ML20072M317
Text
-
No si. ' 8i t, ' A kl S e
LICENSING ADVANCED NUCLEAR POWER PIANTS 1
1
~
Herschel Specter New York Power Authority 123 Main Street White Plains, N.Y. 10601 j
914-681-6994 I
i 3
have occurred and. almost all of the i
ABSTRACT public's nuclear plant risks comes.
g from them.
It.seems illogical to Tedcy's operating nuclear power regulate nuclear power plants using j plcnts, and future ones as well, design basis accidents, which
{ nasd a more modern licensing basis, sometimes can be both far less i This paper suggests one way of probable and far less severe than I precsoding which ties together the events that experience and j sofoty goals, PSAs, and various NRC modern technology have identified.
j Rules.
It describes a technique by
'Such regulation can lead to a skewed i: which items important to safety can distribution with resources poorly i ba precisely identified.
Should distributed, i.e., overbuilt in some I
cuch cn approach be adopted, design areas and missing risk reducing-j 4
i cartification of advanced plants opportunities in others.
i
- could be greatly simplified.
l.
Pronosal-i
['TTRODUCTION In order to improve the regulatory k
From one perspective the process process, it'is proposed to divide i by which nuclear power plants are nuclear power plant operations into i regulcted today in the United States three areas:
! has basa highly successful.
Its l basic purpose is to protect the
- 1. Operations. outside the scope of l haalth and safety of the public.
nuclear regulatory review, i This has been done. Throughout many
- 2. Operations addressing severe lcparctingyearstheseplantshave accident events, and
- bssn casentially consequence-free.
3 Day-to-day safety operations such On tha other hand neither the as_ plant security, implementing the
{ nuclasr industry nor'its regulators ALARA principle, handling and
)eraantirelysatisfied. There are storage of nuclear fuel'and
! crass of conflict between the radioactive wastes, nuclear
] prassne, design basis accident emergency _ planning, and the release
! epproach used to regulate nuclear of-radioactive plant effluents.
1 plents and the ' lessons learned from I expsrience and probabilistic safety
~
Areas 2 and 3, above, would come
' acessaments (PSAs).
Some design under NRC review, but only area 2 i basis cccident analyses are based on will be discussed here.
Four tasks lsavarallayersofconservative should be pursued under area 2:
' casumptions, and if converted to a f PSA-bened approach, their very low
- 1. Establish an Overall Regulatory
! probebilities would cause them_to be Framework,
!~'.saissedasnegligiblecontributors
- 2. Identify the risk significance of
.o risk..Further, the present systems, structures, and components
- licensing approach does not-(SSCs],
icomprehensively consider core melt
- 3. Identify the dominant failure iaccidents.
Yet core melt accidents O
9409010270 940629 i
PDR COMMS NRCC CGtRESPONDENCE PDR 1
m
-~
~ _. - - _ _ - _ -
?
I nodas of thase risk significant SSCs Dafining whr.t is safety-related is j
- and, of critical importance to design -
l
_4. Utilize regulatory and certification of advanced nuclear un-regulatory processes that plants. As reported in the press 3
efficientiv reduce the frequency of "The Commission adopted a staff tha cbove dominant failure modes to recommendation that the agency take
+
l ccespeably low levels, i.e., levels a " graded approach" to the level of that result in conformance with the detail issue. Under this approach, j
Ovarell Regulatory Framework, the level of detail required for l
certification review will vary j
This approach could be applied to depending on a structure's, j
both present plants and future system's, or component's j
en2s.
Indeed, it can be argued that relationship to safety". The first chare should be a regulatory major _ gradation would occur when the
}
continuum that encompasses both PSA/ safety. goal approach is used to j
-pressnt and future plants. One way solve for the boundary of the~
i to establish this continuum is to safety-related SSCs. Additional' l
update the licensing bases of PSA-based ranking methodologies can i
present plants by applying these then be used to produce even finer j
fcur tasks.
gradations within the safety-related SSCs,_if desired.
1, An Overall Rerulatory Framework 4
In order to show conformance with Much has been said about the the safety goals, two other issues
{
potsntial benefits of standardized would have to be considered:
4 d2 signs. Yet an even more operator actions and initiating j
fundamental.need is a " standardized" events.
PSA technology can also regulatory framework, such as rank the risk significance of 4
mational safety goals. Not only various_ operator accident recovery
..nuld such goals answer "How safe is actions.
