United States         Office of          EPA/520/4-78-004
              Environmental Protection     Radiation Programs
              Agency            Washington DC 20460
              Radiation
&EPA        State of Geological
              Knowledge Regarding
              Potential Transport of
              High-level Radioactive

              Continental Repositories

              Report of An Ad Hoc Panel
              of Earth Scientists


-------
  This report was prepared as an account of work sponsored
by the Environmental Protection Agency of the United States
Government  under Contract No. 68-01-4470,  Modification
No.  2.  Neither  the  United  States  nor the  United States
Environmental Protection Agency makes any warranty, express
or implied, or assumes  any legal  liability or responsibility for
for the accuracy, completeness or usefulness of any information,
apparatus, produce or process disclosed, or represents that its
use would not infringe  privately owned  rights.

-------
vvEPA          State of Geological
               Knowledge Regarding
               Potential Transport of
               High-level Radioactive
               Waste From Deep
               Continental Repositories

               Report of An Ad Hoc Panel
               of Earth Scientists

                    JUN    1878

-------
                       TABLE OF CONTENTS







Foreword




Preface




PART I:     INTRODUCTION AND STATEMENT OF THE PROBLEM




PART II:    THE CONTENTS AND PROPERTIES OF THE CANISTER




PART III:   MECHANICAL PROPERTIES OF ROCKS




PART IV:    CANDIDATE LITHOLOGIES




PART V:     SOLUTION TRANSPORT THROUGH ROCKS




PART VI:    RADIONUCLIDE TRANSPORT FORECASTING




PART VII:   GEOLOGIC AND OTHER ACCIDENTAL HAZARDS




PART VIII:  MONITORING THE REPOSITORY




PART IX:    CONCLUSIONS




References




List of Abbreviations




Glossary
Page




 v




 vii




  1




  5




 13




 17




 25




 29




 35




 41




 43




 47




 49




 51
                               iii

-------
                               FOREWORD
A major Federal effort is underway to develop methods of disposal of high-
level radioactive waste.  An important element of this program, scheduled
for early completion, is the development and promulgation by the Environ-
mental Protection Agency (EPA) of environmental standards for such materials.

In conjunction with its efforts to develop these standards, EPA is using
state-of-the-art techniques to estimate the expected and potential environ-
mental impacts from potential waste repositories, including repositories in
deep geologic formations.  It recognizes, however, that there may be signi-
ficant uncertainties and controversy regarding knowledge of rock properties,
hydrogeology, and other factors, especially as they relate to the ability to
provide long-term containment of radioactive wastes.

Therefore, in order to assure the protection of public health, EPA believed
that an expert panel was needed to evaluate objectively the adequacy of
basic knowledge in the pertinent earth sciences for reliably estimating
environmental impacts.

In March 1977, the Environmental Protection Agency contracted with Arthur D.
Little, Inc., Cambridge, Massachusetts for technical studies of high-level
radioactive waste disposal using deep geological repositories.  Directive
of Work No. One under this contract, issued in June 1977, required that the
contractor, in conjunction with the EPA Project Officer, select the Ad Hoc
Panel of Earth Scientists comprised of nationally recognized experts in the
fields of geology, geochemistry, geohydrology, rock mechanics and other
applicable disciplines in the earth sciences.  The Panel's basic charge was
to perform an independent evaluation of the adequacy of the state of know-
ledge in the earth sciences for reliably estimating the environmental impacts
to be expected from the disposal of radioactive waste in deep geologic forma-
tions, to provide guidance to EPA regarding the uncertainties inherent in
estimates based upon such knowledge.  This report was prepared by the Panel
in response to its charge.

The Panel conducted its work and formed its conclusions independently of
Arthur D. Little, Inc., EPA, or any other agency.  EPA and its contractor
provided only administrative services and such procedural assistance as the
Panel requested.

-------
MEMBERS OF AD HOC PANEL OF EARTH SCIENTISTS

Dr. Bruno Giletti, Co-Chairman
Department of Geological Sciences
Brown University
Providence, Rhode Island

Dr. Raymond Siever, Co-Chairman
Department of Geological Sciences
Harvard University
Cambridge, Massachusetts

Dr. John Handin, Director
Center for Tectonophysics
Texas A & M University
College Station, Texas

Dr. John Lyons
Department of Earth Sciences
Dartmouth  College
Hanover, New Hampshire

Dr. George Finder
Department of Civil Engineering
Princeton  University
Princeton, New  Jersey
                                    vi

-------
                                PREFACE

 In August of  1977,  a panel of geologists, geochemists, and  geophysicists
 was brought together by Arthur D. Little, Inc., under assignment by  the
 Environmental Protection Agency (EPA)  to evaluate  the state of knowledge
 in the earth  sciences  relevant to environmental aspects of  the disposal
 of high-level radioactive wastes  (HLW), by deep burial on the continents.
 This report presents the results of the panel's work in 1977.

 This report contains our evaluation, based on the  available written  reports
 of individuals and  research groups in  government,  industry, and the  uni-
 versities; as well  as meetings of the  panel, subgroups of it, or individual
 members with several individuals currently active  in this field.  We have
 met as a group with members of the Arthur D. Little Staff and the Office of
 Waste Isolation  (OWI) at Oak Ridge.

 We have tried to find out the nature of the pertinent efforts at other
 government agencies and their contractors, including groups of the Depart-
 ment of Energy (DOE), formerly the Energy Research and Development Agency
 (ERDA) and the Nuclear Regulatory Commission (NRC), Arthur  D. Little,
 Battelle Northwest  Laboratories (BNWL), the United States Geological Survey
 (USGS), Sandia Laboratories, Los Alamos Scientific Laboratory, and Lawrence
 Berkeley and Lawrence Livermore Laboratories (LBL and LLL).  We do not claim
 to have made a detailed and comprehensive survey of all of  this work; neither
 time nor manpower allowed that.

 We have attempted to assess the research that has been or is being done, or
 that is planned, so as to identify what needs most to be known in the earth
 sciences in order to permit reliable prediction of the behavior of HLW
 following deep burial.  We do not make judgments on the merit of any specific
 research effort of  any organization or individual.  Our aim is simply to find
 out what is known and what is not.

We have taken as our charge an assessment of the state of the art of the
 geological sciences that relate to the disposal of radioactive wastes at some
 depth under the land surface.  How well do we know the physical and chemical
 characteristics of  the rock materials at depths comparable to those en-
visioned for waste disposal?  How reliably can we calculate such diverse
 quantities as rates of chemical reactions around waste containers and hydro-
 logic flow rates in low-permeability rocks?  Can we predict the likelihood
of changes  in the  current geologic regime that might affect the storage
 site in the future?

We include in our charge judgement of the confidence with which we can
predict the extent  of transport of radioactive materials and how this
 relates to certain kinds of geologic formations in various tectonic
 and surface environmental settings,  but we do not assess the merits
of any particular site.  We consider three phases of the problem:

     (1)   what we can do now,
                                   vii

-------
     (2)  what we can find out within the short term—a year or two—
          that will allow us to do much better, and

     C3)  what we can determine only after several years of  well-
          planned, careful research.

Implicit are the questions:  do we know how to formulate the right specific
questions and can we expect to obtain answers of sufficient  accuracy to
suffice for our needs?

Increasingly earth scientists are being called upon to predict future geo-
logic phenomena specifically.  Earthquake prediction is relatively advanced
because of the important research of the past few years; prediction of
climatic change is an actively researched subject.  Short-term prediction
of volcanic eruptions is possible now.  On a less spectacular level, hydrolo-
gists can predict with fair accuracy the behavior of ground  water flow under
various pumping regimes.  Yet most of these predictions are  valid over time
scales of days, or a few tens of years at most.  In contrast, our ability
to predict in any but the most general way over times of thousands of years
is poor, yet this is the interval over which we must forecast if an under-
ground HLW  repository is to be safe for future generations.

Our study is intended to inform the EPA of the nature of the advice that it
is getting and can expect to get from competent earth scientists who are
knowledgeable about the various aspects of, and current developments in, the
background science of the problem.  From such sources the EPA should expect
assessments and predictions that are solidly based upon current knowledge,
not hypothesis and extrapolation from inadequate or inapplicable field or'
laboratory data.  Above all it should be remembered that the Earth—or any
small part of it—is an extraordinarily complex and heterogeneous assemblage
of materials, constantly subject to slowly acting forces.  Attempts to model
its behavior :in this case must rely not simply on the best data that are
readily available but on sufficiently good data that are truly pertinent
and on a consideration of all of the important factors that may affect the
outcome.  In any  case, we must know the reliability of any estimates before
we can use them as a basis  for practical decisions.

In assessing the  possibility of loss of integrity of a disposal site, we
have  considered that  total  containment is desirable.  Realistically, however,
some  loss of  radioactive material, albeit small, appears likely.  We have
addressed the  question  of  estimating  the amount of  contamination  to the
biosphere.  It is worth noting that estimates  of  (and regulations  concerning)
maximum permissible  levels  of  radioactivity  in different phases of  the biosphere
have  been lowered over  the  years since World War  II, and we must  assume that
this  trend will  continue into  the  next few centuries.

In preparing  this report we have been assisted by the many  individuals who
have  talked with  us,  sent  us reports  of research  in progress,  and  informed
us of where work  is  going  on.  We  have consistently found a high  degree of
cooperation, which has made this  task easier,  and for which we are  grateful.
                                    viii

-------
We acknowledge the important contribution of Arthur D. Little, Inc., in
bringing us together and informing us of their work, facilitating our task,
and seeing our report through publication and submission.   We alone, however,
are responsible for the contents and conclusions of this report.
                                    ix

-------
PART  I;   INTRODUCTION AND STATEMENT OF THE PROBLEM

The impetus  for  addressing  the questions we seek to answer is the necessity
for long-term storage or disposal of various kinds of long-lived radioactive
wastes that  already have been generated by nuclear processing plants and
reactors  and that will be produced by current and future nuclear power
reactors.  We consider high-level wastes of two categories:

      (1)   those  generated by the reprocessing of spent fuel rods, and

      (2)   the unprocessed spent fuel rods themselves.

Wastes of  both kinds must be stored in such a manner that leakage from the
deeply buried repository to the biosphere will be inconsequential to public
health over  the  times during which the stored materials remain sufficiently
radioactive  to be hazardous.  Thus we consider a maximum time scale of
hundreds  of  thousands (10^) to a few million (10^) years, but pay special
attention  to the dangers of leakage over shorter terms up to a thousand
years, when  the  levels of radioactivity are highest.

Unlike ordinary engineering problems, there is no experience with long-
term, sealed underground storage of such materials, and thus no foundation
of empirical knowledge upon which to build.  Almost all of our experience
with underground openings relates to technologies designed for removal and
extraction, such as mining and pumping for water, oil, or gas.  There is
some limited knowledge from the underground storage of gas, oil, and water.
However, small leaks of these materials would not be significant and,
therefore, are not studied.

The engineering design of underground openings such as those envisioned
under various HLW storage plans is like nothing ever developed previously.
This basic difference stems mainly from the addition of radioactive heat
to the normal geothermal temperatures at the relevant depths on the order
of 500 m.

We assume that a safe underground repository requires a volume of rock in
a stable geologic environment.   We envision storage in an area that is
subject neither to frequent, high-energy earthquakes nor to volcanic eruptions,
Variations in the hydrologic regime due to climatic variations of rainfall
and evapo-transpiration must not jeopardize the safety of the biosphere
from HLW contamination.   Above all, we assume that a rock environment is
required that is scalable and has a minimal permeability for fluids and
the radionuclides that might become dissolved in them.

The rocks that now seem to be most suitable are salt, shale,  basalt, and
certain rocks of the granitic clan, but anhydrite and some impermeable
tuffs may be worth considering.  Other common rock types,  owing to their
permeability or chemical reactivity, are not worth serious investigation
at this time.

-------
A central consideration for a great many questions is the type of waste
to be stored.  Until recently storage schemes have been predicated on the
reprocessing of spent fuel to HLW and its subsequent combination in con-
centrated form to produce a calcine, ceramic5 or glass cylinder encased in.
a. metal canister.  Small in volume, these canisters develop high temperatures
in the first few years, when short-lived radioactivity is at its highest level.
For the first 10 years or so the material would have to be stored above ground
in order to allow rapid heat loss.  After this time, during the initial period
of underground storage, the canisters may reach temperatures of 300°C or more,
depending on the thermal conductivity of the surrounding rock.  Neverthe-
less, these canisters enclose the HLW in a form that is designed to be
far more resistant to corrosion and leaching by formation waters than
unprotected spent fuel rods would be.

Storage of spent fuel rods presents fewer heat problems:  the radioactive
materials are in less concentrated form and would not be expected to reach
temperatures greater than 100'C.  However, the rate of corrosion of the fuel
rods would be much greater, and great reliance would have to be placed on
the impermeability of the surrounding rock to contain the HLW.  Although
relatively little work has been done on the feasibility of storing spent
fuel rods, current federal policy favors this option.

Reprocessed HLW or unprocessed spent fuel rods both contain a mixture of
short- and long-lived radioactive elements of two kinds, in addition to
the unconsumed 238u an
-------
years they become marginally reliable; and for times greater than a million
years they become so poor that they become little better than guesses.  Of
course the reliability also depends on where the repository is with respect
to the tectonic framework of the continent.  We can be much more confident
of the stability of old granites on a Precambrian shield than of young
basalts near  a tectonically active continental margin such as the Pacific
coastal regions of the U.S.

Finally we must confront the issue of retrievability of the HLW over a
specified time period.  Because high-grade uranium ore resources are limited,
it may be desirable at some future time to recover unprocessed fuel rods and
reprocess them for what may have become much needed Pu and U.  Indeed, it
would appear that a breach of the repository in order to mine the U and Pu
could be a serious problem for the future.

