DRC-1987-001001 - 0901a068807b2ff9

BNL-- 52075 BNL 52075
UC-70
DE88 005242 jNucleir Waste Mamgement —TIC-4500J
LEVEL OF EVIDENCE FOR REASONABLE ASSURANCE
GUIDES TO PREDICTION
Donald G. Schweitzer and Cesar Saslre I^fl llflf?
Stiff!?!II
» o P = 5' s- — -•" 3 aS 3 c 3 - ag-g 3
iaMUfi;
5aS»E i ?;< ? Q
April 1987 'ifji^llll g
all I a.S<5S c
DONALD G. SCHWEITZER. HEAD
DOE RADIOACTIVE WASTE MANAGEMENT DIVISION
DEPARTMENT OF NUCLEAR ENERGY
BROOKHAVEN NATIONAL LABORATORY
ASSOCIATED UNIVERSITIES. INC.
UPTON. L0N6 ISLANO. NEW YORK 11973
Prepared for the
OFFICE OF CIVILIAN RADIOACTIVE WASTE MANAGEMENT
UNDER CONTRACT NO. DE-AC02-76CH00016 WITH THE
UNITED STATES DEPARTMENT OF ENERGY
li'oi V:\'l:i' •'•.;. -i
mm
DISCLAIMER
This report was prepared as an account of work sponsored by an agency of the United
States Government. Neither the United States Government nor any agency thereof,
nor any of their employees, nor any of their contractors, subcontractors, or their
employees, makes any warranty, express or implied, or assumes any legal liability or
responsibility for the accuracy, completeness, or usefulness of any information,
apparatus, product, or process disclosed, or represents that its use would not infringe
privately owned rights. Reference herein to anyspecifio commercial product, process.
or service by trade name, trademark, manufacturer, or otherwise, does not necessarily
constitute or imply its endorsement, recommendation, or favoring by the United States
Government or any agency, contractor or subcontractor thereof. The views and
opinions of authors expressed herein do not necessarily state or reflect those of the
United States Government or any agency, contractor or subcontractor thereof.
Printed in the United States of America
Available from
National Technical Information Service
U.S. Department of Commerce
5285 Port Royal Road
Springfield, VA 22161
NT1S price codes:
Printed Copy: A02: Microfiche Copy: A01
CONTENTS
Page
ABSTRACT , v
1. INTRODUCTION 1
2. DEFENSE-IN-DEPTH 2
3. DIVERSE CANDIDATE MATERIALS 2
4. TAILORED BACKFILLS 2
b. PEER REVIEWS 3
6. BACK-UP POSITION 3
7. LEVEL OF EVIDENCE FOR POSITIVE DECISIONS OF
REASONABLE ASSURANCE 5
7.1 Uncertainty Requirements 5
7.2 Uncertainty and Prediction 6
7.3 Accelerated Tests and Predictions 7
7.4 Conservative Arguments .................. 8
7.5 Potentially Predictable Systems 8
7.6 Potentially Unpredictable Systems 9
7.7 Absence of Adverse Phenomena 11
7.8 Potential for Problem Resolution 12
7.9 Use and Limitations of Peer Review 13
8. CONCLUSIONS 14
- iii -
ABSTRACT
Over the past years, the DOE Contractors have produced a great deal of
work that has been extensively reviewed and criticized by the Nuclear Regula-
tory Commission (NRC), the Materials Review Board (MRB) of the DOE, the
Advisory Committee on Reactor Safeguards (ACRS), and the technical support
group at Brookhaven National Laboratory (BNL). Common aspects of the reviews
and criticisms have provided information on the level of evidence required by
the scientific community to defend performance claims. Important indicators
of the type of evidence that the NRC will require for favorable decisions of
reasonable assurance also can be obtained from 10 CFR 60 and its rationale,
from NRC guides and Technical Position papers, from past reviews of the DOE
programs by NRC Contractors, and from the use of reasonable assurance by the
NRC in its 1984 Waste Confidence Decision.
Thi.-' report describes general concepts related to the acceptability and
unacceptability of the level of evidence needed to defend claims with reason-
able assurance. The concepts were formulated on the basis of analyses of the
NRC position papers, and of common aspects of the reviews and criticisms
dealing with compliance demonstration.
