[DNFSB
LETTERHEAD]
April 24, 2006
The Honorable Samuel W. Bodman
Secretary of Energy
1000 Independence Avenue, SW
Washington, DC 20585-1000
Dear Secretary Bodman:
The Defense Nuclear Facilities
Safety Board (Board) received the Department of Energy’s (DOE) letter dated
March 30, 2006, enclosing the DOE repackaging prioritization methodology, which
is a deliverable in the Implementation Plan (IP) for the Board’s Recommendation
2005-1, Nuclear
Material Packaging. DOE committed to developing a repackaging
prioritization methodology in response to sub-recommendation (3) of Recommendation
2005-1: “Prioritize
implementation of the improved nuclear material packaging requirement
consistent with the hazards of the different material types and the risk posed
by the existing package configurations and condition.” Section 5.0 of the IP describes the resolution
of this sub-recommendation as follows: “Based
on the Department nuclear material risk profile, the Department will ensure
that the highest priority items, as determined by the complex-wide risk ranking
methodology will be qualified or repackaged first at all sites.”
The DOE letter describes DOE’s
commitments under the IP for each site to prioritize repackaging or
qualification of existing packaging based on risk, and for each site to development
a resource-loaded schedule and funding plan for implementing the DOE Nuclear
Materials Packaging Manual, without
emphasizing the subsequent commitment to issue a “DOE-wide schedule for 2005-1
implementation.” The Board believes that
such a DOE-wide schedule cannot meet the commitment made in Section 5.3 of the
IP unless a complex-wide risk ranking methodology aimed at estimating the
overall nuclear material risk profile is developed. As transmitted to the Board, the
repackaging prioritization methodology cannot be used to develop an accurate
nuclear material risk profile for the complex and thus does not meet the IP commitment.
This deficiency was previously noted by
the Board’s staff in its initial review of the draft repackaging prioritization
methodology.
The Board’s staff has reviewed
the repackaging prioritization methodology and found that none of its earlier
comments on the draft document, provided in enclosure 1 to this letter, are addressed
in the final version. In addition to
those comments, the Board’s staff provided a detailed discussion, presented in
enclosure 2 to this letter, which elaborated on the technical bases and
justification for the comments. Comments
1-6 address technical deficiencies in the repackaging
prioritization methodology that adversely affect DOE’s ability to fulfill the
IP commitments cited above. The Board’s
staff also provided suggestions for addressing specific deficiencies in the
risk ranking methodology that, if adopted, would improve DOE’s ability to identify
the highest-priority items at a complex-wide level.
The most serious deficiency in
the risk ranking methodology involves DOE’s adoption of two inconsistent models
for estimating package failure probabilities. Specifically, one of the allowed models does
not account for the vulnerability of the package to known failure mechanisms,
and would therefore be unlikely, from a nuclear safety point of view, to
identify the highest-risk packages accurately. The other model does attempt to account for
the vulnerability of the package, but is presented with no supporting data for
the values used in predictions for package failure probabilities. Use of two different models for determining
package failure probabilities could result in identical packages being ranked
in a different order at different sites. A single
model that consistently accounted for hazards posed by different material types
and risks posed by the existing package configurations and conditions would
allow for meaningful comparisons of the risks posed by packages across the
complex.
The Board is aware that some
information needed for accurate estimation of the integrity of existing
packaging may not currently exist. However, considerable information does exist and
should be incorporated into the package failure probability model at this time.
For cases in which this information is
lacking, values for such parameters can be judiciously selected. The Board believes that it would be
appropriate to formulate the package failure probability model as a best
estimate based on available information, and to refine the model as additional
information is obtained through surveillance or repackaging efforts.
As noted in comment 8 in
enclosure 1, some of the same issues discussed in this letter were
independently identified by DOE’s technical review board for Recommendation
2005-1 before
the repackaging prioritization methodology was finalized. It appears that DOE’s internal process for
comment resolution failed to resolve the DOE technical review board’s key substantive
comments.
