AECL--10277 ^ ^ ^ CA9200680

ATOMIC ENERGY & ^ M L'ENERGIE ATOMIQUE

OF CANADA LIMITED t j & J DU CANADA LIMITEE

A RADIONUCLIDE MASS-TRANSPORT MODEL

FOR THE PERFORMANCE ASSESSMENT OF ENGINEERED BARRIERS
IN A USED NUCLEAR FUEL DISPOSAL VAULT

MODELE DE TRANSPORT DE MASSE DES RADIONUCLEIDES

POUR I/EVALUATION DU COMPORTEMENT DES BARRIERES OUVRAGEES D'UNE
ENCEINTE DE STOCKAGE PERMANENT DE COMBUSTIBLE NUCLEAIRE USE

N. C. Goristo, D. M. LeNeveu

Whiteshell Laboratories Laboratoires de Whitesheli

Pinawa, Manitoba ROE 1LO
August 1991 aoOt
Copyright Atomic Energy of Carvia Limited. 1991.
 AECL RESEARCH

A RADIONUCLIDE MASS-TRANSPORT MODEL

FOR THE PERFORMANCE ASSESSMENT OF ENGINEERED BARRIERS
IN A USED NUCLEAR FUEL DISPOSAL VAULT

by

N.C. Garisto and D.M. LeNeveu

original contains
color illustrations

Whiteshell Laboratories
Pinawa, Manitoba ROE 1L0
1991
AECL-10277
 MODÈLE DE TRANSPORT DE MASSE DES RADIONUCLÉIDES
POUR L'ÉVALUATION DU COMPORTEMENT DES BARRIÈRES OUVRAGÉES D'UNE
ENCEINTE DE STOCKAGE PERMANENT DE COMBUSTIBLE NUCLÉAIRE USÉ

par

N.C. Garisto et D.M. LeNeveu

RESUME

On a réalisé le modèle d'enceinte pour évaluer le comportement des

barrières ouvragées d'une enceinte conceptuelle de stockage permanent du
combustible nucléaire usé. Le concept de stockage permanent évalué est
celui d'une enceinte scellée de façon étanche, exploitée à une profondeur
de 500 à 1 000 m dans la roche plutonique du bouclier canadien.

Dans le présent rapport, nous décrivons la structure conceptuelle et

mathématique du modèle de transport de masse des radionucléides utilisé
pour le modèle d'enceinte et examinons les hypothèses ayant permis d'en
tirer celle-ci. Le modèle représente le transport de masse des
radionucléides libérés des conteneurs de combustible usé et migrant à
travers les matériaux de scellement d'étanchéité à base d'argile et dans la
roche environnante; il applique des équations de diffusion - convection
unidimensionnelles pour le transport de masse des radionucléides d'une
chaîne de désintégration de longueur arbitraire dans des milieux finis; on
résout les équations semi-analytiquement à l'aide des techniques de
transformation de Laplace.

EACL Recherche
Laboratoires de Vhiteshell
Pinawa, Manitoba ROE 1L0
1991

AECL-10277
 A RADIONUCLIDE MASS-TRANSPORT MODEL
FOR THE PERFORMANCE ASSESSMENT OF ENGINEERED BARRIERS
IN A USED NUCLEAR FUEL DISPOSAL VAULT

by

N.C. Garisto and D.M. LeNeveu

ABSTRACT

The Vault Model has been developed to assess the performance of engineered
barriers in a conceptual disposal vault for used nuclear fuel. The dis-
posal concept being assessed is that of a sealed vault mined at a depth of
500 to 1000 m in plutonic rock in the Canadian Shield.

In this report, we describe the conceptual and mathematical framework of

the radionuclide mass-transport model used in the Vault Model and discuss
the assumptions used in its derivation. The model represents the mass
transport of radionuclides released from used-fuel containers, through
clay-based sealing materials and into the surrounding rock. It uses one-
dimensional diffusion-convection equations for radionuclide transport in a
decay chain of arbitrary length, in finite media. The equations are solved
semi-analytically using Laplace transform techniques.

AECL Research
Whiteshell Laboratories
Pinawa, Manitoba ROE 1L0
1991
AECL-10277
 CONTENTS

Page

1. INTRODUCTION 1

2. THE MODEL GEOMETRY 1

2.1 OVERVIEW 1
2.2 VAULT SECTORS FOR INTERFACE WITH THE GEOSPHERE MODEL 2
2.3 ONE-DIMENSIONAL MODEL GEOMETRY BASED ON DETAILED
THREE-DIMENSIONAL FINITE-ELEMENT CALCULATIONS 8

3. SYNOPSIS OF THE MATHEMATICAL METHODS 19

4. THE CONVECTION-DIFFUSION EQUATIONS 24

5. THE SOURCE TERMS 26

5.1 THE CONCEPTUAL SOURCE-TERM MODEL 26

5.2 THE INSTANT-RELEASE MODEL 26
5.3 THE INSTANT-RELEASE SOURCE TERM 27
5.4 THE LONG-TERM RELEASE MODEL 28
5.5 THE SOURCE TERM FOR THE LONG-TERM
SOLUBILITY-CONTROLLED CONGRUENT RELEASE 29

6. THE PRECIPITATION MODELS 33

6.1 PRECIPITATION OF URANIUM-CONTAINING SOLIDS 33

6.2 PRECIPITATION OF INDIVIDUAL RADIONUCLIDES 34

7. RESPONSE FUNCTIONS FOR THE MASS-TRANSPORT EQUATIONS 37

8. THE VAULT-GEOSPHERE COUPLING 38

8.1 THE MASS-TRANSFER COEFFICIENT APPROACH 38
8.2 THE MASS-TRANSFER COEFFICIENT FOR CONVECTIVE
MASS TRANSPORT IN THE ROCK 39
8.3 THE MASS-TRANSFER COEFFICIENT FOR DIFFUSIVE
MASS TRANSPORT IN THE ROCK AND BACKFILL 41
8.4 THE AUGMENTATION FACTOR FOR THE DIFFUSIVE
MASS-TRANSFER COEFFICIENT 43

continued...
 CONTENTS (concluded)

9. THE RELATIONSHIP BETWEEN THE VAULT MODEL AND DETAILED MODELS 46

10. SUMMARY 47

ACKNOWLEDGEMENTS 48

REFERENCES 48
 1. INTRODUCTION

The consequences to man and the environment that may result from the dis-
posal of used nuclear fuel at a depth of 500-1000 m in plutonic rock are
being assessed as part of the Canadian Nuclear Fuel Waste Management
Program (CNFWMP) (Hancox 1986, Dormuth 1991). These consequences are
determined by simulating the disposal system using the System Variability
Analysis Code (SYVAC) (Dormuth and Quick 1980). SYVAC is an executive
program for three models that represent radionuclide migration through the
three major parts of the disposal system: the vault (or the engineered
barriers system), the geosphere (or the host rock, surrounding the vault)
and the biosphere (or the environment accessible to man in everyday life).
SYVAC repeatedly samples model input parameters from given probability
distributions, calculates consequences and produces a histogram of pre-
dicted consequences versus their frequency of occurrence. Thus, to be
practical, the models used by the SYVAC executive have to be economical in
terms of computer resources required for their application. This is an
essential feature of models associated with probabilistic safety analyses,
which often require thousands of calculations to take into account the
variability and uncertainty in the disposal system (Garisto, N. 1986).

This report describes the conceptual and mathematical framework of the

radionuclide mass-transport model used in the Vault Model (VM) version of
1990 October. The VM is used by SYVAC to calculate radionuclide release
from the vault. It represents the mass transport of radionuclides,
released from used fuel upon container failure, through the clay-based
sealing materials surrounding the waste containers and into the geosphere.
Diffusion, convection, radioactive decay and buildup, and radionuclide
retardation effects are included in the model. Large parts of the October
1990 version of the VM, documented in this report, will be used in a com-
prehensive assessment of the nuclear fuel waste disposal concept, carried
out by the Canadian Nuclear Fuel Waste Management Program. The VM itself
is continuously being updated as more experimental and theoretical results
become available.

Verification of the VM was carried out in a series of scoping calculations

using comparison to analytical limits and to other near-field assessment
codes. These analyses, which are documented elsewhere (Garisto and LeNeveu
1989 and 1990, Apted et al. 1989, Engel et al. 1988) enhance our confidence
in the accuracy of the VM results.

2. THE MODEL GEOMETRY

2.1 OVERVIEW

The Canadian conceptual design for an underground nuclear fuel waste

disposal vault consists of a series of parallel rooms excavated at a depth
of 500 to 1000 m in the host rock. Boreholes would be drilled into the
floor of the rooms and waste containers would be emplaced in the holes and
surrounded by compacted buffer material composed of a bentonite/sand
mixture. After emplacement of the waste, the remaining excavated volume
 - 2 -

would be backfilled with a glacial lake olay/crushed-rock. mixture

(Baumgartner et al. in preparation). Schematic side and plan views of the
vault in a hypothetical location are shown in Figures 2-1 and 2-2,
respectively.

The calculated release rates from the vault are used as input rates into
the geosphere model (Goodwin et al. 1987), which calculates the transport
of radionuclides through the rock surrounding the vault and into the bio-
sphere. In the geosphere model, the geosphere is divided into segments
that represent pathways for mass transport in rock with different hydrogeo-
logical flow characteristics. To interface properly with the geosphere
model, the vault in the VM is divided into sectors (see Section 2.2).

Two- and three-dimensional models are very complex and require large compu-
ter resources. Therefore, they are not suitable at present for a direct
application in probabilistic assessments. For -; sample, a three-dimensional
calculation of purely convective mass transport using a very simplified
model for the vault (where the containers, buffer and backfill are assumed
to have the same hydraulic properties) requires approximately 16 000 nodes
in the finite-element code MOTIF (Chan and Scheier 1987). The central part
of a vertical section through the corresponding finite-element grid is
shown in Figure 2-3 and illustrates the complexity of the analysis, which
typically requires one day of CPU time en a FPS M64-140 computer just to
calculate the groundwater velocity field.

The VM described in this report uses one-dimensional planar geometry to

achieve efficiency in computer resources. That is, it uses one-dimensional
mass-transport equations that calculate transport across the various layers
that represent the vault. Therefore, the conceptual vault design must be
converted into a series of layers, or slabs, in each sector, for use in the
model (see Figure 2-4).

2.2 VAULT SECTORS FOR INTERFACE WITH THE GEOSPHERE MODEL

The hydrogeological finite-element code MOTIF (Chan et al. 1987) is used to

perform three-dimensional simulations of groundwater flow in the vicinity
of the vault. This detailed model treats both the fracture zones and the
relatively fracture-free rock units as equivalent heterogeneous media and
includes the backfilled disposal panels, tunnels, shafts and the
excavation-damaged zones. It predicts the spatial and temporal distribu-
tions of piezometric head and groundwater velocity. From this velocity
distribution, a numerical particle-tracking technique is used to derive the
convective travel paths and travel times of water-coincident tracer parti-
cles from the vault, through the geosphere, to the biosphere (Chan et al.
1986).

