The Gulf of Corinth in central Greece is an active neotectonic basin formed by two opposed facing normal fault zones striking WNW–ESE at the northern and southern bounds (Fig. 1). Two tectonic processes dominate basin formation. One is a north–south-extension rifting of 11–$16\mathrm{mm}{\mathrm{yr}}^{-1}$ [1], the other is isostatic compensation with footwall uplift from erosion of horsts and hangingwall subsidence due to the sedimentation on the basin [8]. These combined tectonic activities have resulted in intermittent catastrophic slips along the faults in the form of earthquakes, the latest a M^s=6.2 earthquake, in 1995 [3].

The geology of the study area consists mainly of a thick formation of tight, highly fractured micritic limestone as a bedrock, overlain by Late Pleistocene marine terrace deposits, composed mainly of conglomerate. Separating the two sequences is commonly a laminated calcitic mud (properly, ‘calcilutite’) up to several metres in thickness, providing an unconformable ductile top cap for the limestones. In other areas, the separation between limestone and conglomerate is marked by siliceous radiolarian cherts.

The calcilutite is important both because it can control the hydraulic connectivity between limestone and conglomerate, and because it is an excellent analogue for fine-grained cataclasite expected to be found on major regional faulting through the limestone sequence. Hence its properties are important for the understanding of the hydromechanical response of the Aigion Fault Zone. Within the Aigion borehole (AG-10) itself, the active fault transects the radiolarian cap sequence, so that comparison of material properties of calcilutite and recovered siliceous fault gouge provides insight into both local and regional hydromechanical coupling in the Aigion fault.

Calcilutite was sampled from a roadcut outcrop west of the Selinoutas River, 4 km south of the Aigion Fault Zone (Fig. 1). Fault gouge was obtained from 760 m below ground level in the AG-10 well at the intersection with the Aigion Fault [4]. Remoulded materials were used for the tests, with the full size distribution used for calcilutite. The remoulded fault gouge, kindly supplied by Sulem, CERMES Paris, was a sample reconstituted from original fault gouge in the correct particle binnings, but with all material >2 mm diameter removed.

Compaction tests were performed in a solid, opposed piston oedometric cell, shown in Fig. 2. The 38 mm-diameter samples, 3–4 cm long, were preformed in a mould, and wrapped radially in two layers of thin Teflon tape, which decoupled frictional binding between sample and cell wall. Samples were capped at either end with porous stainless steel frit discs. Loading was applied using a gas-accumulator-buffered hydraulic piston in a load frame. Sample compaction was monitored by averaging a pair of gauge LVDTs mounted across the load pistons of the oedometer cell. Thermal pressure cycling in the hydraulic load system was mitigated by water-bath temperature stabilisation of the gas accumulator system, with loading constant to within ±0.5%.

Tests commenced with the saturated sample assembly in a drained condition subjected to a constant axial stress. After one or two days of drained consolidation at conditions of zero input flow, water was pumped upward through the sample at a constant rate of $1.05\times {10}^{-10}{\mathrm{m}}^3{\mathrm{s}}^{-1}$ for the calcilutite and $4.2\times {10}^{-11}{\mathrm{m}}^3{\mathrm{s}}^{-1}$ for the fault gauge. The resulting pressure difference across the sample was used to estimate the permeability using Darcy's law. Tests took at least 48 h for the calcilutite, and 72 h for the fault gouge, due to their different properties. The compaction/permeability cycle was repeated at steps of increasing axial load to obtain the hydromechanical behaviour of the samples as a function of overburden thickness.

After the last stage of the test at 27 MPa vertical stress for the calcilutite, and at 25 MPa vertical stress for the fault gouge, sample relaxation was measured for 24 h after sample-load release. The wet specimen was removed from the cell, weighed, measured, then dried at 105 °C for 48 h, re-weighed and remeasured. The mineral composition and particle size were analysed using X-ray diffraction and laser-diffraction particle size analysis.

The particle size of the calcilutite sample ranges from >0.4 to 146 μm, with an average of 15.97 μm in diameter. According to British Standards for particle size ranges [3], it comprises 74% silt size particles, 21% clay size, and 5% fine sand. The reconstituted fault gouge sample from the AG-10 well has a larger range of particle size ranging from 0.4 to 3600 μm, with 490 μm in average. It comprises mostly coarse material (43.2% sand and 30.8% fine-grained gravel size particles) with 9.2% of clay and 16.8% of silt size particles as minor constituents. The particle size distribution curves (Fig. 3), show that the two materials differ both in particle size and distribution. A steep S-shaped curve for the calcilutite sample indicates it is well graded (Fig. 3(b)). The flatter, irregular curves of the full fault gouge (Sulem) suggest a large range of particle sizes, poorly graded. However, the fault gouge distribution is multimodal (Fig. 3(a)), and one would expect the hydraulic behaviour to be dominated by the finest fractions, with coarse materials increasing flow tortuosity through bulk samples.

