Development of a new non-linear elastic hydro-mechanical model for the simulation of compacted MX-80 bentonite: application to laboratory and in situ sealing experiments for geo-repository engineered barriers

Abstract

The management of radioactive wastes is a significant environmental issue facing the international nuclear community today. The current international consensus is for disposal of higher activity waste from a variety of sources in deep geological disposal facilities (GDFs). Hydraulic seals, often planned to consist of compacted bentonite-sand blocks, are an important part of the closure phase of a GDF. As such, an understanding of the hydro-mechanical (HM) behaviour of these seals, and the ability to model and predict their behaviour is fundamental to support many planned safety cases and licence applications. Bentonite is well suited for use as a hydraulic seal due to its high swelling capacity that enables it to swell into voids while maintaining a low permeability sealed barrier to advective flow, and to provide structural support by generating a swelling pressure on the excavation walls. The hydro-mechanical process of bentonite hydration is a highly non-linear problem. As such, coupled process models that are able to account for the strong inter-dependence of the hydraulic and mechanical processes are employed to simulate the behaviour of bentonite under repository conditions. This thesis reports the development of an HM coupled model in the open source finite element code OpenGeoSys (OGS), and its application to the simulation of a range of hydraulic seal test conditions. The developed model couples Richards’ equation for unsaturated flow to a new strain dependent non-linear elastic mechanical model that incorporates a Lagrangian moving finite element mesh to inform the material non-linearity. Stress and volumetric dependent water retention behaviour are incorporated through the implementation of the Dueck suction concept extended to take into account non-recoverable strains during consolidation. A number of permeability functions are implemented and tested against experimental data. The mechanical model is extended to account for wetting-induced collapse behaviour by the definition of a failure curve derived from experimental results. Similar in definition to the Loading-Collapse curve in elasto-plastic models, this failure curve triggers the application of a source term to account for wetting-induced collapse. Coupling between the hydraulic and mechanical processes is achieved through the stress dependency of the water retention behaviour, the inclusion of a new coupling factor for the hydraulic contribution to the mechanical process, and the dependency of numerical convergence criteria on net mean stress. An explicit iterative calculation approach is employed. As a result, the hydraulic and mechanical moving meshes are decoupled to allow volumetric dependent parameters to be updated within process iterations. The model is calibrated and compared to experimental data from the SEALEX experiments conducted by the Institut de Radioprotection et de S ˆ uret´e Nucl´eaire (IRSN) at the Tournemire URL, France. The experimental programme comprises standardised laboratory tests, a 1/10th scale mock-up of a hydraulic seal with a uniform technological void, and a full scale in situ performance test with a non-uniform technological void due to its horizontal geometry. Using a model with 5 hydraulic parameters, 8 mechanical parameters with an experimentally defined failure curve, and one coupling parameter, the major trends of behaviour in all the SEALEX experiments can be recreated, including axial stress build up, water uptake, and final deformation. However, the elastic method employed leads to an over prediction of the rebound on loss of axial confinement in the 1/10th scale mock-up test. Simulations suggest that the non-symmetric technological void in the full scale performance test could have lasting effects on the development of heterogeneity in the hydraulic seal. The development of heterogeneity does not adversely affect the permeability with respect to the design criteria, but may have significant consequences for the development of a heterogeneous swelling pressure.