Introduction – Normal faults are part of the elements that control fluid flows in sedimentary basins. They can play the role of a barrier or a drain [Hippler, 1993]. These pathways are anisotropic. The aim of this study is to determine the fluid pathways and to characterise the pore network and its role in the transfer properties.
Petrophysics, petrographics, geochemical and fluid inclusion studies allow us to characterise a Buntsandstein sandstone affected by a normal fault. This sandstone has a fluviatile origin, field evidenced by fluviatile channels, but also by some clay layers. The fault is located in the north east of France, in the Rhine Graben. The vertical displacement is about 3 meters, and the dip is 70o east. The fractured zone is composed of three compartments (the hanging wall and the footwall separated by a gouge) divided by three main faults (fig. 1). Oriented samples were taken from the three blocks and were studied following the procedure figure 2.
Results – The petrographical and mineralogical composition of the three compartments were different. The gouge and the footwall were characterised by quartz overgrowths, authigenic kaolinite (30 to 40 % of the clay fraction) and diagenetic illite (40 to 60 % of the clay fraction). The hanging wall was characterised by 70 to 80 % diagenetic illite of the clay fraction (fig. 3).
The isotopic composition of the footwall quartz overgrowth (fig. 4) was δ18O enriched ranging from 13,4 to 13,6 ‰ SMOW, compared to detritic quartz ranging from 10,7 to 11,8‰ SMOW. Such quartz precipitations originated from fluid circulations, with temperature ranging between 195oC and 225oC according to fluid inclusion data in quartz overgrowth. This occurred mainly in the hanging wall but also in the fault gouge. The isotopic study of minerals and the quartz overgrowth fluid inclusion study showed that these fluids were similar to present day fluids characterized by Pauwels et al.  in the deep Upper Rhine Graben (tab. I).
The fault gouge was first like a drain allowing the fluid to circulate from the deep graben and then it acted as a barrier preventing the fluid from spreading in the hanging wall. This was confirmed by the study of thin sections, that revealed a cataclastic zone in samples located between the hanging wall and the gouge (fig. 5).
The evolution of porosity was characterised along a profile crossing the fault. Porosity values evolved from 12 % in the hanging wall, to 6 % in the fault gouge, and 12 % in the footwall (fig. 6). Oriented mercury injection measurements were carried out on covered (fig. 7) and non covered (fig. 6) epoxy resin samples to compare permeability related to porous network. When the samples were covered with epoxy resin, mercury was injected only into the network which was connected to the injection surface (fig. 7). The process indicated a connectivity of the sample and it could be quantified. High differences between the two porosity values suggest that the porous network was not connected with the surface of the sample. The covered or not covered samples exhibited no porosity variations with orientation.
The lowest mean permeability occurred in the fault gouge (0,1 mD). It increased in the hanging wall (100 mD) and in the footwall (200 mD). The maximum value of oriented permeability measurements occurred in the bedding plane (250 mD) (fig. 8). The direction of this maximum permeability varied in the two blocks with the direction of the fluviatile channels. The minimum permeability in the hanging wall (12 mD) and in the footwall (34 mD) were perpendicular to the bedding. This sedimentary permeability anisotropy disappeared in the gouge (fig. 8).
Discussion– Fault zones are assumed to be fluid pathways and fluid barriers. This study has shown that the same fault can act as a barrier and a drain for fluid circulation. Permeability anisotropy is usually related to fracturation, but only in the case of short time fluids pathways. Indeed, when the fracture network is totally cemented, the matrix plays the role of pathway. The evolution of the porous network depends on the tectonics and on the fluid circulation. Permeability and permeability anisotropy decrease as the distance to the gouge decreases. We also noticed a decrease of pore threshold and connectivity of the porous network. In fact, permeability depends on tortuosity, connectivity, but also on porosity and pore threshold [Katz and Thompson, 1987].
In these sandstones, classical mercury injection did not indicate any significant variations. But oriented and resin covered mercury injection allowed us to distinguished three types of samples response (fig. 9) :
– similar porosity and pore threshold in covered and non resin covered samples indicate a good connectivity , but no preferential orientation of the porous network ;
– similar porosity but different pore threshold indicate a preferential orientation of the structures but also a good connectivity ;
– different porosity and pore threshold indicate either a bad connectivity or a preferential orientation of the microstructures.
In this study, we have clearly shown an evolution of the permeability due to tectonic events and fluid circulations. The decrease of permeability and permeability anisotropy near the fault is principally due to the tectonic event. This decrease was associated with a decrease of porosity and pore threshold due to compaction in the footwall because of the great number of stylolithes. In the hanging wall, the decrease of petrophysical properties was due to precipitation of cement around quartz grains. The permeability reduction near the fault accounted for the role of the microstructures in fluid pathways. They were horizontal in the undeformed rock and became vertical in the faulted rock.