X-ray diffraction analysis revealed that the calcilutite sample is composed mostly of calcite (Table 1). This result is compatible with the limestone bedrock as a source of sediments (50% of calcite, 35% quartz, 15% of albite, and traces of dolomite, muscovite, and clinochlorite). However, the fault gouge obtained from the AG-10 wellbore consists mostly of quartz, as shown in Table 1, related to its origin from cataclasis of radiolarian chert. After the consolidation and the permeability tests, the dimensions and wet and dry weights of the calcilutite sample were measured. The result is shown in Table 2. Based on the phase relationships between the measured unit weight and the calculated unit weight from the average of specific gravity [6], we obtained void ratio (e) and porosity (φ) as 0.58 and 36.9%, respectively, for the calcilutite sample. Data for the initial fault gouge values were taken from Sulem et al. [7] with kind permission, with void ratio (e) and porosity (φ) of 0.32 and 24%, respectively.

An important mechanical property of fine-grained materials is the time-dependent deformation induced by a sudden change of external load. This transient reaction arises from drainage or suction of pore fluid due to the sudden change of pore volume, and continues until the pore pressure excess or deficit has dissipated [5]. In our experiments, we first observed the drained consolidation behaviour, a transient volume reduction resulting from a sudden increase of external load, at several different vertical stress levels. Shown in Fig. 4(a) are typical examples of time-based records of axial strain, equivalent to volumetric strain in our test condition, for the first 24 h. The compaction curves show that strain rate decreases with time for both samples. However, although the calcilutite is initially stiffer than the siliceous gouge, strain rate decreases far more rapidly in the gouge, so that after 22 h the gouge is rapidly stabilising, while the calcilutite continues to compact. This distinctive behaviour can be also expressed in the form of the strain rate change as a function of time, as shown in Fig. 4(b). The strain rate against time is approximated for calcilutite by a bi-linear function in the double logarithm domain with a negative slope. The gouge shows the same relationship, but with a less pronounced transition.

It is instructive to consider the pore volume evolution of the samples. From the consolidation theory [9], the volume of void is a function of fluid pressure dissipating with time, whilst the volume of solid is an elastic response to effective volumetric stress on particles. The variation of void ratio during consolidation at a constant effective volumetric stress was calculated from the axial strain using back calculations from the void ratio measured after testing [2]. Fig. 5 is the cumulative consolidation curve of the calcilutite sample in terms of the void ratio reduction, showing the first 24 h of each one of the five step increases in vertical loading. (The second 24 h of each step were spent in permeability measurement). The rate of void ratio reduction (−de/dt) for each level of the vertical stress was plotted in a double logarithm domain in Fig. 5(b). These plots show that the first compaction stage, to 16.2 MPa, causes maximum perturbation in the calcilutite sample. Subsequent load increases yield a separate family of curves. Each curve is approximated by a bi-linear function in the double logarithm domain, with a decreasing slope for increasing vertical stress, except for the final load. The intersection between different slopes is located around an hour after test commencement. At the final loading, 27 MPa, increasing rate of void ratio reduction signals the onset of pervasive grain crushing in the sample.

Fig. 6 is the full data set of compaction data illustrating the significant behaviour difference between calcitic mud, and siliceous fault gouge. In calcilutite, the first step of load increasing (16.2 MPa) led to a large consolidation with a clear change in volume reduction rate, but each subsequent 2.7 MPa load increase produced much lower consolidation with only gradual stiffness increase. In the fault gouge, 3 MPa compaction steps show rapid stiffness increase with time.

A notable feature of the data are minor discontinuities, arrowed in Fig. 6(b), in the fault gouge compaction curves, missing in the calcilutite curves. These are traceable to minor changes in applied load arising from slight temperature sensitivity of the load frame gas pressure buffers. An example of a correlation between the subtle change (less than 1.5%) of vertical stress and the sudden change in consolidation curve (about 10%) in Fig. 7 shows that transient compaction is extremely stress-sensitive in the siliceous fault gouge. This implies that stress transmission through the gouge is extremely inhomogeneous, with distinct volumes always lying close to critical collapse strength.

