Faulf, fractures and fluids : gulf of corinth (3F-CORINTH)

The mechanical behaviour of faults is strongly coupled with the flow of water. First the fault's normal deformation is non-linear (hyperbolic) with respect to the normal effective strain. Second, the tangential strength depends on the normal effective stress. Further, the flow properties in the fault are dependent on the fault opening, i.e. of its normal deformation. On the one hand, the longitudinal transmittance may be related to the cubic law, which has been adapted to actual fault behaviour. On the other hand, the transverse flow properties (normally, through the fault) are modified by the fault transverse reduced or increased permeability. These concepts have been implemented in a hydromechanical-coupled constitutive model for fault. Moreover, a new fault finite element has been developed, tested, implemented in a finite element code. It has allowed to model some actual cases.

Stress measurements in deep boreholes can be conducted with both the Hydraulic Fracturing (HF) technique and the Hydraulic test on pre-existing Fractures (HTPF) technique. With the HF technique, a section of a borehole is isolated with a straddle packer and the borehole pressure is raised till a fracture is formed. Orientation of the fracture and analysis of the pressure-time curve yield the orientation of the maximum horizontal principal stress and the magnitude of the minimum principal stress. The HTPF technique is based on the measurement of the normal stress supported by fractures with various azimuths. Combined with the HF technique it provides means to determine all the components of the stress tensor at a given depth and their variations along the borehole axis. The Rock Mechanics laboratory at IPG-P has developed a tool combining electrical borehole imaging with a straddle packer device. The tool provides means to determine first zones of intact rock where to conduct HF tests and zones where one single fracture may be isolated with the straddle packer for HTPF tests. Then the tool is used for running both HF and HTPF tests. It provides directly the tested fracture orientation and the normal stress measurement. Previous experience had shown that the device was ill functioning in high salinity waters because of parasitic electrical current. Work has been conducted in the lab to analyse the source of these parasitic currents and solutions have been proposed. Results obtained at Trizonia, in mud based on sea water show that the tool operates presently efficiently in salty environment.

We have developed a new technique for calculating time-dependent hazard due to earthquake-earthquake triggering, and applied it to seismic activity centred on the gulf of Corinth. The results show that the Gulf of Corinth behaves in away that is near the global average. The correlation function has the same form as that expected in a slightly sub-critical point system. The correlation length is on the order of the seismogenic thickness, and the maximum range of influence is around the lithospheric thickness. The results are directly in the form of a probability of an event being triggered, consequent on the occurrence of a previous one, and hence can be used directly in probabilistic time-dependent seismic hazard calculation.

Seismic signals bring more information than first-arrival times at the free surface of the Earth. We have proceeded in a signal analysis for converted phases in seismograms in order to detect sub-horizontal faults which might occurred in the Aigion area during the graben opening mechanism. Both 3D arrival time estimations have been performed for PS and SP phases between P wave and S wave arrival times. This procedure is systematic and could be performed any time we have velocity structure available to us. The potential use of these converted phases is for locating earthquakes and for better characterization of mid crust mechanical behaviour which has an impact in seismic hazard estimation.

In order to deploy passive seismic antennas on appropriated sites and to record usable ground motion induced by natural seismological activity, homogeneous instrumental response of stations inside the arrays on each site of interest is crucial. Thus, a special care has to be taken for site selection knowing that arrays have to be deployed on flat topography layered homogeneous velocity structures. Four sites have been investigated by refraction seismic thanks to dedicated cross profiles. Based on first arrival tomography, the velocity structure of each site has been recovered. As they revealed smooth and flat near surface structure, two sites have been selected when the two others were rejected due to their strong velocity heterogeneities.

