Consolidation and rebound properties of Opalinus Clay : a long-term, fully drained test

A specimen of Opalinus clay from Mont Terri has been subjected to stress testing over a period of 532 days. Testing was undertaken by changing either (or both) of the axial and confining stresses in sharp steps followed by periods of between 4 and 82 days during which time the specimen was allowed to adjust to the new stress state. In this way, the drained consolidation, creep and rebound behaviour of an Opalinus clay specimen was examined. The test material was subjected to a maximum average effective stress of 38.3 MPa. Volumetric strain data for both volume change and porewater displacement measurements indicate a small inflection in the standard geotechnical plot of void ratio against the logarithm of average effective stress at a value between 20 and 22 MPa. The negative slope of the consolidation curve (α) based on volume change measurements exhibits a general trend of increasing magnitude as effective stress rises. Even though the data do not exhibit the sharp increase in α indicative of classic virgin consolidation behaviour, it would appear that plastic yielding is occurring at an average effective stress below 20 MPa. Analysis of net porewater flow measurements suggest original interstitial fluid was not expelled from the specimen until average effective stress exceeded 20 MPa. Given the data available, an estimate for the preconsolidation stress in the region of 20 to 25 MPa seems reasonable. As effective stress rises the duration of the strain transients lengthen. As the induration state of the mudrock increases, strain traces are characterised by less well-defined transients, indicative of time-dependent plastic yielding at high effective stresses. The volumetric strain data for both volume change and porewater displacement shows similar transient behaviour. These results give an average principal strain ratio of 0.252, suggesting the material is either mechanically anisotropic or behaving as a non-ideal elastic medium. Specific storage values derived from porewater displacement measurements show a general decreasing trend with increasing average effective stress and are in the range 1.5 to 12.5 × 10-6 m-1. Data from volume change measurements are less sensitive to changes in effective stress and are in the range 1.2 to 17.5 × 10-6 m-1. Elastic constants derived for undrained quantities are significantly higher than those for drained conditions by approximately one order of magnitude. Data suggests there is a transition in behaviour centred around an average effective stress of approximately 20 MPa. Analysis of creep curves can be broken down into three distinct responses. The Lemaitre model, as applied to Opalinus clay by Boidy (Boidy et al., 2002), was applied to the current test data. However the published model parameters failed to adequately fit the current data. Minor alteration of these parameters enabled modelling of the longer-term volumetric responses to be undertaken. The Lemaitre model did not predict the initial stage of creep very effectively. A much slower response time was seen in the current data, which was absent in the work by previous researchers. A power-law creep model was established. In general the fit was adequate for the volumetric strain observed, although these data exhibited some noise. In contrast, the fit of the axial strain data was not adequate and even the subdividing of the data into the individual creep stages failed to give an acceptable fit. A combination of power-law for the initial response and Lemaitre for the longer response may achieve a better prediction for this test stage. A numerical simulation was run using the 2-dimensional coupled flow and deformation code STAFAN. Two phases of the testing were modelled separately. During Phase 1, the model was used in an attempt to fit the creep data. A reasonable fit was made to the first step axial strain data, but the extrapolation to later stages showed a progressive deviation from the data. In addition, the model made poor predictions for the radial strain and porewater flow data in all steps. These observations indicate that both the assumptions of linear elasticity and isotropic deformation are probably invalid for this specimen. During the second phase of testing, the axial and confining stresses were raised synchronously in a series of seven 4 MPa steps. In view of the results of the Phase 1 modelling, it was decided to treat each step of Phase 2 as a separate test and to use the model to parameterise the changing state of the specimen. Young’s modulus was significantly lower than those derived from volume and porewater displacement measurements, which can be explained by the over prediction of radial strain due to the simple linear-elastic assumption in the STAFAN model. It has been shown that the linear elastic deformation model is not a good analogue for the behaviour of this specimen. There are clear indications of non-linear responses to stress changes in the data and it seems likely that some form of viscoelastic or viscoplastic model should be adopted. In addition, the axial and radial strain responses would seem to be anisotropic, bringing further complexity to the model that should be employed