The geological site selection criteria for and engineering design of re-
positories will be strongly affected if retrievability is deemed to be de-
sirable.  It is unlikely, for example, that spent fuel rods could be safely
recovered from a repository in salt more than a few tens of years after emplace-
ment and backfilling, for by then the salt would have completely sealed the open-
ings.  Storage in granite or shale might be more desirable if long-term mechanical
stability is required.  As yet there is no firm policy.  We must consider
the state of knowledge concerning the reopening of underground workings in
the rock types proposed for storage and how such reopenings—or maintenance
of a tightly scalable, open shaft—affect the choice of storage repositories.

Here again we are faced with lack of experience, for in ordinary mining
operations, openings are usually abandoned after working, with no thought
of returning.  Though there has been some reworking of mines in a few places,
these are highly hazardous because of the danger of roof-falls in a deterio-
rated mine.  It is well known that keeping any underground mine open and
clean requires constant maintenance and checking of rooms and entries.  The
deeper the mine the greater the danger of rock bursts and floor heaving,
the more so because accumulated strain in surrounding rock may build up over
a long period of time and then suddenly give way by failure.

Retrieval may only be feasible so long as an active crew is kept at the
repository site, perhaps then for only a relatively short number of years,
5 to 10, while the repository is being filled.  We assume that constant
underground human surveillance for significantly longer periods of time is
unlikely.

In the sections that follow we discuss specific details of the integrity
of the storage assembly itself, the nature of the various rocks in relation
to their suitability for storage, the transport of radionuclides through
rock barriers, the hydrologic regimes, and the probability of disruptive
geological events.

-------
PART II:  THE CONTENTS AND PROPERTIES OF THE CANISTER

Most planning to date for possible designs of HLW storage and disposal
systems deep underground has included several stages of containment in
case parts of the system were breached.  These stages usually include:
the resistance of the waste, in whatever form, to leaching and transport;
the resistance of the metal canister to corrosion or other loss processes;
possible special materials placed immediately around the canister; the
geological formation into which the repository is to be placed; and the
nearby geological formations surrounding the host formation itself.  These
containment stages have permitted computation of a variety of time delays
prior to possible release of the HLW to the biosphere.  The various model
computations utilized have differed in the nature of the materials employed
and their geometries.

The nature of the HL¥-containing "vessel" cannot be defined clearly because
no firm decisions have been made concerning a number of parameters to be
listed below.  By "vessel" is meant the canister and its contents, plus any
packing placed immediately around the canister on out to the undisturbed
rock wall.   The parameters that have not been defined include:

     •  The composition of the canister waste contents.   That is,
                                                     Olfi_    *) *\ ^
        the relative proportions of fission products ("°U +  •J-3U)
        and TRU elements are not defined because it is not known
        if the materials will have been reprocessed, or even if
        spent fuel rods will be placed in the canisters  directly.

     •  The composition and phases of the canister contents.   This
        includes the inert (radioactively)  materials used as "binders."
        Glasses, calcines, and ceramics have all been suggested as
        possible candidate materials into which the HLW could be
        incorporated.

     •  The "mix" and concentration of the  HLW in the canister.
        This  affects the heat production rate and its  variation
        with  time.

     •  The nature of the canisters.   It  is generally  assumed that
        the canisters will have  approximately a  1-ft inside  diameter
        and an  approximate 8-ft  length.   It  is not  clear what mate-
        rial (s)  will be used to  make the  canisters.   Steel,  stain-
        less  steel,  and molybdenum have been named  as  candidates,
        but this is  clearly not  an exclusive list.   Wall  thicknesses
        appear  to have  been considered for  purposes  of transport of
        the contents to the repository only.

     •   The nature of the packing  around,  the canisters.   This has
        included as  candidates:  metal sleeves useful  if  recovery
        at  a  later date is desired;  zeolites  or  clays with high
        ion-exchange capacity; crushed rock of the  type in -which

-------
        the repository is  placed;  this  same rock or other material
        cemented together; some combination of  these.

It should be clear that much of the uncertainty concerning  the basic
characteristics of the repository  resides  in two areas:   policy decisions
have not yet been made and accepted generally,  concerning whether or not
the materials will have been reprocessed,  and whether  or not to provide
for retrievability of the  HLW for  time  scales of decades.  The consequences
of these policy decisions  concerning the mode of HLW disposal might, in
fact, eventually result in revision of  the choices made with respect to the
form of the waste, storage, rock type,  and site.

II.1  THE INTEGRITY OF THE HLW AND ITS MATRIX

Several materials have been proposed for mixing with the HLW so as to pro-
vide a solid composite that would resist leaching or other loss of the HLW
in the event of canister breaching.  Included are calcines, ceramics, and
glasses.  The  possibility  that the leaching solutions would be acid has
argued against the use of  calcines, since  they  are not very resistant to
such attack.

The use  of  ceramics  may be viable, but  not enough work has been done to
demonstrate that.  In view of  some of  the  difficulties with glasses to be
discussed shortly, ceramics might well  be  worth more intensive study.  They
have  the advantage that they are  already crystallized so that devitrifica-
tion is  not a problem.  Questions  concerning their  stability  and  resistance
 to leaching under the local pressure and temperature conditions  of  the  solu-
 tions  that are likely to  exist need to  be  explored.

 Various glasses have been suggested as  suitable products to be  placed  into
 the canisters.  The  ability to make a  homogeneous  phase that  will fully
 occupy the canister  shape, the good compressive strength,  and the moderate
 resistance to solution, all are positive arguments for  the use of glass.
 There are two phenomena that can  generate  serious problems in the use  of
 glasses, however.  These  are that glasses  are  not stable,  but will tend to
 devitrify; and that  it is possible that solutions of  high  ionic strength
 will leach the glass at the ambient temperatures around the canister.

 Two aspects affecting the possible devitrification of glasses have not been
 adequately investigated.   The first is the high radiation flux from alpha
 particles.  These can create considerably more recoils and damage than beta
 particles  and gamma rays.  It is primarily the latter that have been used in
 tests for  devitrification.  With the new  possibility that unprocessed fuel
 rods will  be  used,  or that the HLW might  be made into a glass without prior
 separation of the U and Pu, the dose of alpha particles will increase
 markedly.  The devitrification will also  be enhanced by the presence of
 water,  particularly high  ionic strength solutions, at somewhat elevated
  temperatures.  We are unaware of any tests conducted on any of the proposed
  types  of glasses that show the effects of high dosages  of alpha  particles
  imposed on glass during  immersion, in  likely corrosive  solutions, at the
  various temperatures from 300°C  on down.   It would not  be at all surprising

-------
to find that the integrity of the glass was lost over time scales of a
decade, instead of the millenia that are now computed on the basis of the
results of dry tests with gamma and beta radiation.

The consequences of devitrification are the formation of one or more
crystalline phases.  Such processes commonly result in the formation of
materials with simpler chemical compositions than the glass.  The new
phase excludes impurities.  If the impurities include the class of nuclides
that are the fission products or TRU's, then these will end up in inter-
granular boundary areas.  Leaching of these would then be relatively easy,
and the bulk of the original glass need not even go into solution, but could
remain as relatively non-radioactive crystals.

The other area that requires demonstration is the leaching of glasses, even
if no devitrification is assumed to occur.  Experiments need to be conducted
with solutions at temperatures as high as those that may occur in the can-
ister, clearly higher than 300°C in some cases and possibly in excess of
500°C.  These solutions should include the likely ones for the proposed
setting:  such as concentrated brines, including the variety of bitterns
that exist as fluid inclusions in some rocks.

At this point, the Panel believes that there is no evidence that incorpora-
tion into a glass will ensure resistance to significant leaching over time
scales of a decade.  We wish to make clear that this is an area in which
experiments can be done.  If carefully controlled, such studies should be
able to answer the question reasonably well.

In the event that containment by the glass was necessarily a short-term
phenomenon, the primary defense against migration of radionuclides to the
biosphere would clearly have to reside in the other containment stages
referred to at the beginning of this section.  We assume that it is of
little importance whether the glass releases some radioactive waste in times
of a decade or hour,  since the effect would be essentially instantaneous
compared with the time scales of interest for HLW.   We shall assume,  there-
fore,  that the glass  presents no significant barrier to  transport  by  leach-
ing solutions.

If the spent fuel rods are placed directly into the canisters, or are not
changed chemically (that is, they may be crushed and pelletized), we know
of no data concerning the ease with which such materials will be leached.
It would be expected that the material is so far  out of chemical equili-
brium with the hot solutions that can reach it, however, that a significant
amount of solution would occur.  We see no safe alternative at this point
but to assume instantaneous access to leaching solutions.  The limiting
constraint we see here, as with the glasses, is the degree to which the
individual ions are soluble, either as those ions or as complexes.

In dealing with questions of devitrification kinetics, solution of the
various compounds and glasses, or solution of the spent fuel rod mate-
rials themselves, some items should be noted.  It is the repeated ex-
perience of those of us who work in this area:

-------
     •   that  rates measured  in dry systems often are orders  of
        magnitude slower  than those measured  in wet systems;

     •   that  rates measured  with  pure water or dilute  solutions
        are often far  slower than those  in concentrated solutions,
        particularly if  the  latter are at high or  low  pH;  and

     •   that  complexing  of ions  (by making chloride complexes,  as
        an example)  may  permit  considerably higher concentrations
        of the cation in the solution.

Transport of  ions  in such cases  would clearly be faster for the same
volume of solution.

The use of some data in this area in the mathematical models that have
been constructed for HLW is justified as the only information then available.
However, we  should not lose sight of the major discrepancies possible between
available data and those that are needed to bear directly on what the models
are  trying to determine.

Until  it  is  demonstrated otherwise, we  conclude tentatively  that a loss of
integrity of the HLW matrix in a repository would result within  a very short
time of the  emplacement.  This short time could well  be less than a decade.
In view of the need to maintain  isolation for time scales  of centuries to
one  million  years,  it appears safest to consider  the  materials in the canis-
ter  primarily from  the  aspect of what is being  dissolved  and transported
rather than  as barriers  of  any  consequence.

Some comments are  in  order  concerning the nature  of  the solutions that might
attack the canisters  and their  contents. It is clear that the rock  type
will affect  to  some extent  the  nature of the solutions.

 A frequently discussed  candidate lithology  for  a repository is  salt.   Ob-
 servation in salt mines has shown that  chambers cut  into  some  salt  deposits
 remain dry  indefinitely, suggesting the long-term safety  from water that is
 desired for HLW.   Closer inspection of  salt, however, reveals that  the
 crystals do contain significant amounts of  water as fluid inclusions and
 along  intergranular boundaries.  Work done by Roedder and by Stewart (to
 be  discussed in Part IV) suggests that  a potential hazard exists here.  The
 fluid  inclusions in the salt decrepitate (burst on heating) at comparatively
 low temperatures, which means that they are reasonably certain to do so in
 the vicinity  of the canister as the temperature rises following emplacement.
 Decrepitation is quite  likely to occur  at temperatures in  the general
 vicinity of 150°C.   If  we  assume that  the wall temperature  of the canister
 will  reach  300°C,  a  significant amount of water might be  available.  This  is
 most  likely in bedded  salts where water may be in excess  of 1%  of the rock.
  It  becomes  imperative  to determine  if  similar  amounts of  water  exist in  the
  salt  of salt domes.

-------
The process of migration, of water up thermal gradients in salt has been
described numerous times.  It is cited here to make the point that the
canister is likely to be bathed in water soon after emplacement.  This
water would, of course, be saturated with salt.  At the temperatures en-
countered at the wall, the concentration is considerable (as will be
discussed).

The availability of water in shale deposits is considerable.  This is
water normally tied up as water of hydration in the clays or as OH ions
in the structures of the clays and micas.  The high temperatures at the
canister wall will suffice to release some of the water.  Again, the water
will contain a variety of salts in solution.  These will derive from the
cations in and on the clay particles.  As a result, the composition will
differ from that of an evaporite unit in which the salt unit itself will
tend to dominate the composition.  If the salt is largely halite, so will
be the solution.  In the case of the shale, the solution may derive from
salts that were present during deposition and diagenesis, plus the con-
tribution from the particular clays in the shale and their degree of dia-
genetic change.

The granite clan rocks and other igneous rocks will contribute water from
two sources.  The rock itself is slightly permeable, and there will have
been long times available for the water to enter the rock as it would in
any other permeable rock.  Most crystalline rocks of this sort are not
totally massive, but have joints and shear zones on numerous scales.  These
will again permit water to enter.  In this connection, it should be noted
that much of the water obtained in New England from "artesian" wells drilled
in these crystalline rocks is actually from fissures in the rock.  The rock
itself is relatively impermeable if a particular specimen is tested, but
the unit of rock is much more permeable owing to these fissures.  In the
event that HLW is stored or disposed of in such rocks, it is not clear what
the maximum surface temperature of the canister would be.   The thermal con-
ductivity of both these rock types is less than that of salt by a factor of
about 3, a difference suggesting some higher temperature unless different
concentrations of the fission component of the HLW were used.

A source of water from minerals themselves would not be a problem in the
"anhydrous" rocks such as basalt or the granulite facies rocks such as
charnockites or anorthosites.  These may contain quartz, feldspars, and
pyroxenes, in various proportions.  Basalts are common, but the granulite
facies rocks are more rare.

Any of these crystalline rocks, when heated in the presence of water, will
contribute ions to form a solution, the composition of which will depend
on the rock composition, including its fluid inclusions.

II.2  THE INTEGRITY OF THE CANISTER

The canister into which the HLW would be placed is usually described as a
cylinder of approximately 1-ft inside diameter and 8-10 ft in length.  It
is assumed that it would be made of some metal, that it would be sealed by

-------
welding, and that it would be able to withstand both the high temperatures
that would result when the molten glass or other molten material was poured
into it, and the stresses during transport to the burial site.  The wall
thickness is not known to us, but would presumably be on the order of
1/4 inch to perhaps 1 inch.

