[Federal Register Volume 71, Number 53 (Monday, March 20, 2006)]
[Notices]
[Pages 14018-14028]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 06-2611]
-----------------------------------------------------------------------
NUCLEAR REGULATORY COMMISSION
Notice of Availability of Draft Interim Staff Guidance Document
for Fuel Cycle Facilities
AGENCY: Nuclear Regulatory Commission.
ACTION: Notice of availability.
-----------------------------------------------------------------------
FOR FURTHER INFORMATION CONTACT: James Smith, Project Manager,
Technical Support Group, Division of Fuel Cycle Safety and Safeguards,
Office of Nuclear Material Safety and Safeguards, U.S. Nuclear
Regulatory Commission, Washington, DC 20005-0001. Telephone: (301) 415-
6459; fax number: (301) 415-5370; e-mail: [email protected].
SUPPLEMENTARY INFORMATION:
I. Introduction
The Nuclear Regulatory Commission (NRC) continues to prepare and
issue Interim Staff Guidance (ISG) documents for fuel cycle facilities.
These ISG documents provide clarifying guidance to the NRC staff when
reviewing licensee integrated safety analysis, license applications or
amendment requests or other related licensing activities for fuel cycle
facilities under 10 CFR part 70.
II. Summary
The purpose of this notice is to provide the public an opportunity
to review a draft ISG, FCSS-ISG-10, Revision 2, which provides guidance
to NRC staff to determine whether the minimum margin of subcriticality
is sufficient to provide an adequate assurance of subcriticality for
safety to demonstrate compliance with the performance requirements of
10 CFR 70.61(d). Additionally, listing of comments received on the
previous draft and the dispositioning of these comments is also
provided. These documents are being issued to support a public meeting
scheduled for April 28, 2006, at the NRC Headquarters Auditorium in
which the NRC will discuss revision of the guidance document and its
resolution of comments received on Revision 1. A separate meeting
notice will be provided shortly, to give specific details regarding the
meeting agenda.
III. Further Information
The documents related to this action are available electronically
at the NRC's Electronic Reading Room at http://www.nrc.gov/reading-rm/adams.html. From this site, you can access the NRC's Agencywide
Documents Access and Management System (ADAMS), which provides text and
image files of NRC's public documents. The ADAMS ascension numbers for
the documents related to this notice are provided in the following
table. If you do not have access to ADAMS or if there are problems in
accessing the documents located in ADAMS, contact the NRC Public
Document Room (PDR) Reference staff at 1-800-397-4209, 301-415-4737, or
by e-mail to [email protected].
------------------------------------------------------------------------
Interim staff guidance ADAMS accession No.
------------------------------------------------------------------------
Draft FCSS Interim Staff Guidance-- ML060260479
10, Revision 2.
Comments on Draft FCSS ISG-10, Rev. 1 ML060470150
and Resolution.
------------------------------------------------------------------------
This document may also be viewed electronically on the public
computers located at the NRC's PDR, O 1 F21, One White Flint North,
11555 Rockville Pike, Rockville, MD 20852. The PDR reproduction
contractor will copy documents for a fee.
For the Nuclear Regulatory Commission.
Dated at Rockville, Maryland this 7th day of March 2006.
Melanie A. Galloway,
Chief, Technical Support Group, Division of Fuel Cycle Safety and
Safeguards, Office of Nuclear Material Safety and Safeguards.
Draft FCSS Interim Staff Guidance-10, Revision 2
Justification for Minimum Margin of Subcriticality for Safety
Prepared by Division of Fuel Cycle Safety and Safeguards Office of
Nuclear Material Safety and Safeguards
Issue
Technical justification for the selection of the minimum margin of
subcriticality for safety for fuel cycle facilities, as required by 10
CFR 70.61(d)
Introduction
10 CFR 70.61(d) requires, in part, that licensees or applicants
(henceforth to be referred to as ``licensees'') demonstrate that
``under normal and credible abnormal conditions, all nuclear processes
are subcritical, including use of an approved margin of subcriticality
for safety.'' There are a variety of methods that may be used to
demonstrate subcriticality, including use of industry standards,
handbooks, hand calculations, and computer methods. Subcriticality is
assured, in part, by providing margin between actual conditions and
expected critical conditions. This interim staff guidance (ISG),
however, applies only to margin used in those methods that rely on
calculation of keff, including deterministic and
probabilistic computer methods. The use of other methods (e.g., use of
endorsed industry standards, widely accepted handbooks, certain hand
calculations), containing varying amounts of margin, is outside the
scope of this ISG.
For methods relying on calculation of keff, margin may
be provided either in terms of limits on physical parameters of the
system (of which keff is a function), or in terms of limits
on keff directly, or both. For the purposes of this ISG, the
term margin of safety will be used to refer to the margin to
criticality in terms of system parameters, and the term margin of
subcriticality (MoS) will refer to the margin to criticality in terms
of keff. A common approach to ensuring subcriticality is to
determine a maximum keff limit below which the licensee's
calculations must fall. This limit will be referred to in this ISG as
the Upper Subcritical Limit (USL). Licensees using calculational
methods perform validation studies, in which critical experiments
similar to actual or anticipated facility applications are chosen and
then analyzed to determine the bias and uncertainty in the bias. The
bias is a measure of the systematic differences between calculational
method results and experimental data. The uncertainty in the bias is a
measure
[[Page 14019]]
of both the accuracy and precision of the calculations and the
uncertainty in the experimental data. A USL is then established that
includes allowances for bias and bias uncertainty as well as an
additional margin, to be referred to in this ISG as the minimum margin
of subcriticality (MMS). The MMS is variously referred to in the
nuclear industry as minimum subcritical margin, administrative margin,
and arbitrary margin, and the term MMS should be regarded as synonymous
with those terms. The term MMS will be used throughout this ISG, and
has been chosen for consistency with the rule. The MMS is an allowance
for any unknown (or difficult to identify or quantify) errors or
uncertainties in the method of calculating keff that may
exist beyond those which have been accounted for explicitly in
calculating the bias and its uncertainty.
There is little guidance in the fuel facility Standard Review Plans
(SRPs) as to what constitutes sufficient technical justification for
the MMS. NUREG-1520, ``Standard Review Plan for the Review of a License
Application for a Fuel Cycle Facility,'' Section 5.4.3.4.4, states that
there must be margin that includes, among other uncertainties,
``adequate allowance for uncertainty in the methodology, data, and bias
to assure subcriticality.'' An important component of this overall
margin is the MMS. However, there has been almost no guidance on how to
determine an appropriate MMS. Partly due to the lack of historical
guidance, and partly due to differences between facilities' processes
and methods of calculation, there have been significantly different MMS
values approved for the various fuel cycle facilities over time. In
addition, the different ways licensees have of defining margins and
calculating keff limits have made a consistent approach to
reviewing keff limits difficult. Recent licensing experience
has highlighted the need for further guidance to clarify what
constitutes an acceptable justification for the MMS.
The MMS can have a substantial effect on facility operations (e.g.,
storage capacity, throughput) and there has, therefore, been
considerable recent interest in decreasing margin in keff
below what has been licensed previously. In addition, the increasing
sophistication of computer codes and the ready availability of
computing resources means that there has been a gradual move towards
more realistic (often resulting in less conservative) modeling of
process systems. The increasing interest in reducing the MMS and the
reduction in modeling conservatism make technical justification of the
MMS more risk-significant than it has been in the past. In general,
consistent with a risk-informed approach to regulation, a smaller MMS
requires a more substantial technical justification.
