho11y's goal. Since the exact definition of this goal has not yet been approved, the staff is unable to address how much more i
constraining such a subordinate goal might be. However, NUREG-Il50 i
information, using one proposed definition of "large," ir,dicates that f
such a goal could be more constraining by one to several orders of magnitude. The Commission in its June 15, 1990, SRM on safety goal 4
implementation acknowledged that in the large release guideline an inherent order of magnitude constraint could exist relative to the j
quantitative health objectives.
]
It must be noted, however, that_ existing risk analyses indicate that there is considerable margin between plant internal event risks and the individual j
early fatality objective, and, therefore, it is possible that plant designs
(
could in concept, have significantly higher risks (for example, a factor of d
thirty greater than that shown for the Surry internal events risk in i
NUREG-1150) and still meet the quantitative health objectives.
i It is also important to reiterate here that it is the body of NRC regulations and NRC's regulatory oversight that act to define and maintain acceptable j
design and operational practices in commercial reactors.
It is these practices that have served to constrain the frequency of core damage accidents a
and the containment integrity requirements against large releases of F
radioactive material, _ not safety goals per se.
If the NUREG-1150 plant and p
site-specific risk assessment is assumed as a representative snapshot of the levels of safety achieved by our regulations and oversight, then the i
quantitative health objectives in the safety goal policy are being already j
attained with a wide margin.
I, 4
)
I
)
I
9 lt +
l-I t
i individual early fatality /ry 1.0E-06 Leeena
!&8afety Goal 1.0E-07 g
- es 1.0E-06 s
'~
6 t
l I
1.0E-09
_s
)'
1.0E-10
=
l 1.0E-11 SURRY PEACH SEQUOMH GRAND ZION i
BOTTOM OULF l
\\
l Individuallatent cancer fatality /ry 1.0E-06
=
Legend
<>=== safety Goal l
1.OE- 06 s
-- = en si.n--
\\
~
l
~
- es 1
I i
1.0E-07 a
i E
1.0E-08 1.0E- 02 a
j 1
1.0E-10 SURRY PEACH SEQUO1AH ORAND ZION BOTTOM GULF-I Figure 1 Comparisons of NUREG-1150 results to safety goals mu--
. ~
=
1 I
l 1
I I
i Probability of a large release 1.0E-06 L sg. s.l
- R.4 0 t n.1198 e.
f ty 9 6 Lee no l
1.0E-06 s -:
-us l
)
1 1.0E-07
=
as l
\\
i 1.0E-08 s 1.0E-09 s 1.0E-10 SURRY PEACH SEQUOYAH GRAND ZION BOTTOM GULF 1
I l
i 1
i i
l Figure 2 NUREG-1150 large release comparisons l
A Lm
..-a u
-.e v
w w.
e ' O.. '
- 54
,[.
_g 1'
w.
3:, * -'
. s.
I p.
' A['s.
. t 4'.
d'.
, ' L. < > e uh ' '
t,
.s;<
~
..) +
e P-y J k-
,,g,y.
y I
.dI
~}'4*, kN f1dP4 -
gaiNff-
-~ *h h.lh.' } i' 1[ d.".EI ' *t.
.bl
'l M 'mNfsip. p.;~,6,,_, JA$.a;4W/ h.,'s,
.,. gh /hTN 7 N
T:-
.c ', ' -~ ^I,4[.+ M' * '
4#1 5
l' 4 ggy
- a..
.e ;., s fp y
m w
.w.
.g a w ;%.g.,
- i (c < 42: hr e
p w.
a'.
Y,:, +
,,+
gf.-
.J(.
g
'p+
,e 4
A a
o?
w
,,.-t-
-a s
q4 m
a r
'44 f
s.
y g
s
- W-^
a a
F l
4 w -,
wem--
r,
- - - -