In all cases, a pure metal or an alloy appears to be the choice for the
material.  Suggestions include:  steel, stainless steel, molybdenum,
titanium, and copper.  Each has different structural advantages and dis-
advantages, which are not part of our concern.  The resistance to corrosion
does, however, concern us.  We know of no tests that have shown that any of
the candidate metals will resist corrosion by the salt solutions that are
likely to be at the canister surface for a significantly long time.  Again
we stress that the solutions will be at temperatures of approximately 300°C,
will have a high ionic strength, may have a very low PH, but probably will
have very low starting concentrations of the metals of the canister.  Under
these circumstances, it  is likely that the canister could be breached within
time scales of a decade  or less.

For this reason, we do not consider the canister to be a significant barrier
to the solutions, at least for the time scales of centuries to a million
years with which we are  dealing.  In passing, we might note that it is not
essential that all components of the canister be resistant to leaching or
attack.  The optimum canister may prove to be one that is highly effective
as a container and transport vessel and one  that does not enhance attack
by solutions on  the canister contents by adding corrosive materials to the
solution that will ultimately breach the canister.

II.3  THE INTEGRITY OF THE CANISTER SURROUNDINGS

According to our nomenclature, the immediate surroundings of  the canister
are part of  the  "vessel."  That  is, the vessel extends from  the  canister,
through  the packing around the canister, and out to  the undisturbed rock.
If  the  canister  is  to be retrievable,  the packing would include  a metal
sleeve  to permit easy withdrawal of  the  canister.  Around this might be
packed  crushed  rock  similar  to the host  rock of  the  repository  or  some
other material.

The metal  sleeve would doubtless behave  in  a manner  similar  to  the  canister
metal.   It  is  assumed  that the same metal would  be used, since  the  use  of
 dissimilar  metals  could  well set up  an electrolytic  cell, with  the  con-
 sequent possibility  of H2 production and  rapid corrosion of  the metal walls.
 The comments above  for  the canister  apply  to the sleeve material as well.

 Any packing would  have  a porosity and permeability  that  are  much greater
 than would the undisturbed rock.  If  grouting materials were used to  cement
 the packing materials,  the result would  be  a less  porous  and permeable
 packing, but the grouting would  still be susceptible to  attack by the
 solutions.
                                     10

-------
The variety of possible packing materials is infinite, and it remains to
be shown that the selected material would enhance the resistance of the
"vessel" to attack by solutions.  The use of zeolites or other ion ex-
changers might help in the chemisorption of the HLW as it moved through
the packing, thus delaying the migration of the radioactivity.  The amount
of such packing might not be sufficient to make a significant difference,
however, because zeolites lose their water (and so their cation-exchange
capacity) at temperatures well below 300°C.  To be effective, the zeolite
packing would have to be far enough from the canister to remain cool.

Based on the above considerations, we assume that the near-in stages of
containment cannot be relied upon to effect any significant retardation
of the release of the HLW.  We include in the near-in stages all the com-
ponents of what we call the "vessel," that is, the HLW in its matrix, the
canister, any sleeve around the canister, and the packing around the
canister or sleeve.  By significant retardation, we mean for times longer
than a decade.
                                    11

-------
 PART III;  MECHANICAL PROPERTIES OF ROCKS

 The introduction of a repository into a geologic unit poses a number of
 mechanical requirements on the rock:  the need for sufficient strength
 to allow safe excavation and occupancy until the repository has been
 sealed; mechanical integrity despite the subsequent high temperatures;
 low permeability; and absence of discontinuities like jointing or bedding;
 or a very small number of these.  Knowledge of the mechanical properties
 for the various candidate lithologies varies considerably and some un-
 certainties remain for all rock types.

 The Arthur D.  Little report on "Assessment of Geologic Site-Selection
 Factors1^1' outlines problem areas fairly well,  but it neither assesses
 nor criticizes the state of the art that is currently available to re-
 solve these problems.   Nor for that matter does  any other recent document
 known to us except for the NAS (National Academy of Sciences)  draft report
 on "Rock-Mechanics Limitations to Energy-Resource Recovery and Development,"
 which should be published in early 1978. (2)  ye  concur with the conclusion
 of its  Subpanel on Nuclear-Waste Disposal that "present rock-mechanics
 technology is  not sufficient to predict  the full range of thermal effects
 on site containment caused by the emplacement of heat-generating radioactive
 waste.   The Subpanel believes, however,  that the improvements  in technology
 needed  to make such predictions within defined and acceptable  bounds  of
 accuracy can be acquired with the proper research and development effort."

 The engineering design of mines and other underground works  is largely
 empirical,  but it is firmly based on many hundreds  of years  of practical
 experience.  Except for  rock bursts in deep mines  in  strong  rock in which
 residual stresses can  be very high  (as in South  Africa),  accidents  that
 threaten human life rarely occur these days because of  failure of  the  rock
 itself.   (The  risks of flooding and explosion of accumulated gas  remain
 all too  high.)

 The expectable configuration  of  a repository  is  essentially  that  of a  room-
 and-pillar mine.   In practice,  excavation  is  designed  for the maximum  rate
 of  extraction  that  is  allowed by  the short-term stability of the roof.  In
 salt mines  the  rooms may  close  eventually,  but gradually and non-cata-
 strophically,  long  after  they have  been worked out  and abandoned.   In
 principle, long-term stability  can  be achieved by increasing support—
 reducing  the open area relative  to  that of  the pillars and, if necessary,
 providing lining  and other reinforcement.   Thus,  the state of  the art is
 such that a mechanically safe cavity could  ordinarily be placed in virtually
 any rock, provided  only that cost effectiveness be judged not by the value
 of  the rock extracted but by the critical need to isolate high-level waste
 from the biosphere.

The facts are,  however, that the design of an HLW underground repository is
not ordinary, and there is no fund of previous experience because the wall
rock will be subjected to temperatures greatly exceeding those due to the
average geothermal gradient.  The temperature will depend principally on
the thermal conductivity of the rock and  the thermal load imposed, which


                                    13

-------
depends in turn on the time allowed for heat dissipation during previous
surface storage and the size and distribution of canisters in the floor
of the repository.  Temperatures as high as 300°C should be expected in
a highly conductive rock like salt.  Heating to as much as 500°C might
well occur in a relatively poor conductor like dry granite.  Our knowledge
of the mechanical properties of rocks under long-term loading in the labo-
ratory at the relevant temperatures (300-500°C) and confining pressures
(100-200 bars) is fairly good for rock salt.  It is practically nil for
all other rock types.

Much experimental work has been done on the high-temperature creep of
single crystals and artificial aggregates of salt, and work on natural
samples  from the  Waste  Isolation Pilot Plant site  is progressing at RE/SPEC
 (a consulting firm  in Rapid  City,  S.D.) and Sandia Laboratories.  The  labo-
ratory data on flow laws  are being incorporated  into non-linear numerical
models in order to  predict the  behavior of  cavities in  salt.   Although
 extrapolation of  laboratory  data must  still be  tested and the  codes must be
validated by careful measurements  in  the  field,  our ability  to predict the
 formation of salt should  soon be adequate, given the current  level of  re-
 search effort throughout  the world.

We are far indeed from the capability  to  predict,  with  sufficient accuracy,
 the behaviors of  any other rocks because:

      (1)  data on mechanical properties under  relevant  conditions
           are lacking and

      (2)  most rocks cannot be treated as continua because dis-
           continuities like jointing  and  bedding are  nearly
           ubiquitous.

 Because the need for underground isolation of HLW has  been recognized for
 some 30 years, the long postponement of pertinent research on rock other
 than salt is unfortunate.  The problems will not be .solved quickly.  The
 research is  inherently time-consuming because the critical data are attain-
 able only from creep  tests of months-long duration.  Furthermore, the required
 testing machines (to accommodate 10-cm specimens,  at temperatures of 500°C,
 under pressure of 200 bars, and for a duration of several thousand hours) do
 not even exist.  It may take a major research effort of  5 years to build  the
 necessary  laboratory facilities;  to collect adequate data; to develop real-
 istic,  three-dimensional, non-linear, large deformation  codes; and to vali-
 date  predictions in  the field.

 In addition to the  strength of rock, which relates to  long-term mechanical
 stability,  we must  know a lot more than  we do  about thermo-elastic expansion,
  thermal conductivity,  and permeability,  particularly as  they  are affected by
  thermal cracking.   These  parameters relate not  only to mechanical but also  to
  hydrologic stability.  A  modest experimental effort is under  way at Terra Tek,
  the U.S.  Geological Survey, and a few universities and DOE  laboratories.
  Several heater tests in  the field are in progress or planned  for the  immediate
  future  (Avery Island salt dome, RE/SPEC; granite, Stripa mine,  Sweden, LBL;

-------
 granite,  Climax  stock, Nevada Test Site, LLL; basalt, Hanford, LBL;
 Conestauga shale, Oak Ridge National Laboratory, Sandia; Eleana shale,
 Nevada Test  Site, Sandia; norite, Sudbury, Canadian Government).  Valuable
 data on thermo-elastic deformation and permeability may be obtained in situ
 where they count.  However, serious questions can be asked about the choice
 of sites  and the scale of these tests, relative to the natural joint spacing.
 Often site selection seems to have been either geologically naive or directed
 politically  rather than technically.

 We feel compelled to emphasize, once again, the significantly different con-
 sequences of the alternatives of disposal and storage of HLW.  Must the
 facility  remain stable only for a decade while it is being filled?  Can the
 cavity then be back-filled with rock and the access shafts permanently sealed?
 Must it stay open for several decades, perhaps indefinitely, so that spent
 fuel rods can be retrieved and ultimately reprocessed so as not to lose a
 valuable store of energy; so that mistakes could be corrected should isola-
 tion be incomplete; or so that canisters could be moved should the site fail
 altogether?  A weak, ductile medium like salt should be mechanically ideal
 for permanent disposal but might be poor indeed for long-term storage with
 the option of retrievability.  Salt will certainly flow readily at 200°C
 under expectable in situ stresses.  Canisters will have to be isolated from
 the surrounding rock, and artificial support of the cavity would have to be
 provided.

 Excavating a strong rock like granite is much more expensive, and because it
 is brittle and always jointed to some degree, it may not be as good a candi-
 date for permanent disposal.   On the other hand, granite, basalt, or other
 strong rocks seem preferable for storage, at least with regard to long-term
mechanical stability.

 In summary, we shall know enough in a year or two to compute the consequences
 of HLW disposal in dry, homogeneous, ductile salt that would be mechanically
metastable for at least a decade, perhaps indefinitely if we are to pay the
 price for artificial support.  On the other hand, we know very little indeed
 about either the long-term strengths of jointed, brittle rock masses at
 temperatures on the order of  500°C, or the effects of thermal cracking on
 the flow of fluids through them.
                                   15

-------
  PART IV;   CANDIDATE LITHOLOGIES

  Several different  rock types  have  been  discussed  as  possible  host  lithologies
  for  the HLW repositories.   Important  candidates are:   salt, shale,  granitic
  clan rocks,  and  basalt.  Because of the undemonstrated security  of  the assem-
  blages  of  materials in and  immediately  around  the canisters,  the containment
  by the  rock  that is the host  of the repository becomes of paramount importance.
  This is probably the last stage of containment of the  HLW prior  to  entry into
  circulating  ground  waters and contact with the biosphere.

  We do not  attempt  to review the properties of the different lithologies.
  Rather, we focus on the physical and chemical properties that bear directly
  on the question  of  HLW transport.

  The  first  category  discussed is salt.   This occurs in  two ways in nature,
  as bedded  salt deposits in a sedimentary sequence, and as salt domes.   Be-
  cause their  characteristics are quite different,  they are treated in dif-
  ferent subsections.

  IV.I  SALT BEDS

 Bedded salt deposits are commonly believed by those unfamiliar with them
  to be massive, thick, pure beds of the mineral halite  (NaCl),  and this is
 the usual basis for model studies.   Though there are a few local places
 where such beds can be found,  most salt beds  are more typically somewhat
 thin-bedded, may alternate with other evaporite deposits such  as anhydrite
  (CaS04)  or other salts (of the KCl-MgCl2 groups),  and are frequently
 separated by thin beds of shale.  Salt is commonly interbedded with lime-
 stones or  dolomites or may interfinger with them laterally.  Salt beds may
 also  grade laterally into sandstones.

 On a  microscopic scale, it  can be  seen that many  halite crystals  contain
 liquid (brine) or solid (anhydrite) inclusions.   A recent report  by Roedder
 and Belkin estimates the volume  of  these inclusions to  be less than 1% but
 "an additional possibly even greater volume %  fluid is  present ^.n situ,
 filling  intergranular pores."W

 There is abundant evidence for post-depositional recrystallization, dif-
 ferential solution,  redeposition in cross-cutting  veins, and mass flowage
 in evaporite  deposits.  The textures produced in salt beds by  these processes
 increase the  heterogeneity produced by depositional processes.   There is also
 ample evidence of solution channels being established along veins and litho-
 logic breaks.

 Because salt  is so soluble it rarely crops out at  the surface;  thus its
 stratigraphy  and detailed lithology are mainly known from drill hole records.
 Though individual distinctive laminae have been traced for long distances
 through the study of  drill cores, most of our information comes from geo-
 physical logs of drill holes or shallow seismic profiles.  Neither is ade-
 quate to show the amount and number of thin shale, anhydrite,  or limestone
breaks in salt beds.  For this  reason,  site testing needs a carefully de-


                                    17

-------
signed drilling program that includes provision for continuous coring that
would detect all lithologic breaks.  No geologist can predict the presence
of a truly homogeneous bed of salt 30 m thick without such a drilling cam-
paign plus detailed, including microscopic, study of the core samples to
determine their brine and solid inclusion content.

Realistic estimates of water and brine content of salt beds are critical
to proper estimation of the fate of canisters sealed in a salt repository,
for, as indicated earlier, such hot briny solutions may effectively corrode
the metal canister, leach the HLW matrix, and transport dissolved radio-
nuclides.  Water content is also important in affecting the creep rate of
salt.