This ISG is only applicable to fuel enrichment and fabrication
facilities licensed under 10 CFR part 70.
Discussion
This guidance is applicable to evaluating the MMS in methods of
evaluation that rely on calculation of keff. The
keff value of a fissionable system depends, in general, on a
large number of physical variables. The factors that can affect the
calculated value of keff may be broadly divided into the
following categories: (1) The geometric configuration; (2) the material
composition; and (3) the neutron distribution. The geometric form and
material composition of the system--together with the underlying
nuclear data (e.g., v, X(E), cross section data)--determine
the spatial and energy distribution of neutrons in the system (flux and
energy spectrum). An error in the nuclear data or the geometric or
material modeling of these systems can produce an error in the neutron
flux and energy spectrum, and thus in the calculated value of
keff. The bias associated with a single system is defined as
the difference between the calculated and physical values of
keff, by the following equation:
[beta] = kcalc - kphysical
Thus, determining the bias requires knowing both the calculated and
physical keff values of the system. The bias associated with
a single critical experiment can be known with a high degree of
confidence, because the physical (experimental) value is known a priori
(kphysical [ap] 1). However, for calculations performed to
demonstrate subcriticality of facility processes (to be referred to as
``applications''), this is not generally the case. The bias associated
with such an application (i.e., not a known critical configuration) is
not typically known with this same high degree of confidence, because
the actual physical keff of the system is usually not known.
In practice, the bias is determined from the average calculated
keff for a set of experiments that cover different aspects
of the licensee's applications. The bias and its uncertainty must be
estimated by calculating the bias associated with a set of critical
experiments having geometric forms, material compositions, and neutron
spectra similar to those of the application. Because of the large
number of factors that can affect the bias, and the finite number of
critical experiments available, staff should recognize that this is
only an estimate of the true bias of the system. The experiments
analyzed cannot cover all possible combinations of conditions or
sources of error that may be present in the applications to be
evaluated. The effect on keff of geometric, material, or
spectral differences between critical experiments and applications
cannot be known with precision. Therefore, an additional margin (MMS)
must be applied to allow for the effects of any unknown uncertainties
that may exist in the calculated value of keff beyond those
accounted for in the calculation of the bias and its uncertainty. As
the MMS decreases, there needs to be a greater level of assurance that
the various sources of bias and uncertainty have been taken into
account, and that the bias and uncertainty are known with a high degree
of accuracy. In general, the more similar the critical experiments are
to the applications, the more confidence there is in the estimate of
the bias and the less MMS is needed.
In determining an appropriate MMS, the reviewer should consider the
specific conditions and process characteristics present at the facility
in question. However, the MMS should not be reduced below 0.02. The
nuclear cross sections are not generally known to better than ~ 1-2%.
While this does not necessarily translate into a 2%
[Delta]keff, it has been observed over many years of
experience with criticality code validation that biases and spreads in
the data of a few percent can be expected. As stated in NUREG-1520, MoS
should be large compared to the uncertainty in the bias. Moreover,
errors in the criticality codes have been discovered over time that
have produced keff differences of roughly this same
magnitude of 1-2% (e.g., Information Notice 2005-13, ``Potential Non-
Conservative Error in Modeling Geometric Regions in the KENO-V.a
Criticality Code''). While the possibility of having larger
undiscovered errors cannot be entirely discounted, modeling
sufficiently similar critical experiments with the same code options to
be used in modeling applications should minimize the potential for this
to occur. However, many years of experience with the typical
distribution of calculated keff values and with the
magnitude of code errors that have occasionally surfaced support
establishing 0.02 as the minimum MMS that should be considered
acceptable under the best possible conditions.
Staff should recognize the important distinction between ensuring
that processes are safe and ensuring that
[[Page 14020]]
they are adequately subcritical. The value of keff is a
direct indication of the degree of subcriticality of the system, but is
not fully indicative of the degree of safety. A system that is very
subcritical (i.e., with keff [Lt] 1) may have a small margin
of safety if a small change in a process parameter can result in
criticality. An example of this would be a UO2 powder
storage vessel, which is subcritical when dry, but may require only the
addition of water for criticality. Similarly, a system with a small MoS
(i.e., with keff ~1) may have a very large margin of safety
if it cannot credibly become critical. An example of this would be a
natural uranium system in light water, which may have a keff
value close to 1 but will never exceed 1. Because of this, a
distinction should be made between the margin of subcriticality and the
margin of safety. Although a variety of terms are in use in the nuclear
industry, the term margin of subcriticality will be taken to mean the
difference between the actual (physical) value of keff and
the value of keff at which the system is expected to be
critical. The term margin of safety will be taken to mean the
difference between the actual value of a parameter and the value of the
parameter at which the system is expected to be critical. The MMS is
intended to account for the degree of confidence that applications
calculated to be subcritical will be subcritical. It is not intended to
account for other aspects of the process (e.g., safety of the process
or the ability to control parameters within certain bounds) that may
need to be reviewed as part of an overall licensing review.
There are a variety of different approaches that a licensee could
choose in justifying the MMS. Some of these approaches and means of
reviewing them are described in the following sections, in no
particular preferential order. Many of these approaches consist of
qualitative arguments, and therefore there will be some degree of
subjectivity in determining the adequacy of the MMS. Because the MMS is
an allowance for unknown (or difficult to identify or quantify) errors,
the reviewer must ultimately exercise his or her best judgement in
determining whether a specific MMS is justified. Thus, the topics
listed below should be regarded as factors the reviewer should take
into consideration in exercising that judgement, rather than any kind
of prescriptive checklist.
The reviewer should also bear in mind that the licensee is not
required to use any or all of these approaches, but may choose an
approach that is applicable to its facility or a particular process
within its facility. While it may be desirable and convenient to have a
single keff limit or MMS value (and single corresponding
justification) across an entire facility, it is not necessary for this
to be the case. The MMS may be easier to justify for one process than
for another, or for a limited application versus generically for the
entire facility. The reviewer should expect to see various combinations
of these approaches, or entirely different approaches, used, depending
on the nature of the licensee's processes and methods of calculation.
Any approach used must ultimately lead to a determination that there is
adequate assurance of subcriticality.
(1) Conservatism in the Calculational Models
The margin in keff produced by the licensee's modeling
practices, together with the MMS, provide the margin between actual
conditions and expected critical conditions. In terms of the
subcriticality criterion taken from ANSI/ANS-8.17-2004, ``Criticality
Safety Criteria for the Handling, Storage, and Transportation of LWR
Fuel Outside Reactors'' (as explained in Appendix A):
MoS >= [Delta]km + [Delta]ksa
where [Delta]km is the MMS and [Delta]ksa is the
margin in keff due to conservative modeling of the system
(i.e., conservative values of system parameters).
Two different applications for which the sums on the right hand
side of the equation above are equal to each other are equally
subcritical. Assurance of subcriticality may thus be provided by
specifying a margin in keff ([Delta]km), or
specifying conservative modeling practices ([Delta]ksa), or
some combination thereof. This principle will be particularly useful to
the reviewer evaluating a proposed reduction in the currently approved
MMS; the review of such a reduction should prove straightforward in
cases in which the overall combination of modeling conservatism and MMS
has not changed. Because of this straightforward quantitative
relationship, any modeling conservatism that has not been previously
credited should be considered before examining other factors. Cases in
which the overall MoS has decreased may still be acceptable, but would
have to be justified by other means.
