a numerical model for the evolution of the carbona numerical model for the evolution of the carbon...
TRANSCRIPT
A NUMERICAL MODEL FOR THE EVOLUTION OF THE CARBON
SYSTEM GEOCEEMETRY AT THE PROPOSED NUCLEAR WASTE
REPOSITORY AT YUCCA MOUNTAIN, NEVADA, USA
William M. Murphy'**, Chrisropher J. Freitas2, Peter C. LichmerlJ
'Center for Nuclear Waste Regulatory Analyses, and
'Southwest Research Institute. 6220 Culebra Rd, San Antonio, TX 78238 USA
Abstract, Carbon system geccnemistry in the near-field zone of the proposed nuclear
waste repository at Yucca Mountain, Nevada, can affect isolation perfonnance through
controls on solution pH. radioeiement solubility, speciation, and sorption. A liquid and
gas phase transport model has Deen developed that couples local equilibrium carbon
system chemistry and txanspon KO an independent heat and two phase fluid flow model
for the partially saturated host rock This new algorithm implements a model of the
rnspon of chemical species and local equilibrium chemistry through an operator
splitting technique which a i l o ~ s for iterative solution of species transport equations
and local equilibrium mass amon relations.
Model results for the p p s e d repository site indicate that major excursions in
carbon system chemistry from ambient conditions may occur as a result of repository
heating and fluid flow. With irsreasing temperature in the near field, CO, exsolves
from the aqueous phase, whch is the major carbon reservoir, to the gas phase.
Consequently, solution pH rises with temperature, the dominant aqueous carbon
1
speciation shifts from HC03- to CO?--, and calcite (CaC03) precipitates in the vicinity
o i the repository horizon and from the edge of the repository horizon down to the
water table. Liberated CO, is transponed in the gas phase to the cooler far field. CO,
fumeroles are predicted to be an ear& chemical manifestation of the repository at the
-ground surface. Bicarbonate and carbonate are dominant aqueous complexing
compounds for many radioactive waste species in the Yucca Mountain groundwater
system. The thermal impact on carbon system chemistry and pH can alter radioelement
solubilities and solid-liquid disaibudon coefficients by orders of magnitude. Carbon 14
rcleased from the engineered bamer system will mix as a trace component in the
multiphase carbon system of the repository environment. Hence, the carbon system
model ais0 permits assessment ot‘ “C migration.
Introduction
Yucca Mountain, Nevada. is under investigation as a proposed high-level
nuclear waste (HLW) repository sire. The repository horizon is in silicic volcanic
strata approximately 375 meters below the ground surface and 225 meters above the
groundwater table (DOE, 1988). In general, bedded and zeolitized rocks at the site
have porosities of 20 to 35 percent which are 20 to 50 percent saturated with water,
and welded units are 5 to 20 percent porous and 60 to 90 percent sanuated (Nelson
and Anderson, 1992; Kwicklis et al.. 1994). Groundwater chemistry is dilute and
oxidizing (McKinley et ai., 1991).
Carbon system geochemisrp at Yucca Mountain may exext multiple controls on
radionuclide migration. Bicarbonate (HCO,-) is the dominant anion in tuffaceous
2
w W
groundwaters at Yucca Mountain (McKinley et al., 1991); calcite (CaCO,) occurs as a
fracture and void filling mineral in many rocks (Bish and Chipera, 1989; Carlos et al.,
1991); and the gas of the vadose zone is enriched in CO, relative to the atmosphere
(Thorstenson et ai., 1990). Thermal loading due to radioactive decay of HLW is
expected to enhance gas and liquid flow (e.g., Buscheck and Nitao, 1994 Pruess and
Tsang, 1994) and chemical reactions (e.g., Murphy, 1993) leading to migration of
aqueous and gaseous carbon species. and precipitation and dissolution of calcite.
Hydrolysis and complexing with carbonate species previously has been shown
to have dominant effects on aqueous speciation of radioactive waste elements in
repository groundwaters (e.g.. Clark et al., 1994). Thermal effects on carbon system
chemistry and pH can alter radioelement solubilities by orders of magnitude with
potentially large effects on the source term for radionuclide migration. Solid-liquid
distribution coefficients also depend strongly on aqueous speciation and pH (e.g..
Beneni et al., 1995) and can vary by large factors due to chemistry changes.
Variations in distribution coefficients affect retardation of radionuclide migration.
Although silicate system chemistry is not included in the carbon system model
developed in this study, depletion of aqueous Ca2+ due to calcite precipitation may
destabilize calcic clinoptilolite, which is a major natural alteration product of the
silicic tuffs at Yucca Mountain with significant sorption capacity.
Radioactive 14C is a constituent of nuclear waste proposed for disposal at
Yucca Mountain (Van Konynenburg, 1994). 14C released from breached waste
containers as I4CO2 would mix as a mice component in the multiphase carbon system
3
W V
of the repository environment W i i n isotopic fractionation having a negligible effect
with regard to radionuclide rmgmon. Given a source term for 14C release (e.g..
Codell, 1993), transport of “ C Arcugh the geologic repository can be modeled for the
purposes of performance assessment to be proportional to tmsport of carbon (e.g.,
Codell and Murphy, 1992).
