MAUMEO Vers finale

3 downloads 0 Views 351KB Size Report
If the role of fluids on earthquake initiation and propagation is now fully recognized (see review in Miller. S.A. 2002 and Fitzenz and Miller 2003), the effect of gas ...
Mechanical effect of the presence of gas on faults modeled as a sandwiched Cam-Clay material MAURY Vincent*, PIAU Jean-Michel**, FITZENZ Delphine*** *IFP School Rueil-Malmaison and Université de Montpellier, France (formerly with Total) **IFSTTAR, formerly Laboratoire Central des Ponts et Chaussées, Nantes, France ***Universidade de Evora, Centro de Geofisica, Portugal

ABSTRACT: The role of fluids on fault stability and earthquake initiation, now undisputed, is studied accounting for several mechanisms : (i) the pore pressure in the fault core may vary according to the tectonic loading (slow process) and/or link with external fluid pressure (slow or rapid process) leading to different and opposite effective stress variation, (ii) the presence of gas in the fluid saturating the fault core material alters its geomechanical properties, cancelling the Skempton's effect (pore pressure variation under volumetric deformation in undrained conditions). Both mechanisms totally change the stress paths on the core fault.

Assuming a poroplastic behaviour of Cam-Clay type for the core fault sandwiched between two elastic pads, we propose a 2D analytical model to appraise the conditions of progressive (stable) or unstable slip displacement and stress evolution, according to the volumetric (contractant or dilatant) deformation of the core fault. The model can be driven by the far-field tectonic displacements and/or far field pore pressure condition. It allows us to evaluate the jumps of displacements and stresses when the instability occurs and addresses the effect of the fluid and in particular of its compressibility which may vary with the gas content. Unexpected effects of gas on instability (earthquake initiation) are shown, even for faults in progressive (stable) plastic deformation. SUBJECT: Earthquake initiation KEYWORDS: fluid flow, gas flow, physical modeling, risks and hazards, rock failure, stability analysis

INTRODUCTION If the role of fluids on earthquake initiation and propagation is now fully recognized (see review in Miller S.A. 2002 and Fitzenz and Miller 2003), the effect of gas is still debated (King and al 2006, Hartmann J et al 2005). Contrary to what was hoped during the seventies after Chinese observations (Heicheng Earthq. St. Del. 1977, Wakita H. 1996), the observation of gas leakage around faults before a large earthquake is far from systematic. That prevents using it as a reliable earthquake precursor, because it could lead to missed earthquakes as well as false alarms. Field observations show that in some cases, during interseismic periods, surrounding zones exhibit dilatancy while the fault core is contractant (due to combined effects of normal and shear stress). This dilatancy in undrained regime for low porosity rocks and/or insufficient hydrogeological water supply is prone to pore pressure falloff, degassing of the fluid of these zones and also the much narrower fault core. Combined to many other possible origins of gas (see in Miller, 2002), this dilatancy of the surrounding zones makes the presence of gas a frequent issue in active fault settings (Kuo et al. 2006) confirmed by numerous observations. This paper reviews some features of the fault zone structure and deformation mechanisms. Main features of fault zones are recalled Part I. The effects of gas on fluid compressibilities and geomechanical properties of porous

rocks are given Part II. The role of fluid and more particularly gas appearance and/or disappearance on the stress paths on faults during interseismic period is presented Part III. To evaluate the occurrence of instability and the effect of gas, an analytical modelling assuming a Cam-Clay type behaviour for the fault core is briefly described and discussed in Part IV. Conclusions and perspectives are presented in Part V. The considerations of this paper would also apply to situations such as landslides, mine roofs, borehole stability. I FAULT ZONE STRUCTURE The geological structure of active faults has been extensively studied in earthquake geology. We will refer here as an example to the simplified model proposed by Tanaka et al. (2001) for the Nojima fault, further studied by Fitzenz et Miller (2003). In this model the following zones are identified : - the fault surface itself (assumed of small thickness), surrounded by one or two highly deformed (broken and crushed) zones not necessarily symmetric, constitutes the fault core (thickness sometimes called the "width" of the fault) can be of metric range or less. A fault core panel is sometimes considered bounded by two low permeability zones, making impervious screens, vertical for strike-slip faults for instance,