Therefore, pursuant to 42
U.S.C. § 2286b(d), the Board requests that, within 30 days of receipt
of this letter, DOE provide a response to the enclosed comments. This response should specifically address
comments 1-6. Comment 7 need not
be addressed at this time, as it also applies to draft DOE Manual M 441.1, Nuclear Materials Packaging Manual,
and can better be addressed
after the Board has had the opportunity to review that document. DOE should also outline in the response the
steps to be taken to ensure more effective and consistent resolution of the
technical review board’s comments.
Sincerely,
A.
J. Eggenberger
Chairman
c: The Honorable C. Russell H. Shearer
Mr. Richard
M. Stark
Mr. Mark
B. Whitaker, Jr.
Enclosures
Enclosure
1
Comments
on DOE’S Draft Repackaging Prioritization Methodology for Recommendation 2005-1,
Nuclear
Material Packaging
1. The vulnerability of package components should be reincorporated into the option 1 model. The vulnerability parameter from the original LA-UR-05-3864 model has been removed from the model adopted as option 1. By not attempting to account for the vulnerability of the package, option 1 currently assigns the same failure probability index to a specific material form, regardless of the type, number, or robustness of the containers. A model that does not attempt to account for the vulnerability of the package to known failure mechanisms is not likely to estimate failure probabilities accurately. Assuming container vulnerability is fully reincorporated in a future revision of the model, the following sub-comments apply:
(a) The vulnerability indices
for unknown inner containers may not be appropriate for the characteristics of
the population. The assignment of maximum
vulnerability for unknown containers results in assigning packages with inner
containers known to be highly vulnerable a lower failure probability than
packages consisting of unknown inner containers within the same outer container.
This may not be appropriate if a large number
of packages having initially unknown inner containers are eventually shown, on average,
to contain much more robust containers than the assumed worst-case scenario.
(b) A listing of standardized
container vulnerability indices for package configurations that are present in
the complex is not provided. There can be no expectation of consistent
choices for container vulnerability indices across sites without an agreed-upon
list. The packaging information
collected under the first Implementation Plan commitment could be used as the
basis for providing expert judgments to form this list.
(c) The use of zero values for
minimum vulnerability indices creates inconsistencies in the predicted results. The
assignment of zero value vulnerability indices to a barrier having otherwise
maximum indices results mathematically in a degenerate total container
vulnerability vector for packages having unknown barriers. For example, three nested slip-lid cans would
have the same vulnerability index as a single such can with unknown inner
containers.
(d) The fifth reactivity parameter for radiation-induced challenges to the package is not utilized. Recent experience with package failure has reinforced the importance of this challenge to the packaging. Assignment of values reflecting the true radiolytic potential of the material, rather than a placeholder value of 1, might better account for potential radiation damage to polymer-based packaging.
2. The reactivity indices
provided for option 1 in LA-UR-05-3864 do not reflect known differences in
reactivity among elements. For example, a highly reactive
material, such as plutonium metal, is currently assigned the same reactivity
indices as a considerably less reactive material of the same form, such as uranium
metal. There may be a need for additional
expert judgments regarding other material forms to account for differences in reactivity
among elements.
3. The assumed linear effect
of package age on failure probability may warrant further refinement. There
may be other time-to-failure relationships that agree better with recent survey
and package failure data. For example, a
survey of the literature suggests nonrandom failures of components that wear or
degrade over time may exhibit a more than linearly increasing failure rate over
time. Appropriate consideration of age
strongly impacts the accuracy of package failure predictions.
4. The value of allowing for
the use of alternative package failure probability models is unclear. Having
two options for determining relative package failure probability could result
in identical packages being ranked in a different order at different sites. In principle, a single methodology is
preferable because it facilitates meaningful comparisons of the risk posed by
packages across the complex. Having a
single methodology would provide an important tool for the Department of Energy
(DOE) to ensure that the highest-priority items are qualified or repackaged
first at all sites, as stated in Section 5.3 of the Implementation Plan.