Recent experiments in the Experimental Geochemistry and Mineralogy Branch
of USGS by D. B. Stewart and others(4) have shown that at 200°C, water
in a rock salt including potassium and magnesium components can contain
approximately 70 wt % of dissolved salts.  This implies that as little as
1 wt % H20 in a salt bed may yield fluid that is 3 wt % because of the high
concentration of the solution.  The dissolving power of such brines should
not be underestimated.  We know from many hydrothermal experiments that be-
cause of changes in hydrolysis constants and the association of ion pair
complexes, highly saline solutions at elevated temperatures are highly
acidic—the  pH can be as low as 2.

It should be noted that the pressure-temperature conditions near the canister
may result in the brine becoming two phases (liquid and vapor), which could
retard brine access to the canisters.  The activity (in a chemical sense) of
the brine components would still be the same in the two phases, however, and
the liquids  would still act as buffers to maintain the corrosive attack on
the canisters.

These solutions would strongly attack any metal and effectively leach glass.
There is also the possibility that permeation of these solutions along glass
grain boundaries and incipient fractures would speed up devitrification
which would  then feed back positively to enhance leaching.              '

Because of the high density of the canister, it might sink  in  any salt bed
that contained  the  above  quantities of saline water solutions, being corroded
and  leached  as  it moved downward.  Many beds of salt overlie permeable lime-
stone formations and the  canisters might end up there, in the  path of ground
water movement.  Any lithologic breaks in  the salt might impede that movement
depending on the thickness and the rock strength.  Bu? solution movem™y
also be  enhanced along  those very  breaks.                        luvcmcuu uiay

We emphasize the importance of knowing the water  content of salt beds proposed
for  repositories,  particularly with  the background and the  experience at Lyons,
Kansas, where considerable volumes of water migrated in an  unpredicted manner.
That problem arose as  a consequence  of dissolution of salt  by  ground water
seeping  into the repository.   Seepage was  along an abandoned  drill hole  that,
like most, had  not been cased  and  plugged.  This  puts a premium on picking  a

                                     ^^ dr±U ^S °*  °ld -S.rg.oSnd


                                    18

-------
 Though remote sensing techniques seem likely to be able to spot some sur-
 face manifestations of these underground penetrations to supplement state
 and company records, it remains to be proven by ordinary double-blind
 experiments that such detection is highly accurate.  In this connection
 it is necessary to prove not only that holes identified from aerial photo-
 graphs are in fact there, but also that all holes actually have been identi-
 fied from the photograph—no mean task.   If this assurance cannot be made,
 then modellers of repository behavior will need to take into account the
 probability of leakage of water into salt through such old holes.

 Any abandoned drill holes discovered may disqualify a site if they cannot
 be properly plugged and sealed.  If the well has penetrated the salt there
 may be a cavernous dissolved region around the hole.   In addition, most such
 wells will show extensive caving of weak rocks in the section,  which might
 provide an efficient vertical transfer path among various  aquifers.   Since
 metal casing cannot be guaranteed against corrosion over the time scales
 we have to consider, some impermeable and chemically inert cement to grout
 and plug the hole must be developed.   This indeed may be an important re-
 quirement for any drill holes designed to explore a site for a  future
 repository.

 IV.2  SALT DOMES

 Of all the geologic media suggested for  the underground  isolation of  HLW
 from the biosphere, the rock salt  in  the dry domes of the  Gulf  Coastal
 Plain seems  mechanically best for  disposal (permanent isolation by back-
 filling and  sealing)  but  probably  not for long-term storage  and ready
 retrievability.   Room-and-pillar mining  in salt  is cheap relative  to  that
 in hard rock.

 Given the time  frame  of  a million  years,  one cannot  claim  that  the proba-
 bility of any geologic  event is absolutely zero.   However, the  risks of
 catastrophic  events like  destructive  earthquakes  and  volcanism,  or of
 sudden,  unpredictable changes  in the  secular rate  of  erosion, or of con-
 tinental  glaciation are evidently  extremely  small  in  the coastal regions
 of  Texas  and Louisiana,  certainly  as  unlikely as  they would be  anywhere
 in  North  America.

 Once a  critical  thickness  of overburden had  been achieved, Gulf  Coast domes
 grew more or less concurrently with the deposition and loading of  the sedi-
ments above them.   Since the rock over onshore domes  is now being eroded,
 further growth  ("diapirism") should not be expected.  There is no obvious
reason why the coastal plain should not remain tectonically stable for at
 least another million years.

Below the superjacent cap rock, the lithology is remarkably uniform.   Al-
 though bedding may have been contorted (it is not always so), the rock is
typically free of open fractures.  Although its lateral extent is limited,
the salt is several kilometers thick, and it should be possible  to identify
a repository site at least 1000 m from the closest water-bearing strata.
                                   19

-------
Some (though not all) salt domes appear to be remarkably dry and to have
been so for some millions of years.  Professor Robert R. Unterberger of
Texas A&M University has been probing domes with 40-cm radar.  In a dry
salt mass like that of Grand Saline in northeast Texas, he is able to map
reflections at the boundaries with surrounding water-bearing strata, as
much as 2 km distant from the transmitting antenna.  He could not do so
if the average water content exceeded on the order of 0.1%.  (In the "wet"
Weeks Island dome, on the contrary, attenuation of the signal is so high
that no reflections are recorded, and he turns to sonar.)  In a dry dome
no lateral migration of ground water appears to have occurred over several
millions of years.  It is hard to imagine how any ground water could reach
the repository and, hence, how any radioactive elements could reach the
biosphere.

In a dry dome, the bulk density of the encapsulated HLW would presumably
exceed that of the salt, so that containers would  tend to migrate downward,
all to the good,  even if the high-temperature viscosity of the salt were
low enough to permit significant movements over the relevant span of time.
Thus disposal in  salt domes is likely to be permanent.

We do not know whether the  sinking will tend to "focus" the  canisters, al-
though this effect should be amenable to  computation.  By focusing we mean
that the originally horizontal array will have not only a temperature
gradient that decreases  away from  each canister site, but also a higher-
than-expected local  temperature at the center of  the condensed array.  As
the  canisters sink in the salt, will they be "refracted" inward toward each
other?   If  they  are, the central temperature would rise, leading to still
 faster sinking  and  focusing until  very high  temperatures indeed might be
reached.   This  possibility  should  be studied.  There is also an important
 corollary:   additional  production  of HLW  by  sub-critical or  even, conceivably,
 critical activity with  the  U,  TRU, and water.  We do not suggest that  this
will happen,  but rather that  the  question be properly  addressed.  Enough  is
now known to permit  a  computational  assessment.

 Salt is  a valuable  mineral  resource, but  worldwide reserves  are virtually
 limitless,  so there is  no logical  reason  why a  single  dome  should not  be
 perpetually withdrawn  and dedicated   to HLW disposal.   The  possibility of
human intrusion at  a far future time cannot, however,  be discounted.

 IV.3  SHALES

 Just as  most salt beds are not pure  and monolithic,  shales are rarely  homo-
 geneous  and uniform.  A typical shale is  a rock that is more or less  silty
 and contains occasional interbeds or laminae of permeable  siltstone or
 sandstone.  Other shales may be interbedded with limestone or dolomite or
 grade into a calcareous shale and then into a shaly  limestone.

 To varying degrees shales may be thinly bedded or show fissility,  the
 ability to break into thin sheets.  They may also be cut by systems of
 joints and fractures,  and, where the shale is traced into areas where the
 rocks are metamorphic, may show incipient fracture cleavage.  It is just


                                     20

-------
now becoming known that fracture cleavage in many shales is related to
original conditions of sedimentation and early diagenesis that make it
susceptible to breaking along certain zones.  In general, the older and
more physically and mineralogically altered the original mud becomes, the
more brittle the resulting shale becomes.

Since approximately 70% of the sedimentary rocks of the earth are shales,
it is not surprising to find a great many thick shale sections.  Yet it
is extraordinarily difficult to find one that is uninterrupted by inter-
beds of other, generally more permeable, lithologies.  As in the case for
salt beds, it is difficult if not impossible to get an accurate detailed
log of a vertical section of most shales from outcrops because of their
rapid weathering in most climates and topographies.  At the same time,
geophysical methods are inadequate to pick up fine interbeds and inter-
laminations.  As thin as these may be, frequently on the order of a few
millimeters in thickness, there is good geological evidence that there
has been sufficient water flow through these laminae to alter the clay
mineralogy in the sandstone whereas the more impermeable shale is unaltered.
Hence a requirement for a shale repository must be the acquisition and care-
ful study of drill cores.

Shales are particularly susceptible to mineralogical alteration that pro-
motes physical change if they come in contact with solutions of a different
kind than they were naturally bathed  in at depth.  Shales are complex mix-
tures of quartz, feldspar, several varieties of clay minerals, zeolites,
carbonates, sulfides and oxides.  Any significant chemical reactions of
these minerals are likely to weaken the physical structure and promote
cracking and disintegration at the relatively low confining pressures of
a repository.  All of these reactions are promoted by increase in tem-
perature.

Before any detailed plans for a shale repository can be drawn, the strati-
graphic and mineralogical-geochemical work has to be done.  The detailed
stratigraphy and petrography we know how to do, but the study of how altered
mineralogy affects physical structure is in its infancy.  Long-term heating
to 300-500°C is likely to induce mineralogical changes including dehydration
of smectites.  This dehydration could liberate significant quantities of
free water and induce textural changes that might be reflected in fracture
patterns, which in turn affect the fluid permeability of the formation.  We
cannot as yet predict such behavior of a specific complex mineral assemblage
from general rules.

The USGS has done a fair amount of work on the merits of the Pierre
shale of the Dakotas as a possible repository.  The stratigraphic and
mineralogical-chemical knowledge of the Pierre shale is possibly greater
than that of any other shale in the world.  We know that there are thick
shale sections in it and that it has a generally stable mineralogy.  On
the other hand, as much as we know, we still are unable to predict with
any high probability what the fluid permeability is of such a section after
long-term heating by buried HLW.  And we are not able at this time to de-
fine how rare and how thin occasional interbedded silt laminae would have
                                    21

-------
to be to prevent significant fluid migration in or out of the formation.
At this time we could not rule out extensive fluid migration via fracture
permeability following a 10-20-year period of moderate heating.

IV . 4  GRANITE

Because they are widespread, underlie large regions of the mountain belts
and shield areas of North America, are relatively homogeneous, have high
crushing strengths (up to a few kilobars at low confining stress) ,  low
porosities and, therefore, low pore-water contents (0.01 to 0.03) 'and low
permeabilities (10-7 to 10-17 Cm2), granitic rocks have good potential for
development as high-level waste repositories.  Large granite plutons range
in thickness from 2 km to 15 km and, unlike sedimentary rocks  are  not
likely to have aquifers at depth.  Retrieval of HLW canisters would be
easier in granite than in most other rocks.  However, granites are  brittle,
and we currently know neither their behavior under thermal stress nor their
sorptive properties.

Many of the important physical properties of granite are site-specific, such
as joint density and state of stress.  Other site-specific features such as
hydrothermal alteration zones, dikes, quartz veins, the irregular three-
dimensional geometry of the granite body itself, and faults will all con-
tribute, along with the joint pattern, to the hydrologic regime.  It is
the Panel s opinion, and apparently that of several foreign countries as
well, that a sizable body of granite underlying a hydrologic basin  of
appropriate dimensions may prove, in the long run, to be an excellent under-
ground repository.  We know of no reasons, as yet, to ^ it out.   Research
on granites should be pushed vigorously, particularly because there may be
either socio-political or geological reasons why burial in salt may be ruled

  "         ^
isn obvious                              ""

IV. 5  BASALT
The presence of tens of thousands of square miles of up to a 5,000-f t-thick
succession of plateau basalts in the northwestern United States  and their
nearness  to the Hanford, Washington and Idaho National Eng^lnTLaSrrt
Fresh massive basalt will have crushing strengths as high or higher than
those of granite, and porosities, permeabilities, and linear thlrmal expan-
sion coefficients that may be as low or lower   Thp nnlT^ ! thermal expan
conductivity for basalt and granite lies    the      rfx 10-3    8 x
10-3  cal/[cm][sec2n°C]), but basalt is likely to be thThigher or the two

Although  the laboratory-measured physical properties of granite and basalt
thus  seem to be roughly equivalent, there are geological differences in their
mode  of occurrence that favor the former rock over the latter.  The typical
basalt flow of the Columbia and Snake River plateaus ranges from 10 m to 45 ni
in  thickness, and is often separated from the overlying and underlying flows '
                                    22

-------
 by an aquifer.   Lower columnar and  upper  fan-type  jointing of  each  indi-
 vidual flow is  characteristic, and  most lavas  have a  5-m  thick vesicular
 zone  at the top,  and  a 1-m thick vesicular  zone  at the base of each flow.
 Thus  it seems  that, on the average, basalt  should  be  far  more porous and
 permeable  than  granite,  and that it would also offer  a higher  risk  of
 contaminating  the ground water if used as an HLW repository.   Advantages
 of some basalts would be their lateral continuity  over hundreds of  square
 miles,  and the  fact that some  of the rocks  are slightly altered and contain
 clays or zeolites, which may enhance the  sorptive  properties of the basalt.
 Unfortunately,  though  there are some data  on  clay minerals and zeolites,
 there are  no published data on the  sorptive properties of well-described
 altered basalts.

 The general feeling of the Panel is that  because of geologic constraints,
 establishing a  safe repository will be more difficult in basalt than it will
 be in granite.  Finding  a  thick unfractured or unjointed basalt flow that
 can be  opened up  and  resealed  without the development of fractures  that will
 communicate with  an actual or  potential interflow  aquifer may  be a  require-
 ment  that  is difficult to  meet.  Again, we  know  enough about the areal
 geology, stratigraphy,  and general  lithologic  character of basalt,  but would
 need  means  to determine  the site-specific fracture permeability or  bulk
 retentivity.  At  a minimum,  a  strong laboratory  reconnaissance program would
 be needed  in order to demonstrate the desirability of further  consideration
 of basalt.