In evaluating justification for the MMS relying on conservatism in
the model, the reviewer should consider only that conservatism in
excess of any manufacturing tolerances, uncertainties in system
parameters, or credible process variations. That is, the conservatism
should consist of conservatism beyond the worst-case normal or abnormal
conditions, as appropriate, including allowance for any tolerances.
Examples of this added conservatism may include assuming optimum
concentration in solution processes, neglecting neutron absorbers in
structural materials, or assuming minimum reflector conditions (e.g.,
at least a 1-inch, tight-fitting reflector around process equipment).
These technical practices used to perform criticality calculations
generally result in conservatism of at least several percent in
keff. To credit this as part of the justification for the
MMS, the reviewer should have assurance that the modeling practices
described will result in a predictable and dependable amount of
conservatism in keff. In some cases, the conservatism may be
process-dependent, in which case it may be relied on as justification
for the MMS for a particular process. However, only modeling practices
that result in a global conservatism across the entire facility should
be relied on as justification for a site-wide MMS. Ensuring predictable
and dependable conservatism includes verifying that this conservatism
will be maintained over the facility lifetime, such as through the use
of license commitments or conditions.
If the licensee has a program that establishes operating limits (to
ensure that subcritical limits are not exceeded) below subcritical
limits determined in nuclear criticality safety evaluations, the margin
provided by this (optional) practice may be credited as part of the
conservatism. In such cases, the reviewer should credit only the
difference between operating and subcritical limits that exceeds any
tolerances or process variation, and should ensure that operating
limits will be maintained over the facility lifetime, through the use
of license commitments or conditions.
Some questions that the reviewer may ask in evaluating the use of
modeling conservatism as justification for the MMS include:
How much margin in keff is provided due to
conservatism in modeling practices?
How much of this margin exceeds allowance for tolerances
and process variations?
Is this margin specific to a particular process or does it
apply to all facility processes?
What provides assurance that this margin will be
maintained over the facility lifetime?
[[Page 14021]]
(2) Validation Methodology and Results
Assurance of subcriticality for methods that rely on the
calculation of keff requires that those methods be
appropriately validated. One of the goals of validation is to determine
the method's bias and the uncertainty in the bias. After this has been
done, an additional margin (MMS) is specified to account for any
additional uncertainties that may exist. The appropriate MMS depends,
in part, on the degree of confidence in the validation results. Having
a high degree of confidence in the bias and bias uncertainty requires
both that there be sufficient (for the statistical method used)
applicable benchmark-quality experiments and that there be a rigorous
validation methodology. Critical experiments that do not rise to the
level of benchmark-quality experiments may also be acceptable, but may
require additional margin. If either the data or the methodology is
deficient, a high degree of confidence in the results cannot be
attained, and a larger MMS may need to be employed than would otherwise
be acceptable. Therefore, although validation and determining the MMS
are separate exercises, they are related. The more confidence one has
in the validation results, the less additional margin (MMS) is needed.
The less confidence one has in the validation results, the more MMS is
needed.
Any review of a licensing action involving the MMS should involve
examination of the licensee's validation methodology and results. While
there is no clear quantifiable relationship between the validation and
MMS (as exists with modeling conservatism), several aspects of
validation should be considered before making a qualitative
determination of the adequacy of the MMS.
There are four factors that the reviewer should consider in
evaluating the validation: (1) The similarity of critical experiments
to actual applications; (2) sufficiency of the data (including the
number and quality of experiments); (3) adequacy of the validation
methodology; and (4) conservatism in the calculation of the bias and
its uncertainty. These factors are discussed in more detail below.
Similarity of Critical Experiments
Because the bias and its uncertainty must be estimated based on
critical experiments having geometric form, material composition, and
neutronic behavior similar to specific applications, the degree of
similarity between the critical experiments and applications is a key
consideration in determining the appropriateness of the MMS. The more
closely critical experiments represent the characteristics of
applications being validated, the more confidence the reviewer has in
the estimate of the bias and the bias uncertainty for those
applications.
The reviewer must understand both the critical experiments and
applications in sufficient detail to ascertain the degree of similarity
between them. Validation reports generally contain a description of
critical experiments (including source references). The reviewer may
need to consult these references to understand the physical
characteristics of the experiments. In addition, the reviewer may need
to consult process descriptions, nuclear criticality safety
evaluations, drawings, tables, input files, or other information to
understand the physical characteristics of applications. The reviewer
must consider the full spectrum of normal and abnormal conditions that
may have to be modeled when evaluating the similarity of the critical
experiments to applications.
In evaluating the similarity of experiments to applications, the
reviewer must recognize that some parameters are more significant than
others to accurately calculate keff. The parameters that
have the greatest effect on the calculated keff of the
system are those that are most important to match when choosing
critical experiments. Because of this, there is a close relationship
between similarity of critical experiments to applications and system
sensitivity. Historically, certain parameters have been used to trend
the bias because these are the parameters that have been found to have
the greatest effect on the bias. These parameters include the
moderator-to-fuel ratio (e.g., H/U, H/X, v\m\/v\f\), isotopic abundance
(e.g., uranium-235 (\235\U), plutonium-239 (\239\Pu), or overall Pu-to-
uranium ratio), and parameters that characterize the neutron energy
spectrum (e.g., energy of average lethargy causing fission (EALF),
average energy group (AEG)). Other parameters, such as material density
or overall geometric shape, are generally considered to be of less
importance. The reviewer should consider all important system
characteristics that can reasonably be expected to affect the bias. For
example, the critical experiments should include any materials that can
have an appreciable effect on the calculated keff, so that
the effect due to the cross sections of those materials is included in
the bias. Furthermore, these materials should have at least the same
reactivity worth in the experiments (which may be evidenced by having
similar number densities) as in the applications. Otherwise, the effect
of any bias from the underlying cross sections or the assumed material
composition may be masked in the applications. The materials must be
present in a statistically significant number of experiments having
similar neutron spectra to the application. Conversely, materials that
do not have an appreciable effect on the bias may be neglected and
would not have to be represented in the critical experiments.
Merely having critical experiments that are representative of
applications is the minimum acceptance criterion, and does not alone
justify having any particular value of the MMS. There are some
situations, however, in which there is an unusually high degree of
similarity between the critical experiments and applications, and in
these cases, this fact may be credited as justification for having a
smaller MMS than would otherwise be acceptable. If the critical
experiments have geometric forms, material compositions, and neutron
spectra that are nearly indistinguishable from those of the
applications, this may be justification for a smaller MMS than would
otherwise be acceptable. For example, justification for having a small
MMS for finished fuel assemblies could include selecting critical
experiments consisting of fuel assemblies in water, where the fuel has
nearly the same pellet diameter, pellet density, cladding materials,
pitch, absorber content, enrichment, and neutron energy spectrum as the
licensee's fuel. In this case, the validation should be very specific
to this type of system, because including other types of critical
experiments could mask variations in the bias. Therefore, this type of
justification is generally easiest when the area of applicability (AOA)
is very narrowly defined. The reviewer should pay particular attention
to abnormal conditions. In this example, changes in process conditions
such as damage to the fuel or partial flooding may significantly affect
the applicability of the critical experiments.
There are several tools available to the reviewer to ascertain the
degree of similarity between critical experiments and applications.