This article reports deveiopment of a coupled thenno-hydro-geochemical
reactive transport model to simuiar,e rhe evolution of the carbon system in the near-
field zone of the proposed reposirory and to aid assessments of radionuclide migration.
The model segregates transport ana equdibrium chemistry by operator splitting.
Independent transport of gaseous CO- and aqueous species is modeled, then mass
action relations for local equiibrium are solved. This paper presents the theoretical
framework of the model along ~1r.h results for a repository scale simulation.
Previous studies
The evolution of the carbon system geochemistry in the proposed nuclear waste
repository at Yucca has been sruciid previously. primarily in relation to 14C release.
Knapp (1990) solved an approxmre equation for transport of 14C as a brealung
kinematic wave. Retardation of “ C transport was assumed to be controlled by the
isotopic equilibrium
H 12~0,- -:‘co2(g) = H 14~0; +12C02(g) (1)
Numerous simplifications in this marhematical model, such as neglecting the effects of
diffusion and temperature variations. lead to approximate results.
Light et al. (1990) calculated gaseous and aqueous carbonate equilibrium
4
relations to yield a dimensioniss dsmbution coefficient for carbon between gas and
aqueous phases. The dsmbuccn coefficient was shown to be a strong function of
temperature (25-100 "C) ma ?H i7-9), with values ranging from less than one to
several hundred for likely r e m ~ r a n u e and pH conditions in the Yucca Mountain
repository. A reference value ci three was used in their calculations of 14C fluxes and
concentrations. Their calculations indicated that diffusion of carbon would tend to
homogenize 14C concentrations m a rock matrix containing a static liquid at a value
near equilibrium with a gas flosmg sufficiently slowly in nearby fractures. They
concluded, therefore, that an eaurvalent porous medium approximation to the fractured,
unsaturated system can be accurately implemented for the appropriate conditions. Light
et al. (1990) performed calcuhons of 14C transport using the nominal values of
hydrologic properties and the quivalent porous medium approximation. Their results
indicate that release of 14C at rt;c land surface depends strongly on the amount of I4C
released from the waste packagm. and only slightly on the timing of waste package
release.
Ross (1988) develop a conceptual model relating I4C transport to the
carbonate system at Yucca Mounmn which emphasizes gas-water-rock equilibria and
gas phase transport of CO,. - X relatively realistic water chemistry model was
recognized in this study as a q m m e n t for modeling 14C retardation.
Codell and Murphy ( 1992) developed a one-dimensional model with gas phase
transport of CO, and local eqdirium chemistry as a function of time and space. The
model for local equilibrium presented by Codell and Murphy (1992) is equivalent to
5
the chemical model presented in tlus paper. Gas flow was approximated for that model
by uruiorm, time dependent vertical flow generally consistent with results of two
Olmensional thennohydrologic modeling. Major excursions from ambient conditions
were predicted to occur in carbon system chemistry due to repository heating and fluid
flow. With increasing temperature, CO, exsolved from the aqueous phase, which is the
major carbon reservoir, to the gas phase. Solution pH was shown to rise with
temperature, and the dominant aqueous carbon speciation shifted €?om HCO,' to CO,,.
Lncreasing pH, increasing temperature, and evaporation all promoted calcite
precipitation. However, the maximum mass of calcite to precipitate was small because
it is limited by the supply of aqueous Ca2'. The aqueous concentration of Ca2+
dropped to low values in areas of elevated tempera- and calcite precipitation.
Gas phase transport in the model of Codell and Murphy (1992) led to carbon
redsmbution on the scale of the mountain. Carbon was shown to be transported
hundreds of meters in hundreds of years. Although transport was limited to gas flow,
most carbon remained in the aqueous phase, so its migration rate was retarded by the
rano of its aqueous to gas phase distribution. Carbon precipitated in calcite did not
migate, but its mass was small compared to that in fluid phases. CO, liberated from
the aqueous phase in the hotter near field was transported in the gas phase to the
cooler far field where it condensed As the near field temperature diminished the water
chemistry slowly reverted toward initial conditions and calcite dissolved.
The present model is an extension of the model developed by Codell and
llurphy (1992) incorporating more realistic gas and liquid phase flow models,
6
W W
diffusive as well as advective transport, use of a radially symmetric geometxy, and an
implementation of higher-order convective modeling techniques developed by Freitas
(1985).
Mathematical Model for Carbon System Chemistry and Transport
The Carbon System T R U I S ~ ~ K Code (CST) is a standalone numerical model
that simulates the transport of aqueous and gas phase species and performs chemical
equilibrium calculations at each grid point. CST uses as input the state variables of
temperature, saturation, pressure, and aqueous and gas phase velocity fields, all as
functions of time and space. Cumntly, these values are obtained through a transient
simulation performed using the CTOUGH code (Lichtner and Walton, 1994), a version
of VTOUGH (Nitao, 1989 and Pruess, 1987). In the course of a CTOUGH simulation.
data at specified instants of time record the spatial distribution of all variables. CST
then reads these data and performs linear interpolations between discrete times to
obtain intermediate values of the state variables at any arbitrary time as determined by
the CST time step scheme. In this way, CST is not restricted to the time scales
inherent in the CTOUGH data files but rather can perform a sub-cycle calculation at a
finer time scale than that represented by the data files. Obviously, the ability of CST
to resolve critical temporal and spatnl events is dictated by the level of temporal
refinement of the instantaneous CTOUGH data fiies and their spatial resolution.