The particle paths determined from the detailed MOTIF model are then used
to construct the geometry of the SYVAC geosphere model, which consists of a
network of one-dimensional segments (see Figure 2-5). For each segment,
the geosphere model solves the coupled convection-dispersion-retardation
equation for a radionuclide decay chain using response function and Laplace
transform techniques (Heinrich and Andres 1985). The geosphere model
requires as input the groundwater velocity in each segment. This velocity
is obtained from the MOTIF calculations. The geosphere network model also
 Surface

iiiiliiiiniiii M11M IIIIM11111111!11111M

Vault
Exclusion
Side view zone
of the The major
boreholes fractures
Schematic vault geometry

FIGURE 2-1: A Schematic Diagram of a Side View of the Actual Vault. Geometry and the Surroundin.i;
Rock. Not to scale.
 N

E 2-2: A Schematic Plan View of the Vault and the Surrounding Area
 Vault
Fracture
Zones
Shaft

FIGURE 2-3: A Vertical Cross Section through the Central Part of the Three-Dimensional Grid for
MOTIF Calculations (Chan and Stanchell 1990). The scale of the vertical distances is
extended with respect to the horizontal distances by a factor of two.
 One-dimensional, one- and two-layer sectors with
different mass-transfer coefficients ( t )

Sector Rock fracture

Backfill

Buffer
Waste form

FIGURE 2-4: A Schematic Diagram of the Vault Model Geometry Showing One- and Two-Layer, One-
Dimensional Sectors. The arrows represent mass transport across layers and do not
signify any particular direction.
 Boggy Boggy
Pinawa Creek Creek
Channel NW SE Well
upper rock zone

intermediate rock zone

lower rock zone

u I t

Cross section with projection of transport network

FIGURE 2-5: A Schematic Example of the Mass-Transport Network in the Geosphere (see Figure .-. for
the cross-section location). LD1 is a fracture zone that intersects the vault. The
lines with the arrows delineate the geosphere segments.
 - 8 -

uses laboratory and field hydrogeochemical data for chemical retardation

and diffusion. The dispersion coefficient input into the geosphere network
model is chosen to be consistent with tracer-test interpretation (see
Section 8 ) .

The rate of input of radionuclides into each geosphere segment is equal to

the release rate of radionuclides from the corresponding vault sector.

An example of the source nodes for the geosphere network model is shown
schematically in Figure 2-6 in relation to the vault structure. In con-
structing the vault sectors, one source node is placed at, or near, the
centre of each sector (see Figure 2-6). The one-dimensional vault mass-
transport calculations are applied to each sector. Most sectors are clus-
tered near the fracture zone because the mass-transport properties vary
more rapidly in the vicinity of the fracture zone. Each sector is allo-
cated a fraction of the waste containers in proportion to its plan area.
The sectors also have different internal properties, such as container
failure rate and mass-transfer coefficients (described below), depending on
the location of the sector within the vault.

The VM calculations are done separately for each sector and intersector
coupling and vault edge effects are neglected. This is an essential fea-
ture of the one-dimensional approximation used in the VM (see Section 2.3).
A large part of the data used in the VM calculations is sector-dependent;
however, to simplify the notation, we do not use a sector-dependent sub-
script in the present report.

2.3 ONE-DIMENSIONAL MODEL GEOMETRY BASED ON DETAILED

THREE-DIMENSIONAL FINITE-ELEMENT CALCULATIONS

The rationale for assuming one-dimensional planar geometry in the VM is

based on preliminary scoping calculations done using the three-dimensional
finite-element hydrogeological code, MOTIF (Chan et al. 1987). A case of
mass transport by diffusion, with no groundwater movement, was analyzed
separately from cases with groundwater movement. A three-dimensional mesh
was set up to represent the buffer, backfill and rock. MOTIF was used to
calculate the concentration profiles of a non-sorbing tracer in space and
time.

The initial conditions used were such that the rock, buffer and backfill
were devoid of tracer while the waste containers had a uniform tracer con-
centration of 100 (arbitrary units). This concentration was assumed to
stay constant with time. The groundwater was considered to be stagnant so
that tracer movement was solely by diffusio..

The symmetry of the problem for diffusion allows us to consider only one
and a half boreholes in the three-dimensional mesh with 3656 nodes. The
centre borehole of an array of three boreholes across a rooa was bisected
through the vertical axis (see Figure 2-7). Tha vertical planar boundaries
of the mesh were located midway between adjacent rooms or boreholes
according to the symmetry of the vault design (see Figures 2-8 and 2-9).
Because of this symmetry, the vertical planes were considered to be zero-
flux boundaries for the tracer transport. Data on the geoaetry of the
various vault components are given by Atomic Energy of Canada Limited CANDU
Operations et al. (1987).
 FRACTURE ZONE

-15
18
6 12 ff-21

SECTOR
NUMBER
-14
5 8 II If "20

-SOURCE
NOL>

13
16
4 10 19

I \

EXCLUSION ZONE
FUIUKE 2 A Schema t. i i Kxample ot the S o u r c e N o d e s for the ileosphere M a s s - T r a n s p o r t Networl- with
Respect to the V a u l t . Plan view. N o t to s c a l e .
 backfill

.. . <±

A • • • . • *

' A ' . - A- •' • -A"

f
0D 0
buffer

waste container

o
r-emplacement room wall I

borehole

: waste container

.. * <**jj> f
FIGURE 2-7: The Emplacement Room and Boreholes Design (Atomic Energy of Canada Limited CANDU
Operations et al. 1987)
lover rock boundary

plane bisecting a room

plane midway between adjacent rooms

*H fracture
•h rnf:' r on.-HOI. ••] ,:. zone

FIGURE 2-8: Three-Dimensional Finite-Element Mesh for the Overall Layout of One and One-Half
Boreholes in a Slab of Rock
 - 12 -

top of the roon

plane bisecting a room

bisected borehole

.-.."'! . •'••••- • .intact borehole

FIGURE 2-9: Detail of the Three-Dimensional Finite-Element Mesh in the

Region of the One and One Half Boreholes
 In t h e f i n i t o c l e m e n t c a l c u l a t i o n s , t h e u p p e r h o r i z o n t a l plain-. .1' i<..iie-<
?() m a b o v e t h e c o n t a i n e r s . T h i s r e l a t i v e l y s h o r t d i s t a n c e w a s ir < <) '« an-
of p r a c t i c a l c o n s i d e r a t i o n s r e g a r d i n g t h e m e s h s i z e , r e s o l u t i o n of t h>
c a l c u l a t e d r e s u l t s , a n d t h e c o m p u t e r e x e c u t i o n t i m e , a n d it c o u l d a t i o t
the time d e p e n d e n c e of t h e l e s u l t s . A z e r o - c o n c e n t r a t i o n bound.11 , ii
tion w a s used at t h i s u p p e r s u r f a c e , i . e . , it v a s a s s u m e d that t h e \ 1 ,ne-1
w a s i n f i n i t e l y d i l u t e d o r s w e p t a w a y a t t h i s s u r f a c e . T h i s s u i f a i c \t\i<<
sents a highly conductive fiacture zone.

T h e lower h o r i z o n t a l p l a n e w a s g i v e n a z e r o - f l u x b o u n d a r y (onditioi.. •;.:

sin face v a s p l a c e d s u f f i c i e n t l y far from t h e c o n t a i n e r s ( 9 4 m be-iov ••••
c o n t a i n e r s ) s o a s n o t to s i g n i f i c a n t l y a f f e c t t h e t r a n s i e n t con<<;i'i .': •••
profiles above the containers.

F i g u r e s 2 -10 to 2 - 1 2 s h o w t h e c o n c e n t r a t i o n p r o f i l e s (O to l o o m :\:'. • • 1 n
u n i t s ) in a v e r t i c a l s e c t i o n t h r o u g h t h e c e n t r e o f t h e c o n t a i n e r • '• i
ferent times.

For d i f f u s i v e m a s s t r a n s p o i t , t h e d e t a i l e d t h r e e - d i m e n s i o n a l fini'< ...-••••>

m o d e l s h o w s d i f f u s i o n t o w a r d s t h e b a c k f i l l a f t e r a b o u t 1 0 0 0 y t a : ' <•••
Figui e 2 - 1 0 ) . T h i s e a r l y t i m e b e h a v i o u r is n o t a f f e c t e d h v t)v :.-••• :
b o u n d a r y c o n d i t i o n . A t l a t e r t i m e s , t h e c o n c e n t r a t i o n p t o f i l e 5 ; ( i:; '••
L o c k ) b e c o m e p a r a l l e l w i t h t h e g r o u n d s u r f a c e a n d t h e lock till: u;. .
the t r a c e r ( s e e F i g u r e s 2 - 1 1 a n d 2 - 1 2 ) . T h e h i g h c o n c e n t i a t ion- . i.: .:,<••
b e l o w t h e v a u l t after v e r y long t i m e s a r e a n a r t i f a c t c a u s e d by the
i m p e r m e a b l e b o u n d a r y c o n d i t i o n i m p o s e d at 9 4 m b e l o w t h e c o m a int-t • .

A laigei s c a l e w a s n e c e s s a r y to s i m u l a t e t h e e f f e c t s ot gi o u i u K a t * : •• . .
merit s i n c e t h e g r o u n d w a t e r f l o w f i e l d d e p e n d s o n h y d r a u l i c giadifMit- '',:;-'
a r e e s t a b l i s h e d o v e r a r e g i o n a l s c a l e o f s e v e r a l k i l o m e t r e s . A nun I.
c o a r s e r m e s h is u s e d f o r t h i s t y p e o f a n a l y s i s . I n t h e c o a r s e r ir.c-i.. •!•
p r o p e r t i e s o f t h e b a c k f i l l e d r o o m s w e r e m o d e l l e d , b u t t h e s m a l l r ; - -.,,
e f f e c t s o f t h e b o r e h o l e s w e r e i g n o r e d . T h e f i n i t e - e l e m e n t a n a l 'si 1 r !.<•-•
t h a t , f o r m o s t o f t h e v a u l t , g r o u n d w a t e r f l o w s f r o m t h e rock int<> 11.- ! , T !
f i l l ( s e e F i g u r e 2 - 1 ^ ) e v e n in t h e p r e s e n c e o f a n e x c a v a t i o n dama^,-. n .••!,»
( C h a n a n d S t a n c h e l l 1 9 9 0 ) . F o r t h e s e p o r t i o n s o f t h e v a u l t it > a n n.
i n f e r r e d that r a d i o n u c l i d e s w o u l d b e t r a n s p o r t e d t h r o u g h t h e b a c k t i . i !•
the p r o c e s s e s of c o n v e c t i o n a n d d i s p e r s i o n . T h e r e i s a s m a l l p < • 1t i > •. <•]
the v a u l t , p l a c e d to t h e r i g h t o f t h e f r a c t u r e that i n t e r s e c t s t h e -,;.!•
( i . e . , a b o v e t h e f r a c t u r e z o n e , s e e F i g u r e s 2-1 a n d 2 - H ) , w h e r e th>
g i o u n d w a t e r f l o w is d o w n w a r d . I n t h i s r e g i o n , w e conservatively." as-n--
thnt t h e r a d i o n u c ] i d e s d o n o t d i f f u s e a g a i n s t t h e f l o w , into t h< b-> :; :
( i . e . , t h e b a r r i e r p r o p e r t i e s o f t h e b a c k f i l l a r e o m i t t e d from tin '»')*. >. \\.
(he p r e s e n t m o d e l , t h e p o s s i b i l i t y that a p o r t i o n o f t h e radionur ii.i- -,.y
m i g r a t e a c r o s s t h e buffer" d i r e c t l y i n t o t h e s u r r o u n d i n g rock i "= not
c o n s i d e r e d in t h e v a u l t r e g i o n to t h e left o f t h e f r a c t u r e ? o n e (,.(
b e l o w t h e f r a c t u r e z o n e ) . T h i s p o s s i b i l i t y w i l l b e s t u d i e d in r !•• t.;T-i:«.
p a r t i c u l a r l y in t h o s e s e c t o r s w h e r e c o n v e c t i o n is f o u n d to domiit.iri 1 ii-
m a s s - t 1 a n s p o r t in t h e s u r r o u n d i n g r o c k .