5. No evidence is presented to
support the option 2 model. Without data to support the judgments
made on individual values used for the parameters or the model itself, there is
no way to assess the validity of the option 2 model for predicting package
failure probabilities.
6. Some parameters and
numerical indices used in option 2 appear inconsistent. While
the values chosen appear to be generally reasonable and attempt to account for
the robustness of the package, the values assigned for unknown conditions do
not appear to be consistent with respect to the parameters for known conditions.
7. The threshold dose
consequence in the repackaging document appears to be inconsistent with the
threshold being proposed in the draft packaging manual. The
chart in Appendix C of the draft repackaging document illustrates the threshold
for repackaging as a potential dose consequence of 5 rem committed effective
dose equivalent or greater, using the methodology of Los Alamos National
Laboratory for calculating the dose to workers. This approach yields considerably different
results from the threshold the staff understands to be proposed in the draft
manual, which is based on the methodology in 49 Code of Federal Regulations
(CFR) 173 and does not use airborne respirable material calculations. This inconsistency results in excluding
packages with sufficient quantities of material to be within the scope of the
manual from the repackaging prioritization process.
8. DOE’s
review process for Recommendation 2005-1 deliverables needs improvement. Many
of the problems identified by the staff ought to have been identified by the
technical review board and resolved before the draft document was transmitted
to the Board. A subsequent staff review of the
comments of the technical review board revealed that in fact the technical
review board had identified some of these problems. Although most of the technical review board’s
comments of an editorial nature were addressed, the more significant comments
were not resolved. The comment resolution
process needs to be improved and better integrated for future deliverables.
Enclosure
2
Discussion
of Comments of the Board’s Staff on DOE’s Draft Repackaging Prioritization Methodology
for Recommendation 2005-1,
Nuclear
Material Packaging
DOE’s IP for Recommendation
2005-1 includes deliverable 5.3-1, “Provide draft of repackaging risk
prioritization methodology to the Board for review and comment.” This commitment was made by DOE in response to
sub-recommendation (3) of Recommendation 2005-1: “Prioritize implementation of the improved nuclear material
packaging requirement consistent with the hazards of the different material
types and the risk posed by the existing package configurations and condition.”
Section 4.0 of DOE’s IP for Recommendation
2005-1 summarizes
the purpose of the risk ranking methodology as follows: “The application of this methodology will
allow DOE to focus resources on the highest risk materials and packages and will
accelerate the reduction in risk to nuclear material handlers.” Section 5.3 of the IP describes the resolution
of this issue as follows: “Based on the
Department nuclear material risk profile, the Department will ensure that the
highest priority items, as determined by the complex-wide risk ranking methodology
will be qualified or repackaged first at all sites.” DOE’s Responsible Manager for Recommendation
2005-1 transmitted the draft repackaging prioritization methodology to the
Board in a letter dated January 30, 2006. This enclosure provides the technical bases of
the comments transmitted to DOE, which were intended to improve the ability of
the risk ranking methodology to meet the above commitments.
Draft Repackaging Risk
Prioritization Methodology. The draft repackaging risk prioritization
methodology is based on a model developed by Los Alamos National Laboratory (LANL)
entitled Risk
Ranking of LANL Nuclear Material Storage Containers for Repackaging Prioritization
(LA-UR-05-3864). DOE’s draft document includes two alternative
models for the package failure index, which is a measure of the relative
probability of package failure. Use of either
risk ranking model is limited to establishing priorities for repackaging of
nuclear material; the models are not to be used for safety basis purposes.
Both models share a common
method for the calculation of dose consequence based on the amount of
radionuclides released as airborne respirable material, using the five-factor formula
from DOE-HDBK-3010-94, Airborne Release Fractions/Rates and Respirable Fractions
for Nonreactor Nuclear Facilities. This amount of material is then
converted to a worker dose using an assumed exposure time and appropriate dose
conversion factor. The LANL model
(option 1) defines
relative risk as the product of the package failure index and dose consequence.