Finally, CST calculates a time step size which maintains the maximum Courant
number in the flow domain to be less than 0.75. This is done primarily to achieve
optimal temporal accuracy in the transport simulations.
7
An operator splitting algorithm is used to coupled transport of aqueous and
gaseous species and chemical reactions. In this approach species YZ transported over a
sinele time step by advection and diffusion according to the nonrcxtive transport
equations. At the end of each time step, the reacting species are equilibrated at each
node. The operator splitting approach restricts the maximum time step that is possible
by the Courant condition which allows the fluid to move at most one node point in a
given time step.
The thermohydrologic model used here was taken from Lichmer and Walton
(19941. The repository was treated as a cylindrically symmetrical system with a 2 km
radius and 600 m height representing the zone from the water table to the ground
surface. Uniform thermal and hydraulic properties characteristic of Yucca Mountain
were assigned. An equivalent continuum representation was used for the tuff host rock
with manix and fracture permeabilities of 1.9~10-l~ m2 and 10'" rn'. respectively.
Porosities of 0.1 for the matrix and 0.0018 for the fracture network were used. Values
for other material parameters are given in Lichmer and Walton I 1994). The initial
thermal loading at the repository horizon was 28,000 kW/km2 ( 1 14 kW/acre) over an
area of 3 km2. Thermohydroiogic model results indicate that the near field zone is
heated to a maximum temperature of approximately 180 "C within tens of years after
repository closure. Heat flow, which is dominated by conduction. raises far field
temperatures as the near field cools over a long time period. Transient heat pipes, with
counter flow of gas and liquid, develop above and below the repository during the first
1 ,OOO yr. Gas flow was downward across the repository horizon for the first 1.000 yr.
8
w
Following this period gas flow was towards the repository horizon in the immediate
neighborhood both above and below the repository. These results could be different
with different assumptions for material properties.
Temperature, gas and liquid velocities, water saturation, and pressure from
CTOUGH wen used as input to the finite volume based diffusive and advective,
reactive transport model. Temperam dependent three-phase reaction chemistry was
coupled to species transport by optrator splitting. Effects of chemical reactions on the
thennohydrologic system, other than evaporation and condensation of water, were
neglected. Complete desiccation oi the system was permitted in the chemistry model,
although high ionic stnngth chemistry was not modeled explicitly. Instead, remaining
solutes were precipitated as unspecrfied salts when saturation dropped below 1 percent.
Five key reactions affecting carbon chemistry in the Yucca Mountain unsaturated
- mundwater system wen considend These reactions account for aqueous carbon
speciation, water hydrolysis (and pH), distribution of CO, between aqueous and gas
phases. and dissolution and precipitation of calcite. The Na+ ion was also modeled as
the dominant aqueous cation. Mass action, mass conservation, and aqueous charge
conservation were used as constramu to compute conditions of local equilibrium as a
function of time and space. Reacuons in this model system are npid relative to most
other changes in geochemical systems, e.g., silicate system reactions. Natural
groundwater and ground gas systems are comonly in equilibrium with respect to
aqueous carbonate speciation, CO? - pressure, and calcite. Therefore, local equilibrium
for h s model system is a reasonable assumption. Initial chemical conditions were
9
selected to closely represent the amDient system at Yucca Mountain. The initial water
was dilute with intermeciute pH and slightly undersaturated with respect to calcite.
Carbon System Chemisrrp
A relatively simpie geochemical model incorporating reactions of primary
si@icance to the carbon system and "C transport at Yucca Mountain is based on
local chemical equilibrium and mass and charge conservation in a representative
elementary volume. Chermcal reactions and corresponding mass action relations
included in the model are:
H' - OH-= H - 0
H20(l) - COJaq) = H + HCO,
CO&) + HzOfl) = HCO; - H'
caco3(cc) + H = Cai* - HCO;
= %' acoj2-
10
where (1). (aq), (g), and (cc) refer to liquid, aqueous, gas, and calcite phases,
respectively, denotes the thermodynamic activity of the subscripted species, fq
stands for the fugacity of gaseous CO,, and PCo2 stands for the reference fugacity of
CO,, which equals 1 bar. In addition to the aqueous species in reactions (2) to (6).
Na' is included in the model to represent generally other basic aqueous cations.
Sodium is the dominant cation in groundwaters from the tuffaceous aquifer at Yucca
Mountain, and calcium is second in molal concentration. Consequently, the simplified
water chemistry closely resembles the natural waters, and the 5 reactions above
represent the carbon system at Yucca Mountain nearly completely.
Local charge balance in the model aqueous phase is represented by
mH+ + mNat + 2 mca2+ - m HCO, + mco:- + mOH- (7)
where m, stands for the molality of the subscripted species. Local mass conservation
for carbon can be expressed as
"COicB, + n COZ(9, +nco;- + ncaco,(cc) = "c (8)
where I+ stands for the number of moles of the subscripted species in the
representative volume, and nc denotes the total number of moles of carbon in the
volume. Similarly,
"ca (9) ca=03(,) nca2- + n
gives the mass conservation condition for calcium. In the case that the aqueous
solution is undersaturated with respect to calcite and no calcite exists locally, ncAco~(cc)
11
_--
is Zero, and Eq. (9) is mvial. The. mass of sodium is conserved in the aqueous phase.
