Modelling Approach For Geochemical Changes in The Prototype Repository Fngineered Barrier System

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 42

Working Report 2004-31

Modelling Approach for Geochemical


Changes in the Prototype Repository
fngineered Barrier System

Ari Luukkonen

July 2004

POSIVA OY
FIN-27160 OLKILUOTO, FINLAND
Tel +358-2-8372 31
Fax +358-2-8372 3709
31.5.2004 1 (1)

TILAUKSET OSALLISTUMISESTA EU:N PROTOTYPE REPOSITORY


-HANKKEESEEN VUOSINA 2000-2004

TEKIJAORGANISAATIO: VTT Rakennus- ja Yhdyskuntatekniikka


PL 1800
02044 VTT

TILAAJA: Posiva Oy
27160 OLKILUOTO
~Jt,jJ{ (
TILAAJAN YHDYSHENKILO: JJa~Pe~a Salo, Posiva Oy
TILAUSNUMEROT: 9655/00/JPS
9517/02/JPS
9564/03/JPS

TEKIJAORGANISAATION
YHDYSHENKILO:

Ari Luukkonen VTT Rakennus- ja Yhdyskuntatekniikka

TARKASTAJA:

J~~
Markku Tuhola VTT Rakennus- ja Yhdyskuntatekniikka

VTT RAKENNU5- JA YHDYSKUNTATEKNIIKKA


Uimpomiehenkuja 2, Espoo Puh. (09) 4561 etunimi. sukunimi@vtt.fi
PL 1800, 02044 VTT Faksi (09) 463 251 www.vtt.fi
Y-tunnus 0244679-4
Working Report 2004-31

Modelling Approach for Geochemical


Changes in the Prototype Repository
fngineered Barrier System

Ari Luukkonen

VTT Building and Transport

July 2004

Working Reports contain information on work in progress


or pending completion.

The conclusions and viewpoints presented in the report


are those of author(s) and do not necessarily
coincide with those of Posiva.
r------------------------------------------------------------------- -- -- --

ABSTRACT

Luukkonen, A., 2004. Modelling approach for geochemical changes in the Prototype repository
engineered barrier system.

The study deals with a full-scale test of a repository concept for nuclear waste. The work is
motivated and targeted to the safety assessment of a final repository for highly active nuclear waste.
Successive rounds of predictive modelling are confidence building steps if modelling results can be
successfully compared to gradually growing up measured test data.

Calculations consider geochemical changes during the wetting of repository tunnel backfill and
canister buffer, and the time-dependent changes at the boundaries of a repository engineered barrier
system (EBS). The backfill and buffer volumes are divided to uniform subsequent reaction cells.
The modelling assumes only schematic coupling between geochemical reactions and hydrologic
transport. During a wetting front advance, instant water saturation occurs in subsequent, initially
undersaturated, cell volumes. However, a water parcel introduced into a cell is expected to stay
within that cell until dissolved components, and solid phases are fully equilibrated.

The approach concentrates on the major element compositions of porewaters, and the changes in
solid phases of the repository. The initial properties of canister buffer resemble the estimations for
compacted Wyoming MX-80 sodium bentonite. The tunnel backfill is assumed to consist of sodium
bentonite (30%), and crushed Aspo diorite (70%) components. The backfill composition has been
estimated in accordance with mineral quantities present in the components of the backfill mixture.
The initial groundwater sucked into the EBS at the repository boundaries is Na-Ca-(HC03)-S04-Cl
-water having a reference to brackish seawater origin.

The reactions considered are cation exchange, surface complexation, and dissolution/precipitation
of certain minerals. Calculations assume also diffuse water layers to occur upon the clay platelet
surfaces. As an initial condition, the undersaturated pore volumes of backfill and buffer contain
entrapped air (0 2 content 20%). In the backfill modelling cases, the porewater speciation and
dissolution/precipitation equilibria are considered at 40°C, and in the buffer modelling cases the
temperature is assumed to be 65°C.

A comparative review to literature is made in relation to the modelling results. Many of the
mineralogical reactions, utilised in the modelling, can be verified from the results of other EBS
related projects launched in the Aspo Hard Rock Laboratory.

Key words: nuclear waste, final disposal, engineered barrier, geochemistry, tunnel backfill, canister
buffer
TIIVISTELMA

Luukkonen, A., 2004. Geokemiallinen mallinnusmenetelma prototyyppi loppusijoitustilan teknisille


paastoesteille.

Tutkimus kasittelee ydinjatteen loppusijoitustilakonseptin taysimittaista koetta. Tyon motiivit ja


paamaarat liittyvat korkea-aktiivisen ydinjatteen loppusijoitustilan turvallisuusarviointiin. Toistuvat
ennustemallinnukset ovat turvallisuusarvioinnin luottamusta lisaavia askelmia, jos mallinnus-
tuloksia voidaan menestyksekkaasti vertailla asteittain kasvavaan mitattuun koeaineistoon.

Laskelmat pohtivat geokemiallisia muutoksia, joita tapahtuu tunnelitaytteen ja kanisteripuskurin


vettymisen aikana, seka ajasta riippuvia muutoksia, joita tapahtuu teknisessa paastoeste-
jarjestelmassa sijoitustilan reunoilla. Tunnelitayte- ja kanisteripuskuritilavuudet jaetaan
yhdenmukaisiin toisiaan seuraaviin reaktioelementteihin. Laskelmat olettavat vain kaavamaisen
kytkennan geokemiallisten reaktioiden ja hydrologisen kuljetuksen valille. Vettymisrintaman
edetessa valiton vedella saturoituminen tapahtuu perakkaisissa, aluksi alikyllaisissa,
elementtitilavuuksissa. Elementtiin saapuneen veden oletetaan kuitenkin viipyvan ko. elementissa
niin pitkaan, etta liuenneet komponentit ja kiinteat faasit ovat taysin tasapainottuneet keskenaan.

Lahestymistapa keskittyy loppusijoitustilan huokosvesien paakomponenttipitoisuuksien Ja


kiinteiden faasien koostumusten tarkasteluun. Kanisteripuskurin alkukoostumus vastaa kokoon-
puristetun Wyoming MX-80 natrium-bentoniitin ominaisuuksia. Tunnelitayteen on oletettu
koostuvan natrium-bentoniitista (30%) ja murskatusta Aspo dioriitista (70%). Tunnelitaytteen
kokonaiskoostumus on arvioitu taytteeseen sekoittuneiden mineraalimaarien perusteella.
Mallinnuksien lahtovesi, joka imeytyy teknisiin paastoesteisiin loppusijoitustilan reunoilta, on
koostumukseltaan Na-Ca-(HC0 3)-S04-Cl -vesi, jonka koostumus viittaa merelliseen murtovesi
alkuperaan.

Mallinnuksissa tarkasteltavia reaktioita ovat kationinvaihto, pintakompleksaatio ja muutamien


mineraalien liukenemis-saostumisreaktiot. Laskennat olettavat myos diffuuseja vesikerroksia
savimineraalien pinnalle. Alkutilassa veden suhteen alikyllaisen tunnelitaytteen ja kanisteripuskurin
huokostilavuudet sisaltavat ilmaa (happipitoisuus 20%). Tunnelitaytteen mallinnustapauksissa
huokosveden ja mineraalitasapainojen reaktioita tarkastellaan 40°C Himpotilassa. Kanisteripuskurin
mallinnustapauksissa lampotilan oletetaan olevan 65°C.

Mallinnustulosten suhteen tehdaan vertaileva tarkastelu kirjallisuuteen. Monet mallinnuksissa


kaytetyista mineraalireaktioista voidaan todentaa Aspon kalliolaboratorion muista teknisiin
paastoesteisiin keskittyvien proj ektien tuloksista.

Avainsanat: ydinjate, loppusijoitus, tekninen paastoeste, geokemia, tunnelitayte, kanisteripuskuri


PREFACE

The full scale testing of the KBS-3 concept for high-level radioactive waste (the Prototype
repository project) is underway in the Aspo Hard Rock Laboratory in southern Sweden. The long-
term test (up to 20 years) is organised and led by the Swedish Nuclear Fuel and Waste Management
Company (SKB). The project received eo-funding from the European Commission during years
2000-2004. Posiva Oy has participated to the project through the work carried out in VTT Building
and Transport.
1

CONTENTS

Abstract
Tiivistelma
Preface

1 INTRODUCTION ......................................... ................................................................. 2

2 WYOMING MX-80 BENTONITE .................................................................................. 4


2.1 Material properties .................................................................................................... 4
2.2 Geochemical properties ............................................................................................ 4

3 TUNNEL BACKFILL ..................................................................................................... 6


3.1 Material properties .................................................................................................... 6
3.2 Geochemical properties ............................................................................................ 6

4 GEOCHEMICAL REACTIONS ..................................................................................... 9

5 MODELLING APPROACH ......................................................................................... 15


5.1 EBS interior ............................................................................................................ 15
5.2 EBS boundary ........................................................................................................ 15

6 MODELLING RESULTS ............................................................................................. 16


6.1 EBS interior ............................................................................................................ 16
6.2 EBS boundary ........................................................................................................ 22

7 COMPARISON OF MODELLING RESULTS TO OBSERVATIONS .......................... 27

8 DISCUSSION ............................................................................................................. 29

9 CONCLUSIONS ......................................................................................................... 33

10 REFERENCES ........................................................................................................... 34
1 INTRODUCTION

Prototype repository is a full-scale simulation of a final repository concept (KBS-3) for spent
nuclear fuel with 6 deposition holes located in the Aspo Hard Rock Laboratory, southern Sweden
(Fig. 1). The 65m-long Prototype Repository drift has been drilled in the end of the Hard Rock
Laboratory tunnel spiral, and it is sited at depth of -450 m.

The Prototype project was initiated in late 1996 (SKB, 1999) with design and planning by the
Swedish Nuclear Waste Management Company (SKB). From the beginning of September 2000, the
project was eo-funded for a 42 months period by the European Commission (EC contract FIKW-
CT2000-00055). However, the Prototype project is expected to provide long-term experience on the
SKB-3V repository concept performance (SKB, 2003), and therefore, the Prototype repository is
planned to be in function for a long time (up to 20 years).

In the Prototype repository, electrically heated canisters cause a heat flow from canisters to the
engineered barrier system (EBS). According to Pusch & Borgesson (2001), the MX-80 bentonite
buffer surrounding the heated canisters experiences a maximum temperature of 90°C. The
temperature in the tunnel backfill is expected to stay below 40-50°C during the execution of
experiment (King et al., 2001; Pusch & Borgesson, 2001 ).

Figure 1. The location of the Aspo Hard Rock Laboratory in southern Sweden.
~------------------- ----------------- -- --

A simplified cross-section of the Prototype drift is presented in Figure 2. The first wetted volumes
of buffer and backfill are created at the boundaries of the Prototype EBS. These porewaters adsorb
deeper in the undersaturated volumes of buffer and backfill, and represent the first wetting of EBS.
At the same time, new porewater compositions are repeatedly generated at the EBS boundary.
These waters, as well, are sucked deeper into EBS until the suction power vanishes to the EBS
water saturation.

This study deals with geochemical changes during the wetting of repository EBS (tunnel backfill
and canister buffer), and the time-dependent changes at the boundaries of EBS. The studied EBS
volumes are divided to uniform subsequent reaction cells. There is no quantitative coupling to
hydrological transport, but the modelling assumes instant full saturation of a cell volume as soon as
infiltrating water first time enters the studied cell. The calculations are batch reaction oriented, and
follow the equilibrium thermodynamic assumption.