The package failure index is calculated
from the product of the square of the reactivity index and package age. The reactivity index is a qualitative measure
of the propensity of specific material forms to challenge the package integrity
through (1) corrosion, (2) internal pressurization, (3) pyrophoric reactions,
(4) oxidation expansion of metals, and (5) radiation damage.
The alternative model (option 2)
assesses the package failure index using numerical indices that are assigned to
various generic material and container characteristics and used for qualitative
assessment of various challenges to the package. The indices are summed to determine a
numerical ranking for container robustness, the inverse of which is multiplied by
the age of the package to determine the package failure index.
Comments on the Draft
Document. The comments developed by the
Board’s staff can be categorized into the following areas:
·
Suggested
improvements to the package failure probability model (option 1)
–
Incorporating
the vulnerability of the package into the model
–
Accounting
for material type in the reactivity indices
–
Increasing
the effect of package age on failure probability
·
Issues
regarding the alternative package failure probability model (option 2)
·
Misapplication
of threshold dose consequence in the repackaging document
·
Need
for improvement in the review process for Recommendation 2005-1 deliverables
Each of these areas is discussed
in detail below.
Suggested Improvements to
the Package Failure Probability Model (Option 1).
Several aspects of the LANL
model could be improved. The
Recommendation 2005-1 working group, with input from the technical review
board, should investigate for potential improvements to the current model in
three principal areas: (1) accounting
for the vulnerability of the packaging configuration, (2) accounting for the
effects of different material types in the reactivity indices, and (3)
increasing the effect of package age on failure probability. A preliminary evaluation by the Board’s staff
suggests that these changes could result in more reasonable predictions for
package failure. The three areas of
suggested improvement to the option 1 model are discussed separately below.
Incorporating
the Vulnerability of the Package into the Model―The original LANL model, as
presented in LA-UR-05-3864, was based on the product of dose consequence, material
reactivity, package age, and package vulnerability. The current LANL model, as adopted in option 1,
calculates risk based only on the first three parameters. LANL chose not to incorporate container
vulnerability in the current model because of the lack of specific container information
on many of its items. However, by not
attempting to account for the vulnerability of the package, the option 1 model
assigns the same failure probability index to a specific material form
regardless of the type, number, or robustness of the containers. For example, a package consisting of plutonium
oxide stored in a plastic bottle, which is known to decompose from ionizing
radiation, would be assigned the same failure probability index as the
identical material stored in a stainless steel container.
The first commitment completed
under the IP for Recommendation 2005-1 was issuance of a request to the sites
to provide DOE with specific information on material packaging. LANL also has been collecting surveillance
data on packages in its inventory as part of its packaging risk reduction
commitments under the Board’s Recommendations 94-1 and 2000-1. The members of the Recommendation 2005-1
working group and technical review board should be able to use this information
as the basis for providing expert judgments on container vulnerabilities. A model that does not attempt to account for
the vulnerability of the package to known failure mechanisms is unlikely to
yield accurate estimates of the failure probabilities of packages and the
resulting risk posed to workers.
The draft repackaging risk
prioritization document includes a figure from LA-UR-05-3864 that is intended
to offer some support for verification of the option 1 model. LANL points out that this plot
of dose consequence versus failure probability index demonstrates that the
model correctly assigns the highest-risk decade to the failed packages that were
the subject of the August 5, 2003, Type B accident investigation. However, these items were assigned the highest
material reactivity after the accident and were among the materials with the
highest dose consequence in the inventory. Since these represent two of the three parameters
LANL uses to assess risk, this result provides only limited verification of the
model. To adequately demonstrate
verification of the model, results based on data obtained after the model has
been developed need to be consistent with model predictions.