Such information would be 1
l safe enough?", they can be used to quite valuable in structuring an i
dstarmine which plant features are accident management program and'in j
safety-related and which are not.
operator training. Similarly, PSAs j
Those systems, structures, and can identify and rank the most risk l
corponents [SSCs) which PSA analyses significant initiating events.
j nhow are necessary to meet the Therefore, PSAs can also be used to j
safety goals, such as a core melt create boundaries between i
fraquency goal, would be labeled as risk-significant and safety-related. All other SSCs risk-insignificant operator actions i
bayond those reeded to meer une and initiating events.
l safety goals would not be j
safocy-related, unless they are part one can then monitor actual plant j
of area 3, above. A conservative performance against PRA input j
asthsd for determining'the parameters, such as calculated j
ocfaty-related/not safety-related safety-related SSC reliabilities and 3
SSC boundary is to set the unavailabilities and the calculated i
untvsilability of selected SSC's to
_ frequencies of the more important i
1.0 in a PSA. As the number of initiating events. This supports f
SSC's with assumed unavailabilities the "living PSA" process, the of 1.0 expands the PSA " bottom-line" monitoring requirements of the NRC's j
nuabar would rise. All SSC's with recent Maintenance Rule, and en assumed unavailability of 1.0 at conformance to the safety goals.
j tha position where the calculated j
- bottom-line" value matched the Another Overall Regulatory i
tafety goal would be not Framework benefit is that it j
safety-related. In order to deal encourages design and operational
{
withtheisgueofPSAuncertainty, optimizations.
For example, one can j
an approach recently suggested by trade off inherent strengths in a the ACRS could be used.
design to offset other areas that are not as capable, so long as the i
I
i 4
d overall safety goals are est.
Such Thus the use of en Overall i
treda-offs may, however, be Regulatory Framework:
) ~ intentionally restricted to retain j
ufficient defense-in-depth.
For
- 1. Establishes how safe is safe
,. purposes of illustration only, the
- enough, i
follcwing criteria might be utilized
- 2. Identifies which plant features 3
as an Overall Regulatory Framework are safety related,
'that offers flexibility, while
- 3. Identifies which operator actions assuring defense-in-depth and a very are most risk significant, thereby Icw risk:
affecting operator training,
- 4. Provides a basis for the' j
MEANOVERALLSIGg/RY Maintenance Rule's monitoring IFICANT RELEASE FREQUENCY - 10'
- program, l
MAXIMUMMEANCOp/RY MELT FREQUENCY,
- 5. Encourages design and operational j
[CMF)
- 10' optimization, while meeting both 1
MAXIMUM MEAN CONDITIONAL CONTAINMENT defense-in-depth and strict release.
I FAILUREFRAC{ ION, criteria, and,
[CCFF) = 10'
- 6. May encourage public support for s
nuclear power.
i Such an arrangement would i
satchlish " floors" on both Innortance Rankins and Dominant
~
I prevantion (CNF) and mitigation Failure Modes (CCFF).
Since the product of the i
"ficor" values of CKF and CCFF is a Plant SSCs can be evaluated
]
factor of ten larger than the according to their importance.to required overall significant release risk by use of a ranking process i
fraqusncy, further risk-reducing based on a PSA technology called stsps might have to be taken.
For importance weighting.
Importance
- 'BVRs further steps might be directed
. weighting provides a numerical
- ehieving a lower CMF, perhaps to measure of each SSC's safety worth.
j 10'g/RY.
FWR's, especially those 4 with large dry containment designs, During the past few years a number f cright favor more emphasis on smaller of importance weighting measures j
CCFF's in order to meet the severe have been developed such as the'
}
relsace frequency criterion.
In any
" risk achievement worth", " risk l case, the plant designer / operator reduction worth", the Birnbaum would have flexibility on how these structural importance measure, and jl edditional steps might be taken, the Fussell-Vesely importance measure.
For the purpose of j
Ccamunicating the strict core melt illustration only, the.importance
! and esvere release frequency inessure chosen here is the " risk critaria and the resultant very low achievement worth".
, early and latent fatality health i risks that nuclear plants are held After completing a PSA analysis of
{
to, may be an additional way to an individual plant, PSA results encourage public support, would be recalculated assuming Furtharmore, if NRC policy requires system X1 was totally absent from j that advanced units meet even more the plant. The absence of system X1 j stringent criteria than present could cause an increase in various
- plants, this too might be more calculated results, such as the core i clectly communicated when Overall damage frequency. This increase j Regulatory Frameworks are used.
For could be called delta XI.
The base example, i present plants need to seata10'{/RYcoremeltfrequency case would then be perturbed again by assuming that a different system, j
tigerionandadvancedplantsa X2, was totally absent and its.
4
..J' /RY criterion, it should be associated delta X2 would be found.
lapparentthatadvancedplantswould This perturbation process of the have to conform to a ten fold base case would then be repeated one stricter CMF criterion.
system at a time and the resultant deltas would then be ranked from a
5
_- = -
i larga to reall. Ths larger the feilure modzs within tha service dalta, the more safety significant water system itself could be i
ths system.
wasteful.
In fact, by concentrating on the dominant failure modes there Ranking SSCs according to their may be justification for reducing ocfsty worth has many benefits.