The calculations attempt to illustrate how the geochemistry evolves in the EBS interior and
boundaries as a function of batch reaction cycles. The calculations model major elemental
compositions for porewaters and related compositional changes in the EBS solid phases. The
modelling code utilised is PHREEQC-2.7 with associated thermodynamic database Wateq4f
(Parkhurst & Appelo, 1999).

BACKFILL
INTERIOR

ROCKIBACKFILL
INTERFACE
BUFFER
INTERIOR
ROCK/BUFFER
INTERFACE

Figure 2. Schematic cross-section of the Prototype repository with indications to the four special
interest areas of current geochemical modelling.
4

2 WYOMING MX-80 BENTONITE

2.1 Material properties

The MX-80 sodium bentonite was used as buffer in the canister deposition holes. Material
properties that can be associated to the bentonite blocks after compaction are presented in Table 1
(Borgesson & Hemelind, 1999).

Table 1. The material properties for initial compacted bentonite (Borgesson & Hernelind, 1999).
Water content (mwatelmsolicd, porosity (Vvoir/V,oJ, and degree of saturation (Vwate!Vvoicd·
Dry density g/cm3 Water content Porosity Degree of saturation
1.57 0.17 0.435 0.61

For modelling purposes, all properties of bentonite are defined for 1 dm3 of pore volume. Initially a
bentonite cell contains 0.61 dm3 water, and 0.39 dm3 air (Table 1). During the first reaction cycle,
external water is assumed to replace completely the earlier pore volume and the cell saturates with
water. In practice, this means that the gas phase either reacts with infiltrating water completely
("closed system") or non-reacted part of the gas phase is pushed away from the pore volume during
infiltration ("open system"). In compacted bentonite, the initial 0.39-dm3 volume of air equals to a
5.89-dm3 volume of water and bentonite, that weights 9.25 kg. The weight of bentonite is 8.64 kg.

2.2 Geochemical properties

The composition of MX-80 bentonite is considered to vary as presented in Table 2. Based on this
variation an average mineralogical composition has been estimated for current purposes. The cation
exchange capacity (CEC) for bentonite is around 787 meq/kg (Bradbury & Baeyens, 2003). The
specific surface area in MX-80 bentonite available for surface complexation is around 31.5 m 2/g
(Wieland et al., 1994; Bradbury & Baeyens, 2002; Wersin, 2003). The amount potentially reactive
species in 8.64 kg of bentonite are presented in the last column of Table 2.

Compared to the recent composition tabulations of MX-80 bentonite by Bradbury & Baeyens
(2002), Table 2 contains only minor differences. Bradbury & Baeyens (2002) report smaller amount
kaolinite and quartz, and slightly more feldspar than Table 2. They also indicate a trace of mica in
bentonite while Table 2 specifies this phase as illite. Interestingly, Bradbury & Baeyens (2002)
specify both calcite (0.7 wt.o/o) and siderite (0.7 wt.%) from MX-80. However, Table 2 assumes that
only the carbonate present in MX-80 bentonite is calcite (1.4 wt.%).
5

Table 2. The composition of Wyoming MX-80 bentonite.


Unit wt Bentonite<a Current Bentonite Buffer
3
g/mol ea. wt.% Estim. wt.% moles/kg moles/{1dm QOre water2
Montmorillonite 744 75.0 75.0 1.01 8.71
Illite 778 0.0-4.0 2.0 0.03 0.22
Kaolinite 258 1.0-7.0 4.0 0.15 1.34
Albite 262 5.0-9.0 7.0 0.27 2.31
Quartz 60.1 10.0 10.0 1.66 14.4
Gypsum 136 0.3 0.3 0.02 0.22
Pyrite 120 0.3 0.3 0.03 0.22
Calcite 100 1.4 1.4 0.14 1.21
CH20(b, (c 30.0 0.4 0.13 1.15
NaCl(c 58.4 1.35e-3 0.012
Cation ocCUQancies in the exchange sites{a
Ca2+ 0.03 0.29
Mg2+ 0.02 0.17
Na+ 0.67 5.77
K+ 0.01 0.11
Surface site caQacities{e
=SwOH 0.03 0.25
According to aJ Bruno et al. (1999), bJ Pirhonen (1986), cJ Bradbury & Baeyens (2002), dJ Bradbury &
Baeyens (2003), and eJ Wieland et al. (1994).
6

3 TUNNEL BACKFILL

3.1 Material properties

The tunnel backfill consist of 30% bentonite mixed with 70% crushed Tunnel Boring Machine
(TBM) muck. The material properties presented in Table 3 are after Borgesson & Hemelind (1999).

Table 3. The material properties for initial compacted tunnel baclifill (Borgesson & Hernelind,
1999). Water content (mwatelmso/ictJ, porosity (Vvoici"VtoJ, and degree of saturation (Vwate!VvoictJ·
Dry density g/cm Water content Porosity Degree of saturation
1. 75 0.19 0.363 0.58

The initial compacted backfill cell with 1 dm3 pore volume contains 0.58 dm 3 of water, and 0.42
3
dm of air in its pores. When this block saturates with external groundwater, the gas phase is
assumed to disappear. The initial 0.42-dm3 volume of air in a compacted block gives a 6.56-dm3
volume, and a weight of 11.48 kg, of water and backfill. The weight ofbackfill is 10.90 kg.

3.2 Geochemical properties

The TBM-muck is mostly of Aspo diorite (Patel et al., 1997) and its average modal composition is
presented in Table 4. Aspo diorite contains frequently narrow fracture and broader hydrothermal
alteration zones. For current purposes, it is estimated that 2% of rock is completely altered. This
increase the amount of secondary minerals (chlorite, goethite, calcite, epidote, fluorite, and quartz)
compared to the fresh rock composition. A judged average composition for the completely altered
diorite rock is presented in Table 5. Finally, an estimated weighted average of the fresh and
completely altered rock mixture is presented in Table 6.

Table 4. The average mineralogical composition offresh Aspo diorite (Pate/ et al., 1997 and refs.
therein).
Amount vol.% 1kg Unit wt. g/mol moles/kg
Quartz 14.0 37.1 0.13 60.1 2.21
K-feldspar 15.0 38.7 0.14 278 0.50
Oligoclase 45.0 119 0.43 265 1.61
Biotite 15.0 45.0 0.16 986 0.16
Muscovite 0.5 1.42 0.005 797 0.01
Epidote 6.0 20.6 0.07 469 0.16
Amphiboles 1.0 3.50 0.01 877 0.01
Sphene 2.0 7.08 0.03 197 0.13
Apatite 0.5 1.60 0.01 422 0.01
Magnetite 0.8 4.16 0.02 232 0.06
Pyrite 0.2 1.00 0.004 120 0.03
100.0 279 1.00
- - - - - - - - -- -- - - - - - - - - -- - - -

Table 5. An estimated average composition for fracture fillings and completely altered Aspo diorite.
Amount vol.% wt./1dm3 1kg Unit wt. g/mol moles/kg
Chlorite 25.0 74.3 0.23 1261 0.18
Goethite 25.0 108 0.33 88.9 3.71
Calcite 25.0 67.9 0.21 100 2.08
Epidote 10.0 34.4 0.11 469 0.23
Fluorite 5.0 15.9 0.05 78.1 0.62
Quartz 10.0 26.5 0.08 60.1 1.35
100.0 326 1.00

The CEC of the TBM-muck is at the most 5-10% of the CEC of bentonite. Therefore, in a 70/30
mixture, bentonite is still mostly responsible (more than 80%) of the exchange properties of the
backfill. In this study, the cation exchange properties of backfill are based on the bentonite fraction
only. The amounts of potentially reactive cation exchange species in 10.90 kg of backfill are
presented in Table 6.

Similarly, the estimate of surface complexation capacity presented in Table 6 is based on MX-80
bentonite measurements only (Wieland et al., 1994). However, there is potentially significant
uncertainty in relation to TBM-muck surface complexation capacity. The specific surface areas
capable for complexation are functions of grain sizes and sub-microscopic surface roughness
(Appelo & Postma, 1996) that is impossible to define for TBM-muck without measurements.
Moreover, during the Prototype repository function neoformed minerals are likely to occur within
the EBS (e.g. amorphous silica, goethite ). The porosity of these poorly crystallised mineral surfaces
are likely high, and correspondingly reactive surface areas are high as well (Davies & Kent, 1990).
8

Table 6. Potentially reactive solids of the tunnel backjill.


Unit wt TBM-muck Bentonite 70/30 Backfill 70/30 Backfill
3
g/mol mol/kg mol/kg mol/kg mol/{1dm water2
Montmorillonite 744 1.01 0.30 3.30
Illite 778 0.03 0.01 0.08
Kaolinite 258 0.15 0.05 0.51
Biotite 986 0.16 0.11 1.22
K-feldspar 278 0.49 0.34 3.72
Albite 262 1.58 0.27 1.18 12.9
Quartz 60.1 2.19 1.66 2.03 22.2
Epidote 469 0.16 0.11 1.21
Amphibole 877 0.01 0.01 0.11
Pyrite 120 0.03 0.03 0.03 0.30
Goethite 88.9 0.07 0.05 0.57
Calcite 100 0.04 0.14 0.07 0.77
Gypsum 136 0.02 0.01 0.08
Fluorite 78.1 0.01 0.01 0.10
CH20 30.0 0.13 0.04 0.44
NaCl 58.4 1.35e-3 4.05e-4 4.41 e-3
Cation occu2ancies in the exchange sites
Ca2+ 0.03 0.01 0.11
Mg2+ 0.02 0.01 0.07
Na+ 0.67 0.20 2.18
K+ 0.01 0.00 0.04
Surface site ca2acities
=SwOH 0.03 0.009 0.09
9

4 GEOCHEMICAL REACTIONS

The bedrock fractures around the repository feed external water into the EBS. The external water
used in the calculations is a salinity median of the samplings March '98 - June '99 (Andersson &
Safvestad, 2000) from the near field of the Prototype repository drift. The ionic strength in this
sample is at level 0.125 eq/kgw. Otherwise, the measured concentrations in water are shown in
Table 7.

Table 7. The main composition of sample KA3542G02/2587 (Andersson & Sa.fvestad, 2000). The pe
value is estimated. All concentrations are in mg/kgw except pH and pe values.
H 7.4 e -3.0
Na+ 1620 HC03- 188
K+ 9.5 er 3890
Ca2+ 624 sol- 296
Mg2+ 79.4 HS- 0.0
Fe2+ 0.23
Si4+ 7.8

The initial undersaturated MX-80 bentonite contains porewater that is in equilibrium with its host.
Based on studies of Bradbury & Baeyens (2003) the ionic strength of porewater at a dry density of
1.5 g/cm3 is 0.314 eq/kgw. The composition of this porewater is presented in Table 8. Current
calculations omit the possible mixings between external water and initial porewater. However,
before the first intrusion of external water (Table 7) the calculations equilibrate the surface sites of
bentonite to the porewater composition presented in Table 8. Similarly, an initial positively charged
diffuse layer (equilibrated to Table 8 composition) is created upon clay surfaces (see below).