- v -
1. INTRODUCTION
In preparing its licensing strategy, the DOE will be required to make
early decisions on what it considers to be acceptable demonstrations of rea-
sonable assurance. There arc likely to be several definitions of reasonable
assurance. Each will be dependent on the type of evidence used in the license
subraittar.ee. The claim that there is reasonable assurance that the corrosion
rate of copper at 60°C is less than a given value will involve different argu-
ments from those used to defend the claim that there is reasonable assurance
that safe disposal of high-level waste is technically feasible.
Important indicators of the type of evidence that the NRC will require
for favorable decisions of reasonable assurance can be obtained from 10 CFR 60
and its rationale, from NRC guides and Technical Position papers, from past
reviews of the DOE programs, and from the use of reasonable assurance by the
NRC in its 1984 Waste Confidence Decision.
The NRC has already made positive findings of reasonable assurance in
dealing with the disposal of high-level waste.
In responding to negative criticisms of the DOE programs, the NRC
accepted and deferred the claims made by the DOE in the draft Mission Plan For
The Civilian Radioactive Waste Management Program (1984) and in the earlier
Statement of Position by the DOE to justify "finding (1)":
"(1) The Commission finds reasonable assurance that safe
disposal of high level radioactive waste and spent fuel in a
mined geologic repository is technically feasible."
The arguments used by the NRC to defend the finding of reasonable assur-
ance were that the DOE claimed, and the NRC agreed, that proposed DOE programs
had the potential to demonstrate safe isolation of high-level waste because
they were going to use and evaluate:
(1) A defense-in-depth approach
(2) Diverse candidate materials
(3) A variety of tailored backfills
(4) Peer review of the DOE program
(5) An acceptable back-up positiony if needed.
The potential of the proposed DOE program as described prior to 1984, was
considered sufficient to justify a favorable finding of reasonable assurance
with the following reservation:
"The Commission's Waste Confidence decision Is unavoid-
ably in the nature of a prediction. While the Commission
believes for the reasons set out in the decision that it can
with reasonable assurance, reach favorable conclusions of con-
fidence, the Commission recognizes that the possibility of
significant unexpected events remains open. Consequently, the
Commission will review its conclusions on waste confidence
should significant and pertinent unexpected events occur, or
- 1 -
at least every 5 years until a repository for high-level
radioactive waste and spent fuel is available."
The specific arguments given by the NRC in the Waste Confidence decision
were:
2. DEFENSE-IN-DEPTH
"To assure long-term containment, DOE's conceptual design
of a waste package is based on a defense-in-depth approach and
involves a number of components including spent fuel,
stabilizer (or filler), waste canister, overpack, and an
emplacement sleeve. The stabilizer is intended to improve heat
transfer from the spent fuel, to provide mechanical resistance
to possible canister collapse caused by lithostatic pressure,
and to act as a corrosion-resistant barrier between the spent
fuel and the canister A variety of candidate materials
is being considered for these elements. The various waste
package components are to be combined in a conservative design
that will compensate for the overall technical uncertainties in
containment capability."
3. DIVERSE CANDIDATE MATERIALS
"DOE is studying over 28 candidate materials for canisters
and overpack. The DOE evaluation program indicates that many
of these materials are promising. For example, iron alloys
have demonstrated long term durability [DOE], and titanium
alloys and nickel alloys show high resistance to corrosion
[DOE]. Ceramics are resistant to chemical degradation and have
many other desirable properties [DOE].... For more demanding
requirements such as brine applications, the alloys of tita-
nium, zirconium or nickel appear to represent alternate choices
[DOE]."
A. TAILORED BACKFILLS
"In DOE's conceptual design, one engineered barrier con-
,sists of backfill materials for filling voids between canister,
overpack, sleeve and host rock. The materials are chosen to
retard radionuclide migration. The task is to design and test
barrier materials which will be effective for very long periods
oi: time. Candidate materials include bentonite, zeolites,
iron, calcium or magnesium oxide, tachyhydrite, anhydrite, apa-
tite, peat, gypsum, alumina, carbon, calcium chloride, crushed
host rock and others [DOE]."