 IV.6  OTHER ROCKS

 It is appropriate at  this  point to comment  briefly on other rock types that
 may have excellent potential as disposal  hosts but have been largely ignored
 up to  the  present.

 Analysis of thousands  of water well records in places such as New England
 demonstrates that the  flow of  ground water  in bedrock in these areas is
 entirely fracture-dependent, and that metamorphic  rocks have very low po-
 rosities and negligible  permeabilities.   With  increasing degree of meta-
 morphism,  such  rocks  become  increasingly  anhydrous and granular textured,
 approaching the igneous  rocks  in their general physical properties.   High-
 grade metamorphic terrains  are  common in  many mountain belts, as well as
 in areas such as the  Minnesota  Shield, the  South Dakota Black Hills, and
 the Adirondacks.  In  the latter region they are associated with igneous
 rocks such  as anorthosite,  gabbro, syenite, and charnockite, all of  which
 are composed of anhydrous  silicates and all of which are potentially usable
 depository hosts.

 Dunite, a  relatively uncommon  igneous rock  occurring in parts of the
 Appalachians and the  Coast Ranges, has the unique  property of being  po-
 tentially  self-healing.  Access of water  to an HLW canister in these rocks
would result under the pressure and temperature conditions of burial,  in
 the development of serpentine;  the large volume increase accompanying this
 reaction would help to make the repository  impermeable.   It would need to
be  shown, however, that  the process itself would not induce fracturing.
                                   23

-------
In a similar way, a proposal by Professor W.S.  Fyfe that canisters be
surrounded by a packing of MgO powder (which would hydrate to brucite-
Mg(OH)2—upon contact with water) is also worth considering as a safe-
guard against possible waste leakage.
                                     24

-------
PART V;  SOLUTION TRANSPORT THROUGH ROCKS

The principal barrier to contamination of the biosphere by HLW is the time
delay expected during the migration of the nuclides through the rock unit
that contains the repository.  This delay can be predicted from experimental
data plus some assumptions regarding the particular setting, or from evidence
of previously completed "experiments."

The first approach is based on data concerning three key factors:  the actual
rock permeability in situ, whether this be true permeability or flow through
fractures; the flow rate; and the retardation, if any, of the dissolved
nuclide relative to the flow of the water.  Part VI will consider the first
two factors.  In this section, the third factor will be discussed and then
the previously completed "experiments."

V.I  SORPTION OF RADIONUCLIDES ONTO ROCK MATERIALS

Critical to the operation of a secondary barrier to radionuclide migration
is the ability of a rock through which fluids may slowly flow to adsorb or
react with dissolved radionuclides.  There is now an abundance of experimental
data on the extent of adsorption of various fission products and transuranics
by soil and rock  materials.   Unfortunately these experimental data are useful
only in a general way and even then may be misleading.

The reasons are that for any such surface chemical process, the nature of the
mineral surface must be specified precisely (mineralogical and chemical com-
position, surface area per unit of mass) and the experiment must be carried
out under conditions similar to those expected underground.  In addition, for
experimental results to be meaningful a uniform methodology would have to be
employed so that various materials and different elements could be compared
on a sound common basis.

The experiments thus far performed have been on materials described variously
as "desert soil", "tuff", "basalt", or "roontmorillonite", to mention a few.
None of these terms is a precise description that would allow anyone to infer
what the chemical behavior of the material might be.  "Montmorillonite" is a
term  used in many different ways by clay mineralogists and rarely to describe
a specific mineral with a fixed composition and crystal structure.  Specific
standard clay minerals (American Petroleum Institute Standards) have been used,
but only for the sake of comparison and not to give the impression that they
are standard mineral types.  Adequate description of any clay is made only by
chemical analysis (either bulk or electron microprobe) coupled with crystal
structure analysis by X-ray diffraction.  Since the methods of sample prepara-
tion strongly influence surface area, these too must be specified.  The ion-
exchange capacity and selectivity of the clays used must also be determined.

The same comments apply to the use of rock terms such as "basalt" or "tuff",
which are just field  terms of gross lithology used for the convenience of
the geologist mapping in the field and not in any sense the same as quotation
of the precise chemical formula of a reagent.
                                    25

-------
Results of experiments conducted at 25°C and 1 atm.  pressure are not a good
guide to what might happen at 100°, 200°, or 300°C.   We know of temperature-
induced reversals in cation selectivity, exchange capacity, and sorption
characteristics that are a result of surface configurational changes.  In
addition, all of these data are almost meaningless unless some realistic
choice of solution is made.  Formation waters of deeply buried rocks tend
to be rich in dissolved solids, and this ionic strength can greatly alter
sorptive properties of mineral surfaces.  Cation adsorption is generally
competitive, and radionuclides may be passed unsorbed if competing cations
of formation waters have already been sorbed strongly onto surfaces.

What we can say with the information at hand is that we would expect cer-
tain classes of materials to be more retentive than others at room tempera-
tures.  Thus desert soils would have high retentivity and granites and salt
low retentivity; with tuffs, basalts and shales lying between, we would ex-
pect notable cation exchangers, such as smectites and zeolites, to be highly
retentive.  But  this is no quantitative basis for estimation of what will
really occur in  a specific formation at depth and at elevated temperatures.
Before that can  be done we need a  carefully planned and executed program of
testing by experimentation.

V.2  FIELD "EXPERIMENTS"

An. alternative method to predict behavior of HLW  transport  through rocks is
to examine natural  settings  where  such  transport  might have occurred.  Two
such oases have  been studied in some detail  and  can  shed  some  light  on the
process.  These  are  the Oklo natural fission reactor discovered in Gabon,
and  the  underground  nuclear  explosion  test,  Cambric, carried  out  in  Nevada.

Approximately  1.8 billion years ago a  uranium  ore deposit experienced  the
appropriate  conditions  of  U  concentration,  235^ concentration,  and H/U ratio
 for  it to go critical and yield fissions  and energy.   This is  known  as the
 Oklo phenomenon.  The four to six local zones  where  this  occurred are  thought
 to have produced 15,000 megawatt  years (MW  yr)  of energy, over a  period  of
 approximately  0.5 to 1  million years.   The  site has  been studied  in  order  to
 determine the extent of loss of the various  fission  products  and  TRU's  that
were produced during and  subsequent to the  time fission was occurring.

 The results indicate that the losses were largely dependent on the  chemistry
 of the nuclides in question. Alkalies and alkaline  earths were lost  either
 completely or in very significant amounts.   These would include the majo
 fission product nuclides  of concern, 90Sr and 137cs  [see Lancelot et a
 On the other hand, the radioactive rare earth elements and the TRU's largely
 remained in or near the reactor zones.(6'  A very precise material  balance
 is not possible.  It would not be surprising if 10% or 20% of the products
 in a given reactor zone had been lost.

 The difficulty with the interpretation of the Oklo  event is the lack of
 precise geological control.  It is not clear to what depth these deposits
 were buried at  the time of the fission reaction.  Some believe that the
 process was going on under fairly shallow burial, because the water that


                                     26

-------
acted as a moderator had  to be vaporized and  this could only happen close
to the surface.  The sediments have been altered by heat and pressure to
a state suggestive of burial  to approximately 4,000 ft.^'J although there
is a suggestion  that the  depth may be as much as 6 or 7 km.  The mineral
assemblages, their compositions, and their fluid inclusions all suggest a
variety of possible depths of burial so that no sure conclusions can be
drawn.  It is likely, however, that the depth was considerably more than
the 1,500 ft envisioned for HLW burial.

The uncertainty  concerning the depth of burial means that it is not possi-
ble to make a direct comparison of the Oklo phenomenon with an HLW disposal
site.  This is so both because of the dependence of the flow of fluids on
depth (and the consequent variation in degree of openness of cracks) and
because the distance that the nuclides would have traveled to reach the
surface is not known.

Whatever the depth, however,  there must be serious concern over the loss
of the major fission products, 90sr and 137Cs.

The second "experiment" is the Cambric nuclear test in Nevada and the subse-
quent water pumping tests conducted using drill holes at the site.   The massivt
pumping tests conducted in the decade since the explosion have shown that the
radioactivity that has entered the water is in very low concentrations.  This
is very encouraging.

Two factors that play a role here, however, make the test less relevant to
HLW containment.  The first is that much of the radioactivity is in a glass
produced by the extreme heat of the explosion.  This glass has had ten years
to devitrify and be leached.  Since little radioactivity is detected in the
pumped water, it can be inferred that the nuclides are being retained.  The
problem is that the water is cold, and the pumping continues to bring cold
water to the site.  This test, therefore, is not using waters that are com-
parable to those anticipated in an HLW repository:  hotter than 300°C and with
high ionic strengths.   As a result, the channels through which the water is
flowing are behaving like pipes.   Further,  the very low concentrations of
nuclides are partly due to the rapid flow rates, which serve to dilute the
nuclides as they are released.
                                   27

-------
 PART VI;  RADIONUCLIDE TRANSPORT FORECASTING

 The objective of the transport modeling considered in this analysis is to
 forecast the subsurface movement and evolution of radionuclides emanating
 from a radioactive waste repository under various hypothetical situations.
 Perhaps the most challenging aspect of this problem is the necessity to
 forecast over long time periods (250,000 years) (°' with uncertain informa-
 tion.   In addition to the elements of uncertainty concerning the transport
 listed previously, (9,10) the following also play a role:

      •  hydrologic characteristics of the site (primary versus
         secondary permeability and porosity, hydrodynamic dis-
         persion) ,

      •  mathematical representation of subsurface transport,  and

      •  solution of  the resulting  equations for realistic physical
         conditions.

 Many of these uncertainties  have been encountered in  petroleum engineering
 and subsurface hydrology.  In these instances,  uncertainty is  reduced
 primarily  through the "calibration" process whereby the mathematical model
 and its input parameters are modified until a historical record (pressure
 history, concentration trends)  is  reproduced within a range of error deemed
 to  be  "acceptable."   Obviously,  the simulation  of radioactive  waste trans-
 port does  not,  in general, lend  itself to a "calibration" form of uncertainty
 reduction.   In summary,  the  problem is one of either  minimizing the uncer-
 tainty outlined above such that  a  deterministic analysis  (inclusive of
 sensitivity  analysis)  can be meaningful,  or alternatively,  incorporating
 this uncertainty directly into the  analysis so  that the  decision-maker  is
 cognizant  of  the degree  of uncertainty inherent in  the  forecasts.

VI. 1  STATE  OF  THE ART OF TRANSPORT MODELING

Two  operational  models are currently  capable  of simulating  the  movement of
multiple radionuclides in the subsurface.   These models have been developed
for  the Nuclear  Regulatory Commission and  the Energy  Research  and Develop-
ment Administration  (now Department of Energy)  by Intera  Environmental
Consultants  and  Battelle Northwest Laboratories, respectively.   (Note that
we have not  included  the model employed by  the  American Physical Society
Study Group on Nuclear Fuel  Cycles and Waste Management because of the
physical and mathematical simplifications  inherent in their approach.)
The  salient features of  the  two models are  summarized in Table 1.  Examina-
tion of this  information reveals that  the Intera numerical model, is the
most physically  realistic and flexible.  The BNWL models, however, have
the  advantage of analytical simplicity and  associated computational
efficiency.
Burkholder et al.     have argued because of the non-reducible uncertain-
ties inherent in some elements of the analysis of radionuclide transport,
"that highly sophisticated transport models should be used with caution or


                                    29

-------
                                 TABLE 1
                   CHARACTERISTICS OF TRANSPORT MODELS
    Model
Characteristic

Dimensionality
Mathemat ical
Formulation

Dispersion
Representation
Convection
Representation

Species
Transported

Transporting
Medium

Sorption

Source
Dissolution

Chemical
Form

Discrete  Fracture
Representation

Stochastic
Capability

Geologic  Strata

Temperature
Dependence

Solubility
                           Model
      INTERA

3D Cartesian
2D Radial

Numerical Finite
Difference

Variable, Function
of Velocity,
Tensor

General Velocity
Field

Multiple
Radionuclides

Saturated
Porous Medium

Linear

Variable
 Single
None
None
     BNWL I

ID Cartesian


Analytical


Scalar, Constant
 Constant Velocity
 Field

 Multiple
.Radionuclides

 Saturated
 Porous Medium

 Linear

 Constant
 Single
 None
 None
 Non-Homogeneous      Homogeneous

 Density,  Viscosity   None


 Limited             ?
     BNWL II
2D Cartesian
Analytical-
Numerical

Scalar, Constant
Spacially Varying
Velocity Field

Multiple
Radionuclides

Saturated
Porous Medium

Linear

Constant
Multiple*


None


None


Homogeneous

None
  Under development
                                     30

-------
perhaps  not be  used  at  all."  They  further  state  that  such models  "could
create unwarranted illusions  of  certainty."  The  significance  and  im-
portance of this  point  of view depends,  of  course, on  the interpretation
of  the phrase "highly sophisticated."

In  other branches of science  and engineering, the evolutionary trend has
been  to  minimize forecasting uncertainty through improved mathematical
representation  of the physical system and careful evaluation of associated
physical parameters  (see, for example, work in structures and mechanics,
hydrology, meteorology,  etc.).  The use  of  a  simpler model, however, is
advantageous when numerous alternative scenarios must  be examined, such as
in  a  sensitivity analysis or  an optimization  algorithm.  A combination of
both  approaches, such as utilized by Intera in their report to EPA,(12)
represents a cost-effective approach to  investigating  the problem  of
radionuclide transport.

VI.2  THE QUESTION OF UNCERTAINTY

There is  little question that the models presented in  Table 1 represent the
state of  the art of  engineering capability  in transport modeling.  Certain
degrees  of sophistication could be included,  such as a better representation
of  the fracture system, but these modifications are not likely to  enhance
the accuracy of the model forecasts materially, because of the uncertainties
that will still remain.