Table 8. The calculated composition of bentonite porewater at a dry density of 1. 5 g/cm 3 (After
Bradbury and Baeyens, 2003). All concentrations are in mglkgw except pH value.
H 8.0
Na+ 5587 HC03- 53.9
K+ 48.1 er 2393
Ca2+ 380 sol- 9990
Mg2+ 172

The first intrusion of external water into a batch cell differs from all subsequent cycles because
initially cells contain entrapped air. An undersaturated bentonite cell contains 2.8 mmol 02 and
0.004 mmol C02 (0.39 dm 3 air at 1 atm pressure) while an undersaturated backfill cell contains 3.3
mmol 0 2 and 0.006 mmol C02 (cf. Fig. 3). Oxygen is consumed during the first batch cycle
(dissolution of pyrite) and C02 is available, if needed, for further reactions.
10

The modelling tries to take into account both cation exchange and surface complexation (Fig. 3).
The constants for equilibria are presented in Table 9. According to Bradbury & Baeyens (2002),
proton exchange is not significant among the cation exchange processes. The surface complexation
follows the approach presented by Wieland et al. (1994). The protonation/deprotonation of a pH-
dependent surface affect to surface charges of solids. At near neutral pH it is expected that wet clay
mineral surfaces (i.e. montmorillonite, kaolinite, and illite) become slightly negatively charged (cf.
Appelo & Postma, 1996; Davis & Kent, 1990).

Gas

~
, . .('\.
~ , ...('\."'1111
""~~~~
,..~ ,~
r'
I(~ ~ I(~ ~ ~~ Fe 2·,s 2- ~ Pyrite
~ ~"V"All1l ~ .... ~ .... ~ .... ~All ~ 0 2 C0 2
~ ~ ~ ~ ~ ~

(±) (±) (±) 3


~ ~ ~
Fe • Goethite
~ ,~ ,~ ,.-~ ,~ ,..~
~
~ ~ ~ ~ ~
Alii! ~Alii ~Alii ~ .... ~AI ~All ..."' Ca 2·,Mg2.,( )
Ca 2·,co32- Calcite
(±) (±) (±)
.. Na·
~

~ ,~
~

,.~
~

,~
'"'
,. ~

~ ,..~
'"'
~ a2·,so42- Gypsum
~ I(~ ~·)I I(·~ ~
All ~All ~ .... ~All ~All ~All ~
Si 4• Quartz
~
(±)

""'~~~~ ""'~~~~
~ ~·~ ~ ~
(±)
,..~
~

,..~
(±)
,..~

I(~
, H• Na·,cr Halite
~
...... ~
~
.... ~
~
.... ~
~
.... ~
~
.... ~
~
.... ~

Figure 3. Schematic illustration of the geochemical equilibrium processes considered in the


modelling study. The tetrahedral-octahedral-tetrahedral layering and interlayer sites of
montmorillonite are presented on the left, and the mineral equilibria on the right. The initial
entrapped oxygen interaction is presented on the top of the illustration.

Table 9. Exchange and surface parameters for the thermodynamic models considered
Parameter Reaction Value
Cation exchange a
logK Ca2+ + 2NaX {::::} CaX2 + 2Na+ 2.6
logK Mg2+ + 2NaX {::::} MgX2 + 2N a+ 2.2
logK K+ + NaX {::::} KX + Na+ 4.0

Surface complexation(b
logK =SwOH + H+ {::::} =SwOH2+ 5.4
logK =SwOH {::::} =Swo- + H+ -6.7

a) According to Bradbury & Baeyens (2003), Gaines-Thomas convention.


b)According to Wieland et al. (1994)
11

The calculations equilibrate porewater compositions to following solids: calcite, gypsum, quartz,
pyrite, goethite and halite (cf. Muurinen & Lehikoinen, 1999; Pusch et al., 1999; Bruno et al.,
1999). Concentrations of these solids in the initial buffer and backfill cells are presented in Tables 2
and 6, and the mineral equilibria are illustrated schematically in Figure 3.

According to Bradbury & Baeyens (2002), Wyoming MX-80 bentonite contains two carbonate
minerals, i.e. calcite and siderite. However, the precipitation of pyrite-siderite assemblage during
the formation and alteration of bentonite deposits seems unlikely. Siderite is stabile compared to
pyrite only at extremely low fluid-sulphide concentrations (e.g. Appelo & Postma, 1996). The mode
of origin of Wyoming bentonite (Smellie, 2001) indicates that this condition is improbable. It is
possible that the dominant carbonate mineral in bentonite is actually ankerite [CaFe(C03)2].
Because there are no widely accepted data available for ankerite thermodynamics, calcite was
chosen as the representative of carbonate mineral phase in current calculations.

Equilibrium thermodynamic calculations make a partial attempt to take into account the effect of
temperature. In the buffer calculations the assigned temperature is 65°C while in the backfill
calculations the temperature is 40°C. With the current knowledge, temperature dependencies are
available for gas-phase and aqueous speciation calculations, and for dissolution/precipitation
reactions of calcite, gypsum, quartz, and pyrite (Parkhurst & Appelo, 1999). However, the cation
exchange and surface geochemical reactions (Fig. 3 and Table 9), and goethite precipitation are
estimated with equilibrium constants defined at 25°C only.

Negative surface charges, created on the basal planes and edges of clay platelets by surface
complexation, lead to compositional gradients within porewater. The charged surfaces together with
counterbalancing diffuse layers created upon the platelet surfaces are called as diffuse double
layers. The porewater in diffuse layer is positively overcharged and this overcharge is proportional
to distance from the platelet surface. Two illustrations of diffuse double layers are presented in
Figure 4. The gradual change in cation and anion concentrations away from the clay surface is often
described with Gouy-Chapman theory. In this theory, both cations and anions are subject to
electrostatic and diffusion forces that act to opposite directions. The negative surface attracts cations
but diffusion tries to push cations towards lower concentrations (towards the electroneutral
porewater). Similarly, diffusion attempts to push anions towards the surface, but the charged surface
repels anions. The thickness of diffuse layer is usually defined with the Debije length 11 K as follows
(c.f. Appelo & Postma, 1996):

(1)

where £= dielectric constant for aqueous porewater (F/m), R = gas constant (J/K·mol), T =
temperature (K), Na =Avogadro's constant (1/mol), qe = charge of proton (C), and I= ionic strength
ofporewater (eq/kgw).

As Equation (1) indicates, the thickness of diffuse layer is inversely proportional to ionic strength
(both concentrations and valencies affect) of porewater, and directly proportional to temperature.
Furthermore, Wersin et al. (2004 and ref. therein) and Parkhurst & Appelo (1999) indicate that the
dielectric constant in solution near the charged surface (i.e. in diffuse layer) can be significantly
smaller than in the "free" porewater.
12

-~urface layer diffuse "free" diffuse surface layer_


a) ----------1 layer porewater layer ~----------
!8 ® ® 8!
[:~ 11

0 N If 0

\:s
i1f
c
.Q ::t: K ~

rns ~
D- ~ ~
i) q t
c ~ K
Q.)
(.)
c
0 I~ :;(
·~ )-

y
~
:>
66
::t:
~
< 0
0
(.)

- ':·
1!1
--- electroneutral
- - -concentration
0 !I: 0
K ;r

.....~_....;Distance between _ _....;.....


clay platelets

b) -~urface layer
----------1
overlapping
diffuse layers
surface layer_
~----------
!8 ® 8!

0 0

(J
c
0

~
:;::;
~
c 0
compaction
Q.)
(.) 0 pressure
c
0
(.)
electroneutral
- - - - - - -concentration
0 0

Distance between
+ clay platelets •

Figure 4. Schematic distributions of cations and anions within clay pores. a) Distance between clay
platelets is greater than the thickness of two diffuse layers. b) Distance between clay platelets is so
small that the diffuse layers overlap.

Based on the external water ionic strength (0.125 eq/kgw) and the given temperatures (40°C and
65°C), Equation (1) indicates that diffuse layer should be around 10 A thick both in backfill and
buffer cases. This means that of the available 1-dm3 pore volumes in backfill and buffer, 0.10 dm3
and 0.27 dm3 would be used, respectively, for diffuse layers. However, the porewater transport
calculations (see Ch. 5) indicate that the initial external water ionic strength will grow to a 10 times
higher level during its transportation within backfill and buffer. This means that the initial thickness
of diffuse layer should gradually drop down to 1/3 from the original 10 A.

In PHREEQC-2.7, the elemental concentration calculations for diffuse layers follow the method of
Borkovec & Westall (1983). The method resolves simultaneously the bulk elemental concentrations
13

within diffuse layers and in "free" electroneutral porewater. Parkhurst & Appelo ( 1999) point out
that there are significant uncertainties related to diffuse layer composition calculations. The method
assumes that speciation within diffuse layer is similar as in the bulk solution, which is unlikely if
changes in dielectric constant occur. Furthermore, Parkhurst & Appelo (1999) indicate that there are
significant uncertainties; how the thickness of the diffuse layer is defined (cf. Eq. 1).

In the Borkovec-Westall method, the ionic strength in the solution and the created net-charge in the
surface layer (cf. Fig. 4) are correlated in the manner that defines lower limit for the diffuse layer
thickness. According to Parkhurst & Appelo (1999), it is possible for solutions with low ionic
strength (interacting with highly net-charged surfaces) that, within diffuse layer, one or more
calculated concentrations of elements become negative. This situation indicates that the assumed
thickness of diffuse layer is too small, and counterbalancing of the surface net-charge within the
specified diffuse layer is not successful.

In the current modelling, this constrain (created by the ionic strength 0.125 eg/kgw and the net-
surface charges) is met around the diffuse layer thickness of 20 A. If calculations are carried out
with lower thickness values (e.g. 10 A) negative concentrations within diffuse layer begin to occur.
Therefore, the current studies assume 20 A for the thickness of diffuse layer despite the value given
by Equation (1 ). With the 20-A thickness, of the 1-dm3 pore volume available, 0.21 dm 3 and 0.54
dm3 are used for diffuse layers in backfill and buffer, respectively.

All calculations presented later on assume that "free" electroneutral porewater exist within wetted
backfill and buffer (Fig. 4a). In accordance with Wieland et al. (1994), the total BET surface area
for MX-80 bentonite is 31.5 m2/g (cf. Ch. 2.2) that comprises of basal planes of montmorillonite
platelets (28.5 m 2/g) and surface edge sites (3.0 m 2/g). For the backfill, the calculations assume that
the total BET surface area originates completely from the bentonite fraction. Therefore, the BET
area for the backfill is 0.3* 31.5 m2/g. The total reactive surface areas for backfill (10900 g) and
buffer (8640 g) are therefore 103,000 m 2 and 272,300 m2, respectively. The available volumes for
"free" porewater in the backfill and buffer are 0.79 dm 3 and 0.46 dm 3, respectively. By assuming
that all pores occur between basal planes of clay platelets (about 90% of the total BET area) the
following assumption can be done. The average "free" apertures, filled with movable porewater,
should be around 170A (0.017 J.tm) and 40A (0.004 J.tm) for backfill and buffer, respectively (cf.
Fig. 4a). These calculated apertures are very small compared to the pore diameters measured from
saturated compacted MX-80 bentonite. According to Pusch (2001, p. 69) pore diameters in
saturated bentonite (bulk density 2.13 g/cm3) are on the average around 0. 7 J.tm.

In order to understand this discrepancy, it must be assumed that in wetted EBS a large part of the
reactive BET area should be eliminated from the available pore volume. One of the reasons for the
BET area reduction may originate from coagulation of clay particles as discussed by Pusch (200 1).
In this gelation process negatively charged basal planes of clay platelets attract slightly positively
charged platelet edges yielding edge-to-face clay aggregates. It has been also shown that salt
diffusion causes shrinkage of the diffuse layers upon the platelets, and this causes enhancing
flocculation (platelets are partially tied together with solute cations) making the bentonite more
permeable. However, it seems that gelation and flocculation solve only partially the BET area
reduction problem.