B a s e d o n t h e p r e l i m i n a r y s t u d i e s , a m u l t i l a y e r g e o m e t r y h a s b e e n .• w•• >'
the V M to r e p r e s e n t m a s s t r a n s p o r t a t l o n g t i m e s . It c o n s i s t s ••( •:.!. •
l a y e r s : b u f f e r , b a c k f i l l a n d host rock ( s e e F i g u r e 2 - 1 4 ) to t h e 1 !< .<1 "•
f r a c t u r e z o n e a n d t w o l a y e r s , buffer a n d r o c k to t h e light of t h e ', \ .1 'si-.
 - 14 -

Surface
260

-4.0 -

Concentration
profiles at
-19.0 t = 1 0 7 9 years

~ -34.0 Rock
c
o

± -49.0
UJ

-64.0

-79.0

-94.0
-13.0 20
Distance (m)

FIGURE 2-10: Concentration Profiles in the Vault and Surrounding Rock fiom
a Constant Concentration Source from the Waste Containers
after 1079 years. Significant migration into the backfill
has occurred. This delays the progress of the diffusing
front to the surface.
 - 15 -

-Surface
26.0

1.0

-4.0

-19.0
'Concentration profiles
at 17 100 years

I "340 20
o

-49.0

-64.0

-79.0

-94.0
-13.0 2.0
Distance ( m )

FIGURE 2-11: Concentration Profiles in the Vault and Surrounding Rock from
a Constant Concentration Source from the Waste Containers
after 17 100 Years. Breakthrough through the backfill/rock
interface has occurred and some migration has occurred into
the rock region in between boreholes.
 - 16 -

Surface
26.0

I 1.0-

-4.0 -
Concentration
profiles of
t = \.7
-19.0

E
c
-34
T
o
o 70
>
-49.0

Rock

-64.0

-79.0

-94.0
-13.0 2.0
Distance (m)

FIGURE 2-12: Concentration Profiles in the Vault and Surrounding Rock from
a Constant Concentration Source from the Waste Containers
after 1.711 x 105 Years. Breakthrough to the surface has
occurred and concentration profiles have become relatively
parallel to the surface.
 - 17 -

z 300
o
H
250
W
W

Velocity Vector 3"

Tlme = Oa 6
Scale (log) 1 0 " 1m/a
400

2 °
H
> -400
W
W

-1200

Velocity Vector 3"

6
Time = 9800c Scale (log) 10~ 1m/a
400

-400

-1200
-1600 -800 0 800 1600

DISTANCE (m)
FIGURE 2-13: Water Table Profile and Calculated Velocity Distribution near
the Vault (Chan et al. 1986). This calculation takes .ieat
generation by the vault into account. The local surface
topography as reflected in the water table drives the flov.
At 9800 a, the heat perturbation of the velocity field is
large. The flow pattern is nearly vertical through almost
the entire vault because of the buoyancy effect of the waste
heat. Note that, at all times, flow is down into the low-
dipping fracture zone intersecting the vault, to the right of
this intersection.
 - 18 -

Buffer ^-Container

FIGURE 2-14: Simplified Vault Model Geometry for Mass Transport to the
Left of the Fracture Zone
 19 -

zone (see Figure 2-15). The use of a one-dimensional model to describe

mass transport across the buffer was justified by Garisto and Garisto
(1988b). In the VM, to the left of the fracture zone, the surface area of
each layer is equal to the surface area of the vault; to the right of the
fracture zone, the surface area of each layer is equal to the surface area
of a borehole . Ttiplied by the number of boreholes in the region. The
thickness of Li.e buffer layer in the model is the actual buffer thickness
and the thickness of the backfill layer is such that the volume of the
layer is equal to the total volume of the backfill in the rooms. The
rationale for this choice of geometrical parameters was discussed by
I.eNeveu (1986).

3. SYNOPSIS OF THE MATHEMATICAL METHODS

As discussed above, the three-dimensional conceptual vault design must

first be divided into a series of layers, representing the various engi-
neered barriers (see Figures 2-14 and 2-15). Mass transport is then calcu-
lated across these layers and into the surrounding host rock (Garisto and
LeNeveu 1989).

To solve the mass-transport equations, the initial concentration of radio-

nuclides in the medium (i.e., buffer, backfill or rock) is taken as zero,
the source boundary conditions are determined by models for the release of
radionuclides from used fuel (Garisto and LeNeveu 1987, 1989), and the exit
boundary condition assumes that the release of radionuclides from the vault
is proportional to their concentration at the vault/host rock interface.
The proportionality factor, referred to as the mass-transfer coefficient
(Bird et al. 1960), represents the resistance to radionuclide transport
across the vault/rock interface (see Section 8 ) . Thus, a large mass-
transfer coefficient approaches a zero-concentration boundary condition at
the vault-rock interface, whereas a small mass-transfer coefficient corre
sponds to a semi-impermeable condition at the interface (LeNeveu and
Garisto 1988). Detailed studies have shown that, with some parametric
adjustment of its value, the mass-transfer coefficient provides a rough
approximation for the peak of a release pulse of one-member decay chains
(Garisto et al., in preparation). The derivation of the mass-transfer
coefficient and a discussion of the validity of the mass-transfer coeffi-
cient approach and of its parametric adjustment are given in Section 8.

To interface properly with the geosphere model for the host rock, the vault
is divided into sectors (see Section 2) representing regions dominated by
different rock properties, as determined from hydrogeological studies of
the hypothetical disposal site in the Vhiteshell Research Area (Hancox
1986, Dormuth 1991). We then perform the calculations in each sector using
the properties of the host rock adjacent to each sector. The mass-
transport properties of the host rock affect the mass-transport processes
in the vault through the mass-transport coefficient (Bird et al. 1960,
LeNeveu 1986, LeNeveu and Garisto 1988).

To describe transport in the buffer and backfill, we have adopted a bulk

medium approach to account for porous medium effects (Eriksen and Jacobsson
1984, Skagius and ''eretnieks 1985, Lever 1986). In this approach,
 - 20 -

f "—
Buffer
Container

FIGURE 2-15: Simplified Model Geometry for Mass Transport to the Right of
the Fracture Zone
 - 21 -

effective steady state diffusion coefficients represent diffusion in the

buffer and backfill, whereas capacity factors, determined from the ratio of
diffusion coefficients measured i.:der transient and steady-state condi-
tions, represent storage and sorption effects (storage refers to ladio-
nuclides being held in the buffer pores even in the absence of sorption).
In the bulk medium approach, the total quantity of a species stored in a
unit volume of porous medium is given by the product of its pore-waler
concentration and the capacity factor of the medium (Lever 1986). The
model is sufficiently general to encompass diffusion via several pathways,
which is believed to occur in compacted clays (Torstenfelt and Allard 1986,
Cheung 1989).

Inherent in this approach is the assumption that all sorption processes

increase linearly with concentration in the pore water. An approximate,
equivalent linear sorption model can be used in the VM (Walker and LeNeveu
1987) if experimental data indicate that sorption processes are non-linear.
However, this equivalent model does not fully account for buffer saturation
effects (i.e., for saturation of sorption and storage sites on the buffer).
Discussions of the validity of the bulk-medium approach and the associated
data for the expected vault conditions are given by Cook (1988) and Cheung
(1989) respectively.

The present mass-transport model in the VM is applicable to mass transport

in porous, uncracked media. A separate mass-transport model was developed
for studying diffusion and convection of radionuclides in cracked, porous
media (Garisto and Garisto 1989, 1990a, 1990b). In addition, analytical
expressions were derived for the enhancement of the flux due to the pre-
sence of cracks in the buffer (Garisto and Garisto 1989). These expres-
sions can be incorporated into the VM to estimate the effects of potential
buffer cracks on the performance of the vault.

A response-function approach is used in the VM to solve the convection-

diffusion equations (LeNeveu 1987). The response functions are the solu-
tions to the mass-transport equations in a finite layer (e.g., buffer,
backfill) for a unit impulse input of radionuclide and are discussed in
Section 7. The response functions are calculated using Laplace transform
methods. Analytical methods (LeNeveu 1987) were used to invert the Laplace
transform solutions.

The release rate of a radionuclide into a layer, called the source release
rate, is convoluted with the response function to obtain the rate of
release from that layer (see Figure 3-1). The source release rate is
obtained by solving the convection-diffusion equation for appropriate
source boundary conditions.

Two source boundary conditions are used: one representing instant release
of )adionuclides upon container failure, and another representing long-
term, solubility-controlled, congruent release (Garisto et al. 1986, 1990;
Garisto and Garisto 1985; Lemire and Garisto 1989). Evaluation of the flux
at the source using these solutions gives the source term into the buffer.

The source boundary condition for congruent release of a radionuclide is

given by the dissolution rate of the used-fuel matrix multiplied by the
time-varying ratio of the mass inventory of the radionuclide in the matri'
 - 22 -

THE OVERALL PROCEDURE IN THE VAULT MODEL

Release rate
from the vault

Backfill
response
function Mass transfer
coefficients
Buffer
Radionuclide response
function
solubility
controlled
source term

Rate of
container Precipitation
failure

\nstant
release Matrix solubility
source term controlled
congruent release
I
Mass transfer
coefficients

FIGURE 3-1: The Overall Procedure in the Vault Model

 to the mass of the matrix. The dissolution rate of the used-fuel matrix ir
determined from a solution of the convection-diffusion equation with a
constant-concent rat ion (the solubility of the UO, matrix in the waste
container (Lemire and Garisto 1989)) source boundary condition and an exit
boundary condition (at the edge of the buffer) determined by a steady-.'tate
mass-transfer coefficient. Detailed studies have shown that the steady-
state mass-transfer coefficient is a reasonable boundary condition to apply
here (see Section 5 ) . Conceptually, this can be explained by the fact that
the transient source release is not greatly influenced by the nature of the
exit boundary condition. At early times, therefore, the solution should be
accurate. At long times, the solution should also be accurate because the
mass-transfer coefficient is evaluated for steady-state conditions that
apply at long times. At intermediate times, the solution could be somewhat
inaccurate because of the presence of the backfill and rock media, which
could prolong the transient regime. This inaccuracy should not pertain to
the dissolution rate of the used-fuel matrix in those cases vhere UO, pre
cipitates within the buffer and thus is not influenced by the rock and
backfill media. Inaccuracies could occur for the dissolution rate of the
Zircaloy fuel sheath and for the dissolution rate of a precipitated radio
nuclide. These inaccuracies are not expected to be larger than the range
of dissolution rates calculated from the uncertainty in solubilities.

The mass-transfer coefficient used to calculate the source term for rongi'i-
ent release is derived from an evaluation of the steady-state flux release
under two different conditions. Under the first condition, it is assumed
that mass transport in the buffer is dominated by diffusion, and mass
transport in the surrounding rock is dominated by convection, with water
moving at a constant velocity (both in space and time) over the surface of
the buffer. The barrier properties of the backfill are neglected in this
case. Under the second condition, all three-layers are included (buffet,
backfill, rock), and it is assumed that there is no water movement in the
system. The larger of these mass-transfer coefficients is used to deter
mine the dissolution rate of the matrix (for conservatism).

For the source boundary condition for instant release, it is assumed that,
at the time of container failure, the instant-release inventory is uni-
formly diluted throughout the available pore volume in the container, which
is assumed to be saturated with water. It is also assumed that the inven
tory is subsequently depleted by mass transport into the buffer. It is
further assumed that the buffer layer is semi-infinite. This is the same
type of approximation that is used to determine the matrix dissolution
rate, namely, that the transient source release is not greatly affected hv
the exit boundary condition. This approximation is valid if most of the
instant-release inventory was released during a short time, before the
occurrence of significant breakthrough into the backfill. A finite layer
model with a steady-state mass-transfer coefficient for the exit boundary
condition is not used for instant release because there is no steady state
in this case. The potential inaccuracy caused by the semi-infinite appro:;
imation is being investigated using an alternative approach that does not
require estimating mass-transfer coefficients (Garisto et al., in
preparat ion).

The derivation of the source terms for instant release and for solubility
controlled congruent release is given in Section 5. The instant-release
 rate and the congruent release rate are summed to give the total source
release of a radionuclide into the buffer.

