Hydromechanical behaviour of Faults and Induced Seismicity under CO2 Injection Conditions – FISIC

The central concept of the project is the classical model of a fault consisting of an inner fault core made of fine material, often impermeable, and an outer damage zone that acts as a hydraulic pathway. Faults are therefore highly anisotropic bodies. To better characterize each of these compartments, we propose:
1. To develop a geostatistical approach providing a statistical model for the fault zone. This will be based on real data collected in site on analogues of fault zones in carbonate contexts. In this setting, different parts of a fault zone will be modeled in a “cluster” form. The advanced meshing tools will be developed for converting complex geometries to finite element models;
2. To design and perform specific innovative experiments to assess the chemical effects of the CO2 storage on the behavior of fault materials (fractured zone and gouge);
3. To develop constitutive models describing the behavior of different parts of the fault zone model (fault core and damage zone);
4. To develop numerical models taking into account the heterogeneous nature of the fault zone on its behavior.

Geological characterization
The selected site for geological investigation of analogues of fault zones is located in Navacelles (Southern France). Though no injection is planned in this region, this site can be considered a good analogue of a CO2 storage reservoir, because it is located in a low-deformed tectonic setting, i.e. it might a priori present very few fractured / faulted zones. However, the in site investigations have revealed a diversity of complex reservoir-scale fault structures, which does not correspond to the traditional vision (model) of a fault as a plane surface. The on-going studies would allow getting better insight in the hydro-mechanical behavior of such reservoir-scale systems, which can hardly be characterized through geophysical surveys.

Experimental studies
The effect of fluid chemistry on the slow propagation of cracks in calcite single crystals was investigated. Time-lapse images and measurements of force and load-point displacement allowed accurate characterization of crack velocities. The results show two regimes depending on the salinity: weakening conditions where the crack propagation is favored, and strengthening conditions where crack propagation slows down. The future lab experiments should focus on the joint effect of fluid composition and of the CO2.

The ultimate goal of the present project is to provide appropriate theoretical and numerical models for the accurate evaluation of the behavior of existing faults.

The complex relationhsip between fluid salinity and subcritical (slow) fracture propagation in a calcite has been published in Tectonophysics (Rostom, F., Røyne, F., Dysthe, D. K., and Renard, F. (2013) Effect of fluid salinity on subcritical crack propagation in calcite, Tectonophysics, 583, 68-75, doi : 10.1016/j.tecto.2012.10.023). This publication is th estrating point for studying the effect of CO2.

The main goal of the present project is to provide appropriate models for the accurate simulation of the behaviour of the fault zones in CO2 storage conditions. Underground engineering works, generally, and deep well injection more particularly change the stress state in the geological formation. The change of stress field affects, in one hand, the hydraulic response of the fractured rock and in the other hand may move the fault system toward failure. Geochemical interactions between the injected CO2 and fault structure affect the fault behaviour too.
In the framework of the present project, we consider a fault zone as a set of rock joints (i.e. fractures) distributed statistically around the fault core. The hydromechanical response of a fault can thus be modelled as that of a set of fractures combined with the fault core. In this framework, we consider that the permeability of the central part of the fault (i.e. gouge material) is assumed to be small regarding the permeability of fractured zone.
A complete framework from field observation to large scale fault modelling is proposed. To develop the fault model a set of experimental and theoretical developments are planned. A probabilistic approach will be incorporated to model the mechanical behaviour of rock joints in a network and then to upscale the global parameters (metric to decametric scale) by defining a statistically representative volume element (SRVE). The chemical effects will be accounted for through a set of state variables affecting the fracture toughness of the rock or provoking sub-critical propagation of the cracks. The results will be recast in the framework of hydro-chemo-mechanical constitutive laws that will be integrated in a numerical model. Finally, a large scale fault model based on an equivalent continuum will be developed. The developed model will be able to evaluate the initiation of a rupture on the fault zone as a precursor of the induced seismicity.
The FISIC project aims at improving the existing models by developing significant scientific innovative approaches. These improvements can be listed as follows:
• Considering a fault zone as a complex heterogeneous structure and not a linear object
• Development of a geostatistical approach for simulating the fault structure
• Experimental identification of chemical effects on the propagation probability of a dense fracture network in different scales
• Developing innovative numerical techniques based on a specified probabilistic non-local damage model. This procedure renders possible an automatic design of a numerical model from the fracture networks randomly constructed and facilitates the topological changes which are associated to the fracture propagation. The numerical tests will be performed using this technique to study the overall behaviour of joint networks in different scales.
• Probabilistic approach for Up-scaling taking into account the heterogeneity of fault zone on its mechanical, chemical and hydraulic behaviour.