Keijzer (2000) has shown experimentally that salt exclusion (i.e. anion exclusion) caused by
overlapping diffuse layers is an important process in the compacted Wyoming bentonite. The
14

process is shown schematically in Figure 4b. The osmotic water intake within clay platelets causes
swelling of bentonite, and consequently high swelling pressures. These pressures together with van
der Waals attractions (among solutes and surfaces) should contract high numbers of inter-platelet
distances to a so small level that diffuse layers begin to overlap. In this situation, bentonite begins to
act as a semipermeable membrane. Anions attempting to migrate through the narrow pathway
between the platelets are repelled by the negative charge of the platelet surfaces. Anion flux through
the pathway is reduced, and since the electrical neutrality must apply cation flux is reduced as well
(the concept is called also as Donnan exclusion). The result is that the migration of neutral
molecules (esp. water) through the pathway is preferred.

The overlapping diffuse layers may offer a solution to the BET area reduction problem considered
above. A large number of inter-platelet apertures may contract to the overlapping diffuse layer
condition, and a smaller number clay platelets are arranged (by gelation and flocculation) to form
tiny channels and local voids. With this heterogeneity, the void sizes visible in electron microscope
(cf. Pusch 2001) could be explained also from the geochemical viewpoint.

The foregoing consideration related to diffuse double layers pointed out several sources of
uncertainties and ambiguities. There are doubts whether diffuse layer thickness calculated with
Equation (1) apply for the current problem, as well as, there are methodological problems to define
the thickness to a desired one. The geochemical constraints that are related to clay platelet surfaces
give much smaller pore volume estimates than observed with electron microscope. Microscopic
observations indicate that there are plenty of "free" porewater in compacted bentonite, but certain
geochemical laboratory experiments indicate that salt exclusion processes apply to compacted
bentonite and little "free" porewater is available.

Calculation results presented in Chapter 6 must be considered as presentation of a modelling


approach. The calculations assume that "free" porewater exists within tunnel backfill and bentonite
buffer. Major reasons for this assumption are the above-mentioned uncertainties related to diffuse
layers. The other reasons are the code limitations ofPHREEQC-2.7. The Borkovec-Westall method
applied assumes the "free" electroneutral porewater to occur. On the other hand, the no-free-
porewater cases require the application of the Donnan concept on the pathways. Currently, this
concept is not implemented in the PHREEQC code.
15

5 MODELLING APPROACH

5.1 EBS interior

In both buffer and backfill cases, porewater parcels are transported through successive EBS cells.
An illustration of the modelling is presented in Figure 5. The calculation starts from the EBS
boundary. The system is equilibrated and then porewater is transported into a next undersaturated
material cell deeper in EBS. In repeated process porewater equilibrates with successive volumes,
and the approach attempts to simulate the first wetting of the EBS.

--1)111• WATER PARCEL MOVING

Figure 5. An illustration of the modelling method for EBS interior. A porewater parcel is
repeatedly transferred to a new undersaturated EBS cell.

During the first wetting, entrapped air is consumed away from the cells. In real, however, formation
of gas bubbles within and around the repository is likely. After the first wetting, the repeated
external recharge at the EBS boundary defines what kinds of water compositions are sent deeper in
EBS. The later cycles of EBS interior transport, however, are not modelled here.

5.2 EBS boundary

These calculations consider porewater and solid phase evolution in one cell at the EBS boundary as
new parcels of external water repeatedly fill the pore spaces of an edge-cell volume. An illustration
of the modelling method is shown in Figure 6. At start of the modelling, the first parcel of external
water fills the undersaturated cell. After the first cycle, next refills occur into the saturated cell. The
porewaters created in successive cycles are sent deeper into the EBS.

EXTERNAL
WATER

MINERAL
REACTIONS

~ TONEXT
~ CELL

Figure 6. An illustration of the modelling method at EBS boundary. The studied cell volume is
repeatedly filled with external water.
16

6 MODELLING RESULTS

6.1 EBS interior

The modelled evolution occurring in the subsequent buffer and backfill cells (cf. Fig. 5) are
illustrated in Figures 7-10. The diagrams in Figures 7 and 9 illustrate how "free" porewater
composition and solids in the reaction cells evolve as wetting front advances as a function of cell
number. The composition of initial external water (Table 7) and the initial concentrations of solids
(Tables 2 and 6) are presented on the left borders of diagrams. The diagrams in Figures 8 and 10
illustrate the bulk geochemical conditions within charged surfaces and diffuse layers after the first
wetting of a cell. The initial conditions at the surface layers and in the diffuse layers (cf. Fig. 4) are
based on the clay platelet equilibration to the initial porewater composition presented in Table 8.

As external water enters the first cell, distinct changes occur in the solid phases and "free"
porewater (Figs. 7 and 9). The undersaturated cell volume contains atmospheric pressure of oxygen
gas that is consumed with pyrite dissolution. The amount of oxygen gas is constant in every
subsequent cell. Therefore, the amount of pyrite dissolution stabilises to a constant level.
Dissolution of pyrite is a constant source of Fe. Significant part of the Fe input is removed from
porewater with goethite precipitation.

In both buffer and backfill cases, all cells indicate complete dissolution of gypsum (Figs. 7 and 9).
Subsequent dissolution cycles increase gradually the sol- concentrations in "free" porewater to a
high level. However, the gypsum solubility limit is never reached. This occurs because of the
existence of diffuse layers upon the clay platelet surfaces. The corresponding changes of first
wetting in the diffuse double layers are presented in Figures 8 and 10. The surface layers of clay
platelets have a negative charge, and attract primarily solute cations. Still, there are significant
concentrations of solute anions within diffuse layers as well. As Figures 8 and 10 exhibit, sol- ions
are the dominant anions within diffuse layers after the couple first wetting cycles. In the chain of
subsequent equilibration reactions of first wetting, a significant part of the gypsum dissolution
generated sol- is left behind in the reacted cell while the equilibrated porewater is poured into the
next reaction cell.

Similar explanation applies to dissolution of quartz. Without the diffuse layers upon the clay
platelets, quartz dissolution would reach its solubility limit during the wetting of the first reaction
cell (Figs. 7 and 9). Thereafter, no quartz dissolution would occur in the subsequent reaction cells.
However, because of the diffuse layers of the left-behind cells, and constant level of Si in these
diffuse layers, quartz is always somewhat sub saturated within "free" porewater while it is poured
into a new reaction cell. Consequently, after a small initial anomaly, quartz dissolution stabilises to
a constant level within every reaction cell.

In accordance with expectations, all halite from the wetted cells is dissolved completely. During the
equilibration of every cell, a part of dissolved Cl is distributed into the diffuse layers. The Cl
concentrations equilibrate to a constant level in "free" porewater as well as in the diffuse layers of
the left-behind cells (Figs. 7-1 0). The dissolution of halite is the main source for new Na in the
wetting front studies (cf. Fig. 3) because the "free" porewater compositions are quickly equilibrated
to Na-bentonite. Essentially no cation exchange occurs after the first reaction cycle. However, the
diffuse layers take care that Na concentrations are gradually balanced to a constant level. The Na
concentrations in left-behind diffuse layers are shown in Figures 8 and 10.
17

BUFFER WETTING FRONT

Composition of Moving Solid Phases


Porewater Parcel after First Wetting

7 .3~-+--+-+-+-+--+-+-+-+--+-+-+-+--+--i-+-+-~
1.2~-.,.--------,-...,.-.,...-,----,---------,...---,-~

1.0 -
-.§.0
0.025
n o
DJ o
0.020 " ~
Jt Q) 0.3 8.705 ~
"""" U.
0.015 g~
_ a. 3
3 N' 0.2 .700 g
0.010 2. ·c.
- (!)
>.
II 0.1
l
l
I
.695
t ~
f 0.005 M"
u
(.)
0.000
I

I
5.0E-6
-
0
.§_0.000 O.OE+O
4.0E-6
N I>
0 -!< "0
0 ., :I:-0.005 -2.0E-4~
:- 0.12 3.0E-6m ~ I>
c;; - <J

N'
ga-0.010 ---.,.--t-------'----+--+-+-+-4. OE-4
0
;:t
..::::"'""
~ 0.08 ~-:-+-ir-+~~~---+-i-~--! 2 . 0E-6 Q. 3
$ 0.015 -6.0E-4_g
Cii
0 <J
T"" 0.04
« 8 -0.020 +
~ <J ~~----+~1-+
: --+--+~+--+~

<( 0.00 .OE+O -0.02 -1.0E-3


0 4 8 12 16 20 24 28 32 36 40 0 4 8 12 16 20 24 28 32 36 40
Cell Number Cell Number
Figure 7. Geochemical evolution of moving ''free" porewater parcel and compositions of material
volumes after first wetting of buffer. The porewater parcel is transferred 40 times to a new
undersaturated buffer cell. The equilibrium temperature assumption is 65°C. Alk. = alkalinity, Cc =
calcite, Gyp= gypsum, Py =pyrite, FeH = goethite.
18