Recent data demonstrate that the
option 1 model
does not assign a particularly high failure probability index to the breached
package that resulted in the December 19, 2005, occurrence of vault worker
contamination at LANL. Nor does the
model assign high failure probability indices to the three packages that LANL’s
surveillance program recently discovered to have inner containers that were
completely corroded, bulged, or severely degraded. Because risk is defined as the product of dose
consequence and failure probability, packages with low dose consequences and
high failure probabilities may not end up being assigned to the highest-risk
decades. For packages known to have
failed by a mechanism intended to be accounted for in the model, there is an
expectation that the model would assign a high failure probability index even
if it did not assign a high risk. The
option 1 model does not meet this expectation for the recent package failures.
The original LANL model
attempted to account for package vulnerability by considering only the three
most robust barriers of a set of nested package containers. Each of these three barriers was represented
by a vector whose indices varied from zero to 3, corresponding qualitatively to
the vulnerability of the barrier to (1) corrosion, (2) internal pressurization,
(3) pyrophoric reactions, (4) oxidation expansion of metals, and (5)
radiation damage. For packages consisting of fewer than three
barriers or packages with unknown barriers, each missing or unknown barrier was
assigned the highest vulnerability indices. The total container vulnerability vector was
determined by vector multiplication of the indices for each of the three barriers.
Although the LANL document listed some
container vulnerability indices for an example package, it did not include a
table of standardized container vulnerability indices for package
configurations that are present in the complex.
An analysis of the container
indices used in the original LANL model suggests that use of zero values for
minimum vulnerability indices can create inconsistencies in the predicted results,
particularly when the vulnerability of a total package consisting of multiple
containers or barriers is calculated. For example, a package consisting of three
taped stainless steel slip-lid cans each assigned vulnerability indices of (0,
0, 3, 3, 0) has the same total container vulnerability indices of (0, 0, 27,
27, 0) as a package consisting of a single taped stainless steel slip-lid can,
since multiplying by the two missing barrier indices of (3, 3, 3, 3, 3) yields
the same result as multiplying by the additional two cans. This inconsistency could be eliminated by assigning
only nonzero indices for container vulnerability.
Another aspect of container
vulnerability in the option 1 model that may require refinement is the
assignment of maximum vulnerability indices to packages with unknown inner containers.
As a result, packages with inner
containers known to be highly vulnerable to decomposition and accelerated
degradation of the outer container (e.g., a plastic bottle containing plutonium
within an outer steel can) would be assigned lower failure probabilities than
packages consisting of unknown inner containers within the same outer container.
This method may be appropriate for
assigning risk to a large population of packages having only a small number of
unknown containers or a larger number of unknown containers believed to have a
high percentage of vulnerable inner containers. However, it may not be appropriate if a large number
of packages having initially unknown inner containers are eventually shown, on
average, to contain much more robust containers than the assumed worst-case
scenario.
Overall, the option 1 model
might benefit from judicious selection of container vulnerability indices and
assignment of numerical values for unknown containers. The staff’s preliminary evaluation of
incorporating nonzero container vulnerability indices into the model and
assuming high but not maximum values for unknown barrier vulnerability suggests
that predictions for package failure derived with this approach are more
reasonable than predictions obtained with the current model.
Accounting
for Material
Type in the Reactivity Indices―The draft risk ranking methodology document refers to
LA-UR-05-3864 for details on material reactivity indices. These indices were determined by the expert
judgment of LANL personnel and an independent review panel. It appears there may be a need for additional
expert judgments regarding other material forms that account for differences in
reactivity among elements. For example,
a highly reactive material, such as plutonium metal, is assigned the same
reactivity indices as a considerably less reactive material of the same form,
such as uranium metal. Furthermore, the option
1 model does not make use of the fifth reactivity parameter in the original
LANL model. If container vulnerability
indices were incorporated back into the model, this fifth reactivity parameter
could be used to account for additional challenges to the packaging posed by
ionizing radiation.