The mass of 5 0 is conserved ma H,O maintains equilibrium between the gas and
liquid phases. Its distribution is caiculated in the thennohydrologic model. No
oxidation-reduction reactions are rtlevant to this model system, which is completely
oxidized.
Thermodynamic activities in the mass action relations (2) - (6) are related to
moles and molalities in Eqs. (7) - (9 ) by
where yi represents the activity cceiiicient of aqueous species i, and WHzo stands for
the mass of H20(1) in the representative volume. Activity coefficients are calculated
according to
I 2 7 -A zi I
log i", = + B O 1 I
where A, ai. B, and Bo are empirical parameters, 3 stands for the charge of species i,
and I denotes the ionic strength (Heleeson, 1969). The activity coefficient for neutral
aqueous CO,(aq) is taken to be one. Calcite and H,0(1) are assumed not to deviate
sigmfkantly from their standard states, so activity coefficients for both are also set at
unity.
Assuming Dalton's law is vahd for the the low pressure gas phase, the CO,
fugacity is related to the number o i moles of gaseous CO, by the ideal gas law
12
W
where pc4 stands for the p d pressure of CO,, R denotes the ideal gas constant, T
represents the absolute temperam, and Vg stands for the volume of the gas phase in
the representative volume. The gas phase volume Vg is related geometrically to the
representative volume V by
v, = v $ (I-S) (13)
where 4 stands for the total (gzs and liquid) porosity, and the saturation, S, denotes
the fraction of porosity occupied by the liquid phase. The mass of water can be
expressed in similar terms by
where pHzo stands for the mass Of q0 per volume of liquid.
Substituting Eqs. (2)-(5) and (10)-(14) in Eq. (8) and solving for - yields COY
Substituting the same equations in (7) and solving for aH+ yields
13
U
Component Aqueous Calcium
Aqueous Sodium Aqueous Carbon
PH Fugacity - fco2
W
Concentration (rn~les/rn~,~,)
0.40
1.17 2.00
8 0.0012 bar
These substitutions reduce the chemical equilibrium problem to a set of two equations
to be solved simultaneously in the absence of calcite. Once solved the saturation state
of the system with respect to calcite is checked, and if the solution is supersaturated,
then Eqs. (6) and (10) are substituted in Eq. (9) and solved
In this case Eq. (16) contains a modified term for the mass
is obtained by sirmlar substitutions yielding
r
for ncaco, *
t 17)
of aqueous calcium, which
If calcite is present Eqs. (15), (17), and (18) must be solved simultaneously. These
Table 1. Initial Geochemical Conditions
of two or three equations are solved for each point in the domain, at each time by a
Xewton-Raphson technique using masses (e.g.. nNa. nca) provided by the transport
model. Activity coefficients are used from the initial condition or previous time step to
simphfy computations, introducing negligible error. Thermodynamic data are derived
from the EQ3/6 data base R16.com (Wolexy and Daveler, 1992). Equilibrium constants
as a function of temperature are consistent with data reported by Johnson et al. (1992).
Initial conditions for the chemistry model are given in Table 1.
Carbon System Transport
The form of the non-reactive transport equation for both gas and aqueous phase
species, in cylindrical coordinates, is
where the independent variables are time t (s), radial coordinate r (m) and vertical
coordinate z (m). The dependent variables are c, the transported species concentration
(rnoles/m3), D the effective diffusion coefficient (m2/s), the Darcy velocity field,
defined by u and v (ds) and 7 is defined as S for liquid phase saturation and (13)
. for gas phase saturation. Equation (19) results from an operator splitting approach in
which the chemical constituents are first transported non-reactively, hence the zero on
the right-hand-side of Eq. (19). This transport step is then followed by a chemical
equilibrium calculation at each node.
The diffusion coefficients are assumed to be isotropic and homogeneous, and
are therefore, scalar valued functions. Mechanical dispersion is neglected in this
15
W U
analysis. For aqueous species uanspoxt, the diffusion coefficient is taken as a function
of tempemure only and is calculated as
E D.?" = A exp (--) RTi.j '.J
where A = 1.02e-06 m h , E = 1.66e44 Jlmole, R - 8.31441 J/mole-K, and Ti j is the
local absolute temperam. This equation is an empirical regression based on data for
sodium ions (Oelkers and Heigeson, 1988). For gas phase species transport the
diffusion coefficient is calculated here as
where the diffusion constant Do = 1.44e-05 m2/s, Po is standard atmospheric pressure
in Pascals (-1.013e+05 Pa). To is the temperature at standard conditions (-273.15 K),
and Pi,j and Ti,j are the local values of pressure and temperam, respectively. This
equation is taken from Bird Stewart, and Lightfoot (1960) and was developed for
calculating the diffusivity of CO, in fke air.