- 2 -
5. PEER REVIEWS
"Underlying the pessimism of some participants is appar-
ently a belief that DOE's past record in solving technical
problems undermines the possibility of finding confidence in
DOE's ability to solve the waste disposal problems in a timely
way.•.• The qualifications and professional experience of the
many scientists and engineers on the overview committees and
peer review groups who advise and consult on the DOE program
should provide confidence in DOE's efforts [DOE]."
6. BACK-UP POSITION
"The National Research Council after reviewing the Swedish
waste disposal work [DOE], concluded that the Swedish waste
package could contain the radionuclides in spent fuel rods for
hundreds of thousands of years [DOE]."
The NRC and their Contractors have stressed the need for uncertainty
analyses and analyses of alternative interpretations. This is likely to be
the most important concept in allowing a positive finding of reasonable
assurance.
In early definitions and defense of reasonable assurance the Commission
in 48 FR 28204 stated (emphasis added):
"In the Commission's view, the 'reasonable assurance*
standard neither implies a lack of conservatism nor creates a
standard which is impossible to meet. On the contrary, it
parallels language which the Commission has applied in other
contexts, such as the licensing of nuclear reactors, for many
years. See 10 CFR 50.35(a) and 50.40(a). The reasonable
assurance standard is derived from the finding the Commission
is required to make under the Atomic Energy Act that the
licensed activity provide 'adequate protection' to the health
and safety of the public; the standard has been approved by the
Supreme Court, Power Reactor Development Co. v. Electrical
Union, 367 U.S. 396, 407 (1961). This standard, in addition to
being commonly used and accepted in the Commission's licensing
activities, allows the flexibility necessary for the Coaalssion
to make judgmental distinctions with respect to quantitative
data which «ay have large uncertainties (in the mathematical
sense) associated with it."
and in 48 FR 28203:
"As explained in the discussion of Reasonable Assurance,
below, uncertainties in the models used in the analysis of
repository performance must be considered in the Commission's
deliberations on the issuance of a construction authorization
or license."
- 3 -
Further elaboration on the content of reasonable assurance was given in
48 FR 28200 and 28201:
"There are two principal elements that will go into the
Commission's application of this 'reasonable assurance* con-
cept. First, the performance assessment which has been per-
formed must indicate that the likelihood of exceeding the EPA
standard is low. Second, the Commission must be satisfied that
the performance assessment is sufficiently conservative, and
its limitations are sufficiently well understood, that the
actual performance of the geologic repository will be within
predicted limits."
and from 48 FR 28204:
"Even if the calculated probability of meeting the Commis-
sion's standards is very high that would not be sufficient for
the Commission to have 'reasonable assurance1; the Commission
would still have to assess uncertainties associated with the
models and data that had been considered. This involves quali-
tative as well as quantitative assessments. The Commission
would not issue a license unless it were to conclude, after
such assessments, that there is reasonable assurance that the
outcome will in fact conform to the relevant standards and
criteria."
"The licensing decisions which the Commission will be
called upon to make involve additional uncertainties—those
pertaining to the correctness of the models being used to
describe the physical systems—which are not quantifiable by
statistical methods.... this means that modeling uncer-
tainties will be reduced by projecting behavior from well
understood but simpler systems which conservatively approximate
the systems in question."
In addressing the acceptability of SCPs, the NRC has recently stated:
"Should the range of uncertainties and alternative inter-
pretations and assumptions that can be reasonably supported by
the existing data not be considered in the SCP development, the
SCP could be deficient in the identification and description
of:
(1) The site including the range of uncertainties in known
site conditions
(2) The issues and information needed to resolve issues
(3) The issue resolution strategies
(4) The performance allocation (i.e. the definition of
performance goals and desired, associated confidence
levels for various components of the repository system)
(5) The investigation and study plans (tests and analyses)
(6) The rationales for investigations and studies with
consideration to various sources of uncertainty."
- 4 -
7. LEVEL OF EVIDENCE FOR POSITIVE DECISIONS OF REASONABLE ASSURANCE
7.\ Uncertainty Requirements
The NRC's use of reasonable assurance, and the DOE's strategy planning
are intrinsically associated with describing and defending predictions and
perfc mance claims.
A severe problem associated with the development of an acceptable strat-
egy occurs when critical predictions and performance claims are stated without
bounds and are based on data without addressing the effects of uncertainties
in the data.