Its-service water system test and chief benefit is that it facilitates maintenance outage frequencies.
tha distribution of licensee and NRC Since this could improve the raecurces, i.e.,
according to an availability of the service water SSC's safety importance. This can system, overall risks could be lead to both safety improvements and smaller.
coct reductions..Such ranking could also be helpful to implementation of Evaluatina Rerulatory Procemmes ths NRC's Maintenance Rule and the Plent I.icense Renewal Rule.
Not only is it valuable to rank Furthsr the level of detail issue SSCs according to their importance in cdvanced plants which is to be of safety and to identify their cddrsosed in a " graded approach",
dominant failure modes, it is also cculd use a safety importance valuable to determine the waighting system.
First the effectiveness of different ecfoty-related/not safety-related regulatory processes in reducing bsundary would be established, then risks. Achieving a quantitative a " rick achievement worth" or some evaluation of different; regulatory othar ranking method would produce processes is a challenging task, but finer gradations. The level of the rewards could be substantial.
detail could then be correlated to For example, consider the regulatory the renking.
impact if it were shown that, in general, operator training was a
' Although ranking itself is hundred. times more cost-effective
.taluable, more should be done.
than QA/QC in reducing risks.
After SEC ranking one should then cencontrate on the dominant failure As a first step in this direction,
)
modes of the risk significant SSCs.
several efforts have been made in PSAs can also be used to identify 1991 that offer insights on the risk thseo dominant failure modes.
significance of QA/QC. 'Ihese efforts are limited in scope and Cencentrating on the dominant preliminary, however, their failure modes is important because implications are immense.
en initial response to an SSC with a 3
high ranking might be to expend Stevenson presented slides at maximum resources, such as maximum
'the July, 1991 ANS topical meeting QA/QC efforts, on this system.
where-he described the very large Hawavor, one must be selective on costs associated with our present what cnd how resources are applied, QA/QC approach. He suggested cvan for highly ranked SSCs.
alternate ways of performing QA/QC which he believen would be both less To illustrate this, consider the costly and more effective, service water system.
It is likely 4
that this system will have a high Rahn has conducted a data ranking in all plants. There might search. He examined the reported tharefore be a temptation to " gold failure rates of various plats the service water system, components.
Some of these o.g. very frequent testing of components were classified as service water components.
- However, safety-related, while similar
.ha PSA would likely show that the components were not safety-related.
d:ninant failure modes of the total Preliminary comparisons of the data ssrvice water system are initiating do not show any statistical svents that cause a total loss of a difference in the failure rates of critical support system.
It follows the safety-related components that efforts put into non dominant
compared to thair not safety-related counterparts.
All of the above concepts of 5
importance weighting, dominant Vasely performed a PSA failure modes, effectiveness of asnsitivity study on an actual various regulatory processes,'and 1
nuclsar plant. He reports "For optimum distribution of resources cxample, with regard to quality are discussed in the NYPA document, centrol and qualification requirements we found that a plant's Conclusion risk performance is insensitive to largs variations in the failure Risk-based regulation provides.
ratsa of a significant portion of opportunities for both present and-the components in the plant, even advanced plants.
Should nuclear whsn these variations are systematic regulation evolve in this direction, end result in higher failure rates the public could bene' fit from both for all components".
safer and lower cost nuclear power.
If QA/QC is very expensive, if it i
doss not produce a discernable REFERENCES statistical impact on SSC failure races, and may still be risk insignificant even if there was a meccureable impact on SSC failure
- 1. D. A. WARD, Chairman ACRS, "The retas, then it is time to reevaluate Consistent Use of Probabilistic Risk ths QA/QC process.
Assessment", Letter to the Honorable
. Ivan Selin, Chairman, USNRC, July NYPA Prorram 19, 1991
- Tha New York Power Authority
- 2. Nucleonics Week, Page 9, Feb. 28, NYPA) has informally submitted a 1991
~ draft document it developed,
' RISK-BASED REGU1ATION", to some
- 3. J. D. STEVENSON, " Resolving c::r.ber:: of the NRC staff. After Current Operational Safe':y furthar discussion and development Challenges", Plenary Session. ANS, of tha ideas in this document, this' Portland, Oregon, July 21-25, 1991 may lead to a NRC/NYPA pilot i
program, using NYPA's James A.
- 4. F. J. RAHN, Personal FitzPatrick (JAF) plant as a Communication, August, 1991 reference plant.
- 5. W. E. VESELY, " Risk-Based Ona of the early objectives would Prioritization of Operational be to develop risk-based technical Activities: Achieving Both Risk epscifications.
In the long term it Reduction and Burden Reduction",
is hepad that JAF would be regulated Plenary Session, ANS, Portland, 4 en a risk basis to a much greater Oregon, July 21 - 25, 1991
- d2groo.
5 4
T i
j 4
i i
1 l_