For the acquisition of well seismic data, we have applied the known field acquisition technique of VSP and Offset-VSP , with an additional feature for the down-hole VSP tool, specified as 4-component, i.e. containing 3 components of geophone and a hydrophone. The standard VSP processing techniques applied did not yield any reflection results, same as the surface reflection profiles. In these conditions, it became crucial to be able to apply older techniques of analysis with respect to the direct, refracted, and diffracted events of high energy evidencing features for abnormal propagation linked to the presence of main accidents associated with the E-W striking Aigion fault or of the secondary fault striking N-S. The main structural results from VSP concern the Aigion fault characterisation, as follows: - Single step accident approximation - Throw 150m - Dip 60?X, consistent with dipmeter log results and geometrical considerations Additionally, the surface refraction results are now coherent with the earth model derived from the VSP survey. Mythological results: The orientation of the 3 component of geophone was delicate as no gyroscope was combined with the VSP tool: hopefully two positions of source were recorded for each run. Azimuthal propagation deviations revealed the close proximity of the Aigion fault to the AIG-10 well. The hydrophone data was interesting to examine and evaluate in terms of potential industrial usefulness for the following purposes: - Discrimination between P-wave events and S-wave events in well seismic data - Assessment and raw quantification of the permeable potential of fluid productive fractures in hard rock, as long as the fracture/fault has been intersected by the well.

Within this project the development of a complex variable boundary element model coupling the mechanics of fractures, holes and inclusions with the fluid pressure and temperature changes in plane strain and anti-plane strain conditions has been accomplished. This model has been implemented in a first step into a numerical algorithm THEPOMECRACK (C++) for the pore pressure, temperature and stress field solutions around multiple curved cracks. The code is capable to accept mixed boundary conditions along the cracks, i.e. Dirichlet and Neumman. This model will be used as a tool for back-analysis and understanding of the: - Down-hole hydraulic fracturing stress measurements (back-analysis of hydro-fracturing results in order to evaluate the in-situ stresses and stress variation with depth in regions exhibiting normal faulting); - Interaction and activation of normal faults and pore pressure as well as temperature changes in seismic regions such as the Gulf of Corinth; - Fluid flow in fractured reservoirs (hydrogeology).

Rock samples from surface outcrops in the Aigion area and from the AG10 well were measured by a special technique in a laboratory to determine grain and bulk densities, porosity and magnetic susceptibility. Moreover, the wet (saturated) density was calculated from the data. In general, limestone from all sites showed similar values, while conglomerates exhibited wide range of density values. They can be divided into two main groups with respect to formation age (syn-rift and post-rift sedimentation). Extremely low density values and high porosity were observed in the group of clay sediments. The results are important both for the processing and interpretation of gravity data and for general understanding of rock physics in the area.

Surface seismic and VSP reconciliation: using the exact velocities of the major layer units encountered by the AIG-10 well, and taking into account the approximate age of the Aigion fault, a 3 layer velocity model of the overburden of the 4km/s refraction maker observed on raw surface seismic records in the southern compartment of the Aigion fault have been defined and validated. As a result, the refraction marker has been identified as the upper platy limestone in a radiolarite bearing layer encountered at 496m in AIG-10, and a refined computation of the depth of the 4km/s refraction marker has been derived: the final depth results are now coherent with the VSP observations and VSP derived Aigion thrust throw.

A model that describes the motion of a rapidly deforming shear-band, starting from first principles, has been formulated. The fault gouge was considered as a two-phase mixture of solids and fluid, and the governing equations were derived from the corresponding conservation laws of mass, momentum and energy. The resulting governing equations are a set of coupled diffusion-generation partial differential equations that contain three unknown functions, the pore-pressure, the temperature and the velocity field inside the shear-band. This set of equations turns out to be in general mathematically ill-posed unless some additional assumptions are met concerning the dependency of the friction coefficient of the gouge on the strain-rate. In the sequel the phenomenon of 'thermal run-away' in clayey shear-bands has been mathematically formulated. The importance of thermally collapsible and thermally softening clay on the overall dynamic thermo-porous-mechanical behaviour of clay-rich gouges has also been demonstrated. Accordingly, conditions for possible softening behaviour of the clay under thermal loading were investigated. The sensitive parameter for the description of the thermo-porous-mechanical coupling is the thermal expansion coefficient of the material. Possible collapse of the clay under thermal loading may activate fluid pressurization inside the fault and lead to substantial reduction of the apparent friction. The thermo-porous-mechanical model for clay-rich gouges developed within the 3F Project has been used as a guide for the targeted experimental characterization of the thermo-porous-mechanical behaviour of Aegion fault drilling cores (cf. project DGLab). Indeed, the heart of the fault is met at a depth of 760m and consists of clay- rich material on a length of about 1m. This zone is surrounded by a cataclastic one. Following the above modelling work developed within the present project the thermo-porous-mechanical behaviour of the clayey core was investigated by means of an odometer and triaxial compression experiments. It was shown that softening behaviour under thermal loading is a real possibility. Thermal collapse of the material may thus activate fluid pressurization inside the fault and trigger fault slip. In addition the behaviour of the interface between the intact rock and the cataclastic zone is studied by direct shear tests.