Some of these are listed below:
1. NUREG/CR-6698, ``Guide to Validation of Nuclear Criticality
Safety Calculational Method,'' Table 2.3, contains a set of screening
criteria for determining the applicability of critical experiments. As
is stated in the NUREG, these criteria were arrived at by
[[Page 14022]]
consensus among experienced nuclear criticality safety specialists and
may be considered to be conservative. The reviewer should consider
agreement on all screening criteria to be justification for
demonstrating a very high degree of critical experiment similarity.
(Agreement on the most significant screening criteria for a particular
system should be considered as demonstration of an acceptable degree of
critical experiment similarity.) Less conservative (i.e., broader)
screening criteria may also be acceptable, if appropriately justified.
2. Analytical methods that systematically quantify the degree of
similarity between a set of critical experiments and applications in
pair-wise fashion may be used. One example of this is the TSUNAMI code
in the SCALE 5 code package. One strength of TSUNAMI is that it
calculates an overall correlation that is a quantitative measure of the
degree of similarity between an experiment and an application. Another
strength is that this code considers all the nuclear phenomena and
underlying cross sections and weights them by their importance to the
calculated keff (i.e., sensitivity of keff to the
data). The NRC staff currently considers a correlation coefficient of
ck >= 0.95 to be indicative of a very high degree of
similarity. This is based on the staff's experience comparing the
results from TSUNAMI to those from a more traditional screening
criterion approach. The NRC staff also considers a correlation
coefficient between 0.90 and 0.95 to be indicative of a high degree of
similarity. However, owing to the amount of experience with TSUNAMI, in
this range use of the code should be supplemented with other methods of
evaluating critical experiment similarity. Conversely, a correlation
coefficient less than 0.90 should not be used as a demonstration of a
high or very high degree of critical experiment similarity. Because of
limited use of the code to date, all of these observations should be
considered tentative and thus the reviewer should not use TSUNAMI as a
``black box,'' or base conclusions of adequacy solely on its use.
However, it may be used to test a licensee's statement that there is a
high degree of similarity between experiments and applications.
3. Traditional parametric sensitivity studies may be employed to
demonstrate that keff is highly sensitive or insensitive to
a particular parameter. For example, if a 50% reduction in the \10\B
cross section is needed to produce a 1% change in the system
keff, then it can be concluded that the system is highly
insensitive to the boron content, in the amount present. This is
because a credible error in the \10\B cross section of a few percent
will have a statistically insignificant effect on the bias. Therefore,
in the amount present, the boron content is not a parameter that is
important to match in order to conclude that there is a high degree of
similarity between critical experiments and applications.
4. Physical arguments may demonstrate that keff is
highly sensitive or insensitive to a particular parameter. For example,
the fact that oxygen and fluorine are almost transparent to thermal
neutrons (i.e., cross sections are very low) may justify why
experiments consisting of UO2F2 may be considered
similar to UO2 or UF4 applications, provided that
both experiments and applications occur in the thermal energy range.
The reviewer should ensure that all parameters which can measurably
affect the bias are considered when assessing critical experiment
similarity. For example, comparison should not be based solely on
agreement in the \235\U fission spectrum for systems in which the
system keff is highly sensitive to \238\U fission, \10\B
absorption, or \1\H scattering. A method such as TSUNAMI that considers
the complete set of reactions and nuclides present can be used to rank
the various system sensitivities, and to thus determine whether it is
reasonable to rely on the fission spectrum alone in assessing the
similarity of critical experiments to applications.
Some questions that the reviewer may ask in evaluating reliance on
critical experiment similarity as justification for the MMS include:
Do the critical experiments adequately span the range of
geometric forms, material compositions, and neutron energy spectra
expected in applications?
Are the materials present with at least the same
reactivity worth as in applications?
Do the licensee's criteria for determining whether
experiments are sufficiently similar to applications consider all
nuclear reactions and nuclides that can have a statistically
significant effect on the bias?
Sufficiency of the Data
Another aspect of evaluating the selected critical experiments for
a specific MMS is evaluating whether there is a sufficient number of
benchmark-quality experiments to determine the bias across the entire
AOA. Having a sufficient number of benchmark-quality experiments means
that: (1) There are enough (applicable) critical experiments to make a
statistically meaningful calculation of the bias and its uncertainty;
(2) the experiments somewhat evenly span the entire range of all the
important parameters, without gaps requiring extrapolation or wide
interpolation; and (3) the experiments are, preferably, benchmark-
quality experiments. The number of critical experiments needed is
dependent on the statistical method used to analyze the data. For
example, some methods require a minimum number of data points to
reliably determine whether the data are normally distributed. Merely
having a large number of experiments is not sufficient to provide
confidence in the validation result, if the experiments are not
applicable to the application. The reviewer should particularly examine
whether consideration of only the most applicable experiments would
result in a larger negative bias (and thus a lower USL) than that
determined based on the full set of experiments. The experiments should
also ideally be sufficiently well-characterized (including experimental
parameters and their uncertainties) to be considered benchmark
experiments. They should be drawn from established sources (such as
from the International Handbook of Evaluated Criticality Safety
Benchmark Experiments (IHECSBE), laboratory reports, or peer-reviewed
journals). For some applications, benchmark-quality experiments may not
be available; when necessary, critical experiments that do not rise to
the level of benchmark-quality experiments may be used. However, the
reviewer should take this into consideration and should evaluate the
need for additional margin.
Some questions that the reviewer may ask in evaluating the number
and quality of critical experiments as justification for the MMS
include:
Are the critical experiments chosen all high-quality
benchmarks from reliable (e.g., peer-reviewed and widely-accepted)
sources?
Are the critical experiments chosen taken from multiple
independent sources, to minimize the possibility of systematic errors?
Have the experimental uncertainties associated with the
critical experiments been provided and used in calculating the bias and
bias uncertainty?
Is the number and distribution of critical experiments
sufficient to establish trends in the bias across the entire range of
parameters?
Is the number of critical experiments commensurate with
the statistical methodology being used?
[[Page 14023]]
Validation Methodological Rigor
Having a sufficiently rigorous validation methodology means having
a methodology that is appropriate for the number and distribution of
critical experiments, that calculates the bias and its uncertainty
using an established statistical methodology, that accounts for any
trends in the bias, and that accounts for all apparent sources of
uncertainty in the bias (e.g., the increase in uncertainty due to
extrapolating the bias beyond the range covered by the experimental
data). Examples of deficiencies in the validation methodology may
include: (1) Using a statistical methodology relying on the data being
normally distributed about the mean keff to analyze data
that are not normally distributed; (2) using a linear regression fit on
data that has a non-linear dependence on a trending parameter; (3) use
of a single pooled bias when very different types of critical
experiments are being evaluated in the same validation. These
deficiencies serve to decrease confidence in the validation results and
may warrant additional margin (i.e., a larger MMS). Additional guidance
on some of the more commonly observed deficiencies is provided below.
The assumption that data is normally distributed is generally
valid, unless there is a strong trend in the data or different types of
critical experiments with different mean calculated keff
values are being combined. Tests for normality require a minimum number
of critical experiments to attain a specified confidence level
(generally 95%). If there is insufficient data to verify that the data
are normally distributed, or the data are shown to be not normally
distributed, a non-parametric technique should be used to analyze the
data.
The critical experiments chosen should ideally provide a continuum
of data across the entire validated range, so that any variation in the
bias as a function of important system parameters may be observed. The
presence of discrete clusters of experiments having a calculated
keff lower than the set of critical experiments as a whole
should be examined closely to determine if there is some systematic
effect common to a particular type of calculation that makes use of the
overall bias non-conservative. Because the bias can vary with system
parameters, if the licensee has combined different subsets of data
(e.g., solutions and powders, low- and high-enriched, homogeneous and
heterogeneous), the bias for the different subsets should be analyzed.