After calculation of the total source release, precipitation is simulated

in the container (see Figure 3-1) by using a so-called compartment model.
For this simulation, individual radionuclide solubility-limited dissolution
rates must be provided. The solubility-limited dissolution rate is deter-
mined by the same method as the matrix dissolution rate, with the radio-
nuclide solubility replacing the solubility of the matrix. Should the
matrix-controlled congruent-dissolution rate exceed the radionuclide
solubility-limited dissolution rate, the compartment model is used to
determine the amount of the accumulated precipitate and to set the net
release rate equal to the solubility-limited dissolution rate. For simpli-
city, isotopes of the same element are treated separately (except for
uranium isotopes, which are treated more rigorously, see Section 6 ) . In
reality, isotopes of the same element would be expected to coprecipitate.
Since coprecipitation would lead to a lower effective solubility of an
isotope, it is conservative to neglect this effect. It is further assumed
that the effective dissolution area of the precipitate is the same as the
dissolution area used for the matrix.

The source release is then convoluted with the container failure rate (see
Figure 3-1), to account for the fact that not all containers fail at the
same time and thus dissipate the release rate into the buffer. This gives
the net release rate into the buffer for the entire vault, taking into
account the effect of precipitation. It is assumed that the buffer is
fully saturated with water at the time of container failure and, conserva-
tively, that the container has failed over the entire surface and does not
restrict the release to the buffer. The convolution is done only in time
and not in space. Thus the containers are being treated as a smeared
source whose input rate to the buffer gradually increases with time as more
containers fail. In other words, the time for the release from one con-
tainer to spread and interact with another container is neglected. This is
a reasonable approach because, at distances away from the containers that
are large compared with the separation distance between the containers, the
release from a series ~e "screte sources is expected to be indistinguish-
able from that of a single smeared source (Ahn et al. 1986).

The net release rate into the buffer is convoluted with the buffer lesponse
function to give the release rate from the buffer (see Section 7) All
contributions from precursors are summed during this process. Finally, the
release rate from the buffer is convoluted with the backfill response func-
tions in the region of the vault shown to the left of the fracture zone
intersecting the vault (see Figure 2-11) to obtain the release rate from
the backfill into the surrounding rock. In the region of the vault to the
right of the fracture zone, the radionuclides are assumed to be released
from the buffer into the rock without entering the backfill.

4. THE CONVECTION-DIFFUSION EQUATIONS

The governing equations used to describe mass transport in the vault are
the set of one-dimensional convection-diffusion equations (Crank 1975) for
a decay chain of arbitrary length:
 - 25 -

dC1 D? 3'Cj vB aCj

+ =0
3t r? 3x 2 rf 3x

3C2 ? 2 2
— + — + X2C2 -
3t rf 3x 2 rf 3x rf

3C, D?
=0
3t r? 3x 2 rB 3x

and the initial conditions are assumed to be

,x) = 0 , 0 < x < a for all radionuclides, 1 < j < i (4.2)

where .(x,t) is the concentration of a radionuclide in the pore water

of the medium (buffer or backfill) as a function of space
and time; 1 < j < i.
is the Darcy velocity in the medium;
is the total intrinsic diffusion coefficient (Lever 1986)
in the medium for the last radionuclide i in the chain;
is the capacity factor (Eriksen and Jacobsson 1984) of
the medium for radionuclide j , 1 < j < i; and
A. is the decay constant of radionuclide j , 1 < j < i.

The terms involving X i _ 1 represent the buildup of radionuclide i from its

parent, and the terms involving Xi represent the decay of radionuclide i.

We have assumed that all the radionuclides in a decay chain have the same
total intrinsic diffusion coefficient, D^, i.e., that of the radionuclide
under consideration. This simplifies the analytical solution of the equa-
tions (see Section 7 ) , which is otherwise extremely complicated. By using
the total intrinsic diffusion coefficient, D^, and the capacity factor, v*.
explicit use of a porosity variable in the equations is avoided. Thus, the
total intrinsic diffusion coefficient includes the porosity (s) as an
implicit variable (i.e., D^ = f(e)), and D^/r? is sometimes called the
transient diffusion coefficient, D^ i . The advantage in using these
variables lies in the fact that D? and D^ i can be obtained from direct
experimental measurements of diffusion under steady-state and transient
conditions respectively.

It is interesting to note that, if a linear sorption isotherm model (Freeze

and Cherry 1979) were used, the capacity factor could be expressed as

rf = e PK (4.3)
d#i

where p is the bulk density, e is the porosity, and K d t is the distribu

tion coefficient of radionuclide i (Eriksen and Jacobsson 1984).
 26 -

The actual diffusion processes in highly compacted clay-based materials are

quite complex. At least two diffusion pathvays have been identified: one
through pores between clay particles and another along the surfaces of tho
layers within the clay particles (Eriksen 1989). Mass-transport processes
in compacted clay-based materials are discussed in more detail by Cheung
(1989).

5. THE SOURCE TERMS

5.1 THE CONCEPTUAL SOURCE-TERM MODEL

The release of radionuclides from used fuel in contact with groundwater is

a complex process. Three overlapping mechanisms, which operate on very
different time scales, are involved in the release kinetics (Johnson and
Shoesmith 1988). First, there is a rapid release of soluble fission
products (such as cesium and iodine) from the fuel/sheath gap upon contact
with groundwater. Second, there is a more gradual release due to leaching
of grain-boundary species (such as iodine, cesium and technetium). Third,
there is a very slow release of radionuclides present in the U0 2 matrix due
to the dissolution of the U02 grains.

The radionuclide source-term model represents all three processes. The

model consists of two components: (i) an instant-release model, represent-
ing the first two processes (which are much faster than the third one under
the expected vault conditions); and (ii) a long-term release model, repre
senting congruent release by the third process, controlled by the dissolu
tion of the U0 2 matrix (Garisto et al. 1986). The conditions within which
the model is valid are discussed in detail by Garisto and Garisto (1985.
1986).

The release model provides boundary conditions for the solution of mass-
transport equations that simulate the migration of radionuclides away from
the used fuel/buffer interface. The solution of these equations provides
the source term, i.e., the radionuclide flow (due to all the release
mechanisms) at the used-fuel/buffer interface as a function of time.

5.2 THE INSTANT-RELEASE MODEL

The instant-release model represents the relatively fast release of soluble

fission products from the fuel/sheath gap and fuel grain boundaries. It
quantifies release in terms of instant release fractions (IRFs), i.e., the
fractions of the total used-fuel inventory of given radionuclides that are
available for instant release. The derivation of the IRFs for used CANDU:
fuel has been described by Garisto et al. (1990).

Immediately upon container failure, the instantly released radionuclides

from the fuel bundles (72 per container) are assumed to dissolve in the
groundwater inside the container and reach uniform, inventory-limited
concentrations (LeNeveu and Johnson 1986, Apted et al. 1987). The time
required for groundwater to flood the container and breach the Zircaloy

x
CANada Deuterium Uranium, registered trademark.
 - 27 -

fuel sheaths is taken conservatively as zero. The resulting radionuclide

concentrations inside the container then provide boundary conditions for
the mass-transport equations, that simulate the subsequent migration of the
radionuclides through the buffer. The corresponding radionuclide flow at
the used-fuel/buffer interface is the "instant-release source term".

Although ve include both gap and grain-boundary radionuclide inventoiies in

the instant-release fractions, it is important to emphasize that this is a
conservative approximation, which neglects the difference between the kine
tics of release from the gap and from the grain boundaries (Garisto et al.
1990). We use this approximation because it is difficult experimentally to
separate the individual releases from the gap and grain boundaries: the
same mechanisms, such as grain growth, bubble formation, and diffusion lead
to the accumulation of both gap and grain-boundary inventories. Further
more, only scarce experimental data are available on the chemical nature
and quantities of grain-boundary constituents and their rates of release
(Johnson and Shoesmith 1988).

5.3 THE INSTANT-RELEASE SOURCE TERM

Using the notation of Section 4, the instant-release source term, or the

input rate (to the buffer) for the instant release, Gj. for radionuclide i,
for a single-member decay chain is

9C.
? 3x I J
(5.1)
x=0

vhere Ci(x,t) is a solution to

dCi 32C1
XlCi = 0 (5.2)
3t

with the following boundary conditions at x = 0:

2 (5.3)
3t - X i p c V A C 1 + DB
3x
where 1° is a unit inventory,
I? is the instant-release fraction for radionuclide i,
I i (t c ) is the inventory of radionuclide i at the time of
container failure, t c ,
pc is the capacity factor in the container (this factor
is radionuclide-independent here because the instantly
released radionuclides do not sorb inside the container.
Consequently, the capacity factor represents here only the
storage (i.e., porosity) effect,
VA is the volume to surface area ratio of the container, and
6(t) is the Dirac delta function.

CjCxjt) remains finite as x -» » and Ci(x,0) = 0.

 - 28 -

The solution for G'(t) was obtained using Laplace transform techniques.
The transformed solution was inverted analytically and the solution is
given below. We define
(r B D J)l/2
b, = . (5.
VApc
If (b?t < 64), then

- b? exp(bft) erfc^t 1 ' 2 ) IJI.a.) . (5.5)

If (b?t > 64), then

G?(t) = . (5.6)
2bit3/2n1/2

5.4 THE LONG-TERM RELEASE MODEL

The major factor controlling the long-term release of radionuclides froai

used U0 2 fuel is the rate of dissolution of the U0 2 matrix, since most of
the radionuclide inventory is contained within the V02 grains (Johnson et
al. 1985). This rate is limited by the solubility of U0 2 (or U 4 0 9 ) if the
groundwater velocity is slow and groundwater chemistry conditions in the
vault are such that U0 2 (U 4 0 9 ) is the thermodynamically stable solid
(Paquette and Lemire 1981, Garisto and Garisto 1985). Under these condi-
tions, the dissolution is reversible and the rate of dissolution of U0 2
decreases as the concentration of uranium at the fuel/buffer interface
approaches the U0 2 solubility. Under other conditions, the kinetics and
mechanism of used-fuel dissolution will control the release of
radionuclides.

If the dissolution rate of U0 2 is solubility-limited, then the uranium

concentration at the fuel/buffer interface is given by the U0 2 solubility.
This (boundary) condition can be used to solve the diffusion-convection
equation describing the transport of uranium solution species away from the
fuel and through the buffer. The solution to this diffusion-convection
equation can be used, in turn, to calculate the flow of the dissolved
uranium species at the fuel/buffer interface, i.e., the U0 2 dissolution
rate. Under these conditions, the used-fuel dissolution rate depends on
the groundwater flow velocity, the diffusion coefficients of the uranium
solution species in the buffer, the sorption of uranium solution species in
the buffer and, of course, the solubility of U0 2 .

The rate at which a particular radionuclide is released from the used fuel,
the solubility-controlled source term, is given by the product of the UO,
dissolution rate and the fractional abundance of the radionuclide in the
used fuel. This congruent release model is supported by long-term leaching
experiments that show that the fractional releases of cesium, technetium,
 29

strontium, and uranium are comparable in spite of the widely varying chemi
cal properties of the leached elements (Johnson 1982, Johnson et al. 1982).

It is emphasized that a solubility-controlled radionuclide release model

may not be valid if U0 2 (or U4Oq) is not the thermodynamically stable
uranium solid in the vault environment. It is, therefore, necessary to
know the regions of stability of the uranium solids that can form in the
environment of an underground vault in order to determine the region of
validity of solubility-limited dissolution models. The formation of some
uranium solids (e.g., U0 3 ) that proceeds through an initial U0 2 dissolution
step could invalidate the concept of a solubility-controlled radionuclide
release rate. This is particularly important in the presence of
a-radiation, which could enhance the oxidative dissolution of the fuel and,
therefore, invalidate the U0 2 solubility-controlled release model. Elec-
trochemical and X-ray photoelectron spectroscopic (XPS) studies indicate
that, although U0 2 is rapidly oxidized in oxygenated water through the
sequence U0 2 -» U0 2+x -> U 4 O 9 -> U 3 O 7 , no significant dissolution occurs until
a composition close to U 3 O 7 is reached (Shoesmith et al. 1984). Conse-
quently, a solubility model, based on U^0g solubility, may be applied to
the dissolution of used fuel in a disposal vault if it can be shown that
under these conditions oxidation does not proceed beyond a composition of
approximately U 4 0 9 (Aronson et al. 1957). Preliminary research on U0 2
dissolution seems to indicate that oxidation beyond U 4 0 9 will not occur
under the geological disposal conditions expected in the Canadian Shield
(Sunder and Shoesmith, in preparation).