If, in claiming performance or describing predictions, the reader is pro-
vided with assumptions without analyses of the consequences of the assumptions
and without analyses of the differences in conclusions that could occur if the
assumptions were modified, then evaluations pose major problems. In order to
judge the validity of claims and predictions made under these conditions,
the reader may be required to be better informed and to do more work than the
authors of the claims. The problem has been noted repeatedly, and arguments
without uncertainty analyses have been declared as unacceptable in the present
regulatory framework.
This leads to an obvious requirement for the justification for a finding
of reasonable assurance:
A. If the uncertainties cannot be defined, the data cannot be defended.
Pitting is a poorly understood localized failure mode that is dependent
on both macroscopic and microscopic properties of the metal and of the envi-
ronment. Initiation of pitting can depend upon chemical, mechanical, and
thermal inhomogeneities on fhe surface of the metal. The rate of pitting and
the depth to which a pit can progress can be dependent on temperature, temper-
ature gradients, Eh, pH, ion species, ion concentration, dissolved gases, sol-
ute concentration gradients, diffusion of reactants and reaction products to
and from the surface of the pit, and the direction of gravity. The environ-
mental factors affecting the rate and depth of pit formation are the micro-
scopic values in and near the pit. These values can be significantly
different from the average values of the macroscopic environment.
This leads to a rule for uncertainties associated with localized failure
and degradation mechanisms:
B. Uncertainties in the microscopic environments associated with localized
failure •echanisas are generally greater than the uncertainties in the
•acroscopic environments.
**********
- 5 -
7.2 Uncertainty and Prediction
Many strategies for containment by corrosion resistant metals and con-
trolled release by dissolution of the waste forms involve non-equilibrium pro-
cesses that approach uncertain final states. Although the science of chemical
kinetics can be used with severe limitations to predict non-equilibrium
changes that tend to approach equilibrium final states, there is no valid
method for long-term prediction of non-equlibrium systems evolving towards
ill-defined end states.
In the absence of the classic test of substituting equilibrium concentra-
tions to check the validity of an isothermal rate expression at long times, a
great deal of highly precise work is necessary to identify and defend an iso-
thermal rate law for a multi-component heterogeneous system. For such sys-
tems, the kinetic expression at any given temperature is a composite from many
processes only one of which is generally rate determining. In principle, each
process has its own temperature dependence, making it unlikely that the same
process is rate determining at all temperatures. In practice, the energy
barrier that prevents the unstable reactants from instantaneously forming more
stable products can be estimated from manipulation of precise isothermal rate
expressions over narrow temperature ranges. Since different processes may
have similar barriers, justification of the use of these techniques over large
temperature ranges, requires that all the factors affecting the isothermal
rate at any one temperature also appear in the same mathematical form at every
other temperature. Experience has shown that for the corrosion, oxidation,
and dissolution phenomena involved in waste isolation, the uncertainties asso-
ciated with the data necessary for chemical kinetics and the uncertainties in
the kinetic expressions themselves, will very likely preclude successful
justification for kinetic predictions of this type.
The problem is even further complicated by the fact that many of the
individual components involved in reactions with other components are subject
to slow long-term internal changes affecting their properties and their roles
in reacting.
This leads to two observations related to the defense of prediction of
property changes through chemical kinetics:
C. Intrinsic uncertainties in data and theory make it unlikely that chemical
kinetics can be used for long-term prediction of property changes in
systems evolving towards ill-defined end states.
D. Intrinsic uncertainties in data and theory make it unlikely that tempera-
ture can be used as an accelerating variable for long-term prediction of
property changes in systems evolving towards ill-defined end states.
**********
- 6 -
7.3 Accelerated Tests and Prediction
Serious problems can be associated with strategies that depend upon
predicting property changes in thermodynaraically unstable materials even when
the end state may be defined.
Consider the problem associated with predicting radionuclide specific re-
leases from spent fuel over 10,000 years in an environment containing mois-
ture, air and changing temperatures. Over the 10,000-year period the engine-
ered portion of the repository will change from above 200°C to below 60°C.
Compliance with the controlled release part of the existing regulation re-
quires that the amount of water leaving the engineered barrier system is known
on an annual basis and that the amount of each radionuclide in the water can
be acceptably upperbounded on an annual basis for the 10,000 years.