In addition, the goodness-of-fit for any function used to trend the
bias should be examined to ensure it is appropriate to the data being
analyzed.
If critical experiments do not cover the entire range of parameters
needed to cover anticipated applications, it may be necessary to extend
the AOA by making use of trends in the bias. Any extrapolation (or wide
interpolation) of the data should be done by means of an established
mathematical methodology that takes into account the functional form of
both the bias and its uncertainty. The extrapolation should not be
based on judgement alone, such as by observing that the bias is
increasing in the extrapolated range, because this may not account for
the increase in the bias uncertainty that will occur with increasing
extrapolation. The reviewer should independently confirm that the
derived bias is valid in the extrapolated range and should ensure that
the extrapolation is not large. NUREG/CR-6698 states that critical
experiments should be added if the data must be extrapolated more than
10%. There is no corresponding guidance given for interpolation;
however, if the gap represents a significant fraction of the total
range of the data, then the reviewer should question whether
interpolation is reasonable. The reviewer should consider, for
instance, how rapidly the underlying physics or neutronic behavior is
changing in the vicinity of the gap (e.g., if interpolation in H/X is
required, is the system fully thermalized, or is the spectrum changing
from a fast-to-thermal spectrum over the gap?) In general, if the
extrapolation or interpolation is too large, new factors that could
affect the bias may be introduced as the physical phenomena in the
system change. The reviewer should not view validation as a purely
mathematical exercise, but should bear in mind the neutron physics and
underlying physical phenomena when interpreting the results.
Discarding an unusually large number of critical experiments as
outliers (i.e., more than 1-2%) should also be viewed with some
concern. Apparent outliers should not be discarded based purely upon
judgement or statistical grounds (such as causing the data to fail
tests for normality), because they could be providing valuable
information on the method's validity for a particular application. The
reviewer should verify that there are specific defensible reasons, such
as reported inconsistencies in the experimental data, for discarding
any outliers. If any of the critical experiments from a particular data
set are discarded, the reviewer should examine other experiments
included to determine whether they may be subject to the same
systematic errors. Outliers should be examined carefully especially
when they have a lower calculated keff than the other
experiments included.
NUREG-1520 states that the MoS should be large compared to the
uncertainty in the bias. The observed spread of the data about the mean
keff should be examined as an indicator of the overall
precision of the calculational method. The reviewer should ascertain
whether the statistical method of validation considers both the
observed spread in the data and the experimental and calculational
uncertainty in determining the USL. The reviewer should also evaluate
whether the observed spread in the data is consistent with the reported
uncertainty (e.g., whether X\2\/N [ap] 1). If the spread in the data is
larger than, or comparable to, the MMS, then the reviewer should
consider whether additional margin (i.e., a larger MMS) is needed.
As a final test of the code's accuracy, the bias should be
relatively small (i.e., bias [lap] 2 percent), or else the reason for
the bias should be determined. No credit should be taken for positive
bias, because this would result in making changes in a non-conservative
direction without having a clear understanding of those changes. If the
absolute value of the bias is very large--and especially if the reason
for the large bias cannot be determined--this may indicate that the
calculational method is not very accurate, and a larger MMS may be
appropriate.
Some questions that the reviewer may ask in evaluating the rigor of
the validation methodology as justification for the MMS include:
Are the results from use of the methodology consistent
with the data (e.g., normally distributed)?
Is the normality of the data confirmed prior to performing
statistical calculations? If the data does not pass the tests for
normality, is a non-parametric method used?
Does the assumed functional form of the bias represent a
good fit to the critical experiments? Is a goodness-of-fit test
performed?
Does the method determine a pooled bias across disparate
types of critical experiments, or does it consider variations in the
bias for different types of experiments? Are there discrete clusters of
experiments for which the bias appears to be non-conservative?
Has additional margin been applied to account for
extrapolation or wide interpolation? Is this done based on an
established mathematical methodology?
[[Page 14024]]
Have critical experiments been discarded as apparent
outliers? Is there a valid reason for doing so?
Performing an adequate code validation is not by itself sufficient
justification for any specific MMS. The reason for this is that the
validation analysis determines the bias and its uncertainty, but not
the MMS. The MMS is added after the validation has been performed to
provide added assurance of subcriticality. However, having a validation
methodology that either exceeds or falls short of accepted practices
for validation may be a basis for either reducing or increasing the
MMS.
Statistical Conservatism
In addition to having conservatism in keff due to
modeling practices, licensees may also provide conservatism in the
statistical methods used to calculate the USL. For example, NUREG/CR-
6698 states that an acceptable method for calculating the bias is to
use the single-sided tolerance limit approach with a 95/95 confidence
(i.e., 95% confidence that 95% of all future critical calculations will
lie above the USL). If the licensee decides to use the single-sided
tolerance limit approach with a 95/99.9 confidence, this would result
in a more conservative USL than with a 95/95 confidence. This would be
true of other methods for which the licensee's confidence criteria
exceed the minimum accepted criteria. Generally, the NRC has accepted
95% confidence levels for validation results, so using more stringent
confidence levels may provide conservatism. In addition, there may be
other reasons a larger bias and/or bias uncertainty than necessary has
been used (e.g., because of the inclusion of inapplicable critical
experiments that have a lower calculated keff).
The reviewer may credit this conservatism towards having an
adequate MoS if: (1) The licensee demonstrates that this translates
into a specific [delta]keff; and (2) the licensee
demonstrates that the margin will be dependably present, based on
license or other commitments.
(3) Additional Risk-Informed Considerations
Besides modeling conservatism and the validation results, other
factors may provide added assurance of subcriticality. These factors
should be considered in evaluating whether there is adequate MoS and
are discussed below.
System Sensitivity and Uncertainty
The sensitivity of keff to changes in system parameters
can be used to assess the potential effect of errors on the calculation
of keff. If the calculated keff is especially
sensitive to a given parameter, an error in that parameter could have a
correspondingly large contribution to the bias. Conversely, if
keff is very insensitive to a given parameter, then an error
may have a negligible effect on the bias. This is of particular
importance when assessing whether the chosen critical experiments are
sufficiently similar to applications to justify a small MMS.
The reviewer should not consider the sensitivity in isolation, but
should also consider the magnitude of uncertainties in the parameters.
If keff is very sensitive to a given parameter, but the
value of that parameter is known with very high accuracy (and its
variations are well-controlled), the potential contribution to the bias
may still be very small. Thus, the contribution to the bias is a
function of the product of the the keff sensitivity with the
uncertainty. To illustrate this, suppose that keff is a
function of a large number of variables, x1, x2,*
* *, xN. Then the uncertainty in keff may be
expressed as follows, if all the individual terms are independent:
[GRAPHIC] [TIFF OMITTED] TN20MR06.019
Where the partial derivatives [part]k/[part]xi are
proportional to the sensitivity and the terms x1 represent
the uncertainties, or likely variations, in the parameters. (If not all
variables are dependent, then there may be additional terms.) Each term
in this equation then represents the contribution to the overall
uncertainty in keff.
There are several tools available to the reviewer to ascertain the
sensitivity of keff to changes in the underlying parameters.