The VM has the capability to simulate either a solubility-limited or a

kinetically limited release, given the appropriate data (i.e., thermodyna-
mic or kinetic data respectively). However, the data on the release kine-
tics under the expected vault conditions are not available at present. In
the current study, we have focused on the solubility-limited release model
while research is under way to determine which of the two approaches is
more suitable for modelling the long-term release of radionuclides from
used fuel (Sunder and Shoesmith, in preparation).

5.5 THE SOURCE TERM FOR THE LONG-TERM

SOLUBILITY-CONTROLLED CONGRUENT RELEASE

The rate of input into the vault of a radionuclide released from the used
fuel matrix by a solubility-controlled, congruent dissolution of the matrix
m, i.e., the source term Flin(t), is given by

1,(0
F im (t) = F»(t) (5.7)
K
where the subscript m is used so that the source-term model can be
generalized to calculate release from several matrices
(e.g., used fuel and Zircaloy),
I1(t) is the inventory of radionuclide i in the matrix,
I is the inventory of the matrix, and
 - 30 -

(5.8)
3x |x=0

where A£ is the surface area of the equivalent layer used to represent the
vault for the matrix, m, and cm(x,t) is a solution to the diffusion-
convection equation

ac. (5-9)
2
=o
at 3x

with a constant-concentration boundary condition

Cm(O,t) = C* at x = 0 (5.10)

and a flux boundary condition with a mass-transfer coefficient (K s )

at the edge of the medium

3C.
-D2 at x = aB (5.11)
3x
In Equations (5.8) to (5.11), C m , D£ and r^ pertain to properties
of radionuclide i (from matrix m ) . See definitions for Cif Df and r?
in Section k.

The derivation of the mass-transfer coefficient is given in Section 8.

The initial conditions are

C.(xf0) = 0 0 < x < aE (5.12)

where C w (x r t) is the concentration of the matrix of the waste

form in the pore water,
Ks is the source mass-transfer coefficient,
A^ is the decay constant of the matrix, and
C* is the solubility of the matrix in the pore water.

The solution for F^J<t) and the absolute error in FjJ(t), denoted E^(t), are
given below. The solution was obtained using Laplace transform techniques
and the transformed solution was inverted analytically using the techniques
described by LeNeveu (1987).

D2t
If < 1, then
rg(a B ) 2

VB
 Jl

£ exp(-ftt - • 4AECJG' e x p ( - f t t ) erfcdjj,)

(Jit)1/2
(5.13)

2C*
Ejj(t) = exp(-4[T*| 2 )
1 2
(nt) '

° exp(-4[T*]2) 4Gm
l/2
l/2
T
l(Yn)

(5.14)
D2t
and if > 1, then
r°(aB)2

a 2 (a 2 + G2 )exp[ ( [ -

K * G2 ') 2 rS/D B ]
(5.15)
where an are the roots of

a cot<<r) + Gm = 0 (5.16)

l/2
ES<t) = erfc (2n - 1) (5.17)
(lit) 1/2

where

-(vB):
(5.18)

(5.19)

(5.20)
DBrB
 - 32 -

~in in m

r/*%B \ 2 »«B T 1 / 2
TjJ = (5.22)

L Dgt J
Cg = C»A£(Dgrg)i/2 (5.23)

GS = (K» - v»/2) (5.24)

Gm = G-aB/D» (5.25)
1/2
Ym = TJ • G-(tS) (5.26)

E^ = erfc(T^) (5.27)

exp[-(T-)2l
(5.28)

XS = r<Em - 2T
n X m) (5.29)

Cg = GJC^Aji (5.30)

0* = exp(-(3j;t) (5.31)

Rg = 4C={E- - exp[Y2 - (T-)2]erfc(Ym)} . (5.32)

It should be noted that the relationship

erfc(x) = jr 1 / 2 e~x ( — - — + -^— - ...) (5.33)

x 2x3 23x5

has been used to make the determination of Rjj; numerically stable (Carslaw
and Jaeger 1959).

For testing the accuracy of numerical routines used for calculating

Fj^(t), it is useful to know the steady-state value, F^, as X, •> 0 and
as vB -* 0 (Garisto and LeNeveu 1989). Under these limits

C
mAm
(5.34)
1 aB
1 ™
+
Ks D?
 - 33 -

6. THE PRECIPITATION MODELS

6.1 PRECIPITATION OF URANIUM-CONTAINING SOLIDS

The enhancement of the dissolution rate of used fuel due to the mass
transport-precipitation coupling effect (i.e., the enhancement factor) has
been investigated in detail using analytical (Garisto, F. 1986, Garisto and
Garisto 1988a) and numerical (Garisto and Garisto 1986, Garisto and Garisto
1988b) mass-transport models. Quantitative estimates of the release
enhancement factor were derived for fuel dissolution under probable dis-
posal vault conditions (Garisto and Garisto 1986, Garisto and LeNeveu
1989). Upper bounds on the release enhancement factor were also determined
(Garisto and Garisto 1988b) and explicit expressions for the asymptotic
(t -» «°) behaviour of the fuel dissolution rate were derived (Garisto and
Garisto 1988a).

The detailed calculations indicate that the enhancement factor is small

(=1) at short times and increases for very long dissolution times. For
finite system geometries (such as the used-fuel/buffer system), the
enhancement factor asymptotically approaches a constant value as t -* °°.
Thus, precipitation of uranium-containing solids in the buffer limits the
release of uranium (and of other coprecipitates) into the geosphere, but
enhances the dissolution rate of the used fuel, thereby increasing the
fission-product release rate.

The detailed calculations also show that the enhancement factor depends
strongly on the magnitude of the uranium solubility gradient near the used-
fuel/buffer interface (Garisto and Garisto 1986, Apted et al. 1989). This
gradient is largely controlled by the extent of local variations in redox
conditions.

In the VM, it is assumed that radiolysis of the groundwater adjacent to the

fuel bundles will produce a redox gradient, such that the system potential,
E z (i.e., E h at pH = 0 ) , is more oxidizing at the used-fuel surface than in
the buffer (Garisto and Garisto 1986). To determine the solubility gradi-
ent, two uranium solubilities, C p and C N , are calculated from the equations
developed by Lemire and Garisto (1989) for the redox potentials at the
used-fuel surface and at a distance, a u , in the buffer respectively. All
other input variables, such as groundwater composition and temperature, are
considered to be the same.

The system potential at the used-fuel surface, Ef, is sampled from a

uniform distribution between 0 and 0.516 V (Lemire and Garisto 1989). The
system potential in the buffer, E p , is then calculated from the following
equation:

E p = RB(min(E*, E p ) (6.1)

where E| is the maximum buffer potential, and

R^ is a random factor chosen so that E p would be a uniform
distribution between zero and E p .
 - 34 -

The conditions for precipitation of uranium species to occur have been

investigated by Garisto and Garisto (1938a). If the uranium solubility in
the buffer is less than the steady-state concentration obtained without
precipitation, then precipitation is considered to occur.

If precipitation occurs, the dissolution rate of the fuel, (F^(t), is

determined from Equation (5.8) with Cjj; = C N - C p and a medium length equal
to * u instead of the buffer thickness a B (at: < a B ) . In the VM, as a u
becomes smaller, the uranium concentration gradient increases and the fuel
dissolution rate diverges. However, the detailed analysis by Garisto and
Garisto (1988a; has shown that for finite system geometries the enhancement
of the dissolution rate due to precipitation increases with time (for
intermediate times) but approaches a constant value in the t •* m limit.
Based on this analysis, the following relationship is used to establish a
t on a",

au = min(aB, D*/au + a 1 ) (6.2)

where D^ is the total intrinsic diffusion coefficient for uranium

species, (see Section 4 ) ,
au is the dissolution rate constant for U0 2 in buffer, and
a 1 is a sampled gradient length.

The limiting value of D^/a" and, hence, of au as a 1 tends to zero is

obtained from Equation (20) of Garisto and Garisto (1988a). When L p , the
precipitation distance in this equation is allowed to approach zero, the
mass release, M p (t), at long times becomes aa(C1 - C p )t, where aa is a
dissolution rate constant. The mass release from the equations used in the
VM at long times and small medium thicknesses becomes D(C: - Cp)t/a, where
a is the medium thickness and D the diffusion coefficient- Equating these
two asymptotic mass releases gives a value of a = D/a2, or Dg/o^ in the
notation of this report.

The two expressions for mass release described above differ because, in the
vault model, it is assumed that a saturation concentration at the source is
obtained immediately, whereas in the report by Garisto and Garisto (1988a)
the time to saturation is determined by the dissolution rate constant, <t1.
This will result in a small error in the VM in estimating the dissolution
rate at short times.

When precipitation occurs, the release of 2 3 8 U into the geosphere is deter-

mined from the dissolution rate of the precipitate. This is determined
once more from Equation (5.8) with C^ = C p .

The calculated release flux of uranium F^(t) is then convoluted with the
container-failure function to account for the spread in the container-
failure time. This convolution is similar to that done for other released
radionuclides and is described in the next section.

6.2 PRECIPITATION OF INDIVIDUAL RADIONUCLIDES

To determine whether or not precipitation (of radionuclides other than

uranium) occurs in the vault, we first sum the instant release, G*(t), and
congruent release, Fiin(t), and the sum (FT(t)) is then convoluted with the
 - 35 -

container-failure function. Note that the convolution integral includes a

correction facto- to account for the changing radionuclide inventory in the
container (due to radioactive buildup and decay) that is available for
release as a function of the container failure time:

FT(t) = (t') + F iB (t')] - t') dt (6.3)

where

= /c(t) V XJq exp (-\t) (6.4)

q=j

j = l = q (6.5)
x? q -
n x. fi-1 >-i , j* (6.6)

where / c (t) is the container-failure function, i.e., the fraction of the

containers failing as a function of time. Estimates of the function / c (t)
are available in the literature for various container designs (Liebetrau et
al. 1987, Garisto and LeNeveu 1989) and, in particular, the container-
failure function for the reference container design in the CNFUMP has been
described by Ikeda and Shoesmith (1990).

The physical interpretation of the convolution in Equation (6.3) is that

the containers are treated as a smeared source whose input rate into the
buffer changes gradually with time because of the spread in the failure
time of the containers.

