Different holes were drilled through the Soultz-sous-Forêts granite to generate a circulation at great depth but only one of them (EPS-1) was cored between 1400 and 2300 m in the granite basement. Seventy-nine samples were selected along this core at different depths (see Table 1 for their locations and their petrophysical properties). Previous to the geothermal stimulation, the Hercynian and Alpine phases and especially the Tertiary extensive episode (Villemin and Bergerat, 1987) structured the granite body. This long geological history is associated with tectonic fracturing, hydrothermal fluid circulations, primary minerals dissolution and secondary minerals precipitation that created not only a network of connected open fractures but also porous and extended permeable zones around them (Genter et al., 1990; Genter and Traineau, 1992; Ledésert et al., 1993; Rosener, 2007; Sausse et al., 1998; Sausse, 2002; Surma, 2003).

The petrographic study revealed three types of material: fresh granite, pervasive alteration granite and a fractured-altered granite, (Genter et al., 1990; Genter and Traineau, 1992; Jacquemont, 2002; Ledésert et al., 1993; Rosener, 2007; Sausse et al., 1998; Surma, 2003). Quartz, K-feldspar, plagioclase, biotite and a high density of K-feldspar phenocrysts compose the fresh granite. The first type of alteration is characterised by transformation of primary minerals like biotite, amphibole and plagioclase. Biotite is mainly altered into chlorite and sometimes in celadonite, epidote, hydrogarnet, and carbonates. Plagioclase is mainly altered into corrensite, calcite or illite. The second type of alteration is located in the matrix of the damaged zone and controlled by fluid circulation in and around fractures. The main minerals in the fractures are phyllosilicates (illite, chlorite, tosudite), carbonates and quartz, where the matrix shows mostly illite and carbonates as secondary phases. Schematically, each facies could be characterized by clay phases; the protolith is defined by chlorite formation, the damage zone by the chlorite-illite association, and the fault core by illite. Other clay phases or secondary phases are also present but we choose this simple characteristic for a quick analysis of the samples and their location in the fault model (Table 1, Fig. 1).

Petrography and chemical data are available through some publications (Genter et al., 1990; Genter and Traineau, 1992; Jacquemont, 2002; Ledésert et al., 1993; Sausse et al., 1998; Surma, 2003). Petrophysical properties are measured on samples selected along the EPS1 core. Four of these are determined: porosity, permeability, thermal conductivity (TC), and specific surface. These parameters control chemical and physical fluid-rock interactions and could be later use to model these processes. All of these measurements are not systematically performed on the same samples, Table 1 summarizes all of obtained data.

Porosity measurement and threshold diameter determination are obtained from a mercury injection test. This technique is based on the Young-Laplace equation for the displacement of a non-wetting fluid in a thin capillary tube, relating the fluid pressure to the throat size. This technique gives access to the total porosity value, but also provides information about the distribution of pore size and pore access diameters. Volume of the samples varies between 3 and 5 cm³. Due to grain size issues, several samples are used to characterize each facies, in a way to analyse a representative volume.

Permeability measurements are made using a nitrogen gas permeameter working at 40 bars confining pressure. Samples are 18 mm in diameter and 25 mm length. For different constant head pressures (lower than 35 bars) and once the steady-state condition is reached, gas flow rates and pressures from the both sides of the samples (the head and foot pressures) are recorded. These values must be corrected for the Klinkenberg effect to obtain the real permeability value (Scheideffer, 1974; Debschutz et al., 1989; Klinkenberg, 1941). For several samples, these experimental conditions are not sufficient to measure flow and they are defined as impermeable, with permeability lower than 10⁻²⁰ m². It is also important to notice that these experimental conditions are much different from the deep in situ conditions.

To determine the TC, we use the optical scanning technique developed by Popov et al. (1999). With a measurement taken every 1 mm, TC normal to the analysed surface is determined as an arithmetic average of local conductivity along an entire scanning line. The relative measurement error is not more than 2% (Popov et al., 1999). Measurements are performed on samples that are 10 to 20 cm long; some of them were re-sampled afterwards for permeability measurements. Compared to other techniques, the optical scanning allows both average and pinpoint evaluations.

