Modelling the dissolution of vitrified HLW
Divyesh Trivedi, Joe Small and Kurt Smith,
National Nuclear Laboratory, 5th Floor, Chadwick House, Birchwood, Warrington WA3 6AE
Introduction
The UK generates vitrified product (or “glass”) at Sellafield as the wasteform for high level waste (HLW) from reprocessing operations. Understanding the
long terms release of radionuclides from the waste glass is essential in order to support geological disposal of the wasteform. Differences in composition,
for instance the higher magnesium content of Magnox HLW glass means that we cannot reply solely upon performance data and models from other
international HLW glass studies.
Signature Research funding from NNL has funded this study to develop models of UK HLW glass dissolution. Ultimately the models will be tested against
single pass flow-through leach testing data from samples obtained during operation of the VTR.
Approach taken by the CEA GRAAL model
The overall processes occurring during waste glass dissolution are relatively
well known (Figure 1). The CEA GRAAL model developed to simulate
SON68 glass dissolution behaviour is a well known current models, and has
been used to simulate the impact of high magnesium context
environments.
Key assumptions are that the glass is at a relatively mature stage of
interaction where a residual gel (called the passive reactive phase, PRI)
controls the diffusion of water to the pristine glass to promote alteration,
and soluble components diffuse through the PRI layer into solution. The
PRI layer itself can dissolve into solution until it is saturated. The
mathematical representation of these processes are shown below.
Figure 1. Reactions occurring during waste glass leaching.
Development of a waste glass dissolution
model by NNL
NNL is implementing a GRAAL-like glass dissolution model in the
code PHREEQC. The mathematical representation above has been
encoded into the model. The PHREEQC tool allows for the
automatic precipitation of secondary solids formed as solutes are
released during glass dissolution. Initial studies showed that a
GRAAL like approach could be implemented in PHREEQC (Figures
2a and 2b) combining glass dissolution with secondary phase
formation. Clearly seen is that soluble components such as boron
continue to be leached as the glass phases reach saturation, which
is close to experimental data concerning leaching behaviour.
Including an improved numerical solver and with realistic
parameter values, pH and temperature dependencies that fully
incorporate the mathematical approach shown on the LHS leads to
the results shown in Figures 3a and 3b. The timescales and
amounts released are similar to experimental data in the
literature, e.g. silica saturation within a few months with continued
release of soluble boron and sodium.
Dissolution of the PRI
into solution
Formation term:
Diffusion is pH and
temperature
dependent
1.00E+00
2
1.80E-03
1.00E-01
1.60E-03
0
Concentration (M)
1.40E-03
-2
1.00E-03
-4
1.00E-04
saturation index PRI
sceondary phase concentration
1.00E-02
-6
1.20E-03
kaolinite
1.00E-03
amrph.silica
8.00E-04
6.00E-04
Further work
4.00E-04
1.00E-05
2.00E-04
-8
kaolinite
1.00E-06
0
si_SiAl_PRI
1.00E-07
0
200
400
600
800
1000
The next steps are as follows:
0.00E+00
amrph.silica
50
100
150
-10
1200
200
250
300
350
400
Time (d)
time
Figure 2a. Proof of concept: solid
phases
1.00E+00
Figure 3a. With GRAAL like pH
dependencies and parameters
included: Solids
9.5
2.50E-03
1.00E-01
4.00E-07
8.5
B
1.00E-04
Na
8
pH
concnetration mol/l
1.00E-03
Al
1.00E-05
Si
7.5
pH
1.00E-06
7
3.50E-07
2.00E-03
3.00E-07
B
1.50E-03
Na
1.00E-03
6.5
0
200
400
600
800
1000
6
1200
2.00E-07
Al
1.50E-07
0.00E+00
50
100
150
200
250
300
350
0.00E+00
400
Time (d)
time
Figure 2b. Proof of concept:
aqueous
• Test the model against NNL experimental data obtained from
single pass flow through tests of vitrified glass product from the
vitrification text rig (VTR).
• Publish this data, if possible including the results from testing
against any available literature/CEA test data.
5.00E-08
0
1.00E-09
Si
1.00E-07
5.00E-04
1.00E-07
1.00E-08
2.50E-07
Concentration Al (M)
1.00E-02
Concentration B, Na, Si (M)
9
• Include better coupling between chemistry and material
accountancy during leaching so that the mass and thickness of
the gel layer can be correctly calculated and displayed.
Figure 3b. With GRAAL like
parameters: aqueous.
Not part of this programme but a suggestion, is that collaboration
with CEA would be of benefit. Development of GRAAL is an ongoing development programme; NNL and CEA could share test
data and support further model development. A key difference is
that the UK interest includes high Mg content waste glass.