Some of these are listed below:
1. Analytical tools that calculate the sensitivity for each
nuclide-reaction pair present in the problem may be used. One example
of this is the TSUNAMI code in the SCALE 5 code package. TSUNAMI
calculates both an integral sensitivity coefficient (i.e., summed over
all energy groups) and a sensitivity profile as a function of energy
group. The reviewer should recognize that TSUNAMI only calculates the
keff sensitivity to changes in the underlying nuclear data,
and not to other parameters that could affect the bias and should be
considered. (See section on Critical Experiment Similarity for caveats
about using TSUNAMI.)
2. Direct sensitivity calculations may be used, in which system
parameters are perturbed and the resulting impact on keff
determined. Perturbation of atomic number densities can also be used to
confirm the sensitivity calculated by other methods (e.g., TSUNAMI).
Such techniques are not limited to considering the effect of the
nuclear data.
There are also several sources available to the reviewer to
ascertain the uncertainty associated with the underlying parameters.
For process parameters, these sources of uncertainty may include
manufacturing tolerances, quality assurance records, and experimental
and/or measurement results. For nuclear data parameters, these sources
of uncertainty may include published data, uncertainty data distributed
with the cross section libraries, or the covariance data used in
methods such as TSUNAMI.
Some systems are inherently more sensitive to changes in the
underlying parameters than others. For example, high-enriched uranium
systems typically exhibit a greater sensitivity to changes in system
parameters (e.g., mass, moderation) than low-enriched systems. This has
been the reason that HEU (i.e., >20wt% \235\ U) facilities have been
licensed with larger MMS values than LEU (<=10wt% \235\ U) facilities.
This greater sensitivity would also be true of weapons-grade Pu
compared to low-assay mixed oxides (i.e., with a few percent Pu/U).
However, it is also true that the uncertainties associated with
measurement of the \235\ U cross sections are much smaller than those
associated with measurement of the \238\ U cross sections. Both the
greater sensitivity and smaller uncertainty would need to be considered
in evaluating whether a larger MMS is needed for high-enriched systems.
Frequently, operating limits that are more conservative than safety
limits determined using keff calculations are established to
prevent those safety limits from being exceeded. For systems in which
keff is very sensitive to the system parameters, more margin
between the operating and safety limits may be needed. Systems in which
keff is very sensitive to the process parameters may need
both a larger margin between operating and safety limits and a larger
MMS. This is because the system is sensitive to any change, whether it
be caused by normal process variations or caused by unknown errors.
Because of this, the assumption is often made that the MMS is meant to
account for variations in the process or the ability to control the
process parameters. However, the MMS is meant only to
[[Page 14025]]
allow for unknown (or difficult to quantify) uncertainties in the
calculation of keff. The reviewer should recognize that
determination of an appropriate MMS is not dependent on the ability to
control process parameters within safety limits (although both may
depend on the system sensitivity).
Some questions that the reviewer may ask in evaluating the system
sensitivity as justification for the MMS include:
How sensitive is keff to changes in the
underlying nuclear data (e.g., cross sections)?
How sensitive is keff to changes in the
geometric form and material composition?
Are the uncertainties associated with these underlying
parameters well-known?
How does the MMS compare to the expected magnitude of
changes in keff resulting from uncertainties in these
underlying parameters?
Knowledge of the Neutron Physics
Another important consideration that may affect the appropriate MMS
is the extent to which the physical behavior of the system is known.
Fissile systems which are known to be subcritical with a high degree of
confidence do not require as much MMS as systems where subcriticality
is less certain. An example of a system known to be subcritical with
high confidence is a light-water reactor fuel assembly. The design of
these systems is such that they can only be made critical when highly
thermalized. Due to extensive analysis and reactor experience, the
flooded isolated assembly is known to be subcritical. In addition, the
thermal neutron cross sections for materials in finished reactor fuel
have been measured with a very high degree of accuracy (as opposed to
cross sections in the resonance region). Other examples of systems in
which there is independent corroborating evidence of subcriticality may
include systems consisting of very simple geometric shapes, or other
idealized situations, in which there is strong evidence that the system
is subcritical based on comparison with highly similar systems in
published sources (e.g., standards and handbooks). In these cases, the
MMS may be significantly reduced due to the fact that the calculation
of keff is not relied on alone to provide assurance of
subcriticality.
Reliance on independent knowledge that a given system is
subcritical necessarily requires that the configuration of the system
be fixed. If the configuration can change from the reference case,
there will be less knowledge about the behavior of the changed system.
For example, a finished fuel assembly is subject to strict quality
assurance checks and would not reach final processing if it were
outside specifications. In addition, it has a form that has both been
extensively studied and is highly stable. For these reasons, there is a
great deal of certainty that this system is well-characterized and is
not subject to change. A typical solution or powder system (other than
one with a simple geometric arrangement) would not have been studied
with the same level of rigor as a finished fuel assembly. Even if they
were studied with the same level of rigor, these systems have forms
that are subject to change into forms whose neutron physics has not
been as extensively studied.
Some questions that the reviewer may ask in evaluating the
knowledge of the neutron physics as justification for the MMS include:
Is the geometric form and material composition of the
system fixed and very unlikely to change?
Is the geometric form and material composition of the
system subject to strict quality assurance, such that tolerances have
been bounded?
Has the system been extensively studied in the nuclear
industry and shown to be subcritical (e.g., in reactor fuel studies)?
Are there other reasons besides criticality calculations
to conclude that the system will be subcritical (e.g., handbooks,
standards, published data)?
How well-known is the nuclear data (e.g., cross sections)
in the energy range of interest?
Likelihood of the Abnormal Condition
Some facilities have been licensed with different sets of
keff limits for normal and abnormal conditions. Separate
keff limits for normal and abnormal conditions are
permissible, but are not required. There is some low likelihood that
processes calculated to be subcritical will, in fact, be critical, and
this likelihood increases as the MMS is reduced (though it cannot in
general be quantified). NUREG-1718, ``Standard Review Plan for the
Review of an Application for a Mixed Oxide (MOX) Fuel Fabrication
Facility,'' states that abnormal conditions should be at least unlikely
from the standpoint of the double contingency principle. Then, a
somewhat higher likelihood that a system calculated to be subcritical
is, in fact, critical is more permissible for abnormal conditions than
for normal conditions, because of the low likelihood of the abnormal
condition being realized. The reviewer should verify that the licensee
has defined abnormal conditions such that achieving the abnormal
condition requires at least one contingency to have occurred, that the
system will be closely monitored so that it is promptly detected, and
that it will be promptly corrected upon detection. Also, there is
generally more conservatism present in the abnormal case, because the
parameters that are assumed to have failed are analyzed at their worst-
case credible condition.
The increased risk associated with having a smaller MMS for
abnormal conditions should be commensurate with, and offset by, the low
likelihood of achieving the abnormal condition. That is, if the normal
case keff limit is judged to be acceptable, then the
abnormal case limit will also be acceptable, provided the increased
likelihood (that a system calculated to be subcritical will be
critical) is offset by the reduced likelihood of realizing the abnormal
condition because of the controls that have been established. Note that
if two or more contingencies must occur to reach a given condition,
there is no requirement to ensure that the resulting condition is
subcritical. If a single keff limit is used (i.e., no credit
for unlikelihood of the abnormal condition), then the limit must be
found acceptable to cover both normal and credible abnormal conditions.
The reviewer should always make this finding considering specific
conditions and controls in the process(es) being evaluated.