- outside the fault core, not necessarily symmetrically, "damaged" or "surrounding zones", between 10 and several hundreds of metre width, - outside of the damaged zones, "virgin zones" in country rocks of low porosities and permeabilities, considered to behave elastically. During interseismic periods, the fault core is submitted to mechanical wear, shear and compaction creep, progressively reducing its permeability, still reduced by recimentation or geochemical processes. From field observations, well before the end of the interseismic periods (Gratier 2010), the fault core is often considered as constituting a true impermeable screen, with an undrained behaviour. Conversely, surrounding zones are considered dilatant, due to a dense pattern of fissures and voids, sometimes well connected and of high permeability, then in drained regime (example of Nojima fault). During the coseismic period (earthquake), the fault core is considered as in dilatant deformation, accompanied by an important increase of the hydraulic conductivity. Surrounding zones may show contractant deformation. The actual fault structure is surely much more complex than presented here (Boutareaud et al 2008, Faulkner et al. 2008 ), but simplifications were made to evaluate the role of some parameters. In the conceptual model we propose hereafter, virgin and surrounding (damaged) zones were set together and assumed without permeability (undrained). Modelling efforts addressing explicitly both the mechanical role of compressible fluids in faulting, but also the role of fluidassisted interseismic processes, and the changes in the hydraulic structure of the fault through time (e.g., Fitzenz and Miller, 2003) are still scarce. Here we want to go further in this direction and investigate in details the effects of the presence of gas in the pore fluid, both on the stress paths and on the stability of faults.

in the liquid, well known for petroleum fluids (William, Mac Cain). In addition, porous rocks may behave in undrained or drained conditions (no exchange or exchange of fluids with outside the volume of rock considered), with all transient states governed by the hydraulic diffusivity of the rock. Compressibilty of rocks in undrained and drained conditions are linked by the Gassman formula:

1 φ(Cfl-Cs) +Cbd-Cs 1 1 1 =K = ≈ + (if Cs ≈ 0)≈ (if Cfl!!Cbd) Cbu φCbd(Cfl-Cs) +Cs(Cbd- Cs) Cbd φCfl Cbd

(here Cfl, for fluid including gas) In case of large (increase of) fluid compressibility, the bulk compressibility and elastic properties of the rock in undrained conditions become close to the drained properties. Thus even in undrained conditions, the very low percentage of gas makes rocks to behave with drained properties. In this "pseudo-undrained regime", the Skempton effect (pore pressure variation ∆Pp function of the effective mean stress increase ∆σ'm) given by the Skempton's coefficient Bs ,with α Biot's coefficient: Bs= (1-Cbu/Cbd)/α =∆Pp/ ∆σ'm ) becomes very small and B s ! C bu /αφC fl (Figure 1 for classical rocks after Rice and Cleary, 1976). Bs Ruhr

Bs Berea

Bs Weber

1,00 0,90 0,80 0,70 0,60 0,50 0,40 0,30

II EFFECT OF GAS ON FLUIDS AND GEOMECHANICAL PROPERTIES OF ROCKS

0,20 0,10

The compressibility of a mixture of immiscible fluids is evaluated as the harmonic average of bulk moduli of 1

constituents

: C fl , mixture = 1 = C w S w + C o S o + C g S g ,

K fl

(named "Wood's approximation" in HC (hydrocarbons) reservoir engineering). Cg varies from 10-1MPa-1(pressure 10MPa, depth 1000m) to 10-2 MPa-1 (at 10000m), in the range of very soft soils compressibilities, much greater than Cw (4.10-4MPa-1) or Co (10-3MPa-1) or the compressibility of stiff rocks. The compressibility of a mixture of liquid and gas therefore increases rapidly as soon as a tiny percentage of gas appears in the liquid, this effect being slightly attenuated when the pressure increases (and Cg decreases). Moreover, it is known that the Wood's approximation underestimates the actual shift (discontinuity) of compressibilities when the first per thousands of gas appear 1

inverse of volumetric moduli of elasticity , subscript b bulk, u undrained, d drained (Cbd is in fact equal to Cbc Zimmerman's notation), s solid matrix, fl fluid, g gas, w water, o oil, φ porosity, S degree (percentage) of saturation

0,00 0

5

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Degree (percentage) of Gas saturation Sg