Constrains on the fault system geometry in the Aigion city area from 2D seismic reflection survey. A seismic reflection survey was conducted in the Aigion area in 2001. One of the main objectives was to characterize the deep seismo-tectonic structure of the main fault system. In addition to standard seismic imaging processing, we focused on the best exploitable signal, meaning first arrivals and diffracted waves. A new model of the shallow velocity field up to 500m depth as well as the associated fault system geometry were proposed.

The project has demonstrated the installation of a newly developed fibre optic technology based on fibre Bragg grating (FBG) strain sensors in the research well TRI5 at Trizonia island in Gulf of Corinth area (Greece). The TRI5 well is an ideal place for in-situ investigations of the mechanics of faulting in extension zones, with particular emphasis on the relationship between faulting, earthquakes, and fluids. The project work has shown the viability of using FBG sensors for recording the deformation of the borehole casing caused by fault movements, supported by fibreoptic thermal logging and resistivity array data acquisition. Important scientific results are the design and development of fibreoptic strain measurement cable with down-hole capability, the development of a very practicable set of installation tools (cable roll-off unit, working platform with two levels, rope controlled gripping unit as part of the working platform for lowering the borehole casing segments). After well completion and finalizing the boundary conditions for on-line data acquisition, continuous strain data were recorded with the FBG surface measurement unit over several days. In the last field experiment (December 2003), the repeated strain measurements delivered remarkable changes in the region of the assumed fault zone at about 200m. The present and future benefit of the joint TRI5 sensor instrumentation carried out together with the project partner Schlumberger, France, is the operational availability as permanent down-hole multi-parameter monitoring system (strain, temperature, fluid flow). This plays a key role for interpretation, modelling and proper understanding of the fault zone behaviour. The key innovative feature of this application is to derive general technological conclusions from the permanent sensor installation to establish a concept of intelligent well monitoring, which can be transferred to other research and commercial boreholes.

UPSL carried out a detailed and extensive investigation in the two sides of Corinth gulf, in order to scout and detect the potential sites for both a semi-permanent seismic network and small aperture antennas. Finally 50 sites were selected for the network and four sites for the antennas. Main criteria for the selection of the sites of the seismic network were: - Distance from the planned locations. - Noise level as low as possible. - Geological conditions. - Public power supply if possible. - Accessibility. Concerning the sites for small aperture antennas, main criteria were: - As much as possible plain areas 300 by 300m - Noise level. - As homogenous as possible subsurface conditions. It must be mentioned that considering the hilly morphology of both areas, and the extended private properties of agriculture cultivation, the achievement of this task was quite difficult. Finally four sites were selected, two on each side of the gulf, and as according to the requirements of WP3, TASK 3.3, two sub-perpendicular seismic profiles were deployed on each site to explore the homogeneity of the subsurface geological conditions. This last action is described in detail in the corresponding section. The selected sites, after primary selection, were classified according the above criteria in order to select the 38 most proper ones, keeping the rest as alternative solutions. It must be mentioned that the task of preparing these sites with cemented structures for further investigations in the future, forced us to select sites on public places, and if possible avoiding private properties where the owners would be unwilling to allow a permanent use for the site. Thus preferably places were small chapels in the country, backyards of old deserted schools, etc. Although these places were most convenient for the purpose, it took some time to obtain the final permissions due to delay of approval of various particular applications, to the responsible authorities. UPSL also supervised the site preparation by Landtech Enterprises in order to face any problem that could occur on site, ensuring that the final configuration was convenient to planned task. Since the first week of September 2001, the seismic network of the 38 Earthdata seismographs was installed and continuously recording till Feb 20, 2002. The visits around the network to control the site and stations and to collect data by shifting the removable hard disks were done in time interval of 10 days. Acting like this the possible not recorded time due to several reasons such us temperature out of range, power supply, damages by animals or else, was optimised. Two of these three objectives have been achieved beyond what has been proposed for specific tasks, namely (1) deployment of seismic network, (2) data structure and data organisation. These tasks have been considered as first priority in order to achieve an important recovery of seismic data for a dense uniform network and this change in strategy suggested during the first-year kick-off meeting in Paris/France has required more work than expected for data recording and data organisation. The third objective has been accomplished for the velocity reconstruction part while the attenuation still should be performed by GEOAZUR group. In further European and national collaborations, these datasets will provide certainly other information we have not yet extracted during this project.