If FT(t) exceeds the solubility-limited release from used fuel, F^(t), or

if any amount of precipitate A A (t) has accumulated, then the release from
the used fuel, F£(t), is given by F?(t), i.e.,

if A ^ t ) > 0 or FT(t) > F?(t), then (6.7)

Fj(t) = Ff(t)

otherwise,

F?(t) = FT(t) + Pi.jd) (6.8)

where P i _ 1 (t) is the contribution to F^(t) from the decay of a precipitated

parent,
 and

A.(t) = Ff(t - t')Ri'(t') dt' F?(t - t')Rf(t') dt'

- t')RP(t') dt' (6.10)

where

Rf(t) = (6.11)

In the right-hand side of Equation (6.10), the first term represents the
time integration of the total radionuclide inflow, including decay. The
second term represents the time integration of the solubility-limited out-
put, including decay, and the third term represents the time-integrated
contribution from the decay of the parent. F?(t) is given by convoluting
Ff(t) with the container-failure function. F?(t) is given by the solutions
to Equations (5.8) and (5.12) using the solubility of radionuclide i
instead of the solubility of the used-fuel matrix, i.e.,

Ff(t) = Ff(t') ;fc(t - t') dt' (6.12)

Note that the correction for inventory changing as a function of time is

not required here because Ff does not depend on inventory. Note also in
this application, the coprecipitation of isotopes is only considered for
uranium isotopes. In this case, the amount of the 2 3 8 U isotope far exceeds
the amount of other uranium isotopes. It is thus assumed that Fj'(t) for
uranium isotopes is given by

F?(t) = (6.13)
12 38

where Ii(t) is the inventory of a uranium isotope as a function of time and

L
U238
is the initial inventory of
constant in time.) For all other
considered (because of difficulty
tive approximation because higher
pitation is neglected.
238
U . (This inventory is assumed to be
elements, isotopic coprecipitation is not
in implementation). This is a conserva-
concentrations are obtained when copreci-

The effect of precipitation outside the container (e.g., in the buffer) on

the mass transport of radionuclides is not included in the VH at present
(except for uranium-containing solids). This effect is currently being
investigated (Garisto and Garisto, in preparation).
 - 37 -

7. RESPONSE FUNCTIONS FOR THE MASS-TRANSPORT EQUATIONS

In the response function approach, the convection-diffusion equations are

solved for a unit impulse input. For this study, a mass-transfer coeffi-
cient has been used to specify the exit boundary condition. Thus, the
boundary conditions for the response functions at x = 0 are

3Cj
v B C j = I^ (7.1)
3x
|x=0

where 1^ is a unit amount of radionuclide j , and at x = a B ,

3C, '
-DB | + vBCj = 1 < j <i (7.2)
9x I BB
|x=a

where K? is the mass-transfer coefficient for radionuclide i and a B is the

thickness of the medium. Note that we assume that all the radionuclides in
a decay chain have the same mass-transfer coefficient, K|, where i refers
to the last member of a radionuclide decay chain. This assumption simpli-
fies the analytical solution, which is otherwise extremely complicated.
However, this assumption is not always valid because the mass-transfer
coefficient could depend on all the members of the radionuclide chain, not
just the last member (see Section 8 ) . The assumption that all radio-
nuclides in a chain have the same mass-transfer coefficient is valid under
conditions of high convective flow in the rock, where the mass-transfer
boundary condition can be replaced by a zero-concentration boundary condi-
tion. Under such conditions, mass transport is essentially independent of
the mass-transfer coefficient (LeNeveu and Garisto 1988). Such conditions
were considered by Neretnieks (1978) who has also used the equal K?
approximation for all members of a decay chain.

The initial conditions for calculating the response functions are

C.j(x,O) = 0 0 (7-3)

The release rate, R A j(t), of radionuclide i per unit impulse input of

radionuclide j from a medium is given by

Ri;j(t) = KfCi^a 8 ,!) (7.4)

where C ij (a B ,t) is the contribution to the concentration of radionuclide i

due to the unit impulse input of radionuclide j , 1 < j < i.

In Equations (7.1) to (7.4), the superscript B refers to the buffer. The

same response functions are used for the backfill, with the superscript B
replaced by F. The response function solution is given by LeNeveu (1987).
 The solution is obtained using Laplace transform methods. The Laplace-
transformed solutions are then inverted analytically (producing infinite
series) using two complementary methods, one that converges quickly for
small values of time and one that converges quickly for large values of
time.

The range of conditions for convergence of the solutions (LeNeveu 1987) has
been extended following the methods outlined by Carslav and Jaeger (1959).

A numerical routine from the SYVAC executive program (Goodwin et al. 1987)
is then used to convolute the analytical response functions with given
time-dependent sources, to solve for radionuclide transport in a finite
one-dimensional region, provided that a suitable value for the mass-
transfer coefficient has been determined (see Section 8 ) . Thus, for the
given source terms, S (t), the flow Ft(x,t) of radionuclide i from the
system defined by Equation (A.I) is given by convoluting the response func-
tion with each precursor source term and summing over all the precursors
including the last radionuclide, i,

S:(t')R15(x,t-t') dt' . (7.5)

8. THE VAULT-GEOSPHERE COUPLING

8.1 THE MASS-TRANSFER COEFFICIENT APPROACH

To solve Equation (4.1), we have to specify the radionuclide mass-transfer

coefficient, K?, for use in the response functions for the buffer; Kf, for
use in the response functions for the backfill; the source mass-transfer
coefficient, K s , for use in calculating the used-fuel matrix release rate,
F]^(t); and K?(t) the precipitation mass-transfer coefficient for use in
calculating the solubility-limited release rate from the used-fuel matrix.
The calculation of these four mass-transfer coefficients is based initially
on a diffusive mass-transfer coefficient, K£, calculated assuming that the
water in the rock surrounding the vault is stagnant; and a convective mass-
transfer coefficient, K c , calculated assuming mass transport in the rock is
dominated by convection. Both K£ and K c ?.ie calculated from an analysis of
steady-state concentration gradients. The subscript n refers to the number
of layers in the system. In our case, n = 1 refers to the buffer, n = 2
refers to the backfill and n = 3 refeis to the rock. The subscript i
referring to the radionuclide for which the calculations are done, is
omitted henceforth from Section 8 for clarity.

For a constant-concentration source boundary condition, radionuclide decay

has the effect of increasing the concentration gradients along the trans-
port path. Thus, Ks and K r , determined for steady-state conditions, are
functions of the radionuclide decay constant and increase with increase in
the decay constant. However, for an instant-release source boundary condi-
tion, the decay constant would not have the same effect because the radio-
nuclide concentration is not continuously replenished to create a constant
 39 -

source concentration. K B and K F are calculated using the asymptotic foim

of the mass-transfer coefficient equations for a zero decay constant.

During transient diffusion, the concentration gradients would generally be

expected to be larger than during steady state. From detailed studies of a
case with stagnant water, a method has been devised for increasing the
value of the steady-state mass-transfer coefficient to account for this
effect in calculating the response functions. The values of both K B and K1
are augmented to account for the transient conditions.

Other methods for solving the mass-transport equations directly, without

using the mass-transfer coefficient approach, have been recently developed
(Garisto et al., in preparation). A comparison of the VM results using the
various solution methods indicates that the mass-transfer coefficient
provides a reasonable approximation for one-member decay chains. Under
certain conditions, however, it is a poor approximation for calculating the
mass transport of radionuclides in a decay chain. The VM code includes an
option that allows the use of direct solutions that do not involve mass-
transfer coefficients. Results from these different solution methods will
be compared in a future document (Garisto et al., in preparation).

8.2 THE MASS-TRANSFER COEFFICIENT FOR

CONVECTIVE MASS TRANSPORT IN THE ROCK

A two-layer model is used to calculate K c . In the first layer (buffer), it

is assumed that mass transport occurs by diffusion. In the second layer
(rock), it is assumed that mass transport occurs by convection in the
direction parallel to the interface between the two layers and by disper-
sion in the direction perpendicular to the boundary between the layers.
The geometry in this situation is two dimensional. The mass-transfer
coefficient, K c , is determined by the average flux along the interface
divided by the concentration at the interface, C B . The geometry is illus-
trated in Figure 8.1 where

DR is the transverse dispersion coefficient in the rock,

q is the Darcy velocity in the rock,
w is the length of the container/buffer interface
CB is the concentration at the interface (assumed constant at
all locations on the interface), and
x and z are the spatial coordinates.

The governing equation is

D R 32C q 3C
kC = 0 (8.1)
r 9z

where r is the capacity factor in the rock.

The boundary conditions are

C(x,O) = 0 0 < x < °° (8.2)

 t
O

S w
£ o
o
QQ
\
K'

FIGURE 8-1: The Geometry for Deriving a Convective Mass-Transfer

Coefficient

C(O,z) = C B 0 < z <w (8.3)

C(x,z) remains finite a s x - * 0 1 5 0 < z < w (8.4)

The mass-transfer coefficient is given by

dz . (8.5)
J0 n n—
9x

It turns out that K c does not depend on C B :

K c = — — erf — (8.6)
v q

As X -» 0, the expression for Kc approaches the usual form for a convective

mass-transfer coefficient (Bird et al. 1960),

(8.7)
 - 41 -

Detailed calculations have shown that fluxes calculated using the mass-
transfer coefficient reach an asymptotic value for values of the mass-
transfer coefficient, K c , greater than 50D/a, where D is the diffusion
coefficient in the buffer and a is the thickness of the buffer. This
corresponds to a C = 0 exit boundary condition.

It should be noted that Kc depends on w; v is taken as the length of the

container, based on regional flow field studies, which indicate that the
groundwater flow is generally vertical (i.e., approximately parallel to the
longitudinal axis of the containers (Chan and Scheier 1987)).

Equations (8.1) and (8.7) are also used to calculate a convective mass
transfer coefficient for the backfill. Based on the same reasoning as
above, the mass-transfer length, w, is taken to be the vertical height of
the backfilled room. These choices of w are not always conservative (see
Equation 8.6). The validity of the present choices of w will be investi-
gated by comparing the VM results using Kc to the results of a mass-
transport model that does not use the mass-transport coefficient approxi-
mation. This fully coupled model is discussed by Garisto et al. (in
preparation).

Because of lack of data for the transverse dispersion coefficient in the

rock, we have used data on the longitudinal dispersion coefficient instead.
Our use of dispersion coefficients based on large-scale field tests may not
be appropriate for the vault conditions, where small-scale variations are
perhaps more important than large-scale averages (Neuman 1987). However,
the approximation of using the longitudinal rather than transverse disper-
sion is generally considered to be conservative because it leads to a
higher value for the mass-transfer coefficient and, hence, a higher release
from the vault.

8.3 THE MASS-TRANSFER COEFFICIENT FOR DIFFUSIVE

MASS TRANSPORT IN THE ROCK AND BACKFILL

The mass-transfer coefficient K£ is determined for a three-layer system

(see Figure 8-2): buffer, backfill and rock undei. ste-*d>-state conditions.
It is assumed that there is a zero-concentration boundary at the exit from
the rock layer. Continuity of flux and concentration at the interfaces
with the other layers is assumed and the water in the system is assumed to
be stagnant. Two mass-transfer coefficients, K® and K°, are defined foi
the exits from the first and second layers (buffer and backfill respec-
tively). The mass-transfer coefficients are the ratios of the steady-state
fluxes to the concentrations at the exits:

at x = an n = 1, 2 (8.8)

where Dn is the total intrinsic diffusion coefficient in layer n,

C n is the concentration in layer n, and
a n is the thickness of layer n.
The governing equation is

n n n = 1, 2, 3 (8.9)
- AC_= 0

where rn is the capacity factor in layer n, and

X is the radionuclide decay constant.

The boundary conditions are (see Figure 8-2)

C = C o (constant) at x = 0 (8.10)

1+1 at x = a. n = 1, 2 (8.11)

at x = a. n = 1, 2 (8.12)
3x

= 0 at x = a. (8.13)

i.e. (8.14)

n=3

K
Q Backfill M n=2
aKfe^^/imt^ff
Buffer n=i

FIGURE 8-2: The Geometry for Deriving a Convective Mass-Transfer

Coefficient
 The solution for K£ is

(D n + 1 b n + 1 ) 2 sinh(b n + 1 ) + D n + 1 b n + 1 a n + 1 K£ + 1 cosh(b n + 1 )
KD = . (8.15)
D n + 1 b n + 1 a n + 1 cosh(b n + 1 ) + (a n+1 ) 2 K£ +1 sinh(b n+1 )

In the limit as K^ -* °° (corresponding to a zero-concentration boundary

condition at the exit from layer 3), the expression for Kf becomes

D 3 b 3 cosh(b 3 )
KD = (8.16)
a3sinh(b3)

where

K=
In the limit as X -> 0, the expression for K£ reduces to the familiar
equation for addition of resistances,

1 1 an+1
— = + . (8.18)

The above equations for K° can be applied to a medium of any number of

layers.

8.4 THE AUGMENTATION FACTOR FOR THE DIFFUSIVE

MASS-TRANSFER COEFFICIENT

The augmentation factor was determined by comparing solutions obtained

using mass-transfer coefficients with numerical solutions obtained using
continuity of flux and concentration at the boundary for a two-layer (e.g.,
buffer or backfill and rock) system.