BUFFER WETTING FRONT

Surficial Charges Diffuse Layer Composition


after First Wetting after First Wetting

1.2E-3

1.0E-3 ~ ~
:::- 0. 21 +---+--t----........____f---'----'---!--+--t----'---t-......_,..-r-+- 06 0.012
· (')
0 0 -
r-~~~~~-+~~~~--~ ~
.§.0.20 8.0E-4 ::I: ~ ~,.
b : : i<t:n 0.4 0.008 c;
~ 0.19 6.0E-4~ ~
0.18
~~~~~~~~--~--~~ -
-
~
z
30
4.0E-4
0.004..::::
0.17

0.16+-+-+----+----+-+-+--+-+--+-+-+--+-+-+-+--+
-96. Or----r--....----r--r------:---.,..--r------..,...-----:--,.q

l - 0.20
~-97.0
!!.,
ai
uQ) -98.0
3
CJ
~::I
r--t---+-r---~~~-+-~-t0.1 0 _g
cn -99 .0
0.05

l '
-100.0 i 10.0 0.00 0.00
0 4 8 12 16 20 24 28 ~ 36 40 0 4 8 12 16 20 24 28 32 36 40
Cell Number Cell Number
Figure 8. Bulk geochemical conditions within charged surfaces of clay platelets and diffuse layer
after first wetting of buffer. The ''free" porewater parcel is transferred 40 times to a new
undersaturated buffer cell.

Both buffer and backfill modelling cases exhibit consistent behaviour in respect of calcite. The first
reaction cell (Figs. 7 and 9) dissolve significantly calcite and consequently the alkalinity level in
"free" porewater rises considerably. The initial sub saturated reaction cells were equilibrated to pH
of 8.0 (Table 8) and the external water introduced in the first cell is at pH level of 7.1 (Table 7). The
first cell wetting releases protons into solution (Figs. 8 and 10). Similarly, the dissolution of pyrite
contributes a constant amount of protons in the solution. Calcite dissolution counterbalances these
proton contributions. In all however, the factors affecting the calcite equilibria are complex. The
initial external water is saturated in respect of calcite at 25°C. The simple temperature rise should
lead to calcite supersaturation. This does not occur, because a part of HC0 3- is distributed in diffuse
layers (Figs. 8 and 10) lowering the HC03- concentrations in "free" porewater. At the same time,
the first buffer and backfill cells dissolve different amounts of pyrite (because of different amounts
of initial 0 2). Finally, as water parcel enters a cell, it meets an atmospheric pressure of C02 gas. The
atmospheric C02 concentration is low compared to porewater alkalinity, and because pH is kept at
constant level, part of dissolved HC03- is speciated to C02 and escapes in the gas phase (Fig. 11 ).
,------
19

BACKFILL WETTING FRONT

Composition of Moving Solid Phases


Porewater Parcel after First Wetting
9.3 -2.4 1.0
I
: : Na·
8.8 -3.4 ~ 0.8 f-'
~
I
J:
a. tI I -4.4 "C
CO
-
en
m
Q)

Q)
0.6 i
!
~

-+l
C)
i c: 0.4 -~
t
CV
II
.c:
(,) Ii I
I -5.4 w
><
0.2 ~ 1 r Oa I
2.1I f
! l Mg2r
1"\
7 . 3~-+--+-+-+-+--+-+-+-+--+-+-+-+--+-+-+-+-~ -6.4 O.v
1.2.--:------,-....,..--,..--,---,---,------,--,..--.----,-_,---"T0.030 0.5 22.20

0
-
0
.§.
1.0

0.8
0.025

0.020 ,!»
O N
.§.
:i:
0.4 22.15

0 ",.u..Cl> 0.3 t-+-+-~~------+-+-+-+-+-..,.._--+--+-.......;-.t 22.1 0 ~


0 I
.."'""'
C)
0.015 ~
0
t - 3
:E -
3 ..
N 0.2 22.05 g
nf 0.010 0 ~ i i : I I '
z
~ _JJ~
..:::: (!)
M' 0.1 22.00
0
u I

-
0
i 4.0E-6 .§. 0.000
N
0 ...
0 =;-o.oo5
:- 0.12 3 . 0E-6~ u..
(ij _ <J
3 0~0.010
N' 0 T"""
G- 0.08 l-tT=::::f::==:===~=$:=:==f2 . 0E-Q5 Q.
Cii ~0 . 015
0 <I
T'"' 0.04 1.0E-6 -
"'~ 8 -0.020 '+-+-+----.-+--+--+--+-+-.....---+-~+-+-I -8.0E-4
<( 1
<I 1 I
0.000 .OE+O -0.02 -1.0E-3
4 8 12 16 20 24 28 32 36 40 0 4 8 12 16 20 24 28 32 36 40
Cell Number Cell Number
Figure 9. Geochemical evolution of moving ''free" porewater parcel and compositions of material
volumes after first wetting of backjill. The porewater parcel is transferred 40 times to a new
undersaturated backjill cell. The equilibrium temperature assumption is 40°C. Alk. = alkalinity, Cc
= calcite, Gyp = gypsum, Py =pyrite, FeH = goethite.
20

BACKFILL WETTING FRONT

Surficial Charges Diffuse Layer Composition


after First Wetting after First Wetting
0.1 Or--r--..----..-----,--.,.......,...--,-----,------,..------,..-~

0.09

0.012
(')
JU

0.008 "
~
>t

30
---+--+--!~-+ -+--:-+--+--+--+---+0. 004 ..::::

0.000
I 0.25

c
3: :::-0.04 f 0.20
~-97.0 70.0 r:::: 0
~
~ .§. en
cri I ~ 0 o.o3 0.15 ~
u-98.0+-t+--+-+- 50.0 ~ 0
Cl) I-

""' ..... 3
u
~ (;0.02 0.10 _g
~
:::::s !l'o
tJ) -99.0 30.0 - :I:
i3. 0.01 0.05

i -r~-r~~--~--~~
-100.0 10.0 0.00
0 4 8 12 16 20 24 28 32 36 40 4 8 12 16 20 24 28 32 36 40
Cell Number Cell Number
Figure 10. Bulk geochemical conditions within charged surfaces of clay platelets and diffuse layer
after first wetting of buffer. The ''free" porewater parcel is transferred 40 times to a new
undersaturated backji/1 cell.

Figure 11 illustrates the final C02 gas concentrations after water/solid interactions within the
reaction cells. The calcite dissolution reactions considered above promote the abrupt increase in
C02 gas concentrations at the start of porewater parcel transport. The gas phase evolution in the
later cells is defined with porewater alkalinity content. Every time porewater parcel enters a new
cell, it speciates part of its dissolved carbon to gas phase (initial C02 content in the cells is low).
However, at the same time mineral equilibration contributes new dissolved carbon to the system.
The result is that gradually all phases ("free" porewater, diffuse layers, solids and gas phase) fmd
constant equilibrium levels for carbon mass balance (cf. Figs. 7-11 ).

Figure 11 represents the final amount of C02 gas left behind or pushed away from the migration
path of moving water parcel. In both cases, the C02 gas (and atmospheric N2 that is not considered)
has implications to further modelling considerations. The left-behind gases indicate that the cells are
not completely saturated with porewater after first wetting as it is assumed in the current
calculations. Furthermore, Figure 11 indicates that the left-behind gas phase is not inert after first
wetting, but contains still reactive species (i.e. C02).
21

EVOLUTION OF C0 2 GAS CONCENTRATIONS


IN THE WETTING FRONT STUDIES

Buffer Backfill
2.0E-4
!I
I , I I
I

t -- c-
I
-~· i-t

('
1.6E-4 ! ! :

-.§_0
!

-0 ~ ~
I
I !

1.2E-4 1.2E-4
~
f
.§. 1 i !
UJ
ns m
ns I iI I! ! I
I
I
i
i
I
:

~8. OE-5 t-f-.-~+-+-i--"--'---.;_4-----t--i-~---t---+-+--l---.---J ! i


C) t I I I
8.0E-5 - I
i
0 o"' i
I

0 0 !

.
I
I
4.0E-5 +
I
t ! 4.0E-5
1
I
I
I :
! I i i
O.OE+OO 4 8 12 16 20 24 28 32 36 40 O.OE+OO 4 8 12 16 20 24 28 32 36 40
Cell Number Cell Number
Figure 11. Final C02 concentrations after the wetting of EBS cells. The initial C02 gas
concentrations in undersaturated cell volumes are 0. 004 mmol/L and 0. 006 mmol/L in buffer and
backfill, respectively.

Considering all modelled phases ("free" porewater, diffuse layers, solids and gas phase) together, it
becomes evident that while the system is defined with a number of constants, the modelled results
tend to approach constant levels. There are no temperature gradients during water parcel transport.
The amount of reacting gases and the volumes of diffuse layers are kept constant. The final
equilibria (after 40 transfers of the water parcel) for the "free" porewater look remarkably similar
between the buffer and backfill cases. The significant differences originate mostly from the
differences in the amounts of oxygen gas content and temperature in the water sub saturated cells.
For example, because backfill cells contain a little more oxygen compared to buffer cells, they
dissolve a little more pyrite. Due to temperature difference, the equilibrium pH finds a slightly
lower level in the backfill case than in the buffer case, and therefore a little more carbon is speciated
into the gas phase in the backfill case (Fig. 11 ).

However, the wetting front calculations rise certain potentially significant issues. The sulphate
concentrations grow in the "free" porewater, as well as in diffuse layers, to quite high levels. The
sulphate storage into diffuse layers indicates that the sulphate pulse in "free" porewater during the
first wetting is not a single pulse but the following porewater parcels carried through the wetted
transport path will inherit sulphate from diffuse layers. Similarly, the Cl barrier created in the
diffuse layers of the couple first cells will begin to move if calculations would be continued with
calculations of the following sets of EBS interior transport.
22

6.2 EBS boundary

The simulated geochemical evolutions at the EBS boundaries (buffer and backfill) are illustrated in
Figures 12-15. The studied boundary cells (Fig. 6) are refilled 40 times with external water (Table
7). The diagrams in Figures 13 and 15 show how the bulk geochemical conditions within charged
surfaces, and in diffuse layers evolve as a function of external water refills. The first batch reaction
cycle is similar to the first reaction cycle of the previous wetting front studies (Figs. 7-1 0). The
initial undersaturated cell contains entrapped oxygen that is consumed away. Subsequent reaction
cycles occur in the anoxic saturated conditions.

During the first reaction cycle, pyrite dissolution is evident. The consumption of 0 2 and
contemporaneous release of protons from the surface sites (Figs. 13 and 15) do not cause drop in pH
because of counterbalancing effect of calcite dissolution. The one-time dissolution of pyrite
produces Fe in solution that is removed with goethite precipitation. During the following batch
cycles, the surface with negative charge still releases small amounts of protons into the solution.
However, this has little effect to pH that stays at relatively high level due to calcite dissolution.

Similarly, all gypsum is dissolved from the boundary cells during the first reaction cycle. This leads
to significant production of dissolved sulphate and Ca in water. The role of cation exchange is
significant during high Ca production. Cation exchange is an almost complete sink for produced Ca
and it releases other major cations into solution. Similarly, because sulphate runs out from
porewater after the first cycle, the pe values have a chance to find a lower level.

The difference in CEC between the buffer and backfill is evident while comparing the cation
occupation in the exchange sites. In the backfill case (Fig. 14) the fewer exchange sites are filled
faster with Ca than in the buffer case (Fig. 12) In the backfill case, the exchange sites have smaller
capability to regulate cation concentrations in water. However, this feature is screened relatively
well with counteractions caused by diffuse layer (Figs. 13 and 15). Cation exchange releases a lot
Na in the solution, However, because there is much more diffuse layer volume available in the
buffer cases, a great deal of the released Na is bound into the diffuse layers. Consequently, the
major cation concentrations in "free" porewater seem the stay approximately at same level in both
buffer and backfill modelling cases. This simulation result indicates that the compositional changes
within bentonite exchange sites cannot be followed very clearly with compositional changes
occurring in "free" porewater.

The distribution of elements into the diffuse layer has also other significant features. In the case of
quartz, the temperature rise leads to quartz sub saturation in external water during the beginning of
every batch reaction cycle. This would lead to a constant dissolution level of quartz during every
cycle without the diffuse double layers. However, because of diffuse layers, there is smoothly
evolving variation in the dissolution amounts of quartz as a function of batch reaction cycles.
Similarly, the diffuse layers contribute to the Cl concentrations of the equilibrated "free" porewater.
During the first batch reaction cycle, a significant part of the Cl input (external water and halite
dissolution) is distributed into the diffuse layers (Figs. 13 and 15). During the following cycles,
some Cl is released from the diffuse layers into the "free" porewater. At the same time, new Cl is
introduced with new parcels of external water. The result is that the Cl concentrations equilibrate