A Vertical Seismic Profile (VSP) and several Offset well Seismic Profiles (O-VSP) were planned to be acquired under IFP's coordination. The VSP tool includes 4 components, 3 orthogonal geophones ( Z,X,Y) and a hydrophone. In order to orient the horizontal components, each VSP run was recorded with TWO seismic surface vibrators placed in different positions in the well vicinity and successively activated at each downhole tool station. In order to increase the field flexibility and lower the costs, the VSP acquisition was achieved after removal of the drilling machine. An additional VSP run was initially planned to be repeated after injecting gas in the mud column and demonstrate the attenuation of tube waves, however it had to be cancelled in order to avoid the risk of restarting an uncontrollable water eruption from the well by gas lift effect. Fortunately, the well was not eruptive when the surface valve on top of the 7"5/8 casing was opened to lower the VSP tool. Unfortunately, although the drilling ended 10 days prior to the VSP acquisition, the VSP tool was stopped at a depth of 755m (a probable clay plug naturally built up in a few days) and could not be lowered deeper in spite of multiple attempts: therefore the 750-1000m interval could not be logged and it was decided to run more Offset VSP positions. VSP recording took place by day only in order to avoid disturbing the Aigion city neighbours with repeated vibrator noise during the night.

Specific software has been developed during this project for first-arrival time tomography using both P and S waves. Careful attention has been paid for mixing different parameters as velocity and earthquake location parameters. The forward problem is solved by a finite difference scheme of the Eikonal equation allowing to deduce a ray between earthquake and station. Both travel-time and partial derivative with respect to velocity or location parameters are estimated with great accuracy. The inverse problem is solved by a LSQR iterative procedure after specific preconditioning. Double difference information for neighbouring earthquakes could be added to the data set to be interpreted increasing accuracy in earthquake location.

The installation of the seismic network has been planned with our main collaborator, the Seismological group of Patras University, in such a way that we can re-occupy same positions in the future. This will allow us a differential approach for more accurate and time-dependent results. After the seismic experiment in 1991, we have proceeded in 2002 and we may foresee that a new experiment could be conducted in 2012 or later.

The AiG-10 well has crossed an active fault, the Aigion fault, in the Gulf of Corinth. about 100 meters of core have been extracted on both hanging and footwall of the fault. We scan the cores and made systematic thin-sections (on the 3 directions). They are oriented and borehole images (DSI, FMI) are also available on the same well. More than 200 thin sections are available for the various partners; all have been done twice to allow exchange between the partners (IGP-Strasbourg, Uni of Grenoble, Uni of Montpellier). One may note that a lot of these partners are not official partners of the EEC 3F_Corinth project, they are involved in the study thanks to other funds (especially from the INSU in France). Very precise sampling has been done also in the carbonated veins to characterise the fluids that have precipitated in term of isotopes (C, O). The thin sections show numerous veins with clear kinematical relationship. Some of these veins developed likely during the last N-S opening phase whereas the N140-N180 degree oriented veins could be inherited from the compressive phase of the Hellenides thrusting. The presence of the veins, clearly related to the current extension far away from the fault plane and not included in the 3 cataclastic bands described at the macroscopic scale on the cores has been a surprise. At the microscopic scale, the features related to the extension are present in all the thin sections. Usually the veins are sealed above the fault except the more recent ones. Modelling has been done to quantify the rate of fracture sealing in this context. Results gave value around hundred year for a 1 mm thick fractures.