The coupled equations are

= XCn n = 1, 2 (8.19)
3t rn dx*

with the boundary conditions (8.10)-(8.13) and either

-Di = Iu6(t) (8.20)

3x „.„
 or
= Cfl at x = 0 (8.21)

The equations were solved numerically using Laplace transform techniques

and Talbot's inversion algorithm (Talbot 1979). Based on the numerical
calculations, we found that it is necessary to augment the value of the
steady-state diffusive mass-transfer coefficient by an increasing amount
(exp(y n )) as the ratio between the breakthrough time to steady state in the
remaining medium (i.e., the rock.) to the duration of the response to unit
impulse in the first medium (i.e., the buffer or backfill) increased. The
value of the augmentation factor was determined by fitting the peak release
from the second layer, calculated with the mass-transfer coefficient to the
numerical solution. The fitting calculations were all carried out without
radioactive decay (i.e., X = 0 ) .

Thus, for use in the response functions, the steady-state mass-transfer

coefficients, K ° , (see Section 8.3) are augmented by exp(y n ), i.e.,

K B or K F = K£ exp(y n ) (8.22)

where

rbxn
(8.23)

(8.24)

gn = (8-25)
L n d_ . -, \J\,
.n n
k=n+l k

d = 2.131926, (8.26)

r a = 2.030576, (8.27)

r b = 2.045500, and (8.28)

n L is the number of layers.

Note that in Equations (8.21)-(8.26), d, ra and r b denote fitting para-

meters, i.e., the values of the parameters, d, r s and r b vere determined by
a parametric analysis of the augmentation factor required to match the peak
release rates of the two solutions at the exit from the second layer. The
calculation of the augmentation factor has been generalized to describe
release from a multilayered system. The resulting augmentation factor is
shown in Figure 8-3.
 12 1
1 • 1 1
i • i i

j
i
•
i
10k j.

i
r
i
-D 8 _

i
i_

o i _
i

o i

& 6 —
f
1
I
o /
f

- 4 _

/
/ -

t
— -
, *

21-
~ _._•£ • - - - * —
-!-•*---!— 7 i i l i I i

-3.0 -2.0 •1.0 0.0 1.0 2.0

!og e (ratio)
FIGURE 8-3: The Augmentation Factor as a Function of the Ratio Between the Breakthrough Time to
Steady State in the Rock to the Duration of the Response Function in the Buffer
In the VM, the value of the steady-stat- mass-transfei <oetfieient is
generally augmented by exp(y n ). However, the value of the augmented mass-
transfer coefficient was often found to exceed 50 Dj/a (the asymptotic
value corresponding to zero-concent rat ion exit boundaiy condition). In
these cases, the mass-transport calculations aie carried out using a trun-
cated value, i.e., K" or K F = SODj/a. A lover limit of vB/2 or v r /2 is
also applied to the value of K D or K F respectively, to simplify the mathe-
matics (LeNeveu 1987). This allows us to extend the lange of applicability
of the VM code to cases with substantial groundwater flow in the buffer
and/or backfill.

9. THE RELATIONSHIP BETWEEN THE VAULT MODEL AND DETAILED MODELS

The results of research programs are applied in the safety assessment of

nuclear fuel waste disposal through the use of mathematical models. Tvo
types of models are generally involved: detailed models, and assessment
models. The detailed models are based on experimental observations and
theoretical studies and can be used directly to interpret and support
experiments. Assessment models, such as the components of the VM, are
based on experiments and on detailed models, and are primarily used in
safety assessments of complex systems. To be practical, assessment models
have to be economical in terms of computer resources required for their
application. This is essential for assessment models associated with prob-
abilistic safety analyses, which often require thousands of calculations to
take into account the variability and uncertainty in the disposal system
(Dormuth and Lyon 1985).

In general, the procedure for producing an assessment model involves the

development and validation of detailed models in order to analyze processes
to be considered in the assessment. The results of these analyses are then
used to develop assessment models, which pioduce results consistent with
those of the detailed models over the sets of conditions required for the
assessment. The present report has illustrated this procedure uith
detailed models and assessment models representing mass-transport processes
in the vault.

Thus, the major detailed models supporting the mass-ttansport model in the
VM are as follows:

(i) a hydrogeological model (MOTIF1, Chan and Scheier 1987) is used to

derive the one-dimensional VM geometry and data required to cal-
culate the mass-transfer coefficients;

(ii) geochemical and reaction path models (e.g., CHEMP, Garisto and
Garisto 1984) are used to evaluate groundwater/buffer intei-
actions and to calculate radionuclide solubilities for the source
terms and the precipitation models (Lemite and Garisto 1989):

(iii) a radiolysis model (MAKSIMA-CHEMIST, Carver et al. 1978) is used

to evaluate the effect of radiolysis on the geochemical condi-
tions in the vault (Tnit and Johnson 1986)—these conditions
affect the data for the source terms and the mass-transport
model;
 - 47 -

(iv) fuel performance models (e.g., ELESIM, Notley 1979) are used to
derive the instant-release fractions for the instant-release
source term (Garisto et al. 1989);

(v) diffusion-reaction models can be used to derive equivalent sorp-

tion models for use in the VM (e.g., Garisto and Garisto 1990c);

(vi) coupled diffusion-precipitation models are used to derive the

enhancement factors for estimating the effect of precipitation of
uranium-containing solids on the release flux (e.g., Garisto, F.
1986; Garisto and Garistc 1988a);

(vii) fractured-media transport models (e.g., CBC, Garisto and Garisto

1989) can be used to estimate the effect of potential buffer
cracks on the performance of the vault;

(viii) a thermal model (HOTROK, Mathers 1985) is used to calculate the

spatial and temporal temperature distribution in the vault,
which, in turn, influences the mass-transport data and the
container-failure model; and

(ix) a model for deriving diffusion data for the buffer and backfill
(Cheung 1989).

10. SUMMARY

The overall computational procedure in the VM is described in qualitative

terms in Section 2. Using the mathematical formalism outlined in
Sections 3 to 8, we can now summarize the overall procedure in quantitative
term?. For each sector of the vault and for each computer run, the VM
determines whether mass transport in the surrounding rock is dominated by
convection or diffusion. This is done by calculating the convective and
diffusive mass-transfer coefficients, i.e., the buffer -* rock mass-transfer
coefficient for the convective case and the backfill -> rock mass transfer
coefficient for the diffusive case (Sections 8.2 and 8.3 respectively), and
choosing the larger of the two (for conservatism).

The instant-release source term, G?(t), and the long-term source term,
F i m (t), are then calculated and summed and precipitation is taken into
account (see Section 6 ) . The net source term is then convoluted with the
container-failure function.

The resultant release rate into the vault is then convoluted with the
buffer response function, R?v ff * r (t), and with the backfill response
function, R^ c k £ i l l ( t), to produce the release rate from the vault into the
rock, for each radionuclide and in each sector.
 - 48 -

ACKNOWLEDGEMENTS

We wish to thank T. Chan, Y. Jin, T. Melnyk, F. Garisto, F. Stanchell,

L. Johnson and K. Nut tall for helpful suggestions and constructive criti-
cism during the preparation of this report. We also thank T. Melnyk for
preparing Figures 2-2 and 2-5. The Canadian Nuclear Fuel Waste Management
Program is jointly funded by AECL and Ontario Hydro under the auspices of
the CANDU Owners Group.

REFERENCES

Ahn, J., P.L. Chambre and T.H. Pigford. 1986. Radionuclide dispersion
from multiple patch sources into a rock fracture. Lawrence Berkeley
Laboratory Report, LBL-23425.

Apted, M.J., D.W. Engel, N.C. Garisto and D.M. LeNeveu. 1989. Source-term
comparison using the AREST and SYVAC-vault models: effects of decay-
chain in-growth and precipitation. Materials Research Society
Symposium Proceedings 127 (Scientific Basis for Nuclear Waste
Management XII), 597-6047

Apted, M.J., A.M. Liebetrau and D.W. Engel. 1987. Spent fuel as a waste
form: analysis with AREST performance assessment code. In Waste
Management '87, Proceedings of the Symposium on Waste Management,
Tucson, AZ, Volume 2, 545-554.

Aronson, S., R.B. Roof, Jr. and J. Belle. 1957. Kinetic study of the
oxidation of uranium dioxide. Journal of Chemical Physics 27,
137-144. ~~

Atomic Energy of Canada Limited CANDU Operations, in association with

J.S. Redpath Ltd., Golder Associates, and Ralph M. Pearson Co. 1987.
Used fuel disposal centre. A reference concept. Draft Report.

Baumgartner, P., J.L. Crosthwaite, M.N. Gray, L.J. Hosaluk, P.H. Seymour,
G.B. Wilken, C.R. Frost, J.H. Gee and J.S. Nathwani. In preparation.
Technical specifications for the concept assessment engineering study
of nuclear fuel waste disposal centres. Atomic Energy of Canada
Limited Technical Record, TR-410.*

Bird, R.B., W.E. Stewart and E.N. Lightfoot. 1960. Transport phenomena.
John Wiley and Sons, New York.

Carslaw, H.S. and J.C. Jaeger. 1959. Conduction of heat in solids, 2nd
edition. Clarendon Press, Oxford.

Carver, M.B., D.H. Hanley and K.R. Chaplin. 1978. MAKSIMA-CHEMIST: A

program for mass action kinetics simulation by automatic chemical
equation manipulation and integration using stiff techniques. Atomic
Energy of Canada Limited Report, AECL-6413.
 - 49 -

Chan, T. and N.W, Scheier. 1987. Finite-element simulation of groundvater

flow and heat and radionuclide transport in a plutonic rock mass. In
Proceedings of the 6th International Congress on Rock Mechanics,
Montreal, PQ, 1987, 41-46.

Chan, T., J.A.K. Reid and V. Guvanasen. 1987. Numerical modelling of

coupled fluid, heat and solute transport in deformable fractured rock.
In Coupled Processes Associated with Nuclear Waste Repositories,
Proceedings of an International Symposium, Berkeley, CA, 1985, 605-
625.

Chan, T., N.W. Scheier and J.A.K. Reid. 1986. Finite-element thermohydro-
geological modeling for Canadian nuclear fuel waste management. lr\
Canadian Nuclear Society 2nd International Conference on Radioactive
Waste Management, Conference Proceedings, Winnipeg, MB, 1986, 653-660.

Chan, T-. and F. Stanchell. 1990. A numerical study of some effects of

nuclear fuel waste vault construction and closure on evolution of
groundwater flow paths in the geosphere. In Proceedings of the First
Annual International High-Level Radioactive Waste Management
Conference, Las Vegas, NV, 1989, Volume 1, 525-536.

Cheung, S.C.H. 1989. Methods to measure apparent diffusion coefficients

in compacted bentonite clays and data interpretation. Canadian
Journal of Civil Engineering ^6, 434-443.

Cook, A.J. 1988. A desk study of surface diffusion and mass transport in
clay. British Geological Survey Technical Report, VE/88/34.

Crank, J. 1975. The mathematics of diffusion. Oxford University Press,

London, England.

Dormuth, K.W. and R.B. Lyon. 1985. The link between detailed process
models and simplified models. In System Performance Assessments for
Radioactive Waste Disposal, Proceedings of an NEA Workshop, Paris,
France, 1985, 81-87.

Dormuth, K.W. and R.D. Quick. 1980. Accounting for parameter variability
in risk assessment for a Canadian nuclear fuel waste disposal vault.
International Journal of Energy Systems 1^ 125-127. Also Atomic
Energy of Canada Limited Reprint, AECL-6999.

Dormuth, K.W. 1991. Demonstration of a system performance assessment

process. Materials Research Society Symposium Proceedings 212
(Scientific Basis for Nuclear Waste Management XIV).

Engel, D.W., M.J. Apted, N.C. Garisto and D.M. LeNeveu. 1988. Comparison
of source term calculations using the AREST and the SYVAC vault
models. Radioactive Waste Management and Nuclear Fuel Cycle 13, 281-
296 (1989). —

Eriksen, T.E. 1989. Some notes on diffusion of radionuclides through

compacted clays. KBS Technical Report, KBS-TR-38-2a.
 - 50 -

Eriksen, T.E. and A. Jacobsson. 1984. Diffusion in clay - experimental

techniques and theoretical models. KBS Technical Report,
KBS-TR-84-05.