relatively quickly to constant levels in "free" porewater.
23

BUFFER BOUNDARY EVOLUTION

Porewater Composition Solid Phases


in a Boundary Cell in a Boundary Cell
9' 3:.,----,.--,- ,,---,-...,..1--,-
'1! ........-r-""'T""--r,-"T"""-r
1 ...,.1,--...-- 1. . ,. . . . .,,......,...--,--..-,--,---. -2.4 1.0

I 1 1I I! i I I i
I 1 : I I I !
~
8.8 ~v~
I , - r-ti-! ~~lh~
! j · I+ -3.4 e...
f/)
....
'-~
Q)
1
(ii 0.6

zQ. 8.3 - .·-


: i
I
~--t- - I - l - -
I 1
I! ·- f- -4.4 "'0
C1)
Q)
C)
c:
· 1 I i CO 0.4
i I ~

lJ f\1
11. . . . .1-....._~
(.)
I I ><
?.Bv I \~--
-5.4 w
...,......
,
I -+--1--o.--+-+--r--1:--r\ I I I
Ii ! j I I I I i I ! !
7.3
I
-6.4
1.2 0.030
-.§.0
1.0 0.025
-0 C)
o o0
:I: ~
--!---l--;--i--~8 . 710

.§.. 0.8
0
0
0.020 w

"..... Q)
* U.
0.3 ~
8.705 g
~

"'C)
0.6 0.015 0
0
~
0.. 3
~ 3 N' 0.2 .700 ,g
ri 0.4 0.010 0 "'c.
z -
- (.!)
>-
M' 0.1
0.005
(J
(.)

0.000
.OE-6
-0
.§..0.000
N [>
0 "'~-0 . 005 ·-+----J~--+---.~ -2. 0E-4~
0
:- 0.12 3 . 0E-6~ u. C>
(ii _ <1
N~
3 o""0.010 I -4.0E-4 g
o ~
~ 0 . 08 2.0E-O::: Q.. I -
Cii $-0.015 - - -6.0E-4 g
I .:::
0 <1
:- o.o4 c.io 020 -8.0E-4
~ u ·
~ <1
0 OO ~..::.P....._.................,_.p.....j......j......_~_....._..._.
. 0 4 8 12 16 20 24 28 32
..........
36 40
Ul.OE+O -0.02
0 4 8 12 16 20 24 28 32 36
-1.0E-3
40
Batch Reaction Cycle Batch Reaction Cycle
Figure 12. Geochemical evolution in material properties and resulting ''.free" porewater
compositions in an EBS cell volume at the rock-buffer boundary. The cell volume is refilled 40
times with water presented in Table 7. The equilibrium temperature assumption is 65°C. Alk.
alkalinity, Cc = calcite, Gyp = gypsum, Py = pyrite, FeH = goethite.
24

BUFFER BOUNDARY EVOLUTION

Surficial Charges Diffuse Layer Composition


in a Boundary Cell in a Boundary Cell

1.0E-3 ~ ~ 06
·
o- JU
8.oE-4 ::I: ~ ~
..
: . : itC) 0.4 0.008 c;
6.0E-4 3 :lE
0
-- z"'
~
30
0.18 I 4.0E-4
0.004..::::
0.17

c
3: :::-0.04
~-97 .0+---+iH-+-'- 70.0 c 0
~ g: .§.
a:i I; u 0.03
uG,) -98.0 50.0 ~ 0
........ 3
0. .0.02
CJ
~ ~ 0.10 ,g
:l tou
cn -99.0 30.0 ..:.... :I:
~ 0.01 0.05

-100.0 10.0 0.00 0.00


0 4 8 12 16 20 24 28 32 36 40 0 4 8 12 16 20 24 28 32 36 40
Batch Reaction Cycle Batch Reaction Cycle
Figure 13. Bulk geochemical conditions within charged surfaces of clay platelets and diffuse layer
in an EBS cell volume at the rock-buffer boundary. The cell volume is refilled 40 times with water
presented in Table 7.

Both the buffer and backfill simulations indicate similar behaviour in respect of calcite during the
batch reaction cycles. The first reaction cycle is similar to considerations made in the previous
Chapter 6.1. During the second cycle, calcite is still dissolved due to C02 gas left behind in the
reaction cell during the first cycle, and because the bentonite exchange sites act as a Ca sink. The
bentonite keeps the system calcite sub saturated during the later cycles as well. Relatively high
alkalinities inherited from the diffuse layers, created during the previous batch cycles, together with
relatively low Ca available keeps the calcite dissolving. However, according to current calculations
there is a calcite precipitation tendency in the repository tunnel boundaries during the later cycles.
At the batch reaction cycle 20, the alkalinity concentration in "free" porewater drops below a limit,
and the system begins to precipitate calcite. Approximately at the same time, bentonite becomes Ca
dominated, and cycle-after-cycle a little more Ca is left in the final "free" porewater. Calcite
precipitation enhances the pH drop in the backfill modelling case (Fig. 14 ). In principle, the
temperature rise should increase the calcite precipitation tendency. However, this aspect is
completely screened with the bicarbonate distribution into the diffuse layers. Although the
temperature is higher in the buffer case (65°C) it does not precipitate calcite.
: - - - - - - - - - -- - - - -- - -- -- - - -- -- - -- -~- -

25

BACKFILL BOUNDARY EVOLUTION

Porewater Composition Solid Phases


in a Boundary Cell in a Boundary Cell

7.3 0.
1.2 0.5

0
-.§.
0
1.0

0.8 0.020 .!»


O N
.§.
J:
0.4 • 22.15

0
0
;"i &f
*
0.3~--i-.........-"!"""""i'......,..."""""""+-t~- . . . . . .-"!'-!'-1-22.10 w
..... 0.6 0.015 ~ t-
-le
0) 3
~ -
0
3 •
N 0.2 22.05 ,g
tO 0.4 0.010 0 g;
z ...:::: <!>
M' 0.1
0
u
l.j.....L.io_ _ _ _ _ __.__._-+-+-+--+--............__.........---421.95
.-r--r--r--r--------:---,--,--"T'""""""'T"--,----.,.--,--.......--.---. 2.0 E-4

-.§.
0
0.000
4.0E-6
N
I>-
g ~0.005 f1 -2 . 0E-4 ~
:- 0.12 .. t-+~-,--~------;---+-+--+-+--+-~-~-+--i--+-+- 3 .0E-6~ u. 1
00 ~
I>-

N' '[ ;!...o.o1 o I -4.0E-4 w


G- 0.08
;;;
0
2 . 0E-~ Q.

$ o.o15
<] I
Ti
I
,

I
-6.0E-4_g 3
..... 0.04 1.0E-6
• I
8 --0.020 ' I; ,'
-8.0E-4
~ <l I I
C( O.Oo ~:...----+-~-_,__..._ _,.......___...~.OE+O -0.02 -1 .0E-3
0 4 8 12 16 20 24 28 32 36 40 0 4 8 12 16 20 24 28 32 36 40
Batch Reaction Cycle Batch Reaction Cycle
Figure 14. Geochemical evolution in material properties and resulting "free" porewater
compositions in an EBS cell volume at the rock-baclifill boundary. The cell volume is refilled 40
times with water presented in Table 7. The equilibrium temperature assumption is 40°C. Alk.
alkalinity, Cc = calcite, Gyp = gypsum, Py =pyrite, FeH = goethite.
26

BACKFILL BOUNDARY EVOLUTION

Surficial Charges Diffuse Layer Composition


in a Boundary Cell in a Boundary Cell

-,-~r---t--t---+ 0.012
(')
~

0.008 "c;,.
30
+-+-1---'---- - 0.004 ..::

c
==
c
(/1
:-0.04
0
E
t-r I 0.20

CJ)
ai (;CD u
-
0.03 0.15 ~
u4) -98.0 '< 0 I 3
..,CD .....
u
f) ()0.02 0.10 g
"'
't:
::::J !l'u
en -99.0 - :I:
~ 0.01 0.05
~

-100.0 10.0 0.00


0 4 8 12 16 20 24 28 32 36 40 4 8 12 16 20 24 28 32 36 40
Batch Reaction Cycle Batch Reaction Cycle
Figure 15. Bulk geochemical conditions within charged surfaces of clay platelets and diffuse layer
in an EBS cell volume at the rock-backfill boundary. The cell volume is refilled 40 times with water
presented in Table 7.

Considering the overall behaviour of equilibrium minerals, it can be concluded that the 40 batch
cycles, in most cases, have negligible effect on the total mineral amounts available in the cells. The
exception is gypsum that is dissolved away from the reaction cells during the first wetting cycle.
Quartz dissolves to its solubility limit during each batch reaction cycle but the reserve is large.
Based on Tables 2 and 6, it can be concluded that the amounts of most silicates are so high that
dramatic changes in the silicate reserves are not probable in the conditions studied. However,
silicate dissolution reactions might contribute to resulting porewater compositions.
- - - - --- - - - - - -

27

7 COMPARISON OF MODELLING RESULTS TO OBSERVATIONS

Yet, there is no porewater analytical data available from the Prototype repository EBS. Partially this
is due to delays in the repository operation (malfunctions in the heated canisters). In the course of
time, the on-line porewater samplings will also be challenging tasks. In the view of solid phase
observations, the alteration studies cannot be done until the repository is dismantled. For these
reasons, all following considerations rely on the literature review.

At the EBS boundaries, the current calculations predict quick and complete gypsum dissolution.
Small calcite precipitation is predicted at the repository tunnel boundaries after a number of external
water refills. Quartz is constantly dissolved from the repository boundary due to temperature
gradient effect. However, as soon as temperature of porewater drops quartz will precipitate, i.e. a
temperature driven hydraulic convection shell may be generated around the repository that
redistributes e.g. quartz. In the view of cation exchange, the calculations predict that Ca from
external water, and from dissolved gypsum, will be bound to cation exchange sites ofEBS.

During the first wetting of EBS interior, models dissolve gypsum and halite, and precipitate
goethite. Small amounts of calcite and pyrite are constantly dissolved into porewater as wetting
front advances from the EBS boundaries. Calculations indicate that significant concentrations of
sulphate are carried within "free" porewater. Likewise, the calculations let presume that
considerable concentrations of sulphate are left within diffuse layers of the clay platelets. Likely,
these reserves begin to move if the following cycles of EBS interior evolution would be modelled.
Considering the Cl concentrations and constant dissolution of halite, the wetting front calculations
indicate that Cl concentrations will not grow in the moving "free" porewater parcel, but a part of
dissolved Cl is constantly distributed in the diffuse layers.

The silica redistribution in the repository has been predicted by e.g. Carnahan (1987). Also the
mineralogical observations of "long term test of buffer material (LOT)" -experiment (Karnland et
al., 2000) verify the redistribution of Si in the canister buffer system. Kamland et al. also report the
gypsum precipitation in the buffer interior. According to observations by Muurinen (2003) from the
LOT-experiment, the sulphate levels in porewater are gypsum solubility limited. Results of
Muurinen (2003) and Kamland et al. (2000) confirm the cation exchange trends modelled for
bentonite buffer (and tunnel backfill). Part of Ca coming with external water is exchanged with Na
in the cation exchange sites, and Na is released into porewater. According to Muurinen (2003), a
part ofNa in analysed porewaters originates from mineral dissolution (i.e. halite).

The observations of Karnland et al. (2000) and Muurinen (2003) are based on the real buffer
experiment where true temperature and dehydration gradients occur towards the canister surface (cf.
Kamland et al., 2000; Pusch, 2001 ). In this canister-heat-driven convection shell system, porewater
is vaporised near the canister surface and vapour migrates towards colder regions. This is a strong
extraction process leading to precipitation gypsum and redistribution of Si. These results cannot be
duplicated with the current modelling calculations, because the temperature rise and water
saturation were held constant. However, the modelling calculations indicate that the pieces (high
dissolved S04 and Si content) for the precipitation phenomena are available in the modelled "free"
porewaters.
28

The results of Muurinen (2003) indicate very low Fe concentrations in porewater. This is in
accordance with the modelled Fe concentrations. If Fe is produced in porewater e.g. with pyrite
dissolution then significant part of Fe has to be removed as well e.g. with goethite precipitation.

The calculations predicted relatively moderate pH values for "free" porewaters in the EBS interior.
The calculated pH equilibrates to the value of 7.6 in both buffer and backfill wetting front
calculations. According to Muurinen (2003), pH values varied in the LOT samples from 7.6 to 8.5,
the high values being from found close to the copper tube (i.e. dehydration affects). In the EBS
boundary calculations, pH values reach relatively high readings. In the backfill modelling case, the
highest pH values reach 9.2, while in the buffer modelling case the highest values are around 8.9.
These results can be compared to two flow-through experiments done with the MX-80 bentonite
(Vuorinen et al., 2003). In these experiments, slowly flowing external water passes by and
equilibrates with a compacted bentonite surface. After 300 days the fresh water (ionic strength
0.003 eq/kgw, initial pH 8.8) experiment gives pH values around 9.6 from the out-flow water, while
the saline water (ionic strength 0.5 eq/kgw, initial pH 8.3) gives values around 8.0. The gap
between the experimented 8.0 and 9.6 is wide. Nevertheless, the experimented results give the
impression that the less internally buffered (i.e. low ionic strength) is the initial external water the
higher pH values the final out-flow water seem to get. The external water ionic strength (0.125