(4) Statistical Justification for the MMS
The NRC does not consider statistical justification an appropriate
basis for a specific MMS. Previously, some licensees have attempted to
justify specific MMS values based on a comparison of two statistical
methods. For example, the USLSTATS code issued with the SCALE code
package contains two methods for calculating the USL: (1) The
Confidence Band with Administrative Margin approach (calculating USL-
1), and (2) the Lower Tolerance Band approach (calculating USL-2). The
value of the MMS is an input parameter to the Confidence Band approach,
but is not included explicitly in the Lower Tolerance Band approach. In
this particular justification, adequacy of the MMS is based on a
comparison of USL-1 and USL-2 (i.e., the condition that USL-1,
including the chosen MMS, is less than USL-2). However, the reviewer
should not accept this justification.
The condition that USL-1 (with the chosen MMS) is less than USL-2
is necessary, but is not sufficient, to show that an adequate MMS has
been used. These methods are both statistical
[[Page 14026]]
methods, and a comparison can only demonstrate whether the MMS is
sufficient to bound any statistical uncertainties included in the Lower
Tolerance Band approach but not included in the Confidence Band
approach. There may be other statistical or systematic errors in
calculating keff that are not included in either statistical
treatment. Because of this, an MMS value should be specified regardless
of the statistical method used. Therefore, the reviewer should not
consider such a statistical approach an acceptable justification for
any specific value of the MMS.
(5) Summary
Based on a review of the licensee's justification for its chosen
MMS, taking into consideration the aforementioned factors, the staff
should make a determination as to whether the chosen MMS provides
reasonable assurance of subcriticality under normal and credible
abnormal conditions. The staff's review should be risk-informed, in
that the review should be commensurate with the MoS and should consider
the specific facility and process characteristics, as well as the
specific modeling practices used. As an example, approving an MMS value
greater than 0.05 for processes typically encountered in enrichment and
fuel fabrication facilities should require only a cursory review,
provided that an acceptable validation has been performed and modeling
practices at least as conservative as those in NUREG-1520 have been
utilized. The approval of a smaller MMS will require a somewhat more
detailed review, commensurate with the MMS that is requested. However,
the MMS should not be reduced below 0.02 due to inherent uncertainties
in the cross section data and the magnitude of code errors that have
been discovered. Quantitative arguments (such as modeling conservatism)
should be used to the extent practical. However, in many instances, the
reviewer will need to make a judgement based at least partly on
qualitative arguments. The staff should document the basis for finding
the chosen MMS value to be acceptable or unacceptable in the Safety
Evaluation Report (SER), and should ensure that any factors upon which
this determination rests are ensured to be present over the facility
lifetime (e.g., through license commitment or condition).
Regulatory Basis
In addition to complying with paragraphs (b) and (c) of this
section, the risk of nuclear criticality accidents must be limited by
assuring that under normal and credible abnormal conditions, all
nuclear processes are subcritical, including use of an approved margin
of subcriticality for safety. [10 CFR 70.61(d)]
Technical Review Guidance
Determination of an adequate MMS is strongly dependent upon
specific processes, conditions, and calculational practices at the
facility being licensed. Judgement and experience must be employed in
evaluating the adequacy of the proposed MMS. In the past, an MMS of
0.05 has generally been found acceptable for most typical low-enriched
fuel cycle facilities without a detailed technical justification. A
smaller MMS may be acceptable but will require some level of technical
review. (No specific guidance on the appropriate MMS for other types of
facilities, such as high-enriched or plutonium fuel cycle facilities,
is provided. Rather, the MMS for these facilities should be evaluated
on a case-by-case basis using the criteria in this ISG; an example of
the consideration of sensitivity and uncertainty for high-enriched
uranium is given in the section on ``System Sensitivity and
Uncertainty.'') Also, for reasons stated previously, the MMS should not
be reduced below 0.02.
An MMS of 0.05 should be found acceptable for low-enriched fuel
cycle processes and facilities if:
1. A validation has been performed that meets accepted industry
guidelines (e.g., meets the requirements of ANSI/ANS-8.1-1998, NUREG/
CR-6361, and/or NUREG/CR-6698).
2. There is an acceptable number of critical experiments with
similar geometric forms, material compositions, and neutron energy
spectra to applications. These experiments cover the range of
parameters of applications, or else margin is provided to account for
extensions to the AOA.
3. The processes to be evaluated include materials and process
conditions similar to those that occur in low-enriched fuel cycle
applications (i.e., no new fissile materials, unusual moderators or
absorbers, or technologies new to the industry that can affect the
types of systems to be modeled).
The reviewer should consider any factors, including those
enumerated in the discussion above, that could result in applying
additional margin (i.e., a larger MMS) or may justify reducing the MMS.
The reviewer must then exercise judgement in arriving at an MMS that
provides for adequate assurance of subcriticality.
Some of the factors that may serve to justify reducing the MMS
include:
1. There is a predictable and dependable amount of conservatism in
modeling practices, in terms of keff, that is assured to be
maintained (in both normal and abnormal conditions) over the facility
lifetime.
2. Critical experiments have nearly identical geometric forms,
material compositions, and neutron energy spectra to applications, and
the validation is specific to this type of application.
3. The validation methodology substantially exceeds accepted
industry guidelines (e.g., it uses a very conservative statistical
approach, considers an unusually large number of trending parameters,
or analyzes the bias for a large number of subgroups of critical
experiments).
4. The system keff is demonstrably much less sensitive
to uncertainties in cross sections or variations in other system
parameters than typical low-enriched fuel cycle processes.
5. There is reliable information besides results of calculations
that provides assurance that the evaluated applications will be
subcritical (e.g., experimental data, historical evidence, industry
standards or widely-accepted handbooks).
6. The MMS is only applied to abnormal conditions, which are at
least unlikely to be achieved, based on credited controls.
Some of the factors that may necessitate increasing (or not
approving) the MMS include:
1. The technical practices employed by the licensee are less
conservative than standard industry modeling practices (e.g., do not
adequately bound reflection or the full range of credible moderation,
do not take geometric tolerances into account).
2. There are few similar critical experiments of benchmark quality
that cover the range of parameters of applications.
3. The validation methodology substantially falls below accepted
industry guidelines (e.g., it uses less than a 95% confidence in the
statistical approach, fails to consider trends in the bias, fails to
account for extensions to the AOA).
4. The validation results otherwise tend to cast doubt on the
accuracy of the bias and its uncertainty (i.e., the critical
experiments are not normally distributed, there is a large number of
outliers discarded (>=2%), there are distinct subgroups of experiments
with lower keff than the experiments as a whole, trending
fits do not pass goodness-of-fit tests, etc.).
[[Page 14027]]
5. The system keff is demonstrably much more sensitive
to uncertainties in cross sections or other system parameters than
typical low-enriched fuel cycle processes.
6. There is reliable information that casts doubt on the results of
the calculational method or the subcriticality of evaluated
applications (e.g., experimental data, reported concerns with the
nuclear data).
The purpose of asking the questions in the individual discussion
sections is to ascertain the degree to which these factors either
provide justification for reducing the MMS or necessitate increasing
the MMS. These lists are not all-inclusive, and any other technical
information that demonstrates the degree of confidence in the
calculational method should be considered.
Recommendation
The guidance in this ISG should supplement the current guidance in
the nuclear criticality safety chapters of the fuel facility SRPs
(NUREG-1520 and -1718). However, NUREG-1718, Section 6.4.3.3.4, states
that the licensee should submit justification for the MMS, but then
states that an MMS of 0.05 is ``generally considered to be acceptable
without additional justification when both the bias and its uncertainty
are determined to be negligible.'' These two statements are
inconsistent. Therefore, NUREG-1718, Section 6.4.3.3.4, should be
revised to remove the following sentence:
``A minimum subcritical margin of 0.05 is generally considered to
be acceptable without additional justification when both the bias and
its uncertainty are determined to be negligible.''