Increasing
the Effect of
Package Age on Failure
Probability―LANL
has assumed that the relative probability of package failure increases linearly
with age. The source of this assumption
appears to be data presented in an earlier report entitled A Risk-Based Prioritization Methodology for Legacy Fissile
Material Disposition at
LANL
(LA-UR-00-5111).
The author of this report attempted to
establish a relationship that would adequately predict time to failure of packages
using data from inspection of inner containers. When all of the available data for inner containers
noted to have failed inspection were compiled as a function of age of the item,
an approximately linear trend was observed. In this study, containers that failed
inspection were assumed to be representative of containers that could
eventually fail. Even though the linear
age relationship concluded from the limited data was tenuous, the author did
not provide support for such a relationship based on known constitutive
relationships or generally accepted principles of component failure.
The treatment of the data in
that report may not be fully justified. There are, as noted in LA-UR-05-3864, at least
five independent mechanisms for failure of containers: (1) corrosion of the container wall from
residual water and any chlorides present in the material; (2) over
pressurization of the container by internal gas generation from volatilization
of residual water and/or polymers; (3) breach of the container from expanding
gas caused by a pyrophoric reaction between oxygen and metal fines; (4) rupture
of the container from oxygen in-leakage, resulting in oxidation and volume
expansion of metal items; and (5) loss of integrity by radiolytic embrittlement
in plastic containers. The kinetics of
these reactions would not be expected to be the same, and therefore the
corresponding container failure times would be expected to vary depending on
which mechanism of failure was rate-controlling. A set
of containers found to have failed inspection because of degradation by one of
several different possible failure mechanisms would not necessarily be a
homogeneous population. The ideal method
for determining the variation of failure probability with package age would
have been to stratify the data into homogeneous populations based on failure
mechanism.
However, because there may be
insufficient data to stratify in this manner, a more defensible relationship
for time to failure could perhaps be estimated from commonly accepted statistical
analyses of failure probability. In
general, the failure rate for nonrandom failures of components without initial
quality control problems is known to increase monotonically with age. For example, components that wear out over
time, such as automobile parts, are known to exhibit low failure rates after
the infant mortality period in the early years of operation, followed by
rapidly increasing failure rates during the later years. For the case of nuclear material packages, as
the container degrades by one of the above mechanisms, the likelihood of
failure will probably be very low initially, since it takes time for
substantial degradation of the packaging to occur. Depending on the kinetics of the
rate-controlling mechanism(s) of failure, it will take a certain amount of time
before degradation of the packaging is extensive enough to cause failure. The failure rate for a population of similar
packages will begin to increase as a greater number of packages degrade
sufficiently to approach incipient failure. The probability of failure might be expected
to increase substantially beyond this time. Therefore, it may be reasonable to assume that
the relative probability of package failure increases more than linearly with
age.
One method that is commonly used
to model failure rates empirically is the Weibull distribution. The change in failure rate (or probability of
failure) with time is a function of a shape parameter, which can be varied to
account for constant, linearly increasing, or power law relationships with time.
Using different versions of LANL models
based on linear age and the square of age (i.e., power law relationship with an
exponent of 2), the Board’s staff evaluated predictions of failure probability
for the packages involved in the Type B accident at LANL, the recent breached
package at LANL, and the recently discovered inner container failures. Results showed that for all of these cases,
the models based on the square of age yielded consistently high failure
probability indices for all of the actual failed packages. In contrast with these results, the option 1
model based on linear age predicted high failure probabilities for the Type B packages,
but lower failure probability indices for the recently failed containers. Additionally, hypothetical robust packages of
the same material and age as the failed containers, modeled using age squared,
resulted in much lower failure probabilities than those of the failed packages.
Therefore, incorporating
container vulnerability into the model and using age squared instead of linear
age improves the consistency of package failure predictions between the limited
set of package failures and various hypothetical robust packages that would be
expected to have lower failure probability indices. DOE may be able to develop other, more
accurate time-to-failure relationships, which likewise could be evaluated using
recent survey data or additional container failures.