It  is more important to consider the possibility of quantifying the uncer-
tainty in the model  so  that decision-makers are aware  of the degree of un-
certainty associated with the analysis.  The  simplest  approach to  quantifying
uncertainty in simulation modeling is through a sensitivity analysis.  Such
an  analysis was performed by  Intera for EPA. t1^

In  this approach, a range of  parameter values is used  in the predictions.
These are selected so as to encompass the entire range of physically mean-
ingful parameter values.  Such an analysis provides an insight into the
role of each model parameter  on the calculated radionuclide distribution.
To  conduct such an analysis generally requires a tremendous amount of com-
putational effort.  To analyze two levels of  the 12 parameters considered
by  Intera unambiguously would require 21  independent simulations.   Because
this  is prohibitive, even with use of analytical models, a subset of the
general problem is normally considered and  the results, of necessity, are
indicative rather than definitive.

While a sensitivity analysis provides the decision-maker with a range of
possible  radionuclide distributions, it does not give any insight into
the probability of occurrence of any particular forecast.   In simulations
wherein large residual uncertainties in input are unavoidable, the  uncer-
tainty in the predicted radionuclide concentrations should be recognized
and quantified.   In other, words, in an analysis of the radionuclide trans-
port problem, the solution should appear as a confidence region (i.e., we
are 95% certain the real concentration of 90Sr lies between x and y,  given
the probability distributions of the input parameters).  Thus, the  re-
                                    31

-------
liability and importance of the forecast can be directly related to the
reliability of the input parameters.  To conduct such an analysis requires
estimates of parameters and their uncertainty, and a suitable mathematical
model.

Standard mathematical models such as presented in Table 1 can be modified
for simulation under uncertainty.  The procedure, known as Monte Carlo
analysis, involves conducting a series of simulations in which parameter
values are selected at random from their respective probability distribu-
tions.  These values are used in simulation runs and the output variables,
in this case radionuclide distributions, are assembled into probability
distributions.  From these distributions, confidence intervals can be com-
puted.  Such analyses, however, require excessive computational effort,
particularly when several different parameter distributions are involved.
Alternative schemes are available to circumvent these computational diffi-
culties, but these cannot be considered at this time to be state-of-the-art
engineering methodologies.

While uncertainty in the partial differential equations may be accounted
for by use of new mathematical methodology, one must still consider the
problem of estimating these uncertain parameters.  Whether intentional
or by default, these parameters will be subjective probability estimates.
Tools are available and currently being developed to compute these estimates
by use of objective and subjective information (see, for example, Bayesian
methods).  The credibility of forecasts made under condition of uncertainty
will depend upon a reasonable representation of the probabilistic parameters,
The impact of parameter uncertainty, that is the resultant uncertainty in
the projections, is dependent on problem and boundary conditions.  In ex-
ample problems that have been considered/13) situations were encountered
in which the uncertainty increased, reached a maximum, and asymptotically
decreased to zero as a function of time.  Another case resulted in an
exponential increase in uncertainty throughout the simulation period.
Thus it is apparent that generalizations regarding the impact of parameter
uncertainty on forecasts of radionuclide concentrations are not normally
possible.

It should also be noted that uncertainties associated with parameters in
the governing equations interact in such a manner that their individual
contributions to the spread in the confidence interval of the projections
cannot be discerned.  Thus it is not possible, in general, to delineate
unambiguously the importance of the uncertainty associated with, for ex-
ample, the source term.

VI.3  CREDIBILITY OF THE ANALYSIS

At this point it is evident that uncertainty is the distinctive element
of radionuclide transport analysis.  While uncertainty is associated to
some degree with forecasts emanating from all mathematical models of sub-
surface phenomena, this uncertainty is generally minimized through cali-
bration with known historical data.  Even where historical data are scarce
	 generally reserves the option of measuring in situ hydrologic and trans-
one
                                    32

-------
port parameters.  Because the disposal site has not yet been selected,
such measurements are not currently available.  Moreover, even after site
selection, the in situ measurement of parameters must be minimized to
ensure the integrity of the site.  Because of the apparent inability to
decrease significantly the uncertainty associated with the transport
parameters, the only viable alternative is to incorporate as much of the
uncertainty as possible directly into the analysis so that the decision-
maker is aware of at least the minimal uncertainty inherent in the fore-
casts .

In evaluating the suitability of available radionuclide transport analyses
as a basis for establishing radiation standards for high-level radioactive
waste management, EPA should particularly consider the following:

     •  The existing simulations of radionuclide transport emanating
        from high-level waste are limited to a small number of hypo-
        thetical problems associated with selected waste-disposal
        scenarios.  The results represent qualitatively the movement
        of the waste under these conditions.  Although limited in
        number relative to the spectrum of hydrological settings
        possible for a repository, the problems selected represent
        a reasonable and relatively unbiased choice.

     •  The problems examined should be considered indicative rather
        than representative of situations to be encountered in the
        storage of radioactive waste.

     •  The relationship between the parameters and processes
        employed in the model and those to be encountered in a
        disposal site is very uncertain.

     •  The Intera model incorporates directly or indirectly (as
        in the case of fractures) those attributes of the proto-
        type physical system most likely to play an important
        role in radionuclide transport from a repository.  A
        possible exception is unsaturated transport.

     •  The credibility of the model and the associated analysis
        would be enhanced by a satisfactory simulation of radio-
        nuclide transport at the Nevada Test Site.

     •  Current models do not incorporate parameter or process
        uncertainty in their forecasts.

     •  The scenarios considered in the analyses are, in large
        part, arbitrary and may or may not encompass the hydro-
        geological systems to be encountered at a specific site.

     •  The general approach currently used by some consulting en-
        gineers for forecasting radionuclide transport, though strongly
        limited by the inherent uncertainty in the hydrogeologic, geo-
                                   33

-------
        chemical and chemical data input, nevertheless represents the
        best means currently available for establishing the movement
        of radionuclides if and when the proposed containment is
        breached and water comes in contact with the waste.  We
        have considered the quality and applicability of the simu-
        lation methods but not their adequacy.  The adequacy of the
        forecasts depends on the sensitivity of the decision-making
        process to the state of the art of transport modeling.

VI.4  PROPOSED RESEARCH

We have been asked to suggest techniques that would result in an improved
capability for risk assessment regarding the HLW repository concept.  In
transport simulation, three areas appear to be prime candidates for further
research:

     •  In order to incorporate the uncertainty associated with
        parameter identification, a program should be initiated
        that is directed toward the development of an analytical
        capability for forecasting under uncertainty.  This would
        not involve great sums of money inasmuch as the primary
        goal is software development, which is not usually capital
        intensive.  In the short term, it would appear reasonable
        and appropriate to employ existing techniques for simula-
        tion under uncertainty in order to analyze one or more
        mathematically simple scenarios.  This would at least pro-
        vide a ball-park estimate of  the uncertainty associated with
        the forecasts.

     •  Before  the methodology developed in the preceding step can
        be utilized,  estimates of parameter uncertainty must be
        available.  These estimates can be obtained through an
        effective utilization of subjective and measured information.
        The analytical apparatus for  this  type of analysis is incom-
        pletely developed and should  be  considered a priority ele-
        ment of the research effort.

     •  Once uncertainty  can be quantified and incorporated into
        transport  forecasts, the next major objective  is to mini-
        mize this  uncertainty.  This  can be achieved primarily
        through additional  field observation.  Thus, a  primary
        goal of a  research  program  should  be  the  in situ measure-
        ment of hydrological and geochemical  parameters.   Because
        of  the  nature of  the repository,  this methodology  must be
        compatible with  the maintenance  of repository  integrity.
        The  liberal  use  of  subjective information will  minimize  the
        cost and  negative impact  to the  repository  of  data collec-
        tion and,  it  would  seem, provide a cost-effective  program
        of  parameter  identification.
                                    34

-------
 PART  VII;   GEOLOGIC AND OTHER ACCIDENTAL  HAZARDS

 Assuming  that  physical  and hydrologic  requirements  for  a  safe HLW reposi-
 tory  have  been established,  one must still  consider whether all other
 factors that may  disrupt the integrity of the  site  have been taken  into
 consideration,  and  whether these  factors  are tractable  to numerical risk
 analysis.   As  one report states,  "At some point in  the  repository site
 and emplacement method  selection  and safety assessment, whether willing
 or not, we will quantify in  order to make decisions."(14)  Without  attempt-
 ing to assess  all non-geologic and geologic hazards, we shall comment
 briefly on some of  them,  primarily to  illustrate the extreme numerical
 uncertainties  attached  to most.

 VII.1 NON-GEOLOGIC ACCIDENTS

 VII.1.1  Intrusion  by Man

 The development of  an underground waste repository  involves the breaching
 of what was probably initially an intact  geologic container.  After the
 repository has  been backfilled, heating combined with the possible access
 of ground  water will set  up  a  convective  hydrological system in which
 ionic transport rates will also be greatly  accelerated.  These features,
 as well as  uncertainty  about  the  long-term  effectiveness of having sealed
 shafts in  bedrock,  are  matters on which there  are very  few bases for judge-
 ment.

 Intrusion  of a  repository at some  future  date  in the search for mineral
materials,  including the  uranium  and TRU  elements that were buried, or to
 satisfy archeological or  other curiosities  appears  to us also to be a risk
 for which no trustworthy  probability estimates may be applied for the time
 span  over which the integrity  of  the repository is  to be assured.   Material
 demands shift with  technology.  The most  successful civilizations last for
 only  a few millenia at  best, and  even  these are not without their disastrous
 interludes.  Man's unpredictability far outstrips most of the imagined geo-
 logic hazards we  can foresee,  and we doubt  that it is amenable to meaningful
 probability analysis.

VII.1.2  Meteorites

An astronomic risk that is generally cited is meteorite impact.   The earth
annually receives an infall of approximately 10^ tons of cosmic  dust,  but
meteorites large  enough to produce craters the size of Meteor Crater in
Arizona (1.2-km diameter), or  larger, are estimated to have a frequency on
the order of 1  per 20,000 years or less.  The bedrock in a meteorite crater
is pulverized to  a depth of approximately one-tenth of the crater diameter,
but the surrounding and underlying rock will be highly fractured.   If  one
assumes conservatively  that a repository at a depth of 0.5 to 1  km might be
breached if it  lay within a radius of 10 km from the point of meteorite
impact, and also  that the probability for impact is equally distributed
over the earth's  surface  (5.1 x 10^ km^), the likelihood of meteorite
damage is  approximately 3.3 x 10~H/yr, or 1/30,000 averaged over a
                                    35

-------
million-year time span.  Meteorites, therefore, should not be regarded as
significant factors in evaluating repository safety.  A more detailed
treatment by Arthur D. Little, Inc.'  ' yields similar results except for
secondary effects such as stream diversion by impact.  Given the need to
make provision in the repository for major water table variation, this
problem can also be given low significance.

VII.2  GEOLOGIC ACCIDENTS

The structure of the earth and energy sources that drive its physical
processes are such that stochastic models are not widely applicable, unless
one closely restricts the area of. inquiry, and probably the time as well.
We are aware that fault tree analysis is the numerical basis for risk
assessment in the Rasmussen report on nuclear reactor safety.(l^)  However
the question of assessing engineering failure by a surface facility over
a 50-year time span is not the same problem as assuring geologic and
hydrologic integrity at an underground site over a million-year time span,
especially because there is information on reactor performance in the past.

VII.2.1  Extrusive and Intrusive Magmas

Aside from gradual subsidence of the Gulf Coast and the Mississippi Embay-
ment commencing approximately 70 million years (m.y.) ago, and the isostatic
depression and rebound of the northeastern and north-central part of the
country in response to continental glaciation, the United States east of
the Rocky Mountains has been geologically stable for approximately 100 m.y.
In this vast region the youngest dated igneous rocks (dikes) have an age
of approximately 70 m.y.  For this entire area, then, the probability for
a magmatic incursion into a waste repository during the next million years
has to be taken as asttonomically low, but numerically uncertain.

 Along  the  entire Cascades  chain, by contrast,  the probability for a volcanic
 eruption may be on  the order  of 10~2/yr.   If  there  are an estimated 65,000 km2
 of  potential volcanic  terrane and  a repository within a  10-km radius of  a
 volcano might be adversely affected by an  eruption,  the  average rate of  risk
 per  300 km2  is on  the  order of  10~Vyr.   However, it obviously makes a differ-
 ence whether we designate  one of these 300 km2 areas as  lying close to Lassen
 Peak,  Mt.  Baker, Mt.  St. Helens or Mt. Ranier, or remote from them.  Even for
 the  remote  areas,  it  is well  to recollect  that no one predicted or could have
 predicted  the May,  1943  eruption of a new volcano at Paricutin, in the western
 Mexico volcanic belt.

 For the area of  intermediate  magmatic  risk lying between the Rocky Mountain
 Front  and  the  Cascade-Coast  Range  Belt,  there is no very good basis for  specu-
 lating on  numerical risk probabilities,  nor  is there any ready  answer  to why
 areas  such as  Sunset  Crater,  Arizona,  and the Craters of the Moon,  Idaho, have
 been active within the past  few thousand years.

 The situation with respect to magmatic hazard typifies what  the Panel  believes
 to  be  true of most  other geologic  risks.   Within different specified regions
 we  may estimate  a  relative order of danger,  but  for very few areas  can we have
                                    36

-------
any confidence in orders-of-magnitude estimates of risk probabilities.
Of all the world's volcanoes now under surveillance, Kilauea is one of
the few whose short-term behavior is predictive.  Even here, projections
for a million years in the future are well beyond the current state of
the art.

VII.2.2  Earthquakes and Faults

Even in regions of transform faulting where great earthquakes occur re-
latively frequently on the geologic time scale, like the active, western
continental margin of North America, we cannot yet predict specific events
of magnitude 7 or more with precisions better than about 100 years and
100 km.  We do know, however, that the recurrence rate of events on the
San Andreas fault has been on the order of 200 years over the last few
thousand years, that geologically rapid displacements are still going on,
and hence that California is obviously a high-risk region where we would
not place a repository if we fear the hazards of earthquakes.