The uanspon equation for a species concentration c is solved by integrating
Eq. (19) over a nodal control volume. This control volume surrounds a single
computational node which is centered within the volume (Figure I). The result of this
integration is a semi-discrete form of Eq. (19) and is written as
where the subscripts e, w, R, s refer to different faces of the control volume. The
volume (rArAz) and facial areas (rAz or r&) of the control volume are scaled by the
porosity and saturation, such that only that area or volume open to flow are used for
16
V W
n rArAz (c.? - Ciqj) - + (rAz c , - (rAz u ) ~ t, I*J At
either liquid and gas phase transport.
The tlux coefficients represent the volume flu across each 5ce of the control
voiume. They are defrned as (refer to Figure 1 for definition of gecmemc terms and
nodal locations o i variables)
where again. the flux cross-sectional areas have been scaled by the prosity and
saturation ievels.
The spaual gradient terms in Eq. (22) are defined by cenreza differences of the
form
which resuits in the following definitions for the diffusion coeffcicm
17
Substim&I L ----,Tc& 'ps into Eq. (22) results in the discrete form of
Eq. (19), which is
Once c,, c,, c - . .. r . . =x XE k r l and substituted into Eq. (27). terms associated
with each of the n x i ;--re f : LY collected, resulting in a system of algebraic
equations of the f r
ap.: :-- = 2 L - - >=: ..j . c. I- 1.j + ani,j Ci . j+~ + aSij Ci,j-1 + S%.j (28)
where we defme
18
rArAz + FLE - FLW + FLN - FLS + -
At + aRj - aeij + awij + aq,j
n L ~ A Z s q j = c.. - At
The definition of the coefficients aeij, awiVj. aq j , and asij are determined by the
convective interpolation scheme.
Two methods are used to represent the facial values of the species
concentration associated with the convective fluxes in Eq. (27). The first uses a
combination of centered interpolation (second-order) with first-order upwinding, and
the second uses first-order upwinding only. This first method is the HYBRID scheme
of Spalding (1972) and is also hown as the “high-late& flux mMication” to the
second-order centered interpolation scheme. A third and more promising method was
developed here based on a third-order convective scheme. This method is a derivative
of the QUICK Scheme (Leonard, 1979; Freitas et al., 1985). However, typical of
higher-order convective methods, spurious overshoots and undershoots at steep
gradients have limited the applicability of this method to solving reactive transport
problems. The authors are presently working on a monotomicity-preserving form of
this scheme which should allow for solution of reactive transport problems.
The HYBRID method segregates the interpolation process into three regions,
defmed by the magnitude of the cell Peclet number. The cell Peclet number is defined
as the ratio of the flux coefficient to the diffusion coefficient; i.e., on the e face of the
control volume the Peclet number is FLE/DIE. The HYBRID scheme uses fxst-order
upwinding if the magnitude of the Peclet number is greater then 2; and for Peclet
19
v W
numbers between 2 and -2, it uses second-order central differencing. Mathematically,
c, is calculated as
A compact method to represent this scheme, which eliminates conditional
checking on the direction of the velocity across a face, is to define a set of flux
parameters for each face of the ccnuol volume. For example, the e face flux
parameters are
FPE = 0.5 (FLE + IFLEI) (31) FDE = 0.5 (FLE - IFLEI)
For a positive flux, FLEA), FPE is equal to FLE, and FDE is zero. For a negative
flux, FLE<O, FDE is equal to FLE. and FPE is zero. These flux parameters turn on
and off the various terms for each component of the interpolation function which are
dependent on velocity direction. With these definitions, the HYBFUD scheme may be
compactly represented by
dmy.
&i+l.j - max [-FDE, DIE - FLE ( 2 1 1 a%j
dmq . &i,j
awij = max I p W , DIW + FLW (>)I
a q j = max [-FDN, DIN - FLN (>)I dmv. .
dzi,j+l dmv- .
. - ma^ [Fps, DIS + FLS (L)] dzij
1.J
The first tern within the brackets of these expressions represents the first-order
20
W W
upwinding contribution. um z: =End term is the second-order central interpolation
contribution. By simply 50~,3-~- -2 expressions of Eq. 32 one can use either central.
interpolation, fmt-order u p - z g . 3r HYBRID to model the convective fluxes. The
equivalent expressions fcr 5z-c~~ upwinding only are
z:. 1 DIE - FDE L X : : - DIW + FPW 3: -DIN - FDN 5: - DIS + FPS
-1 -,
-J
-1
and for second-order cenrn i ==ia.rion they are
(33)
In the expressions for h-xx :?-Inding, a physical diffusion contribution is
present, while this term 1s e;--y-w m the HYBRID formulation. This is done in the
HYBRID scheme in an sump 12 compensate for the spurious numerical diffusion
introduced by fust-order up-. In the simulations that follow, calculations are
made using the HYBRID Sciz Eq. 32) and fmt-order upwinding (Eq. 33). The
algebraic equation systems xii x i - e i using an SLOR matrix solver.
Model results
To illustrate the capc!;-rrt c i the CST model, results are presented for times
21
W
ICQ and 600 yr following inuoducuon of the heat source. This is a period of rapid and
m:or changes in the near field zone. Temperature, pressure, and saturation distribution
arc 2ustrated in Figs. 2-4. These results are initially derived from CTOUGH.