Freeze, R.A. and J.A. Cherry. 1979. Groundvater. Pientice-Hall, Inc.,

Englewood Cliffs, NJ.

Garisto, Frank and N.C. Garisto. 1985. A U0 2 sc.e.Dility function for the
assessment of used nuclear fuel disposal. Nuclear Science and
Engineering 90, 103-110. Also Atomic Energy of Canada Limited
Reprint, AECL^8515.

Garisto, F. 1986. Solid dissolution: Effect of mass transport-

precipitation coupling. Chemical and Engineering Science 41L, 3219-
3222. Also Atomic Energy of Canada Limited Reprint, AECL-9032.

Garisto, N.C. 1986. Modelling aspects in vault chemistry. In Canadian

Nuclear Society 2nd International Conference on Radioactive Waste
Management, Conference Proceedings, Winnipeg, MB, 1986, 712-714.

Garisto, N.C. and Frank Garisto. 1984. Reaction path calculations of

mineral alteration products: Application to nuclear fuel waste
management. Nuclear and Chemical Waste Management 5, 17-25. Also
Atomic Energy of Canada Limited Reprint, AECL-8151.

Garisto, N.C. and F. Garisto. 1986. The effect of precipitation on the

long-term release of radionuclides from used fuel. Annals of Nuclear
Energy 1_3(11), 591-596- Also Atomic Energy of Canada Limited Reprint,
AECL-9035.

Garisto, Nava and Frank Garisto. 1988a. Mass transport-precipitation

coupling in finite systems. Atomic Energy of Canada Limited Report.
AECL-9562.

Garisto, N.C. and F. Garisto. 1988b. The effect of model dimensionality

on source terms for the assessment of used fuel disposal. Annals of
Nuclear Energy 1^5(9), 471-475. Also Atomic Energy of Canada Limited
Reprint, AECL-9589.

Garisto, N.C. and F. Garisto. 1989. Analysis of diffusive mass transport

in a cracked buffer. Atomic Energy of Canada Limited Report, AECL-
10003.

Garisto, N.C. and F. Garisto. 1990a. The effect of cracks on diffusive

mass transport through a clay barrier. Materials Research Society
Symposium Proceedings 176 (Scientific Basis for Nuclear Waste
Management XIII), 733-739. Also Atomic Energy of Canada Limited
Reprint, AECL-10098.

Garisco, N.C. and F. Garisto. 1990b. Transport of uranium through a

cracked clay barrier. Annals of Nuclear Energy r7, 183.

Garis*o, N.C. and F. Garisto. 1990c. A preliminary model for carbon-14

transport in a clay buffer. !En Proceedings High Level Radioactive
Waste Management, Las Vegas, NV, 1990, Vol. 1, 130.
 - 51 -

Garisto, N.C., F. Garisto and D.M. LeNeveu. In preparation. Radionuclide

mass transport in a finite, multilayer medium. Atomic Energy of
Canada Limited Report, AECL-10384.

Garisto, N.C., K.B. Harvey, F. Garisto and L.H. Johnson. 1986. Source
term models for the assessment of nuclear fuel waste disposal in
Canada. In Waste Management '86, Proceedings of the Symposium on
Waste Management, Tucson, AZ, 1986, Volume 2, 397-402.

Garisto, N.C., L.H. Johnson and W.H. Hocking. 1990. An instant release
for the assessment of used nuclear fuel disposal. In Second Inter-
national Conference on CANDU Fuel, Conference Proceedings, Pembroke,
ON, 1989, 352-368.

Garisto, F. and N.C. Garisto. In preparation. The effect of precipitation

on the diffusion of reaction products.

Garisto, N.C. and D.M. LeNeveu. 1987. A vault model for the assessment of
used fuel disposal in Canada. In Materials Research Society Symposium
Proceedings 112 (Scientific Basis for Nuclear Waste Management XI),
313-322. Also Atomic Energy of Canada Limited Reprint, AECL-9468.

Garisto, N.C. and D.M. LeNeveu. 1989. The Vault Model for the disposal of
used CANDU fuel: documentation and analysis of scoping calculations.
Atomic Energy of Canada Limited Report, AECL-9578.

Garisto, N.C. and D.M. LeNeveu. 1990. Analysis of mass transport in an

engineered barriers system for the disposal of used nuclear fuel.
Materials Research Society Symposium Proceedings 176 (Scientific Basis
for Nuclear Waste Management XIII), 775-783. Also Atomic Energy of
Canada Limited Reprint, AECL-10099.

Garisto, N.C., D.M. LeNeveu and F. Garisto. In preparation. The mass

transport of radionuclides in a multilayered medium. Atomic Energy of
Canada Limited Report, in preparation.

Goodwin, B.W., T.H. Andres, P.A. Davis, D.M. LeNeveu, T.W. Melnyk, G.R.
Sherman and D.M. Wuschke. 1987. Post-closure environmental assess-
ment for the Canadian Nuclear Fuel Waste Management Program.
Radioactive Waste Management and the Nuclear Fuel Cycle 8, 241-272.

Hancox, W.T. 1986. Progress in the Canadian Nuclear Fuel Waste Management
Program. In Canadian Nuclear Society International Conference on
Radioactive Waste Management, Conference Proceedings, Winnipeg, MB,
1986, 1-9.

Heinrich, W.F. and T. Andres. 1985. Response functions of the convection-

dispersion equations describing radionuclide migration in a semi-
infinite medium. Annals of Nuclear Energy 1^2(12), 685-691. Also
Atomic Energy of Canada Limited Reprint, AECL-8691.

Ikeda, B.M. and D.W. Shoesmith. 1990. Development of a container failure

function for titanium. Atomic Energy of Canada Limited Report, AECL-
10121.
 - 52 -

Johnson, L.H. 1982. The dissolution of irradiated U0 2 fuel in ground-

water. Atomic Energy of Canada Limited Report, AECL-6837.

Johnson, L.H. and D.W. Shoesmith. 1988. Spent fuel. In Radioactive Waste
Forms for the Future (W. Lutze and R.C. Ewing, editors), Elsevier
Science Publishers B.V. Also Atomic Energy of Canada Limited Reprint,
AECL-9651.

Johnson, L.H., N.C. Garisto and S. Stroes-Gascoyne. 1985. Used-fuel dis-

solution studies in Canada. Iji Waste Management '85, Proceedings of
the Symposium on Waste Management, Tucson, AZ, 1985, Volume 1,
479-482.

Johnson, L.H., D.W. Shoesmith, G.E. Lunansky, M.G. Bailey and

P.R. Tremaine. 1982. Mechanisms of leaching and dissolution of
uranium oxide fuel. Nuclear Technology 56, 238-253. Also Atomic
Energy of Canada Limited Reprint, AECL-6992.

Lemire, R.J. and F. Garisto. 1989. The solubility of U, Np, Pu, Th and Tc
in a geological disposal vault for used nuclear fuel. Atomic Energy
of Canada Limited Report, AECL-10009.

LeNeveu, D.M. 1986. Vault submodel for the second interim assessment of
the Canadian concept for nuclear fuel waste disposal: Post-closure
phase. Atomic Energy of Canada Limited Report, AECL-8383.

LeNeveu, D.M. 1987. Response function of the convection-dispersion equa-

tions describing radionuclide migration in a finite medium. Annals of
Nuclear Energy ^4(2), 77-82. Also Atomic Energy of Canada Limited
Reprint, AECL-9166.

LeNeveu, D.M. and N.C. Garisto. 1988. Sensitivity surfaces for the
assessment of waste package performance. Proceedings of the Materials
Research Society Symposium 112 (Symposium on the Scientific Basis for
Nuclear Waste Management XI), 323-330. Also Atomic Energy of Canada
Limited Reprint, AECL-9584.

LeNeveu, D.M. and L.H. Johnson. 1986. Reducing radiation doses from 12Q I
disposal by isotopic dilution. In Canadian Nuclear Society Inter-
national Conference on Radioective Waste Management, Conference
Proceedings, Winnipeg, MB, 1986, 661-666.

Lever, D.A. 1986. Some notes on experiments measuring diffusion of sorbed

nuclides through porous media. Harwell Laboratory Report, AERE-R-
12321.

Liebetrau, A.M., M.J. Apted, D.W. Engel and D.H. Alexander. 1987. AREST:
A probabilistic source-t«rm code for waste package performance
analysis. In Waste Management '87, Proceedings of the Symposium on
Waste Management, Tucson, AZ, Volume 2, 535-544.

Mathers, W.G. 1985. HOTROK, a program for calculating the transient

temperature field from an underground nuclear waste disposal vault.
Atomic Energy of Canada Limited Technical Record, TR-336.*
 - 53 -

Neretnieks, I. 1978. Transport of oxidants and radionuclides through a

clay barrier. KBS Technical Report, KBS-TR-79.

Neuman, S.P. 1987. Waste isolation experiments from the hydrologist's

point of view. In Coupled Processes Associated with Nuclear Uaste
Repositories, Proceedings of an International Symposium, Berkeley, CA,
1985, 765-769.

Notley, M.J.F. 1979. ELESIM: A computer code for predicting the perfor-
mance of nuclear fuel elements. Nuclear Technology 44, 445-450.

Paquette, J. and R.J. Lemire. 1981. A description of the chemistry of

aqueous solutions of uranium and plutonium to 200°C using potential-pH
diagrams. Nuclear ocience and Engineering 79_t 26-48. Also Atomic
Energy of Canada Limited Reprint, AECL-7037.

Shoesmith, D.W., S. Sunder, M.G. Bailey, G.J. Wallace and F.W. Stanchell.
1984. Anodic oxidation of U0 2 . IV. X-ray photoelectron spectro-
scopic and electrochemical studies of film growth in carbona'e con-
taining solutions. Applied Surface Science 20, 39-57. Also Atomic
Energy of Canada Limited Reprint, AECL-8174.

Skagius, K. and I. Neretnieks. 1985. Diffusion measurements of cesium and

strontium in biotite gneiss. KBS Technical Report, KBS-TR-85-15.

Sunder, S. and D.W. Shoesmith. In preparation. Chemistry of U0 2 fuel

dissolution in relation to the disposal of used nuclear fuel. Atomic
Energy of Canada Limited Report, AECL-10395.

Tait, J.C. and L.H. Johnson. 1986. Computer modelling of alpha radio-
lysis of aqueous solutions in contact with used U0 2 fuel. In Canadian
Nuclear Society International Conference on Radioactive Waste
Management, Conference Proceedings, Winnipeg, MB, 1986, 611-615.

Talbot, A. 1979. The accurate numerical inversion of Laplace transforms.

Journal of the Institute of Mathematics and Its Applications 23, 97.

Torstenfelt, B. and B. Allard. 1986. Migration of fission products and

actinides in compacted bentonite. KBS Technical Report, KBS-TR-86-14.

Walker, J.R. and D.M. LeNeveu. 1987. Nonlinear chemical sorption iso-
therms in the assessment of nuclear fuel waste disposal. Atomic
Energy of Canada Limited Report, AECL-8394.

Unpublished report available from SDDO, AECL Research,

Chalk River, Ontario KOJ 1J0.
 ISSN 0067-0367 ISSN 0067-0367

To identic individual documents in the series, Hour identifier les rapports individuals
we have assigned an AECL- number to each. faisant partie de cette serie, nous a\on>
affecte un numero AECL- a chacun d"eu\.

Please refer to the AECL- number when Veuillcz indiquer le numero AfcCL- Iorsque vou
requesting additional copies of this document demandez d'auires exemplaires de ce rappon

from

Scientific Document Distribution Office Service de Distribution des Documents Scienlilique*

AECL Research EACl. Recherche
Chalk Ri\er, Ontario, Canada Chalk River. Ontario. Canada
KOJ 1J0 KOJ I JO

Price: B Prix: B