Ignoring spent fuel gap inventories, the bulk of the the radionuclide
inventory is initially retained in a thermodynamically unstable matrix. At
temperatures of concern, pure UO2 is unstable and should convert to UO3.
Irradiated UO2 is more unstable than pure UO2 because of chemical potential
increases resulting from the presence of non-equilibrium amounts of fission
products. Additional thermodynamic instability arises from the excess lattice
energies and crystalline stresses resulting from the quenching of thermal
spikes, from the fission products and from the non-stoichiometric amounts of
excess oxygen resulting from the fission process. All the above factors
depend upon the degree, and to some extent, on the nature of the burn-up. In
principle, the matrix should change with time from non-stoichiometric UO2 to a
more stable form of UO3. This conversion goes through a number of intermedi-
ate uranium oxides which result in large volume changes and in crystalline and
internal surface changes redistributing the radionuclides and consuming the
excess oxygen from fission. In agreement with theory, small amounts of a
hydrated form of UO3 have been determined to form from unirradiated uranium
dioxide over a 15-year period in a laboratory experiment at room temperature
(Wadsten, T., J. Nuc. Mater. 64, 315, 1977). Nevertheless, it is not possible
at present to predict the time dependence of such transformations in spent
fuel because of the very large number of ill-defined factors that can affect
such changes. For irradiated UO2 containing appreciable but varying amounts
of fission products, it is, in fact, not clear what the thermodynamically most
stable states are ovar this temperature range.
Leaching and matrix dissolution depend upon the oxidation state of the
matrix, the microscopic crystalline surface areas and surface energies, the
distribution of the radionuclides, the possible catalytic or inhibitive
effects of impurities and/or excess oxygen, the grain boundary compositions
etc. Recognizing that all of these can vary with burn-up and with time,
defense of predictability of radionuclide specific release rates on an annual
basis for a 10,000-year period in a system of such complexity will require
more than conventional extrapolation of high temperature oxidation rates to
lower temperatures. Predictions of changes in reaction rates are not equiva-
lent to predictions of the changes in properties associated with the rates.
Claims that there may be long periods of time where releases vary from below
acceptable values to above acceptable values during the 10,000 years may be
impossible to disprove. The existing regulation requires compliance with
- 7 -
release rates on an annual basis. The use of longer times to allow fluctua-
tions within an acceptable average release has been rejected because it is not
possible to defend unequal releases to different generations.
E. Accelerated tests used to predict reaction rates, may not be valid
predictors for the property changes associated with these rates.
**********
7.4 Conservative Arguments
The NRC in Staff Technical Position Papers have warned against the indis-
criminate use of temperature as an accelerating variable and indiscriminate
use of short-term tests to predict long-term performance. On the other hand,
they have also stated (10 CFR 60, 1984 Waste Confidence Decision) that they
expect the DOE to use short-term tests as part of the evidence to justify
long-term performance if the short-terrc tests can be shown to be conservative.
F. The acceptability of explicit arguments and evidence defining and defend-
ing "bounding" or "conservative" conditions are prerequisites for the
acceptability of short-tern accelerated experiments.
**********
7.5 Potentially Predictable Systems
Systems in equilibrium do not change.
This leads to an obvious rule for acceptable long-term prediction:
G. The absence of changes is predictable.
**********
Thermodynamically stable materials are materials that can achieve equi-
librium with one or more environments. Thermodynamically unstable materials
undergo internal changes and cannot achieve equilibrium with any environment.
Many non-equilibrium systems consisting of thermodynamically stable mate-
rials approach equilibrium with monotonically decreasing rates. Isothermal
diffusion from a source that is depleted by the diffusion process decreases
with time. Fluid flow under pressure through a fixed orifice decreases as the
pressure decreases. Isothermal oxidation of graphite in a fixed amount of air
decreases as the oxygen or graphite is reacted. The isothermal reaction rate
of fixed amounts of hydrogen and oxygen decreases as water is formed. In many
examples the mechanisms involved in the reactions are not understood but the
end states are defined. For such systems short-term measurements of the time
rate of change can be used to upperbound long-term predictions.