Four particular sites were investigated on both sides of the Corinthian Gulf. Particularly, two sites on the south side close to Akrata and south of the city of Aegion, and two on the north part, close to village Elea and south of the city of Galaxidi correspondingly. The task of these surveys was to carry out seismic profiles (two on each site sub-perpendicularly orientated) in order to map the principal geological units and faults and also the variation with depth of the seismic velocity field. According to that, the seismic lines were going to be processed both as refraction and reflection lines.

An interesting question which arises in connection with sets of parallel faults is that of fault spacing as observed for example in the Corinth Rift. Faults are shear failures in the crust and many mechanical aspects can be dealt with by Mohr-Coulomb theory of brittle failure. An important limitation of the theory is that it cannot provide information on the distance between neighbouring faults. To explore this subject, we develop here two different mechanical analyses. In a first approach, we have proposed a conceptual model for subsequent faulting in a brittle layer as the result of fault weakening and stress redistribution. It is shown that fault spacing is then controlled by the extension of the unloading zone and we show that the distance between to adjacent fault is proportional to the thickness of the upper brittle layer with a ratio which depends on the softening behaviour of the fault and of the mechanical properties of the layer. A different mechanism for fault spacing based on the assumption that the localisation process of the brittle upper crust is triggered by folding instabilities of the lower ductile base. This folding mechanism will cause periodically in space relative uplift and subsidence. It is assumed that such a periodic deformation of the substratum is sufficient to provide selection of critical faults in the brittle block. In that case fault spacing is controlled by the selected wave length of the buckling mode. Selection of a dominant wave length arises as the result of several competing length scales. The vertical distribution of applied stresses, the mechanical properties of the overburden and the detachment layer, the thickness of the overburden and the substratum, the fluid pore-pressure are the key parameters that control the selection of the instability mode (wave-length with the fastest growth rate).

If a rock has a stress-sensitive permeability, then there is a feedback between fluid flow at the present time and that of the future. We developed a finite element code to predict the pressure response in space and tie to a standard well test. In theory these standard curves could be used to interpret actual well tests in the sister DG-lab project, or in any down-hole application in a well drilled into a hydrocarbon reservoir or deep aquifer.

Network of gravity points was established around the Aigion city in the North Peloponnesos and Trizonia island in 2002/2003. Four campaigns of repeated gravity measurements were carried out around Aigion, and three in Trizonia. Observed gravity changes within this short period of 1 year indicate certain dynamics in the vicinity of the Kerinea fault derived from gravity survey data. Low level dynamic signal was observed also in Trizonia, where the measuring points are located on both sides of the main tectonic feature in the island. Local GPS measurements revealed northwards relative horizontal displacements over the Aigion fault.

A seismic tomography experiment has been conducted from November 2001 until June 2002 in the Aigion area in the Gulf of Corinth in Greece. This experiment has allowed an original reconstruction of P wave and S wave velocity structure in a 20 x 20 x 10km3 zone. This new data collection improves strongly available data in this zone when compared to the available one acquired in 1991 mainly during two months. An output of this project is a velocity structure specified on a grid which could be provided at request for anyone who wants to use it. Both P and S wave velocity grid is available in an UTM coordinate system. Better constraints have been obtained on velocity structures which could be used for improving earthquake location in the future with true 3D location codes as NonLinLoc software. This improvement on earthquake location by providing a 3D velocity structure is an scientific output of this program. New European projects (selected STREP on Corinth) may use these data for further analysis.

The seismic data collected by the French National portable seismic network from February 18th until June 26th is available under a database structure upon request at the present moment. A web interactive request of the data will be available after 2004 for further investigations. Precise Acknowledgment will be asked for using this original dataset.

Gravity survey with GPS support was performed in the area of about 300 km2 around the Aigion city in the North Peloponnesos. The field data were processed with two reduction densities. Various filtering techniques were applied in order to extract useful information. Blocks of positive and negative rock densities were defined from residual anomalies. Besides adjusting the position of known faults, other discontinuities and density boundaries were proposed from gravity indications. 2.5-D gravity modelling contributes to the estimation of the depth of sedimentary cover.

Deliverables

Share this page

Download