Issues Regarding the
Alternative Package Failure Probability Model (Option 2). In
principle, a single methodology for package failure probability is preferable
because it facilitates meaningful comparisons of the risk posed by packages
across the complex. Having two different
methodologies may result in ranking packages differently from one site to
another. As previously discussed, there are
inconsistencies and weaknesses in the option 1 model. The most serious
weakness of the option 1 model is the absence of any parameters that account
for the vulnerability of the package to failure. The option 2 model does attempt to account for
the robustness of the package using parameters and numerical indices that
appear to be generally reasonable; however, the values assigned for unknown
parameters do not appear to be applied consistently with respect to known
parameters. No data have been presented
to support the judgments made about individual values used for the parameters
or the model itself, so there is no way to assess the validity of the option 2
model. Combining the best aspects of
each approach into one model for calculating failure probabilities may
represent an improvement over using two separate methodologies.
Misapplication of Threshold
Dose Consequence in the Repackaging Document.
DOE had earlier notified the
Board that the packaging and storage criteria document will be in the form of a
manual supporting 10 Code of Federal Regulations 835, Occupational
Radiation Protection. DOE’s Recommendation 2005-1 working
group has developed a draft methodology for defining nuclear material
thresholds to be associated with specific packaging criteria based upon the
material’s potential radiological consequence. This methodology uses the same calculation
employed by the Department of Transportation in determining the A2 values presented
in the table in 49 CFR 173.435, General Requirements for Shipments and Packagings.
These values are used to determine the need for the more
robust Type B (instead of Type A) packaging
for transport of radionuclides. The A2
values are based on worker doses from an uptake scenario using a net intake
factor of 10-6. Currently,
the working group has set the de
minimis quantity
(threshold to be in the scope of the draft manual requirements) to be equal to
a worker committed effective dose equivalent (CEDE) of 5
rem.
DOE’s draft repackaging
prioritization methodology states that all packages exceeding the threshold
defined on the plot of dose versus failure probability index are deemed to need
repackaging. This plot illustrates the
threshold as those packages determined to have a dose consequence of 5
rem CEDE or
greater, using the LANL methodology for calculating the dose to workers. LANL’s dose methodology yields significantly
different results than those obtained with the method proposed in the draft
manual. As a result, packages with
sufficient quantities of material to be in the scope of the draft manual may be
excluded from the repackaging prioritization process. The threshold dose values in the repackaging
prioritization methodology ought to either be clarified to convey the
definition used in the draft manual or removed.
Need for Improvement in the
Review Process for Recommendation 2005-1 Deliverables. The
process for completing a deliverable, such as this draft repackaging risk prioritization
document, consists of three steps. First, the Recommendation 2005-1 working group
drafts the document and sends it to the technical review board. Second, the technical review board provides
comments on the draft document. Third,
the working group and technical review board resolve the comments, and a final
draft version of the document is sent to the Board. Any comments that cannot be resolved by the
working group and technical review board are to be resolved by the Responsible
Manager for Recommendation 2005-1. The
Board’s staff has observed that the technical review board members are highly
effective at identifying weaknesses and suggesting critical clarifications in
the draft work products.
Many of the problems identified
by the Board’s staff ought to have been identified by the technical review
board and resolved before the draft document was transmitted to
the Board. The Board’s staff conducted a subsequent
review, which revealed that some of the issues raised by the staff had in fact
been identified by the technical review board. Although most of the technical review board’s
comments of an editorial nature were addressed, the more significant comments
were not resolved in the draft risk ranking methodology that was transmitted to
the Board. This situation calls into
question the thoroughness of the comment resolution process. The document development and review processes
of the Recommendation 2005-1 working group need to be improved and better
integrated before future draft deliverables arc transmitted to the Board.