Terzperature rises to over 150°C at the repository horizon in 100 yr, but is still close
to ;%tial conditions at the model boundanes. At 600 yr temperature gradients have
mc&rated as the near field cools. Saturations go to zem around the repository horizon,
k a v q a nearly desiccated interval approximately 50 m thick at 600 yr. Pressure
incp3es to 1.5 times ambient at 100 yr at the repository horizon, but decreases to 1.2
tirzlts ambient at 600 yr.
Similar excmions from ambient conditions occur in carbon system chemistry
due :o repository heating and fluid flow as observed in the model of Codell and
M-ny (1992). With increasing temperature. CO, exsolves from the aqueous phase to
the gas phase. Solution pH rises with temperature, and aqueous carbon speciation
sh15~ from HC03- to CO,*-. Increasing pH, increasing temperature, and evaporation
all ?mote calcite precipitation. The maximum mass of calcite distribution to
pE:pitate is limited by the suppiy of aqueous Ca2'. The aqueous concentration of
Ca-' drops to iow values in areas of elevated temperature.
Aqueous and gas phase carbon and calcite concentration are illustrated in Figs.
5- - . These calculations were pe~ormed using the HYBRID scheme. Evaporation
(voiatllization) and gas phase nanspon cause rapid carbon redistribution on the scale
of ke mountain resulting in steep _madents in carbon. Carbon is transported hundreds
o i ;~efers in hundreds of yean. Although transport is dominated by rapid gas flow,
22
V w
most carbon remains in the aqueous phase in the absence of dessication, so its
migration rate is retarded by the ratio of its aqueous to gas phase distribution. The
mass of carbon precipitated in calcite is small compared to that in fluid phases except
in nearly desiccated areas. CO, liberated horn the aqueous phase in the near field is
transported in the gas phase to the cooler far field both above and below the repository
horizon where it panially condenses in the aqueous phase. Evaporation and gas phase
flow nearly decarbonate the near field.
Figure 5 illustrates that the total aqueous carbon is not strongly affected in the
region above the zone of drying, although the dismbution of carbon species shifts due
to pH changes (see below). Below the repository a greater increase in aqueous carbon
is predicted. In part this effect is due to the no-flow boundary condition at the water
table, which prohibits escape of carbon purged from the near field and the effects of
gravity on water dismbution.
Figure 6 illustrates that a CO, enriched plume of gas moves upward from the
repository horizon. Maximum concentrations in the gas phase are over 10 times greater
than ambient values at 100 yr in an area above the advancing temperature and
dessication fronts. (Ambient CO, pressures in the subsurface are themselves
approximately 4 times atmospheric values.) Similarly at 600 yr the CO, gas
concentrations are over 20 times ambient values near the ground surface over the
center of the repository in an area minimally affected by temperature or saturation
variations. These results indicate that an early surface manifestation of the repository
will be gas springs of CO, enriched gas eminating at the surface, Le., repository
’
23
W W
induced CO, fumaroles. Hetemgeneous gas flow in fractures is likely to accelerate and
m a w local manifestations of CO, fumeroles. This early chemical manifestation of
the repository at the ground surface may have biological consequences. The initial
CO, plume will also tend to acidify groundwater solutions promoting reactions in the
silicate system, which are not represented in the model.
The distribution of calcite (Figure 7) is a complicated function of retrograde
solubility, pH variation with CO, pressure changes, and calcium transport by liquid
flow and concentration by evaporation. Solutions are initially undersaturated with
respect to calcite, but by 100 yr quantities of calcite approaching the initial aqueous
calcium content precipitates within tens of meters of the repository horizon. By 600 yr
notable calcite precipitation occurs in the immediate area of the repository horizon and
also in a zone extending from the repository edge to the groundwater table. This edge
effect is apparently a consequence of liquid flow shed off the edge of the repository
carrying calcium into regions of higher temperature and higher pH. Concentration of
calcite precipitation due to this edge effect could have consequences for fluid flow.
However, such effects were not included in the model.
Finally, Figure 8 displays a comparison of the pH distribution at 600 yr as
calculated by the HYBRID scheme and the fmt-order upwind scheme. Clearly seen in
these figures are the spurious dissipative effects of fmt-order upwinding on pH
gradients. This behavior is exhibited in all the transported quantities. pH gradients
above and below the repository an smeared out by fmt-order upwinding, while the
HYBRID scheme maintains them to a greater extent. Although the HYBRID scheme is
24
W II
not fully second-order. it does h i t , to a degree, the artificial smoothing of gradients
introduced by first-order upwinding. Certainly a third- or higher-order convective
scheme would resolve these p d i e n t even more accurately.
Conclusions
Carbon system chemisn?; an affect performance of the proposed nuclear waste
repository at Yucca Mountain, Szvada, through controls on solution pH, radioelement
solubility and speciation. retardation of xadionuclide migration by sorption, stability of
sorptive phases, and “C mi-&on. A gas and liquid phase reactive transport code
(CST) has been developed to model the carbon system chemistry and pH. It has the
capability to provide constraints for performance modeling of radionuclide release and
retardation of radionuclide migation. The CST code incorporates HYBRID and fmt-
order upwind schemes for convective transport modeling. The results presented here
demonstrate the superior ability o i the HYBRID scheme to resolve sharp gradients.