Because the hydraulic properties of the materials are of importance, the volumetric reduction with time for each level of the axial stress was converted into the variation of porosity as a function of the axial stress (Fig. 8), using consolidation theory [9]. Because the porosity is also a function of time, we plotted in Fig. 8, the porosity data at different times elapsed after the change of axial stress: 0, 24, and 48 h for calcilutite, and 0, 24, 48, and 73 h for fault gouge. We find that porosity at a given time is a linear function of external stress in both samples, with the same dependency, but different zero intercept.

Permeability evolution with compaction was measured by pumping degassed-deionised water upward through the sample at constant low flow-rate ($1.05\times {10}^{-10}{\mathrm{m}}^3{\mathrm{s}}^{-1}$ for calcilutite and $4.2\times {10}^{-11}{\mathrm{m}}^3{\mathrm{s}}^{-1}$ for fault gouge), after the one-day drained consolidation interval at each load step. The evolution of differential fluid pressure across the samples is shown in Fig. 9.

Calcilutite and fault gouge had very different hydraulic responses. The calcilutite, which is a well-sorted granular powder with low clay content, showed a steady pressure rise for the first 15 h toward a plateau value. In contrast, the gouge, which is a poorly sorted powder with moderate clay content, showed a rapid pressure rise, until establishment of an irregular fluctuating differential pressure. The rate of initial pressure rise is load dependent.

Permeabilities were estimated using Darcy's law with an assumption that the stabilised upstream pressure at each vertical stress level comes from a steady-state condition of flow. The evolution of permeability versus porosity is usually expressed in the literature in a double logarithm domain, as shown in Fig. 10. Both samples are approximated by a linear function in such a scale. The relationship between permeability and porosity of calcilutite and fault gouge are compatible with other silt-size mud-rich slurries in the literature [2], except for the horizontal shift due to particle size differences.

The different pressure rise behaviour of the calcilutite and gouge has important implications for the sealing style and evolution of the cap rock and fault filling they represent. The regular slow pressure rise in the calcilutite implies that this material, and by implication, all limestone gouges, act as homogeneous materials, transmitting water by homogeneous plug flow. Thus the permeability measured in the compaction studies detailed above is a true bulk property of the material. Hence the expected sealing capacity of the caprock or fault gouge can be calculated by direct spatial scaling from the experimental data. In contrast, the siliceous fault gouge, perhaps because of poor sorting, but more probably, because of the clay content, exhibits two permeabilities. The extremely rapid initial pressure rise at flow commencement indicates that the ‘static’ permeability of the gouge is extremely low. However, the sudden establishment of a fluctuating, irregular differential pressure under low flow conditions indicates that there is a ‘breakthrough’ differential pressure, which represents the onset of ‘packet’ flow through local inhomogeneities in the gouge. This local nature of permeability in the gouge correlates with the extreme local stress sensitivity of the material, noted in Section 4. Thus the measured flow permeability is not expected to scale spatially, but probably represents an absolute property of this type of gouge. It is notable that the fluid pressure measured below the gouge in the actual borehole is 1 MPa, which is predicted from the flow data in Fig. 9.

Two important poorly consolidated units from the Aigion Fault area have been characterised in terms of hydraulic and mechanical behaviours. One, a calcilutite, is a common fine-grained calcitic cap formation between the basal limestones and overlying conglomerates. This unit is important as a potential topseal to the limestone, and because it is a good analogue for limestone-derived fault gouge. The other is siliceous active fault gouge from the AG-10 well, a cataclasite derived from radiolarian chert. For mechanical deformation, both samples show a clear transient behaviour for a sudden change in external loading. However, calcilutite is more time-dependent, and fault gouge is more pressure-dependent, and the latter is notably sensitive to small perturbations in loading. For hydraulic characteristics, both have low, load-dependent permeability, 9–15×10⁻¹⁸ m² for the calcilutite and 0.9–2×10⁻¹⁸ m² for the fault gouge. Hence both are flow sealing compared to the heavily fractured limestone bedrock in the area. However, values for the calcilutite are shear-insensitive global values, whereas those for the siliceous gouge are local limits under cross-flow, and must be shear rate sensitive. We conclude transient response to sudden changes in effective stress may play a crucial role in the time-dependent triggering of fault movement in the Aigion region.