Bentonite Pore Waters (BPW) for SGT-E3: Model Development, Testing and Final Calculations

  pdf NAB 22-43 Bentonite Pore Waters (BPW) for SGT-E3: Model Development, Testing and Final Calculations(2.92 MB)

In the framework of the Sectoral Plan for Deep Geological Repositories, stage 3 (“Sachplan Geologische Tiefenlager”, Phase 3, shortly SGT-3) PSI was in charge to derive composition and chemical characteristics of pore waters in equilibrium with compacted bentonite under the conditions expected in the planned HLW/SF repository. These pore waters are a pre-requisite to derive solubility limits and sorption coefficients (KD) of dose-relevant radionuclides for safety assessment calculations. These NAB is a summary of TM-44-21-02 Rev. 1 authored by E. Curti. In the TM, additional information for the model setup are presented.

Bentonite pore waters were defined for the three preceding Swiss safety assessments Kristallin-I, Entsorgungsnachweis and SGT-2 (Curti 1993, Curti & Wersin 2002, Bradbury et al. 2014) reflecting the specificity of host-rock type (crystalline basement rocks or Opalinus Clay), available underground water data and the evolution of concepts used to model the interaction between aqueous solutions and compacted clay. This report presents in detail the development of an updated thermodynamic model to derive Bentonite Pore Waters (BPWs) for SGT-3. Moreover, a comparison is made with other approaches and previous BPW models. Finally, preliminary pore water calculations based on provisional water input data derived from the at that time ongoing drilling program were carried out using GEM-Selektor v. 3.7 with the PSI-Nagra database v. 12/07 and discussed.

To date, there is no consensus on a scientifically sound approach on how to model aqueous solutions in equilibrium with compacted clay. This is mainly due to the intrinsic difficulties in measuring in-situ properties and in unravelling the complex nature of water/clay interactions under narrow confinement. For instance, an important proportion of water saturating compacted bentonite enters the interlayer space of montmorillonite, where exchangeable cations reside. In all but the first BPW model applied so far (Kristallin-I), interlayer water was considered to be unreactive and thus excluded from equilibrium calculations. This implies that only the external interparticle water is considered to be able to undergo complexation or dissolution/precipitation reactions. Even more importantly, due to the high degree of compaction and the negatively charged surface of the clay, interparticle water splits into two significant fractions:

1. Water associated to the electrical diffuse double layer (DDL) characterized by a chargecompensating cation excess and anion exclusion.
2. “Free” electrically neutral water, the volume of which is defined by the so-called anion porosity.

Classical aqueous chemistry involving complexation and dissolution/precipitation reactions can be safely applied only in electrically neutral “free” water. The inclusion of DDL water in the equilibrium calculations always involves simplifying assumptions and approximations (see e.g. Wersin et al. 2004). Because the relative volumes of the aforementioned three types of water depend on electrolyte concentration (ionic strength) and clay compaction (dry density), and because the salinity of the interparticle water may be affected by the nature of the bentonite resaturation process ¹, one is faced to the non-trivial problem of defining volume and composition of reactive water. The relative volumes of DDL and free water may even vary during an equilibration step, e.g. due to precipitation reactions which tend to reduce both ionic strength and porosity.

In the BPW calculations developed for Kristallin-I (Curti 1993) and “Entsorgungsnachweis” (Curti & Wersin 2002), all water saturating bentonite (i.e. interlayer + external) was considered to be reactive. However, as a spin-off of the “Entsorgungsnachweis” safety assessment this issue was discussed and a variant was calculated, in which complexation and dissolution/precipitation equilibria were limited only to the interparticle water (Wersin et al. 2004). Finally, for the definition of BPWs in the context of SGT-2, interlayer water was consistently excluded from the equilibrium calculations (Bradbury et al. 2014). In the two latter cases, it was assumed that the Opalinus Clay pore water (OPAw) reacts with bentonite pre-saturated via vapour adsorption, implying that the salinity of intruding OPAw does not increase due to uptake of pure water in the interlayer.

For SGT-2, two model variants were applied, denoted as “conventional” and “new” model. In the conventional model, montmorillonite was treated as a pure ion exchanger. Any dissolution or precipitation of montmorillonite was neglected. This (still customary) treatment is usually justified by the very low solubility of this mineral at low temperatures and by the absence of reliable thermodynamic data (solubility products) for montmorillonite. The “new” model (Berner et al. 2013, Bradbury et al. 2014) takes into account minor dissolution/precipitation of montmorillonite by calibrating thermodynamic data against in-house solution data from batch experiments. Moreover, an ill-defined soluble form of magnetite (“hydro-magnetite”) was used in the “new” model, whereas in all “conventional” calculations the thermodynamic data for crystalline magnetite (those present in the PSI-Nagra database) was used. This choice has consequences on the calculated redox potential, as discussed later.

Finally, the “new” model calculates equilibrium without constraining carbon dioxide partial pressure (pCO2) to a pre-defined fixed value. In the “conventional” model this parameter was fixed to values derived from borehole data. It is important to realize that these two alternatives reflect distinct assumptions on the mobility of dissolved CO2 through the near-field. A constrained pCO2 implicitly assumes fast diffusion and a permanent connection with a large external CO2 reservoir in the geosphere. In contrast, leaving the pCO2 unconstrained implies that the bentonite system is essentially treated as a closed system which equilibrates only internally and is completely separated from external CO2 sources. The two approaches thus represent two limiting cases of an equilibration process that probably proceeds in reality in-between. The results of SGT-2 calculations nevertheless showed only minor differences in the results of these two limiting calculations, indicating that other powerful buffering mechanisms play a more important role in the system of interest.

The SGT-3 model developed here is in-line with the “conventional” variant of the SGT-2 approach, with a few conceptual and technical modifications that will be described later in this report. All (preliminary) calculations presented in this report simulate a one-cycle reaction at 25 °C and 1 bar total pressure between normative amounts of bentonite and interparticle water corresponding to a predefined compaction of the bentonite. As for SGT-2, variants with both constrained and unconstrained pCO2 were calculated in order to assess the impact of the opposite assumptions on CO2 diffusivity.

As an alternative to the aforementioned models (all treating equilibrium in compacted clays as interaction between distinct solid phases and a separated aqueous solution) models based on Donnan-type equilibria have been recently proposed and applied (Birgersson & Karnland 2009, Tournassat 2016, Gimmi & Alt-Epping 2018), in which the entire compacted clay-water system is treated thermodynamically as a single homogeneous phase. The debate on which type of model is scientifically sounder, is still ongoing (Birgersson 2017, Birgersson et al. 2017). However, Donnan treatments of compacted clays are so far limited to major electrolyte ions. No complete description of trace solutes (e.g. radionuclides) inside the saturated bentonite is currently available and therefore this approach is presently not applicable to safety assessment purposes.

Consequently, the classical and still customary modelling method involving separate aqueous and solid phases is used here.

No reactive transport calculations to determine the temporal and spatial evolution of the bentonite pore water system are presented in this study. Such calculations (focussed on the interaction of an external low-pH concrete with bentonite and Opalinus Clay) were extensively carried out in the framework of SGT-2 (Berner et al. 2013) and an update will be presented soon in a separate report. The calculations showed that the predicted modifications in exchanged cation distribution, mineralogy and porosity induced by the concrete liner will affect only a small part of the bentonite external boundary, leaving most of the backfill material intact. The conclusions drawn from that study are anticipated to be still valid and can be taken over for SGT-3.