eq/kgw) utilised in the calculations and the pH results received are in loose concordance with this
deduction.

As a conclusion, many processes considered in the modelling calculations do agree with the recent
experimental results. The complete modelling system is, however, sensitive to modifications.
Significant changes to porewater modelling results are easily done e.g. by changing the CEC of
reaction cells, amounts of reactive minerals (esp. gypsum, halite), or allowing temperature gradients
and water saturation changes in the first wetting modelling.
29

8 DISCUSSION

It seems unlikely that organic matter would be the principal reducer of the entrapped oxygen in the
studied systems. As Tables 2 and 6 indicate there is carbon (CH 20) available. However, if oxygen
is consumed in aerobic respiration instead of dissolution of pyrite, alkalinities in the modelled
porewaters grow to unreasonable levels. In order to compensate this, at least partially, a strong
precipitation of calcite would be needed and consequently a significant drop in pH should occur.
Therefore, the sulphide oxidation is considered the dominant process consuming oxygen in the
calculations. This is in accordance with the results of Pedersen (2000) and Karnland et al. (2000).
The amount of viable microbes decreases rapidly during swelling of bentonite and leads to
eradication of life in water saturation.

The surface complexation mechanism utilised in this study is simplified. The currently adapted
model is single sited. However, it was discussed already in Chapter 4 that the basal planes of clay
platelets are usually charged negatively while the edges of clay platelets tend to have slight positive
charge (cf. also Fig. 16). Recently, Bradbury & Baeyens (1997) introduced their detailed
mechanistic surface complexation model for Na-montmorillonite that considers three different
surface sites available for complexation. Of these sites, the significance of the strongly bound site
(=S 50H) is negligible for most studies. However, at moderate pH the two weak protolysis sites
imitate well the currently modelled reality. For example at pH of 7, the first weak site (=Sw 10H) is
negatively charged while the second weak site (=Sw20H) is positively charged, but the total net
charge of the surface sites is negative. The decision to use the single site model was practical.
Calculations with two complexation sites lead to two separate diffuse layers and increasingly
complex illustrations and considerations.

During the first cycles of cell reactions, the wetting front studies (Ch. 6.1) concentrate strongly the
ionic strengths of solutions. In the buffer modelling case, the ionic strengths soon stabilise to the
level of 1.27 eq/kgw, and in the backfill case to the level of 1.29 eq/kgw. Both figures are about 10
times higher than in the initial external water. The constant dissolution of halite is the one reason for
the ionic strength increase, and the other one is the existence of diffuse layers. During the first
transport cycles of porewater parcel, ions from diffuse layers are released because the initial diffuse
layers were equilibrated to higher ionic strength (Table 8) than the external water (Table 7) poured
into the studied transport path.

There are other processes, as well, that likely increase the ionic strengths along a cumulative
transport path illustrated in Chapter 6.1. Wersin et al. (2004) consider the incorporation of water in
the interlayers of Na-montmorillonite. The interlayer spacing of bentonite varies a lot as a function
of bentonite water content. In the completely collapsed (anhydrous) state, the smectite interlayer
spacing is about 9.6 A (Fig. 16). Na-montmorillonite may contain one (doo 1 ~ 12.5 A), two (15.5 A),
three (19 A) or even four layers of water per an unit cell within its interlayers (Deer et al., 1995).
Wersin et al. (2004) estimate that, while compacted bentonite swells, the collapsed interlayer of
smectite expands to the form where an unit cell contains two layers of water. The result is removal
of water until 11-12 % of the volume of initial external water is left in the final "free" porewater.
This is a strong extraction process (Wersin et al., 2004) that doubles the ionic strength in the
residual "free" porewater compared to the case where no bentonite swelling occurs (and no water is
distributed within interlayers ). It is evident that doubling the ionic strength along a cumulative flow
path soon runs into difficulties in the currently considered calculation cases. Therefore,
counteracting processes against the ionic strength increase must be considered.
30

a) b)

Figure 16. Structural sketches on Na-montmorillonite structure. a) Internal structure of a clay


platelet at a fully collapsed state when the interlayer spacing dooJ is about 9. 6 A. b) Na-
montmorillonite structure having one layer of water per unit cell (d001 is ea. 12.5 A). The negative
charge (-) on the basal planes of di-octahedrallayers are shown in blue, while the slightly positive
charge (+) at the layer edges are shown in red.

The water incorporation within smectite interlayers and consequent swelling of bentonite is
chemical osmotic process (Keijzer, 2000) where neutral water molecules are driven by a chemical
potential towards the higher ionic concentrations occurring within smectite interlayers. The other
reason for bentonite swelling is the diffuse layers generated upon the basal planes of clay platelets.
If the resulting expansion of bentonite is prohibited, these hydration processes give rise to the
bentonite swelling pressure. It is well known that the bentonite swelling pressure is a function of
porewater salinity. According to Kamland ( 1997), a swelling pressure model fitting best to
experimental data postulates the zero swelling pressure for MX-80 bentonite with approximately 7
% NaCl solution. This limit corresponds to the ionic strength of 1.2 eq/kgw (Na and Cl
concentrations are 1.2 mol).

In the case of interlayer swelling, the swelling originates from the osmotic pressure difference
caused by the differences in water activities within the smectite interlayers and porewater (Keijser,
2000). The activity of water is related to molalities of solutes, ionic strength and temperature.
Within "free" porewater, the activity of water decreases as a function of ionic strength increase,
while within interlayers the activity of water increases with water intake. Both activities are related,
as well, to the diffusion driven cation exchange ability of smectite. The osmotic pressure difference
becomes zero as the water activities within the interlayers and porewater become equal.

The swelling related to diffuse layers has been recognised earlier to correlate with the ionic strength
of "free" porewater. It was pointed out in Chapter 4 that the increase of ionic strength from 0.12
eq/kgw to 1.2 eq/kgw should drop the thickness of diffuse layer from 10 A to around 3 A,
respectively.

The above consideration causes a speculation whether the swelling of compacted bentonite buffer is
an important process during the initial wetting cycle (Ch. 6.1 ). If the ionic strengths of first
porewaters rise to the level of 1.2 eq/kgw bentonite should loose its swelling ability. The modelling
results at the EBS boundary (Ch. 6.2) indicate that the ionic strengths in final "free" porewater
equilibrates quickly to the level of 0.11 eq/kgw in both buffer and backfill considerations. As these
31

waters are repeatedly transported deeper in the EBS, the swelling should gradually occur
propagating from the EBS boundaries towards the interior. However, the repeated cycles of
porewater transport within the EBS interior were not considered in the current calculations. In
concordance with the results of W ersin et al. (2004 ), it can be further speculated that the water
intake in the smectite interlayers is a major process that increases ionic strength during these
repeated transport cycles within the EBS. The increase rate of ionic strength within "free" moving
porewater would then define the swelling rate of bentonite rather than only the conservative water
transport through the EBS.

The physical studies of bentonite wetting indicate that wetting is a heterogeneous process (e.g.
Pusch, 2001 ). The non-uniform expansion of clay grains (glomerates of clay platelets) form small
channels and voids in the compacted solid. This indicates that there are likely two types of water
migrating in the compacted bentonite. Water migrating within clay grains is stuck into the slow
transport, while water in the channels is subject to more quick transport. Therefore, it seems that the
homogenous wetting processes considered above are simplified. However, the same geochemical
processes apply to channels as to homogenous bulk considered. It can be assumed that the relation
between the homogenous block of bentonite and the channelled bentonite is similar to the buffer
and backfill calculations. The available amounts of minerals and capacities of smectite should be
reduced for the channelled bentonite calculations.

The backfill contains 70o/o crushed TBM -muck. It was assumed that all cation and surface exchange
properties of backfill originate from bentonite. This judgement has its justification, but it is
undeniable, that the crushed rock contributes at least at small extent to cation exchange properties of
backfill. The uncertainties related to TBM-muck surface complexation properties were pointed out
already in Chapter 3. Likely, uncertainty related to protonation/deprotonation capacity of crushed
silicate surfaces is much larger than uncertainty related to the TBM-muck CEC.

The effect of temperature was taken partially into account in the calculations. However, no
temperature dependency is available for e.g. goethite dissolution/precipitation. Elevated
temperatures may have unexpected effects to the systems because a part of batch reaction times are
potentially extensive. Considering the silicates available in the buffer and backfill (Tables 2 and 6)
there are several phases susceptible for partial alteration or gradual decomposition (e.g.
montmorillonite, illite, amphibole, plagioclase, and K-feldspar). Stumm (1992) indicates that
temperature rise has quite clear effect on surface complexation processes. According to Borkovec &
Westall (1983), the Poisson-Bolzmann equation applied for diffuse layer concentration calculations
is temperature dependent.

The evolution of repository as a function of time was handled qualitatively only. At the beginning
of the repository operation, batch cycles follow each other more quickly than at the later stages of
operation. External water infiltration causes high porewater pressures at the repository boundaries
creating strong gradients in the saturated part ofEBS, and drives water deeper into EBS (Borgesson
& Hernelind, 1999). At the same time, there are high negative porewater pressures in the
undersaturated EBS. Therefore, the suction gradients created are possibly strong. Later the gradients
begin to diminish and finally vanish. This modifies the later stages of the systems and probably
thermal/geochemical gradients will be then the significant driving forces. In any case, near the
saturation level, porewater moves practically only by diffusion meaning very slow transport.
32

Possibly, the most significant simplifications made in the calculations are related to the repository
saturation. In the case of canister buffer studies the strong temperature gradient coupled with
degreasing degree of water saturation towards the canister surface is an important omission. Likely
due to this drawback, the modelling calculations cannot duplicate the gypsum precipitation close to
the canister observed in the LOT experiment.

A further saturation simplification is related to entrapped air. At the initial condition, the studied
EBS cells were undersaturated and pore volumes were partially filled with air. Calculations