References
ANSI/ANS-8.1-1998, ``Nuclear Criticality Safety in Operations with
Fissionable Materials Outside Reactors,'' American Nuclear Society.
ANSI/ANS-8.17-2004, ``Criticality Safety Criteria for the Handling,
Storage, and Transportation of LWR [Light Water Reactor] Fuel
Outside Reactors,'' American Nuclear Society.
``International Handbook of Evaluated Criticality Safety
Experiments,'' NEA/NSC/DOC (95) 03, Nuclear Energy Agency,
Organization for Economic Co-operation and Development, 2003.
IN 2005-13, ``Potential Non-Conservative Error in Modeling Geometric
Regions in the KENO-V.a Criticality Code,'' May 17, 2005.
U.S. Nuclear Regulatory Commission (U.S.) (NRC). NUREG-1520,
``Standard Review Plan for the Review of a License Application for a
Fuel Cycle Facility.'' NRC: Washington, DC March 2002.
U.S. Nuclear Regulatory Commission (U.S.) (NRC). NUREG-1718,
``Standard Review Plan for the Review of an Application for a Mixed
Oxide (MOX) Fuel Fabrication Facility.'' NRC: Washington, DC August
2000.
U.S. Nuclear Regulatory Commission (U.S.) (NRC). NUREG/CR-6698,
``Guide for Validation of Nuclear Criticality Safety Calculational
Methodology.'' NRC: Washington, DC January 2001.
U.S. Nuclear Regulatory Commission (U.S.) (NRC). NUREG/CR-6361,
``Criticality Benchmark Guide for Light-Water-Reactor Fuel in
Transportation and Storage Packages.'' NRC: Washington, DC March
1997.
Approved:-------------------------------------------------------------
Date:-----------------------------------------------------------------
Director, Division of Fuel Cycle Safety and Safeguards, NMSS.
Appendix A--ANSI/ANS-8.17 Calculation of Maximum keff
ANSI/ANS-8.17-2004, ``Criticality Safety Criteria for the
Handling, Storage, and Transportation of LWR Fuel Outside
Reactors,'' contains a detailed discussion of the various factors
that should be considered in setting keff limits. This is
consistent with, but more detailed than, the discussion in ANSI/ANS-
8.1-1998.
The subcriticality criterion from Section 5.1 of ANSI/ANS-8.17-
2004 is:
Ks <= kc - [Delta]ks - [Delta]kc - [Delta]km
Where ks is the calculated keff corresponding
to the application, [Delta]ks is its uncertainty,
kc is the mean keff resulting from the
calculation of critical experiments, [Delta]kc is its uncertainty,
and [Delta]kk is the MMS. The types of uncertainties
included in each of these ``delta'' terms is provided, and includes
the following:
[Delta]ks = (1) statistical uncertainties in computing
ks ; (2) convergence uncertainties in computing
ks, (3) material tolerances; (4) fabrication tolerances;
(5) uncertainties due to limitations in the geometric representation
used in the method; and (6) uncertainties due to limitations in the
material representations used in the method.
[Delta]kc = (7) uncertainties in the critical
experiments; (8) statistical uncertainties in computing
kc; (9) convergence uncertainties in computing
kc; (10) uncertainties due to extrapolating kc
outside the range of experimental data; (11) uncertainties due to
limitations in the geometric representations used in the method; and
(12) uncertainties due to limitations in the material
representations used in the method.
[Delta]km = an allowance for any additional uncertainties
(MMS).
To the extent that not all 12 sources of uncertainty listed
above have been explicitly taken into account, they may be allowed
for by increasing the value of [Delta]km. The more of
these sources of uncertainty that have been taken into account, the
smaller the necessary additional margin [Delta]km. As a
general principle, however, the MMS should be large compared to
known uncertainties in the nuclear data and limitations of the
methodology. However, a value of the MMS below 0.02 should not be
used.
Frequently, the terms in the above equation relating to the
application are grouped on the left-hand side of the equation, so
that the equation is rewritten as follows:
Ks + [Delta]ks <= kc - [Delta]kc - [Delta]km
Where the terms on the right-hand side of the equation are often
lumped together and termed the Upper Subcritical Limit (USL), so
that the USL = kc - [Delta]kc - [Delta]km.
Relation to the Minimum Subcritical Margin (MMS)
The MoS has been defined as the difference between the actual
value of keff and the value of keff at which
the system is expected to be critical. The expected (best estimate)
critical value of keff is the mean keff value
of all critical experiments analyzed (i.e., kc), including
consideration of the uncertainty in the bias (i.e., [Delta]kc). The
calculated value of keff for an application generally
exceeds the actual (physical) keff value due to
conservative assumptions in modeling the system. In terms of the
above USL equation, the MoS may be expressed mathematically as:
MoS = kc - [Delta]kc - (ks - [Delta]ksa) - [Delta]ks
Where the term in parentheses is equal to the actual (physical)
keff of the application, ksa. A term,
[Delta]ksa, has been added to represent the difference
between the actual and calculated value of keff for the
application (i.e., [Delta]ksa = change in keff
resulting from modeling conservatism). In terms of the USL:
MoS = USL + [Delta]km - ks + ksa - [Delta]ks
The minimum allowed value of the MoS is reached when the
calculated keff for the application, ks +
[Delta]ks, is equal to the USL. When this occurs, the
minimum value of the MoS is:
MoS >= [Delta]km + [Delta]ksa
Thus, adequate margin (MoS) may be assured either by
conservatism in modeling practices or in the explicit specification
of [Delta]km (MMS). This is discussed in the ISG section
on modeling conservatism.
Glossary
Application: calculation of a fissionable system in the facility
performed to demonstrate subcriticality under normal or credible
abnormal conditions.
Area of Applicability (AOA): the ranges of material compositions
and geometric arrangements within which the bias of a calculational
method is established.
Benchmark experiment: a critical experiment that has been peer-
reviewed and published and is sufficiently well-defined to be used
for validation of calculational methods.
Bias: a measure of the systematic differences between
calculational method results and experimental data.
Bias uncertainty: a measure of both the accuracy and precision
of the calculations and the uncertainty in the experimental data.
Calculational method: includes the hardware platform, operating
system, computer algorithms and methods, nuclear reaction data, and
methods used to construct computer models.
[[Page 14028]]
Critical experiment: a fissionable system that has been
experimentally determined to be critical (with keff [ap]
1).
Margin of safety: the difference between the actual value of a
parameter and the value of the parameter at which the system is
expected to be critical with critical defined as keff =
1--bias--bias uncertainty.
Margin of Subcriticality (MoS): the difference between the
actual value of keff and the value of keff at
which the system is expected to be critical with critical defined as
keff = 1--bias--bias uncertainty.
Minimum Margin of Subcriticality (MMS): a minimum allowed margin
of subcriticality, which is an allowance for any unknown
uncertainties in calculating keff.
Subcritical limit: the maximum allowed value of a controlled
parameter under normal case conditions.
Upper Subcritical Limit (USL): the maximum allowed value of
keff (including uncertainty in keff), under
both normal and credible abnormal conditions, including allowance
for the bias, the bias uncertainty, and a minimum margin of
subcriticality.
[FR Doc. 06-2611 Filed 3-17-06; 8:45 am]
BILLING CODE 7590-01-P