- 8 -
H. Theoretical arguments can be used to justify •onotonically decreasing
changes in soae systems evolving towards known final states. In such
systeas short tiae aeasureaents can be used to conservatively bound
long—tera predictions.
**********
7.6 Potentially Unpredictable Systems
Models which assume the radionuclide specific release from the engineered
system can be bounded by thermodynamic solubility arguments also are subject
to great uncertainties. The required data on mutual solubility effects for
systems of this complexity do not exist and are difficult to come by. Such
models suffer from the implicit assumption that non-equilibrium phenomena such
as colloid and particulate transport processes are inoperative or negligible
over the required time periods. To date, this assumption has not proven
defensible.
Consider a second example such as brittle failure. Brittle failure has
been studied extensively since the late 1800s and is still not understood.
The following discussion is taken verbatim from Juvinall, "Stress, Strain, and
Strength", McGraw-Hill, Series in Mechanical Engineering, (1967):
"Brittle fracture is the expected mode of failure of materials
like cast iron, glass, concrete,and porcelain. On the other hand,
most steels, wrought aluminum, magnesium, brass, etc., are expected
to show considerable distortion (and to absorb substantial energy)
before fracture can occur. That such is not always the case, at
least with steel, has caused engineers considerable embarrassment.
Consider, for example, the plight of the engineers associated with
the design of the ship illustrated in Fig. 5.3, which fractured at
the fitting dock...." (the figure shows the "S.S. Schenectady" bro-
ken in two at the fitting dock in Portland Oregon, January 16,
1943. ".....This is not an isolated incident; U.S. government
reports cited in [Tipper, C.F., 'The Brittle Fracture Story,'
Cambridge University Press, New York, 1962] list 233 ship casualties
which so weakened the main hull structure that the vessel was lost
or was in a dangerous condition. About two-thirds of the ships
involved were of welded construction, with most of the balance being
partly welded and partly riveted. It is interesting to note that
one riveted tanker broke in two twice and was repaired each time.
..." "....The brittle-fracture problem has been greatly reduced in
ships of modern design. The problem is not completely solved yet,
however as evidenced by the breaking in two of two large modern tan-
kers. One was the 'African Queen,1 a German ship built in 1955. In
December, 1958 she ran aground on a soft sand bar at 45°F and frac-
tured 30 min later. More recently, the 'World Concord,' a British
tanker of around 40,000 tons, fractured in two in the Irish Sea...."
"...Brittle fracture of structural steel has by no means been limit-
ed to ships. In fact, this problem became evident in various struc-
tures during the very early use of structural steel [Shank, M.E., 'A
Critical Survey of Brittle Failure in Carbon Plate Steel Structures
- 9 -
Other Than Ships,' Welding Research Council of the Engineering Foun-
dation Bull. 17, January 1954]...." " Besides ships, brittle
fractures have occurred in such steel members as tanks, bridges,
pressure vessels, smokestacks, penstocks, power shovels and gas
transmission lines. One of the more notable examples occurred in
January, 1919, when the 2,300,000-gal Boston molasses tank, then 3
years old, collapsed. Twelve persons and many horses were killed,
most by drowning in molasses. Forty more persons were injured, and
extensive property damage occurred. After years of litigation which
involved the testimony of many leading authorities, the court audi-
tor stated in his summary that he had 'at times felt that the only
rock to which he could safely cling was the obvious fact that at
least one-half of the scientists must be wrong.1"
The first large scale fracture of a steel bridge was that of a vierendeel
welded truss bridge over the Albert Canal at Hasselt, Belgium, in March 1943.
Shank discusses the brittle failure of 21 steel bridges, the most recent being
that of the then four-year old Duplessis Bridge, in Quebec, Canada in January
1951. Subsequent to Shank's study, The King's Bridge of Melbourne Australia,
failed in July, 1962 about 14 months after its completion.
