Gas flow modelling through clay and claystones

Large scale gas flow experiments conducted in connection with nuclear waste disposal research have shown the dominant effect of "minor" details such as interfaces, contacts and layer boundaries. Even if the scale of the analysis is highly reduced, in search of homogeneous point-like conditions, a systematic development of preferential paths is very often reported. Small size samples become boundary value problems. Preferential paths, when their thickness is modified by the stress-strain response of the media, under the combined action of stress and fluid pressure changes, become highly conductive features for gas flow. The development of preferential paths for fluid flow has been approached in a simple manner by embedding a discontinuity feature into an otherwise continuous element which models clay or claystone matrix behavior. The joint is activated when tensile strains develop in the continuous element. Then, hydraulic properties (permeability, retention behavior) are modified by means of laws derived from the physics of flow in discontinuities. The outlined idea was incorporated into a full Thermo-Hydro-Mechanical finite element code (CODE_BRIGHT) which has a wide range of capabilities for the modeling of two-phase flow in elasto-viscoplastic porous materials. A particular aspect which required attention is the modeling of expansive and shrinkage behavior induced by suction changes. In this way, healing effects during re-saturation may be simulated. Two experimental programs on clay shale samples, performed under triaxial stress conditions will be discussed. In the first case samples of Opalinus shale were subjected to a series of gas pulse decay tests during the application of stress paths involving a particular sequence of confining stress and shearing up to failure. In the second experimental program, performed on a tertiary mudstone from the Norwegian shelf, attention was paid to the effect of bedding-induced anisotropy. Experimental results will be presented and compared with model calculations. In a third example, a large scale gas flow experiment performed in the Grimsel Rock Mechanics Laboratory (Switzerland), will be described. The experiment was designed to evaluate the sealing performance of a sand-bentonite compacted mixture. The experiment was located in a cavern excavated in a fractured granite formation. Gas was forced through the seal into the granite rock. Measured test performance and model calculations will be discussed.