assumed instant full saturation of a cell as soon as infiltrating water first time enters the studied cell.
Considering the repository as a whole this is impossible. Although, 0 2 is consumed from the
entrapped air, calculations indicate that a significant part of C02 speciates to gas phase during the
first wetting ofEBS. Moreover, the major gas component in entrapped air is N 2 that is not expected
to react during the repository wetting. Evidently, the repository wetting is gradual process where
migration of two phases (solution/gas) should be taken into account.

Finally, initial water composition utilised in calculations has potentially large significance to
simulation results. If the water composition presented in Table 7 changes with time, it has effect on
the geochemical evolution. In the view of long-term stability considerations these aspects have an
important role in safety assessment (e.g. Guimera et al., 1998; Puigdomenech et al., 2001).
Similarly, concrete, cast or injected into walls of repository tunnels and deposition holes, likely
alters the infiltrating external water composition (e.g. Huertas et al., 2000). However, these
problems are not topical in the view of current modelling or Prototype experiment.
33

9 CONCLUSIONS

The current calculations present a modelling approach that considers geochemical changes in the
Prototype repository EBS. The predictive calculations were batch reaction oriented. Infiltrating
water was put into an EBS (reaction) cell in an instantaneous manner and then water stays in the
cell as long as required for full chemical equilibration in respect of minerals, cation exchanges and
diffuse layers considered. The prediction concept contains no exact estimates how fast water
saturation (or partial saturation) front advances within the EBS (tunnel backfill or canister buffer).
The gradual wetting of EBS is a hydrological TH problem that was omitted in the present
considerations.

The predictive modelling calculations within the EBS rely on the constant near-field groundwater
(external water) composition. Presently, the hydraulic drawdown remains constant in the Aspo
Island due to the open tunnel conditions caused by the Hard Rock Laboratory. Therefore, it was
assumed that there would not be essential changes in the near-field groundwater compositions in the
near future (following 20 years) either.

The geochemical modelling within the repository EBS focused on the major element compositions
of porewaters, and on the certain solid phases of EBS. The model computations followed the
equilibrium thermodynamic assumptions and took into account the known cation exchange and
surface complexation properties of the EBS. The available chemical information was tried to take
into account as accurately as possible. Furthermore, calculations assume two types of porewaters.
One porewater is "free" capable to migrate within the EBS, while the other porewater is "stagnant"
suck within the diffuse layers of clay platelets. The calculations considered both reactions that occur
during the wetting of EBS interior and reactions that are related to time-dependent changes at the
EBS boundaries.

The reactive EBS solid concentrations were defined for 1 dm 3 of pore volume. Therefore, the
physical size of a single reaction cell became relatively large (an undersaturated bentonite cell
weights 9.25 kg, and backfill cell 11.5 kg). For more detailed coupled reactive transport
calculations, it may be preferable to define the reaction cell size suitable for other physical
dimensions of EBS, and model smaller porewater volumes.

Up do date, no analytical data (porewater compositions or solid phase alterations) are available
from the Prototype repository. However, certain mineralogical and porewater observations from
other experiments (esp. the "long term test of buffer material (LOT)") give conceptual support to
the modelling concept utilised in this study.
34

10 REFERENCES

Andersson, C & Safvestad, A (2000) Aspo Hard Rock Laboratory. Compilation of groundwater
chemistry data from the Prototype repository. March 1998 - June 1999 (unpubl. report).
Swedish Nuclear Fuel and Waste Management Co (SKB), Stockholm, Sweden.
International Technical Document 00-04: 6 p.
Borkovec, M. & Westall, J. (1983) Solution of the Poisson-Bolzmann equation for surface excesses
of ions in the diffuse layer at the oxide-electrolyte interface. Journal of Electroanalytical
Chemistry 150, 325-337.
Bradbury, MH & Baeyens, B (1997) A mechanistic description of Ni and Zn sorption on Na-
montmorillonite. Part II: modelling. Journal of Contaminant Hydrology 27: 223-248.
Bradbury, MH & Baeyens, B (2002) Porewater chemistry in compacted re-saturated MX-80
bentonite: Physico-chemical characterisation and geochemical modelling. Paul Scherrer
Institute, Villingen, Swizerland, PSI Bericht 02-10: 41 p.
Bradbury, MH & Baeyens, B (2003) Porewater chemistry in compacted re-saturated MX -80
bentonite. Journal of Contaminant Hydrology 61: 329-338.
Borgesson, L & Hemelind, J (1999) Aspo Hard Rock Laboratory. Prototype Repository.
Preliminary modelling of the water-saturation phase of the buffer and backfill materials.
Swedish Nuclear Fuel and Waste Management Co (SKB), Stockholm, Sweden.
International Progress Report 00-11: 99p.
Bruno, J, Arcos, D & Duro, L (1999) Processes and features affecting the near field
hydrochemistry. Groundwater-bentonite interaction. Swedish Nuclear Fuel and Waste
Management Co (SKB), Stockholm, Sweden. Technical Report 99-29: 56 p.
Camahan, CL (1987) Simulation of chemically reactive solute transport under conditions of
changing temperature. In: (ed. Tsang, C.-H.) Coupled processes associated with nuclear
waste repositories. Academic Press Inc., Orlando, USA., 249-257.
Davis, JA & Kent, DB (1990) Surface complexation modeling in aqueous geochemistry. In: (ed.
Ribbe, P.H.) Mineral-water interface geochemistry. Mineralogical Society of America,
Washington D. C., USA. Reviews in Mineralogy 23: 177-260.
Deer, WA, Howie, RA, Zussman, J (1995) An Introduction to the rock-forming minerals, 2nd ed.
Longman Group Limited, Essex, England. 696 p.
Guimen1, J, Duro, L, Jordana, S & Bruno, J (1998) Effects of ice melting and redox front migration
in low permeability media. In: Characterization and evaluation of sites for deep geological
disposal of radioactive waste in fractured rocks. Proceedings from the 3rd Aspo
International Seminar, Oskarshamn, June 10-12, 1998. Swedish Nuclear Fuel and Waste
Management Co (SKB), Stockholm, Sweden. Technical Report 98-10: 253-262.
Huertas, F, Farias, J, Griffault, L, Leguey, S, Cuevas, J, Ramirez, S, Vigil de la Villa, R, Cobefia, J,
Andrade, C, Alonso, MC, Hidalgo, A, Pameix, JC, Rassineux, F, Bouchet, A, Meunier, A,
Decarreau, A, Petit, S & Viellard, P (2000) Effects of cement on clay barrier performance.
Ecoclay project. European Commission, Luxembourg. Nuclear science and technology
series EUR 19609: 140 p.
Kamland, 0 (1997) Bentonite swelling pressure in strong NaCl solutions. Correlation between
model calculations and experimentally determined data. Swedish Nuclear Fuel and Waste
Management Co (SKB), Stockholm, Sweden. Technical Report 97-31. 30 p.
Kamland, 0, Sanden, T, Johannesson, L-E, Eriksen, TE, Jansson, M, Wold, S, Pedersen, K,
Motamedi, M, Rosberg, B (2000) Long term test of buffer material. Final report on the
pilot parcels. Swedish Nuclear Fuel and Waste Management Co (SKB), Stockholm,
Sweden. Technical Report 00-22. 131 p.
35

Keijzer, T. (2000) Chemical osmosis in natural clayey materials. Geologica Ultraiectina 196,
Universiteit Utrecht. Utrecht, The Netherlands. Ph. D. thesis, 166 p.
King, F, Ahonen, L, Taxen, C, Vuorinen, U & Werme, L (2001) Copper corrosion under expected
conditions in a deep geologic repository. Swedish Nuclear Fuel and Waste Management
Co (SKB), Stockholm, Sweden. Technical Report 01-23: 176 p.
Muurinen, A (2003) Chemical conditions in the AO parcel of the long-term test of buffer material
in Aspo (LOT). Posiva Oy, Olkiluoto, Finland. Working Report 2003-32. 30 p.
Muurinen, A & Lehikoinen, J (1999) Porewater chemistry in compacted bentonite. Posiva Oy,
Eurajoki, Finland. Posiva Report POSIVA 99-20: 46 p.
Parkhurst, DL & Appelo, CAJ (1999) User's guide to PHREEQC (Version 2) - A computer
program for speciation, batch-reaction, one-dimensional transport, and inverse
geochemical calculations. US. Geological Survey, Denver, Colorado. Water-Resources
Investigations Report 99-4259: 312 p.
Patel, S, Dahlstrom, L-0 & Stenberg, L (1997) Aspo Hard Rock Laboratory. Characterisation of the
rock mass in the Prototype Repository at Aspo HRL, Stage 1 (unpubl. report). Swedish
Nuclear Fuel and Waste Management Co (SKB), Stockholm, Sweden. Progress Report
HRL-97-24: 62 p.
Pedersen, K (2000) Microbial processes in radioactive waste disposal. Swedish Nuclear Fuel and
Waste Management Co (SKB), Stockholm, Sweden. Technical Report 00-04: 97 p.
Pirhonen, V (1986) Bentoniittitayteaineen ominaisuudet. Puristetun bentoniitin koostumus ja
fysikaaliset ominaisuudet - koe-era R11-85 (unpubl. report in Finnish). Teollisuuden
Voima Oy, Eurajoki, Finland. TVOIKPA - Turvallisuus }a tekniikka, Work Report 86-03: 6
p.
Puigdomenech, I (ed.), Gurban, I, Laaksoharju, M, Luukkonen, A, Lofman, J, Pitkanen, P, Rhen, I,
Ruotsalainen, P, Smellie, J, Snellman, M, Svensson, U, Tullborg, E-L, Wallin, B,
Vuorinen, U & Wikberg, P (200 1) Hydrochemical stability of groundwaters surrounding a
spent nuclear fuel repository in a 100,000 year perspective. Posiva Oy, Eurajoki, Finland.
Posiva Report POSIVA 2001-20: 46 p.
Pusch, R. (2001) The microstructure of MX-80 clay with respect to its bulk physical properties
under different environmental conditions. Swedish Nuclear Fuel and Waste Management
Co (SKB), Stockholm, Sweden. Technical Report TR-01-08. 111 p.
Pusch, R & Borgesson, L (2001) Aspo Hard Rock Laboratory. Prototype Repository.
Instrumentation of buffer and backfill in Section I. Swedish Nuclear Fuel and Waste
Management Co (SKB), Stockholm, Sweden. International Progress Report 01-60: 28 p.
Pusch, R, Muurinen, A, Lehikoinen, J, Bors, J & Eriksen, T ( 1999) Microstructural and chemical
parameters of bentonite as determinants of waste isolation efficiency. European
Commission, Luxembourg. Nuclear science and technology series EUR 18950: 121 p.
SKB (1999) Aspo Hard Rock Laboratory. Current Research Projects 1998. Swedish Nuclear Fuel
and Waste Management Co (SKB), Stockholm, Sweden. 35 p.
SKB (2003) Aspo Hard Rock Laboratory. Annual Report 2002. Swedish Nuclear Fuel and Waste
Management Co (SKB), Stockholm, Sweden. Technical Report TR-03-10. 203 p.
Smellie, J (200 1) Wyoming bentonites - Evidence from the geological record to evaluate the
suitability of bentonite as a buffer material during the long-term underground containment
of radioactive wastes. Swedish Nuclear Fuel and Waste Management Co (SKB),
Stockholm, Sweden. Technical Report TR-01-26. 24 p.
Stumm, W (1992) Chemistry of the solid-water interface. Processes at the mineral-water and
particle-water interface in natural systems. John Wiley & Sons, New York. 428 p.
36

Vuorinen, U, Lehikoinen, J, Luukkonen, A & Ervanne, H (2003) Effects of salinity and high pH in
crushed rock and bentonite - Experimental work and modelling in 2001 and 2002. Posiva
Oy, Eurajoki, Finland. Posiva Working Report 2003-22: 33 p.
Wersin, P (2003) Geochemical modelling of bentonite porewater in high-level waste repositories.
Journal ofContaminant Hydrology 61: 405-422.
Wersin, P., Curti, E. & Appelo, C.A.J. (2004) Modelling bentonite - water interactions at high
solid/liquid ratios: swelling and diffuse double layer effects. Applied Clay Science (in
print).
Wieland, E, Wanner, H, Albinsson, Y, Wersin, P & Kamland, 0 (1994) A surface chemical model
of the bentonite-water interface and its implications for modelling the near field chemistry
in repository for spent fuel. Swedish Nuclear Fuel and Waste Management Co (SKB),
Stockholm, Sweden. Technical Report 94-26: 64 p.

You might also like