The BET method (Brunnauer et al., 1938) is used for specific surface area (SSA) measurements. At low temperature and low pressure, gas is adsorbed on the porosity network surface (Gregg and Sing, 1982). By measuring the quantity of gas adsorbed at different pressures, it is possible to estimate the specific surface of the sample (in m² g⁻¹ or m² m⁻³). Nitrogen is used for our measurement.

Using SEM observations, void shapes and voids relationships with mineral components are determined to define the link between the different physical properties. Observations are performed on ships of rocks under a GEOL scanning electronic microscope.

All of the results are summarized on the Fig. 2. Two main types of information come from the mercury injection technique, the porosity value and the distribution of the pore volume versus the threshold value. Porosity ranges from 0.5% to up to 15% and its variation is not depending on the depth of sampling. It is controlled by the location of the sample within the fault zone. Samples from the protholith have porosity as low as 0.5%. In the damage zone, porosity can increase up to 15%, and lower values can also be measured near the fault core. Porosity in the fault core varies over two orders of magnitude.

With regard to the EPS-1 well, four high porosity zones are recognized, three of them linked to fracture zones at 1620, 2070 and 2120 m depth (Fig. 2). The fourth is located at the upper part of the batholith on the Hercynian unconformity, where alteration of the granite's top and post Permian/Triassic fluid flow induced high porosity at the granite-sedimentary cover contact. Zones with lower porosity are also localized along secondary fracture zones, as at 1800 m depth. Low fracture density characterises these secondary fracture zones.

The porosity increase could be linked to an increase of the threshold values (Fig. 3). Different porosity spectra could be recognized, from relatively flat spectra for the low porosity material to a porosity network controlled by a particular threshold value, where the porosity increase is associated with the threshold increase. These porosity variations are measured on small volume samples (3 or 5 cm³) and describe matrix porosity and microcracks porosity development.

Permeability varies over five orders of magnitude from 10⁻²⁰ m² to 10⁻¹⁵ m² (Fig. 2). Several high permeability zones are recognized with permeability values higher than 5.10⁻¹⁷ m². In several cases, these zones could be related to high porosity zones like at 1620 or 2120 m depth, but in other cases, these high permeability zones are not linked to high porosity like at 1800 or 1900 m depth. Conversely, high porosity zones are not related to high permeability volume. This could be due to lack of permeability samples or to low cracking of the matrix around different fault zones. In the alteration zone of the upper part of the batholith, a large variation of the permeability is measured over three orders of magnitude. The permeability–porosity relationship is complex (Fig. 4) and depends on the pore structure, voids geometry and location (Fig. 5). Three types of behaviour could be distinguished depending on the porosity shape (Rosener and Géraud, 2007). For the cracks network (group 1), a sharp increase of the permeability is measured for a low porosity increase. The second group (group 2) is composed of samples where alteration and secondary phase precipitation modify the pore structure. Under these conditions the permeability decreases with porosity. Samples of the group 3 are intensively cataclased and illite and chlorite precipitate in the enlarged cracks and pores by mechanical and chemical processes. For low porosity samples, the permeability increases with porosity, and for porosity above 4% the permeability decreases. These particular relationships (decrease) between the porosity and the permeability are clearly the illustration of the tortuosity effect controlled by the infilling development in the cracks networks and the alteration of the primary phases. Especially for the group 3, the threshold value being constant for all the samples of the group, porosity parameters (volume and threshold) increase and the permeability decreases. They illustrate the influence of the tortuosity increase.

The SSA is not measured for all the samples, but a similar tendency like the ones described for permeability and porosity is observed, i.e. a large variation is measured for samples coming from the fractured zone. However, a simple relationship with permeability or porosity could not be found. The highest values are measured on samples coming from two fracture zones at 1660 and 2100 m depth. The altered zone in the upper part of the batholith gives also samples with high SSA value. These high values are directly linked to the clay content of the sample, whereas the lowest SSA is measured for the unaltered and little fractured samples (Rosener and Géraud, 2007).