Preliminary results demonstrate the capabilities of the numerical model and
give indications of extensive ma large scale variations in carbon system chemistry due
to heat and fluid flow in the reFsitory. The results presented here could change if, for
example, different mated properties were used to represent the tuff host rock.
Furthermore, the applicability o i the equivalent continuum model needs to be more
fully investigated for species uanspon. A more detailed and longer term repository
model is in preparation for pexformance assessment applications. And, the author are
working on the formulation of a thud-order convection scheme that eliminates
spurious overshoots and underSnoots that are typical of these higher-order methods.
’
25
Acknowledgments. Richard B. Codell assisted with model development. The report
documents work performed at rhe Center for Nuclear Waste Regulatory Analyses
(CNWRA) for the Nuclear Regulatory Commission (NRC) under contract NRC-02-93-
005. The paper is an independent product of the CNWRA and does not necessarily
reflect the views or regulatory position of the NRC.
References
Bertetti, F.P., Pabalan, R.T., and Turner, D.R. (1995) Neptunium (V) sorption behavior
on clinoptilolite, quam and montmorillonite. Proceedings Scientific Basis for
Nuclear Waste Management XIX (Murphy. W.M. and Knecht, D.A., eds.).
Materials Research Society, Pittsburgh, PA. In preparation.
Bird, R.B., Stewart, WE., and Lightfoot, EN. (1960) Transport Phenomena. New
York, John Wiley and Sons, 780 p.
Bish, D.L., and Chipera, S.J. (1989) Revised mineralogic summary of Yucca
Mountain, Nevada. Los Alamos National Laboratory, LA-11497-MS. Los
Alamos, NM.
Buscheck, T.A., and Nitao, JJ. (1994) The impact of buoyant, gas-phase flow and
heterogeneq on thenno-hydrolocical behavior at Yucca Mountain. High Level
26
W
Radioactive Waste Proceedings, p. 2450-2474. American Nuclear Society, La
Grange Park, IL.
Carlos, B.A., Chipera, S.J., and Bish, DL. (1991) Fracture-lining minerals in the lower
Topopah Spring Tuff at Yucca Mountain. High Level Radioactive Waste
Proceedings, p. 486-493. American Nuclear Society, La Grange Park, IL.
Clark, DL., Ekberg, S.A., Moms, DE., Palmer, P.D.. and Tait, C.D. (1994)
Actinide(1V) and actinide(VI) carbonate speciation studies by PAS and NMR
spectroscopies. Los Alamos National Laboratory, LA-1 2820-MS. Los Alamos,
NM.
Codell, R.B. (1993) Model for nlease of gaseous 14C from spent fuel. High Level
Radioactive Waste Proceedings, p. 22-29. American Nuclear Society, La
Grange Park, IL.
Codell, R.B., and Murphy, W.M. (1992) Geochemical model for C-14 transport in
unsaturated rock. High Level Radioactive Waste Proceedings, p. 1959-1965.
American Nuclear Society, La Grange Park, IL.
27
V
DOE (1988) Site Characterization Plan Yucca Mountain Site, Nevada Research and
Development Area, Nevada. Depanment of Energy, Washington, D.C.
DOEYRW-0 199.
Freitas, CJ., Street, RL., Findikakis. AN., and Koseff, J.R. (1985) Numerical
Simulation of Three-Dimensional Flow in a Cavity. Int. J. Nun. Methods in
Fluids, voL 5 , pp. 561-575.
Helgeson, H.C. (1969) Thermodynamics of hydrothermal systems at elevated
temperatures and pressures. American Joumai of Science, V. 267, p.
729-804.
Johnson, J.W., Oellcers, E.H., and Helgeson, H.C. (1992) SUPCRT92: A software
package for calculating the standard partial molal thermodynamic properties of
minerals, gases, aqueous species. and reactions from 1 to 5,000 bars and 0" to
l,OoO°C. Computers and Geosciences, v. 18, p. 899-947.
'
Knapp, R.B. (1990) An approximate calculation of advective gas-phase
transport of I4C at Yucca Mountain, Nevada. Journal of Contaminant
Hydrogeology, v. 5, p. 133-154.
28
W v
Kwicklis, E.M., Flint, A.L.. 2: SAY, R.W. (1994) Simulation of flow in the
unsaturated zone b e n c 3:gany Wash, Yucca Mountain. High Level
Radioactive Waste h z a g s , p. 2341-2351. American Nuclear Society, La
Grange Park, E.
Leonard, B.P. (1979) A Stablc ~ .Accurate Convective Modeling Procedure Based on
Quadratic Upstream k-z=iauon. Computer Methods in Applied Mechanics
and Engineering, vol. :r. 2.59-98.
Lichmer, PL., and Walton, 1.1. :34) Near-field liquid-vapor uansport in a partially
satulrated high-level n c c k - waste repository. Center for Nuclear Waste
Regulatory Analyses. DTRA 94-022, San Antonio, TX.
Light, W.B., Pigford, T.H., k ~ . P.L., and Lee, W.W.-L. (1990) Analytical
models for C-14 t rans~z = a partially saturated, fractured, porous
media. Proceedings S.xk Waste Isolation in the Unsaturated Zone
Focus '89. American Xxzar Society, LaGrange Park, IL., p. 271-277.