On the other hand, we cannot claim that the probability of an earthquake
is identically zero anywhere in the world because the origins of earth-
quakes in the interior of the continent (New Madrid, 1811) or at its
passive margin (Charleston, 1886) remain pretty much a mystery.  Had  ,
these events occurred a few generations earlier, before written records
became available, we would not now be worried about destructive earth-
quakes in the central and eastern United States.  About all that we can
confidently say is that the earthquake risk is much less in some regions,
say the Gulf Coastal Plain, than it is elsewhere, say along the trace of
the San Andreas fault and its subsidiaries.

The hazards of earthquakes may well be overly exaggerated.  Recent sur-
veys by Professor H. Dowding of Northwestern University and Dr. Howard R.
Pratt of Terra Tek reveal that damage to underground openings is very
much less than that to surface structures immediately above.  Disruption
would occur if a seismogenic fault actually transected an opening, but
adequate site characterization should have eliminated this eventuality.
Thus, once a repository has been sealed and its surface facilities
abandoned, the effects even of large earthquakes are likely to be negli-
gible.                              ,   .

It is well to recall that the earthquake history of the United States is
based chiefly upon diaries or newspaper accounts, which may cover a time
span varying from 300 years to a few decades, depending upon the area.
The development of a widespread instrumented seismic network with sophis-
ticated equipment and data processing is partly an outgrowth, during the
past three decades, of the interest in detecting nuclear bomb blasts.
Instrumental seismology has existed'for decades, but systematic data com-
pilation is relatively recent.  On either basis, whether from an incomplete
historical record stretching backward for up to three centuries, or from a
compilation of seismic record data, perhaps adequate for only three decades,
there is little reason for confidence in forward predictions covering a
million years.


                                   37

-------
During the past three decades a series of qualitative seismic risk maps
has been produced by the Coast and Geodetic Survey, the most recent of
these in 1969.  Responsibility for the next set of risk maps lies with
the U.S. Geological Survey, but their own estimate is that a reliable
seismic probability risk map lies a decade or more in the future.  Even
this assessment may be optimistic.

VII.2.3  Glaciation, and Climatic Changes

Geologic records of glaciation in Alaska extend backwards for at least 10
million years and in the coterminous United States for more than 2 million
years.  Despite the generally adopted statement that there were only four
major cycles of glaciation, it has now been demonstrated that there have
been numerous glacial epochs, that they follow a cyclicity on the order
of 1C)5 years, and that they are predominantly controlled by the earth's
orbital eccentricity, as proposed long ago by Milankovitch.  Both theory
and the now established record show that the bulk of these lO^-year cycles
represents glaciation for the Northern Hemisphere, with interglacial
periods like the present one lasting on the order of 10,000-20,000 years.
The current interglacial was established between 16,000 and 11,000 years
ago.  It is now predicted that the long-term trend for the next 20,000
years is toward extensive glaciation in the Northern Hemisphere.  Although
this forecast is in accord with climatic analyses that show a consistent
cooling trend for the past three decades, the latter trend is too short to
permit  interpretation that the two are coupled.

The near certainty that re-glaciation will occur long before any disposal
site will have fulfilled its assigned role implies that there will be other
complications.  The depth of bedrock scouring by ice sheets averages only
20 ft.  However, re-routed streams have incised bedrock for depths of up to
200 ft; Niagara Falls, the best known example, has excavated a 170-ft-deep
trench  for a  length of 7 mi  during the past 10,000 years.  Waxing icecaps
cause isostatic depression of bedrock amounting to hundreds of feet, po-
tentially producing faulting;  waning icecaps allow isostatic rebound, more
potential faulting and, with extreme melting, a good possibility for flooding
continental margins to depths of  100 ft or more above present sea level.

Mean  precipitation and evaporation differ markedly between glacial and inter-
glacial times.  We cannot  assume  that an area that is arid today will remain
so  during a glacial cycle.   This  is most obviously the case in Utah and
Nevada  where, during  the recent glacial epochs, there was a pronounced plu-
vial  environment.  What such a pluvial environment might do to the hydrologic
regime  in the site of an HLW  repository seems to have been largely ignored in
current risk  assessments of  repositories such as Hanford and the Nevada Test
Site.

The Panel believes  that continental  re-glaciation has a very high probability
of  occurring  within  the time period  of concern for HLW, and that  the associ-
ated  climatic changes will have differing  effects on all waste repositories,
depending on  location.  There may, however, be certain advantages to locating
repositories  in areas likely to be re-glaciated, because  this would effec-
tively  seal  them  from intrusion several  thousands of years hence.

                                    38

-------
A different problem arises from the potential combustion of coal and
petroleum products.  This, the so-called greenhouse effect, would be a
warmer world-wide climate created by the C02 resulting from fuel burning.
Even if the effect were not total melting of the Antarctic and Greenland
ice-sheets, sea level would rise significantly.  The evaluation of salt
domes as repository  sites should include consideration that such an
eventuality is possible.  If the top of the dome were accessible to sea-
water, extensive dissolution of the salt could occur, possibly exposing
the HLW.  Further, even if the greenhouse effect did not play a role, it
is known that sea level has varied for different interglacial epochs and
could be higher next time than it is now.

VII.3  FUTURE RESOURCE EXPLORATION

Although the energy supply appears to loom as the major mineral resource
problem for the next few decades, this is only because its present urgency
has diverted attention from the far more important question of how the
world's rapidly expanding population can maintain the raw material supplies
necessary for an industrialized society.  Current annual consumption of
newly mined mineral products is 3.75 MT per person, and by 1985 this figure
will have doubled.  Geochemical and economic arguments have been adduced to
show that when the average ore grade of a metal slips below a critical level,
the amount of energy needed to extract that mineral will far exceed any
costs that society can reasonably pay for it, so the mineral will no longer
be available.  Mercury, silver, gold, copper, and lead are candidates for
such early scarcity.  The 1975 National Academy of Sciences Report of the
Committee on Mineral Resources and the Environment has concluded that
technology cannot always close the gap between rising demands, and available
resources, so we shall be inexorably driven to lower ore grades, larger
tonnages, and other materials.

As an illustration of the enforced shift to increasing tonnages of decreasing
grade we may cite uranium, for which the 1990 projected demand of UaOg is
700,000 MT; production of this amount of UaOs will be possible only if the
present ore (economically mineable) grade (0.22% ^03) falls to 0.075% in 1990.

From what is known of the world's mineral resource picture, it is 'difficult
to be optimistic that industrialized society can sustain itself at anything
approaching its current levels in the next millennium.  A steady decline in
the quality of life appears inevitable, as well as an increasingly desperate
exploitation of raw materials.  What the mineral exploitation might be like
a thousand years from now is impossible to predict.

Of the rock types currently under consideration as hosts for HLW repositories,
shale and basalt seem unlikely to assume an important economic role.  Granite
is somewhat different because there are some granites that potentially are
mineable because they contain relatively large amounts of thorium,  uranium,
and rare earth.  Several other types of ores are also associated with granites.
                                    39

-------
The most likely targets for near-term exploitation, however, are salt domes
because of the potential productivity of petroleum, halite, and sulfur; and
bedded salt deposits because of their potash, halite, and gypsum.  The
United States has only 4% of the world's total proven potash reserves, and
most of these are concentrated in the New Mexico area now being evaluated
as an HLW repository.  Future conflicts between the demand for HLW reposi-
tories in bedded salt and the needs of agriculture for potash seem in-
evitable, and may even now constitute an important negative socio-economic
factor in the development of some repositories.  Such conflicts could be
minimized in the case of salt domes by selection of domes that are barren
of petroleum and sulfur.

VII.4  OTHER GEOLOGIC HAZARDS

Some data on rates of uplift and denudation may be used to underscore the
unpredictable nature of so-called steady-state geologic processes.  The
average rate of regional denudation in the United States (Q.063 m/103 yr)
is such that it would require approximately 15 million years to uncover an
HLW repository at the burial depths now planned.  This seems to be such a
large safety factor that we could easily relegate denudation to the status
of a relatively minor risk.  However, denudation rates vary by factors of
up to 75, depending on climatic and topographic factors, so there is a
large uncertainty in applying averages that are not site-specific.

Assuming that a repository is planned for a stable area of the continent,
we may now ask how certain we are of this stability.  The Adirondacks,
long regarded as an ancient stable shield area, are now known to be rising
at the rate of 3.7 mm/yr in the center of the mountains.  The beginning of
the doming of the Adirondacks is not well dated, but at the present rate
of uplift the 1,600 meters of maximum uplift, and subsequent deep erosion,
could have taken place within a half-million years; but it could also, of
course, have taken millions of years longer.  Prior to the initiation of
this recent uplift, there would have been no reason to predict that this
large area of the United States would be rapidly uplifted and eroded.
There is also no basis for predicting where, or when, other such uplifts
may occur on the stable shield of the continent.  The point is that there
is a finite but very low probability that a repository buried at a depth
of 0.5 km almost anywhere might, within a relatively short time, unpredict-
ably reappear at the earth's surface.  How one can make an orders-of-
magnitude conjecture concerning these probabilities is beyond the Panel's
ability to resolve.
                                   40

-------
 PART VIII;   MONITORING THE REPOSITORY

 Part of the responsible establishment of an HLW repository  is  the  provision
 for a warning system in the event  of  loss  of  some of  the HLW  to the  biosphere.
 If we assume that  some form of  monitoring  for radioactivity will  be  mandatory,
 there are two aspects to this problem:   early warning to permit some attempt
 to repair the site or confine the  spread of the HLW,  and longer range sur-
 veillance to provide warning in sufficient time to reduce  exposure to the  con-
 tamination.   The first type of  monitoring  would be done  close-in  to  detect
 early signs  of the breach of the repository.   For the second, more widely
 spaced monitors would be used.

 Any close-in monitoring would,  of  necessity,  include  some  connection with
 a  surface-recording device,  a means for  replacement or repair of  the sensor,
 and due consideration of the possibility that the sensing  scheme  itself could
 provide an  additional pathway for  migration of  radionuclides.  That  monitor-
 ing be continued for a period of time on the  order of at least several cen-
 turies for  the fission products  and  one million years for the TRU wastes
 is  a commitment that it is  not  reasonable  to  expect future inhabitants to
 fulfill.

 There is  no  known instrumentation that has  demonstrable reliability for dec-
 ades,  much less for centuries.   Beyond those  times, the  radioactivity is
 much  less, will have a large alpha component, and  the  sensitivity of  any
 currently available instrumentation to the  losses would  decrease markedly
 just  when the  sensitivity needed for  detection would  increase.  The  commit-
 ment must be to a  repository that  may be monitored remotely for some period
 of  time,  but we shall  not know what is going  on within the general confines
 of  the repository  once the monitors left in place malfunction.

 Remote monitoring  can  be effected  both from the surface  and from the drill
 holes  made to  help  define the site to begin with.  However, such monitoring
 cannot define  any  activity such  as migration  that is occurring in the
 repository.  A new  technology would be needed to achieve this, and such a
 technology would first  have  to be  shown to be possible.

 An  analysis  of  the  nature of'remote monitoring arrays pinpoints the very
 difficulty that predicting the direction and rate of migration of the HLW
 poses  in  the first  place.  Even if the overall hydrologic regime were well
 determined,  small deviations from  the average flow might still result in
 the arrival  of  some HLW at the surface.  Such short-circuits  could contain
 only a  small fraction of the total HLW, but they could still  have serious
 local  effects.  It  is rarely possible to identify where such  effects  might
 occur,  thus  they are potentially hazardous  even beyond the  confines of the
 fenced-off reservation of a repository.  We see no way to predict  such
occurrences  either in time or space.  It is not part of our charge to con-
 sider risk tradeoffs that might  arise in the case of small  numbers of people
being at potentially serious risk in such cases.  Consideration of this
problem should be included in any monitoring scheme.
                                   41

-------
If our prediction of the behavior of a repository were quite accurate,
we could allow for small losses that take sufficiently long times to
arrive at a biosphere reservoir so as to be below presently specified
maximum safe levels.  However, owing to the variable hydrologic regimes
that would almost surely occur over the next million years, the leaks of
the HLW might occur significantly earlier than predicted, but still much
beyond any time when continuous monitoring can be anticipated.  The TRU
wastes that enter the biosphere undetected at such time in the far future
might impose serious risks on our descendants.
                                   42

-------
PART IX;  CONCLUSIONS

It should be emphasized that our intention is not to recommend what may
be the best host rock or geographic location for HLW storage or disposal.
Our intention is to identify gaps in the state of geological knowledge
that may prohibit reasonable predictions to be made about events and
processes that might occur in certain host lithologies.  In some in-
stances we feel compelled to indicate potential problems.

     (1)  Knowledge concerning the properties of the salt in dry domes
          is good.  If the HLW canisters are not to be retrieved after
          the 5 or 10 years needed to complete the filling of the
          repository, then the short-time deformations in dry salt domes
          around the canisters can be computed and reasonable predictions
          are possible.  Disposal would be permanent because at the ex-
          pectable high temperatures, the high-density canisters would
          probably sink into thetsalt.  The usual assumption seems to
          be that the salt is indeed dry.  This is clearly not so in
          certain bedded deposits or in all domes.  Careful measure-
          ments of water content will be most important.  Further,
          even dry salt contains some water, so that the sinking of
          canisters computations, with the possibility of "focusing,"
          should be done with this in mind.

     (2)  Retrievability of HLW in other rock types is not so much
          a question of locating the canisters because they have
          bodily moved elsewhere, but being able to collect all
          of the waste because corrosion and leaching might so
          disintegrate the canisters that much of it is dispersed.
          This problem is further exacerbated by the possibility
          of dispersal by fluids far away from the site.  We be-
          lieve that the appropriate computational methods for
          assessing such possibilities are known, but our knowledge
          of the access of the repository system to fluids via
          transport through cracks is not now adequate.  Research
          is needed on how to determine the extent to which a
          repository is an "open" system.  As a consequence, while
          several lithologies are potentially viable candidates
          for repositories that permit retrieval, not enough is
          known to permit a selection or priority assignment.