TC varies between 2.3 W K⁻¹m⁻¹ and 3.7 W K⁻¹m⁻¹. TC for the fresh granite lies between 2.5 W K⁻¹m⁻¹ and 2.7 W K⁻¹m⁻¹, in accordance with the literature (Cermak and Rybach, 1982; Pribnow and Sass, 1995; Rummel, 1991; Surma and Géraud, 2003). As for the other properties, TC presents large variations within the fractured zones. Decrease of the TC value is the result of the porosity development by fracturing or by mineral alteration. This process is particularly obvious across the altered zone at the top of the batholith, where the lowest values are measured. On the other hand, a large part of the tested samples shows an increase of the TC value with alteration. This evolution is clearly associated to secondary phase precipitations, especially quartz with a TC value of 7.69 W K⁻¹m⁻¹ and calcite with a mean TC value 5 W K⁻¹m⁻¹. This TC increase could mask the effect of the porosity development.

From these data, physical properties of a theoretical fault zone may be proposed. A first qualitative geometry of the fault zones in the Soultz's batholith was proposed by Genter et al.(1995). Here after, our model is the synthesis of data coming from different Phd theses (Jacquemont, 2002; Rosener, 2007; Surma, 2003), their mineralogical and chemical point of view is developed in Fritz et al. (this volume). Using the model proposed by Caine et al., 1996 and Sibson and Rowland, 2003, a more refined synthesis of the fault zone could be drawn and analysed samples could be localised on this model. Essentially, a set of altered conjugate structures with lower offset is associated to the principal fault.

The central part of the main structure is described as the fault core: a highly strained material, with large ranges of porosity and permeability depending on the degree of cementation. It can also include locally open sections that could support high fluid flow after being reconnected during the stimulation. The second part is the damage zone, where previously fractured material is subject to alteration and secondary precipitations. Generally, the increase of porosity induced by fracturing and mineral dissolution is higher than the decrease of porosity associated with secondary mineral precipitation. However, due to their nature (mostly clays) secondary precipitations generate high tortuosity, and so tend to decrease the permeability of the rock (group 3 material). In these damage zones, there are also some preserved rock volumes of which some small porosity increase related to sharp permeability increase is unclear but probably induced by a crack network development (group 1). These volumes (fault core and damage zone) are generally not connected to the protolith or other preserved blocks. The last part of the system is the protholith that remains largely unaffected by the fault.

The global geometry and physical properties profile across the fault zone could be described as follows: within the damage zone, from the protolith to the fault core, permeability increases due to fracturing and primary phase alteration. Then, toward the fault plane, it could be reduced by secondary mineral precipitation (Lockner et al., 2000). Between the two surfaces of the fault plane, occasional free space could represent a high fluid flow conduit with high permeability; where in other sections, this volume could be filled more or less completely, possibly showing a lower permeability than the surrounding material. In the last case, the damage zone would then be the section showing the highest permeability. Finally, connections between the fault core and the damage zone are made possible via several secondary structures that could, at their own scale, present a similar layout as the main fault zone.

This geometry explains the high frequency variation of the different physical properties measured in the fault zones or the fractured zones and also the large variations of transport properties recognized along the borehole. It is also in accordance with the conceptual model proposed by Caine et al., 1996, where the size of the fault core and the damage zone could vary as well as their transport properties.

Petrophysical characterizations of different samples coming from the EPS1 core are used to build a fault zone model of the deep geothermal reservoir. The different elements of this fault zone are the open space between the fault planes, the fault core, the damage zone and the protolith. The characteristics of each part point out the role of the damage zone as a high porosity and permeability zone. In particular, it could be a pathway for the fluid if the fault core is filled by impermeable secondary phases. Across the fault zone, even if structure sizes are variable, petrophysical properties of the damage zone seem to be more consistent than in the fault core, where fluid-rock interactions are more intense and induce bigger variations.