McKinley, P.W.. Long, MP.. zz Benson, L.V. (1991) Chemical analyses of water
from selected wells am s ~ g in the Yucca Mountain area, Nevada and
Southeastern Cal i fom l- 5. Geological Survey Open-File Report 90-355,
Denver, CO.
29
W W
Murphy, W.M. (1993) Geochemical models for gas-water-rock interactions in a
proposed nuclear waste repository at Yucca Mountain, Nevada. Proceedings
Site Characterization and Model Validation Focus '93, p. 115-121. American
Yuciear Society, La Grange Park, IL.
Nelson, PH., and Anderson, L.A. (1992) Physical properties of ash flow tuff from
Yucca Mountain, Nevada. Joumal of Geophysical Research, v. 97, p. 6823-
6841.
Nitao, JJ. (1989) VTOUGH-An Enhanced Version of the TOUGH Code for the
Thermai and Hydrologic Simulation of Large-Scale ProbIems in Nuclear Waste
Isolation, U r n - 2 1954, Livermore, CA, Lawrence Livermore National
Laboratory.
OelkeIs, E.H., and Helgeson, H.C. (1988) Calculation of the thermodynamic and
nanspon properties of aqueous species at high pressures and temperatures:
Aqueous tracer diffusion coefficients of ion to 1,OOO"C and 5kb. Gemhimica et
Cosmochimica Aeta, v. 52, p. 63-85.
Pruess, K. (1987) TOUGH User's Guide. NUREG/CR-4645, Washington, DC, Nuclear
Regulatory Commission.
30
'v V
Pruess, K., and Y. Tsang (1994) Thennal Modeling for a Potential High-Level Nuclear
Waste Repository at Yucca Mountain, Nevada, LBL-35381, UC-600, Berkeley,
CA, Lawrence Berkeley Laboratory.
Ross, B. (1988) Gas phase transport of carbon-14 released from nuclear waste
into the unsaturated zone. Materials Research Society Symposium
Proceedings, v. 112, p. 273-284.
Spalding, D.B. (1972) A Novel Finite-Difference Formulation for Differential
Expressions Involving Both First and Second Derivatives. Int. J. Num. Methods
Eng., vol. 4, pp. 551.
Thorstenson, D.C. Weeks, El.. Haas, H., and Woodward, J.C. (1990) Physical and
chemical characteristics of topographically affected airflow in an open borehole
at Yucca Mountain, Nevada. Proceedings Nuclear Waste Isolation in the
Unsaturated Zone Focus '89, p. 256-270. American Nuclear Society, La Grange
Park, IL.
Van Konynenburg, R.A. (1994) Behavior of carbon-14 in waste packages for spent
fuel in a tuffrepository. Waste Management, v. 14, p. 363-383.
31
V W
Wolery, TJ., and Daveler S.A. (1992) EQ3/6, A software package for
geochemical modeling of aqueous systems. UCRL-MA-110662 PT I-IV.
Lawrence Livermore National Laboratory, Livermore, CA.
32
'u,
Fi-eue Captions.
1. Definition of geomemc terms for a cell centered control volume.
2 . Temperature ("C) distribution at 100 years (A) and 600 years (B). Vertical scale is the same as horizontal scale.
3. Pressure (Pa) distribution at 100 years (A) and 600 years (B). Vertical scale is the same as horizontal scale.
4. Saturation distribution at 100 years (A) and 600 years (B). Vertical scale is the same as horizontal scale.
5 . Distribution of total aqueous carbon (moles/&) at 100 years (A) and 600 years (B). Vertical scale is the same as horizontal scale.
6. Distribution of gas phase carbon (moles/&) at 100 years (A) and 600 years (B). Vertical scale is the same as horizontal scale.
7. Distribution of calcite (rnoles/d) at 100 years (A) and 600 years (B). Vertical scale is the same as horizontal scale.
8. Comparison of pH distribution at 600 years, calculated by HYBRID (A) and first-order upwind (B). Vertical scale is the same as horizontal scale.
33
I I 1 I I I I I I I I I I 1 I I I I I I I I I I I I I I I I I I I L
-
0 0
v)
3 cu 0
I I I I I I I I I :
I
I I I I I I I I I I I 1
1
I I I I I -
V
I I I I I I I I I I I I I I I I I I I I I I I I 4
a m
I I I I
I I
I I
I I
I I
I I
I I
I I
I I I I I I I
I I I I
4
';
I
: 0
1
I ;
- c -
0
? I
I I
I I 1 I
I I
I I
I I I
0 250 500 750 lo00 1 250 1500 1750 2000 Length Scale (m)
Figure 5a
0
0
v)
0
v)
cv
0
J ' " ' ' " ' ' " ' " " ' " " ~ " ' " ' " ' ' " " l 0 250 500 750 1000 1250 1500 1750 2000
Length Scale (m) Figure 6a
-\
n
9
0 0
0
cu
0
m
b
7
0
F
0
m
cu F
8 0 r
0
v)
b
0 0
v)
0
m
cu
0
0
/w 0
R 'c 0 0
m
F
0
ua N
'c
0
8 'c 0
R 0
5: 5: N
0
R
0
ln
b
F
0 0
ln
F
0
u)
cv F
0
0
0
.-
0
v)
b
0
0
ln
0
v)
cv