The above examples illustrate:
I. There is no theory for the tine evolution of internal property changes In
theraodynaaically unstable Materials. Any process dependent on these
properties is likely to be unpredictable. •
**********
Corrosion of metals and oxidation of porous materials can involve non-
monotonic rate changes. Corrosion is a kinetic, non-equilibrium heterogeneous
process. In general, there are more difficulties in predicting rate changes
in heterogeneous systems then there are in homogeneous systems. For metals
corroding in liquids, the factors that can lead to increases in the corrosion
rate with time are:
• increases in surface area
• stress induced fracture of adherent corrosion product films
• precipitation or isothermal annealing in the metal causing increased
surface or grain boundary reactivity
• formation and subsequent breakaway of a surface film formed by the
corrosion reaction
• depletion of surface components leading to increased surface
reactivity
• changes in local pH and chemical composition of the corroding medium
due to depletion from reactions with the metal or secondary reactions
of the corrosion products
• development of local concentration or aeration cells, etc.
J. A large number of processes can cause non-aonotonic rate changes in
•etallic corrosion. These processes may preclude acceptable defense of
long-tera prediction of corrosion rates for •ulti-coaponent metals.
**********
- 10 -
Equilibrium compounds which form as solid reaction products have unit
thermodynaraic activity. Variations or uncertainties in the quantities of
solid reaction products do not affect the final states or rate of a reaction
through chemical theory. Their roles are mechanical, physical, or chemi-
physical if they adhere to the solids from which they were formed and develop
diffusion barriers for further reaction. Solid reaction products in the form
of films can cause non-monotonic reaction rates in isolated systems. Adherent
films are known to "breakaway" as they grow and develop stress. Film forma-
tion is often associated with changes in the surface composition of the under-
lying reacting metal. The new surface can react slower or faster than the
original surface. Neither theory nor experience provides arguments that can
be used to predict long-term changes for rates dependent on film formation.
K. Long-tem predictions of rates determined by film formation may not be
defensible.
**********
7.7 Absence of Adverse Phenomena
For the present DOE design choices, the evidence needed to determine
whether it is possible to achieve containment of HLW by a Waste Package and
controlled release from an Engineered Barrier System requires state-of-the-art
evaluation of the kinetics and thermodynamics of many complex processes.
These include corrosion, dissolution and leaching of glass and spent fuel,
metastable phase transformations in steels and UO2, heterogeneous radiation
reactions, surface sorption and reaction phenomena, and radionuclide movement
through diffusion, leaching, and colloidal or particulate transport.
In the existing regulatory framework, the burden of proof is on the
applicant for the repository license. The logic for justifying a performance
claim requires the explicit description of the temporal changes in the envi-
ronment and in the physical and chemical conditions affecting the expected
performance. The regulatory agencies evaluating the adequacy of this evidence
invoke analyses of the uncertainties in the data and in the models, and
require the applicant to provide analyses of the consequences of both favor-
able and unfavorable predicted performance. Implicit in every claim for
favorable performance is the assumption that adverse processes do not occur.
The present conservative regulatory tenor has imposed a severe burden on the
applicant by requiring evidence for the absence of potentially adverse proces-
ses and by limiting the use of experiments in which the adverse process was
not observed. The absence of potentially adverse phenomena must be defended
by theoretical arguments which are supported by confirmatory null experi-
ments. Because of the inordinately long times over which performance
predictions are required, null experiments by themselves bear little weight.
With the important but restrictive exceptions of thermodynaraic arguments
on perpetual motion, conversion of heat to work, efficiency of Carnot cycles,
etc., and obvious violations of physical laws, it is in the nature of science
to be skeptical about claims of phenomena that will not occur.
This leads to a rule for the elimination of potential, failure mechanisms,
degradation mechanisms, or adverse processes:
- 11 -
L. The absence of potentially adverse phenomena aust be defended bj
theoretical arguments supported by conciliatory null experiments.