     (3)  There is a fundamental paradox to be encountered in the
          design and construction of a "closed" repository.  It is
          desirable to avoid disturbance of the rock "mass by ex-
          ploration drilling as this provides extra pathways for the
          HLW to reach the surface.  However, one must determine very
          precisely the geometric distribution of rock properties
          throughout the future repository site and its immediate
          surroundings.  Prior to excavation, only careful examination
          of many drill cores can possibly delineate these properties.
          These two contradictory demands must somehow be resolved.
          Proper assessment may have to await excavation of shafts


                                    43

-------
     and adits,  despite the high  risk of  the  capital  investment
     should the  site then be found to be  unsuitable.

(4)   If the HLW  is not reprocessed,  U and Pu  will  pose  long-
     term radiation hazards, as well as a different chemistry
     of transport.  It is unlikely,  however,  that  the integri-
     ties of the canister, its  contents,  and  its immediate
     surroundings will last very  long, whether  or  not repro-
     cessing is  carried out. We  have seen no evidence  of
     survivals longer than a decade.   The question of repro-
     cessing does relate to the mix produced  and thus the
     temperatures to be expected  in the canisters  as  a  result
     of the heat production from  that mix.

(5)   We are surprised and dismayed to discover  how few  rele-
     vant data are available on most of the candidate rock
     types even  30 years after wastes began to  accumulate
     from weapons development.  These rocks include granitic
     types, basalts, and shales.   Furthermore,  we  are only
     just now learning about the  problem  of water  in  bedded
     salts, and  the need for careful measurements  of  water
     content In  domes.

(6)   The state of knowledge concerning total  permeability  in
     jointed shales, granites,  and basalts is still inadequate
     for meaningful forecasting.   Total permeability  includes
     all paths for fluid migration through the  rock,  including
     cracks, joints, faults, and  inhomogeneities of lithology.
     The lack of such knowledge does not  necessarily  imply
     that a rock is unsuitable.   Comparatively  anhydrous rocks
     in the granitic clan or basalts might well be the  best
     rock types  for storage,-and  perhaps  disposal, even though
     they are jointed and salt  is not. Transport  properties
     depend on both the chemical  and mechanical nature  of  the
     rock.

(7)   Two principal questions arise for all lithologic types
     other than the salt in domes:

        1.  How does one determine the real "permeability"
            of the rock mass surrounding  the  repository
            and extending to the  surface, and

        2.  How*does the permeability of  the  fissures
            affect solute-retardation factors relative
            to the flux of water  throughout the rock  mass?

     Development of-methods to answer the first question will
     be very hard, but must be undertaken.  The plethora of
     distribution coefficient and retardation coefficient  data
     now available provide a good start,  but  they  are not
                              44

-------
           adequate to address the critical problems of retardation
           under conditions of high ionic strength.

      (8)   Existing computational methods for assessing  transport
           may well be adequate,  and the accuracies  of  these  models
           can be specified in principle.   However,  until the un-
           certainties of  the  input parameters are  included in the
           numerical models, the  reliability of any  model remains
           undemonstrable.   further,  while mathematical simulation
           of  radioactive  ion  transport  from the repository is an
           important and very  useful  tool in the risk analysis of
           an  HLW repository,  such simulations can be dangerously
           misleading unless careful  attention is directed toward
           inherent physical and  mathematical assumptions.  Of
           particular concern  here is the relatively superficial
           attempt to date  to  incorporate parameter  uncertainty
           in  the contaminant  transport  forecasts.   It  is  of  para-
           mount  importance that  the  analyst present unambiguous
           estimates of uncertainty in such forecasts so  that
           decision-makers  are not misled  into believing  that such
           forecasts are independent  of  parameter uncertainty.

      (9)   It  seems clear that the uncertainties  of  forecasting
           the behaviors of conceptual HLW repositories are due
           principally to inadequate  knowledge of the relevant
           mechanical, radiochemical,  and  hydrologic properties
           of  the candidate rock  types.  Most  of  these  can be
           measured by well established  methods,  but  times required
           even for adequately funded  research efforts  are likely
           to  vary widely—from a  year or  so  to a decade or more.

As noted in the  text, there are  also  several  questions, notably the de-
termination of real permeabilities and  porosities in the rocks at a site,
or the nature of  the long-term monitoring  systems, answers  to which must
await the  invention of new technology.  The time scale for such research
is much less  readily determined.

Except for the modest effort  on salt, the geological aspect of the HLW
repository problem had largely been neglected by our generation until a
year or so ago.   It will not be solved without a strong commitment of
money and manpower,  lasting beyond 1985.,
                                   45

-------
                                REFERENCES


  1.   Arthur D.  Little,  Inc.   Technical  Support for Radiation Standards  for
      High-Level Radioactive  Waste Management.   Sub-task C-l.   Assessment
      of  Site Selection  Factors  for Deep Geologic  Disposal  of High-Level
      Radioactive Waste.   Draft  Report.   Performed by  D'Appolonia Consulting
      Engineers, Inc.  Contract 68-01-4470,  Office  of Radiation Programs,
      U.S.  Environmental  Protection Agency,  1977.

  2.   National Academy of  Sciences.   Rock-Mechanics Limitations to Energy-
      Resource Recovery and Development.   Draft to be  published.

  3.   Roedder, E., and H.D. Belkin.   Fluids  Present During  the Diagenetic
      History  of the Salado Formation, Delaware Basin, Southeastern New
      Mexico,  As Recorded  by  Fluid  Inclusions.   Abstracts of  Midwest
      American Geophysical Union Meetings,  1977.

  4.   Bredehoft,  J.D., et  al.  Geologic  Disposal of High-Level Radioactive
      Waste-Earth Science  Perspectives.   Draft  U.S.  Geological Survey
      Circular,  1978.

  5.   Lancelot, J.F.,  A. Vitrac, and  C.J. Allegre,   The Oklo Natural Reactor:
      Age and  Evolution Studies by U-Pb  and  Rb-Sr  Systematics.  Earth & Planet.
      Sci. Letters 25:189-196, 1975.

  6.   Loubet, M., and  C.J. Allegre.  Behavior of the Rare Earth Elements in the
     Okla Natural Reactor:  Geochimica et Cosmochimica Acta 41:1539-1548,  1977.

 7.  Bryant, E.A.,  G.A.  Cowan, W.R. Daniels, and W.J.  Maeck.  Oklo,  An Exper-
     ment in Long-Term Geologic Storage.  LA-UR-76-701,  Los Alamos Scientific
     Laboratory, 1976.

  8.   Logan, S.E.  Workshop on Geologic  Data Requirements for  Radioactive Waste
     Management Models:   University of New Mexico, Report No. NE-27(76),
     Union Carbide Report No. 297-1, 40, 1976.

  9.   Merritt, W.F.   High-Level Waste Glass:  Field Leach Test, Nuclear
      Technology 32, 1977.

10.  Arthur D.  Little, Inc.  Technical  Support  for Radiation  Standards for
     High-Level Radioactive Waste Management.  Task B.  Effectiveness of
     Engineering Controls.  Draft Report.   Contract 68-01-4470, Office
     of Radiation Programs, U.S. Environmental Protection Agency, 1978.

11.  Burkholder, H.C., J.A. Stottletnyre, and J.R.  Raymond.   Safety Assess-
     ment and Geosphere Transport Methodology for the Geologic Isolation
     of Nuclear Waste Materials, OECD/NEA Sponsored Workshop on Risk
     Assessment and Geologic Modeling, Ispra, Italy, 1977.
                                    47

-------
12.  Arthur D. Little, Inc.   Technical Support for Radiation Standards
     for High-Level Radioactive Waste Management.  Sub-task C-2.  Analysis
     of Migration Potential.  Draft Report.  Performed by Intera Environ-
     mental Consultants, Inc.  Contract 68-01-4470, Office of Radiation
     Programs, U.S. Environmental Protection Agency, 1977.

13.  Tang, D.H., and G.F. Finder.  Simulation of Groundwater Flow and Mass
     Transport Under Uncertainty;  Advances in Water Resources, 1 (1), 1977,

14.  Logan, S.E.  Workshop on Geologic Data Requirements for Radioactive
     Waste Management Assessment Models.  University of New Mexico Report
     #NE-27 (76) Union Carbide 297-1, 30, 1976.

15.  Arthur D. Little, Inc.   Technical Support for Radiation Standards.
     Task D.  Assessment of Accidental Pathways.  Draft Report.  Contract
     No. 68-01-4470, Office of Radiation Programs, U.S. Environmental
     Protection Agency, August 1977.

16.  Reactor  Safety Study.  An Assessment of Accident Risks in U.S.
     Commercial Nuclear Power Plants.  WASH-1400, U.S. Atomic Energy
     Commission, Washington, D.C., August 1974.
                                    48

-------
                         LIST OF ABBREVIATIONS






EPA        U.S. Environmental Protection Agency




HLW        High-level radioactive wastes




OWI        Office of Waste Isolation




DOE        U.S. Department of Energy




ERDA       Energy Research and Development Administration




NRG        Nuclear Regulatory Commission




BNWL       Battelle Northwest Laboratories




USGS       U.S. Geological Survey




LBL        Lawrence Berkeley Laboratory




LLL        Lawrence Livermore Laboratory




TRU        Transuranic




m.y.       Million years




m          Meter




MX         Metric ton




ft         Foot
                                    49

-------
                                GLOSSARY
 Adit:  A nearly horizontal opening by which a mine is entered, drained,
      or ventilated.

 Alpha particle:  A positively charged particle emitted by certain
      radioactive materials.  It is made up of two neutrons and two
      protons bound together,  hence it is identical to the nucleus of
      a helium atom.  It is the least penetrating of the three
      common types of radiation (alpha, beta, gamma) emitted by
      radioactive material, being stopped by a sheet of paper.
      Alpha particles are dangerous to plants, animals, or man only if
      the alpha-emitting substance has entered the body.

 Anhydrous;  Free from water,  especially water of crystallization.

 Aquifer;   A subsurface formation or geological unit containing
      sufficient saturated permeable material to yield significant
      quantities of water.

 Bars (pressure);  Absolute unit of pressure in the cgjs system
      equal to 1 dyne per square centimeter.

 Beta particle:   An elementary particle emitted from a nucleus
      during radioactive decay,  with a single electrical  charge and
      a  mass equal to 1/1837 that of a proton.   A negatively
      charged beta particle is identical to an electron.   A
      positively charged beta  particle is  called a positron.

 Biosphere;  Zone at  and adjacent to the earth's, surface where  all
      life  exists.

 Bittern:   The residual liquids  left  after  crystallization  is  complete.

 Calcine;   Product of  calcination wherein materials  are heated  to a
      higher temperature  under oxidizing conditions  but without
      fusing.

Diagenesis;  The physical and changes  that occur  to sedimentary
      rocks.  In  the history of a sedimentary rock  there is no
      point  at which change stops,  such as occurs in the solidification
      of a molten  igneous rock.  Thus are changes that occur with
      sedimentary  accumulations are known collectively as diagenesis.

Diapirism;  The forceful intrusion  of one geologic material into
      another; overlying geologic material.

Devitrification;  Crystallization  from a glassy phase.  Glass is an
      unstable substance.  Because  its components are spaced without
     plan, at unequal distances,, the forces of attraction that surround
     a "particle" are unbalanced.  Ultimately, the components respond
      to the unequal stresses and are drawn together to form crystals.
     This process is known as  devitrification.

                                   51

-------
Fission products;  The nuclei (fission fragments) formed by the
     fission of heavy elements, plus the nuclides formed by the
     fission fragments' radioactive decay.

Formation waters; Water that exists in the intergranular or fracture
     space in rock formations.

Gamma rays;  High energy, short wavelength electromagnetic radiation.
     Gamma radiation frequently accompanies alpha and beta emissions
     and always accompanies fission.  Gamma rays are essentially
     similar to X-rays but are usually more energetic and are nuclear
     in origin.

Half-life^ ^radioactive); The time required for one-half of an initial
     radioactive material to undergo nuclear transformation, the
     half-life is a measure of the persistence of a radioactive
     material and is unique to each radionuclide.

Halite; Impure common salt, NaCl.

High-level waste; The highly radioactive waste resulting from the
     reprocessing of spent fuel to separate uranium and plutonium
     from the fission products.  The term includes the high-level-
     liquid-wastes (HLLW) produced directly in reprocessing, and
     the solid high-level-wastes (HLW) which can be made therefrom.

Laminae;  Thin geologic layers, plates, or scales disposed in
     layers like the leaves of a book.

Lithologic; Pertaining to the characteristics of rock formations such
     as layering.

Magma; Molten igneous rock.

Montmorillonite;  A mineral resembling clay consisting of an
     hydrous aluminum silicate with considerable capacity for
     exchanging part of the aluminum for magnesium, alkalies
     and other bases.

Pluton;  Very large masses of igneous rocks that extend along the
     cores of most major mountain ranges and underlie vast areas of
     the ancient shield or central stable areas; sometimes called
     batholiths.

Salt dome; A type of geologic structure resulting from the upward
     thrust of a great salt mass through overlying rock layers.  The
     resulting salt form is roughly cylindrical and in some cases
     has resulted in observable uplift at the earth's surface.

Smectite minerals;  Clay minerals.
                                     52

-------
Tecjtonic: Of or pertaining to the formation  of  the  earth1 s crust;
     the forces involved in or producing  such deformation and the
     resulting forms.

Transuranic elements^: Elements with atomic number greater than 92.
     They include neptunium, plutonium, americium,  curium, and others.

Zeolite:  Any of a family of hydrous silicates, which have capacity
     to act as ion exchangers.
                    720-335/6105

-------