7.8 Potential for Problem Resolution
Almost no information exists at present which is both necessary and suf-
ficient to quantitatively estimate the likelihood of the occurrence of many
potential failure mechanisms in repository environments. Literature dealing
with the likelihood of occurrence of potential failure mechanisms generally
concludes with recommendations for future studies without discussions of the
likelihood that the studies will produce the information needed for problem
resolution. Consider the following conclusions obtained from an NRC Contrac-
tor study ["Stress-Corrosion Cracking of Low-Strength Carbon Steels in Candi-
date High-Level Waste Repository Environments," NUREG/CR-3861, BMI-2147, 1987]
in which the objective of the study was a literature review to identify poten-
tial stress-corrosion cracking agents for low-strength carbon and low-alloy
steels in repository environments:
"On the basis of the literature survey, it is recommended
that experimental and modeling studies be undertaken to assess
the likelihood of stress-corrosion cracking of low-strength
carbon steel in waste repository environments. For the identi-
fied cracking agents that are present in waste repositories,
lower limits in concentration needed to promote cracking must
be established. It is important to establish these data under
realistic conditions, where synergistic effects of other spe-
cies in the waste repository are considered In addition,
physical modeling of potential concentrating mechanisms needs
• to be performed in order to assess more accurately the likeli-
hood that these proposed mechanisms operate and to bound the
upper limits of concentration for each mechanism. Identifying
and quantifying mechanisms where selective concentration of a
single or of a few species occurs is especially important in
relating the modeling to the laboratory studies. The models
should then be verified through experimentation."
Much of the emphasis on problems associated with the DOE Contractors'
waste package programs addresses claims related to containment by metal con-
tainers. In BNL reviews we have concluded that claims needed to demonstrate
controlled release may be more difficult to defend than those needed for con-
tainment. This view is supported by the National Research Council of The
National Academy of Sciences. In stressing the advantages of long-term con-
tainment vs controlled release in the Swedish Plan for Final Storage of Spent
Fuel, the Council claimed:
"The most troubling uncertainties remaining in the KBS
proposal relate to the part least likely ever to be relevant to
actual repository performance, i.e., the migration of radio-
nuclides through buffer and bedrock in the improbable event of
- 12 -
early canister rupture. Uncertainties here are greater simply
because of the number of variables that must be considered in
any model used to describe the dissolution of radionuclides
from fuel rods and their movement through a variety of materi-
als to the point where they may become hazardous to humans.
Solubilities, effects of local oxidation due to radiolysis,
retardation by sorption of some radionuclides as they migrate
through buffer and rock, effect on retardation of the formation
of complexes and colloids, dispersion of nuclides into rock
along fractures where flow is concentrated—all these must
enter into a system of models for radionuclide behavior."
M. Estimates of the likelihood of problem resolution require definitions of
both the necessary and the sufficient work needed and evidence that the
needed work can be accomplished.
**********
7.9 Use and Limitations of Peer Review
Past experience has shown that the the peer review process is resulting
in an expanding list of problems requiring resolution. New experts tend to
add their special concerns to an already unmanageable list of issues that the
Contractors are required to put to rest. Many of these issues have challenged
science for decades without significant progress in understanding or identify-
ing the mechanisms involved. Indiscriminate and generic peer review without
concomitant procedures for problem resolution is likely to result in continu-
ing negative criticism from the scientific community. Much, if not all of
this criticism is accumulating in the public domain. If negative criticism of
the DOE programs by members of the scientific community continues to accumu-
late, and if the existing criticism is not negated through appropriate change*
very stringent standards for "Reasonable Assurance" may be predetermined by
reviewers and not by the DOE or the NRC. It is unreasonable to expect the NRC
to accept and defend DOE claims in license hearings if reputable scientists
have already rejected them in prior peer reviews.
Productive use of peer review by the DOE should be focused on specific,
high priority issues with explicit methods for problem resolution as prerequi-
sites for conclusion of the review. If the panel of experts used for the
review cannot define the necessary and sufficient conditions for problem reso-
lution along with the likelihood of accomplishing resolution to the problem,
it should be assumed that the problem is insoluble.
I. Unless the peer review process provides resolutions to the problems being
addressed, the number of controversial issues will increase and the
standards for reasonable assurance will become increasingly stringent.
If the peer review panel cannot provide means for problem resolution, it
should be assumed the problem is insoluble.
- 13
8. CONCLUSIONS
A great deal of information exists related to the level of evidence that
will be required to demonstrate and defend claims with reasonable assurance.
Additional information in the form of reviews by peer groups, by NRC staff and
Contractors and by the ACRS can be used to develop guidelines for future
programs. In some cases DOE Contractors have taken exception to comments in
the reviews. Resolution of the differences is necessary to the development of
a successful license application. We believe elaboration and review of the
concepts discussed here will be needed by the DOE in developing its license
application and in making decisions as to whether or not compliance with
existing regulations is technically feasible.
- 14 -