Fig.1 Skempton's coefficient for Ruhr, Berea, and Weber sandstones versus gas saturation A variation of mean effective stress does not induce significant pore pressure variation, even for rocks in theoretically undrained conditions. This has an important effect on the stress path imposed to the fault core as we are going to see further on. III EFFECT OF FLUID ON THE STRESS PATHS (INTERSEISMIC PERIODS) IIIa Case of fully liquid saturated fluid The effective stress variation (the "stress path") in the fault core depends on its pore pressure evolution that may result from several processes (Fig.2): (a) its deformation; in this case, pore pressure depends on the Skempton's coefficient

of fault core material, (b) a link with an external under or overpressurized fluid source; in this case the pore pressure depends on this external pressure, possibly constant, (c) a combination of both, due to diffusivity of the fault core and adjacent zones. The distinction between these main processes in the field during interseismic periods is surely a difficult but fundamental issue, although hydrogeological and geochemical studies as well as variation of rates may help. The process (b) may be extremely rapid (scale of hours or days), while the process (a) due to tectonic loading of the fault core should be slower (scale of several years) (Cappa et al. 2009) and possibly minimized (even cancelled) due to diffusivity. The distinction between various processes nevertheless appears critical as illustrated Figure 2 in a usual Mohr-Coulomb's diagram (in spite of the plastic dilatancy associated to this criterion, and with compressive stress here as positive as usual in this representation). In process (a) (slow tectonic deformation, cf Point P in Figure 2), assuming a fault with a Coulomb's failure criterion, inactive and working in elastic conditions (Point P), a compressive ∆σn and shear ∆τ tectonic loading can be represented by PR, corresponding to an increase of effective normal stress equal to ∆σ'n= ∆σn –α!P= ∆σn-αBs ∆σn (in case of pore pressure increase due to the Skempton's effect). U' Coulomb Criterion τ R true undrained: Bs ≤ 1

∆τ

R'

U

S P σ’n0

∆σ’n= ∆σn−Bs ∆σn

P'R' Gas disappearance

pseudo-undrained: Bs ~ 0

IIIb Case of fluid containing gas The presence of gas reducing the Skempton's coefficient to ~ 0 in process (a) (Fig. 2), the total stress increase induces a greater effective normal stress ∆σ'n#∆σn corresponding to the stress path PU, with the following consequences: (i) the presence of gas delays the occurrence of rupture, (ii) it increases the shear stress at which the rupture occurs (U'), (iii) according to actual geomechanical properties, and under certain stress conditions, gouge material can compact more easily, and result in geomechanical properties improvement of the fault core (with all the geochemical processes to introduce later). When gas disappears (point P'), a stress path such as P'R' is followed, and the later this disappearance, the smaller the normal stress and shear increase to reach the failure criterion and obtain rupture; these conditions of rupture may recall some observations made around "strong" faults. IV EFFECT OF FLUID (LIQUID AND GAS) ON THE STABILITY

P'

T

scenarii are very common in oil and gas drilling since the usual practice for safety reasons is to impose slight overpressures of the drilling fluid compared to hydrostatic pressure. This is sufficient to trigger shear release and lateral shift of boreholes, even along inactive faults. The same process is responsible for casing shear induced in cased holes by overpressurized fluids chanelling along the cement annuli around the casings. This early and fully liquid fluid supply recalls the conditions sometimes encountered for "weak" faults.

σ'n

∆σ’n~ ∆σn (Bs ∼ 0)

Fig. 2 Stress path PR slow tectonic deformartion, no gas PU slow tectonic deformation, with gas P'R' after gas disappearance ST after rapid linkage with external overpressure Let's assume now (b) a link set up with an overpressurized (liquid) fluid source (process (b) occurring at Point S), rapid compared to the rate of tectonic loading. The shear and normal total stress can be considered as quasi constant, but an increase of pore pressure ∆Pe is induced in the fault core. Therefore ∆σ'n decreases by α∆Pe (stress path ST). The earlier the link with external (liquid) fluid source, and the greater the external pore overpressure, the smaller the shear stress at which the rupture is going to happen. Conversely, the later the link with external fluid source, the greater the shear stress at which rupture occurs, and the smaller the overpressures needed to reach it. Although recognized only in 1989 (Maury et al. 1989, 1996), these

We describe here the 2D model that we developed to get a deeper insight into the initiation of fault instability, with a particular emphasis on the role of gas. The model features the fault core either dilatant or contractant deformation, Skempton's effects, links with external sources of fluids, and change of the pore fluid compressibility due to gas appearance2. The plasticity criterion of the fault core is of Cam-Clay type, able to describe the dilatant or contractant plastic deformation of the gouge material (the Mohr Coulomb criterion might be more appropriate for the representation of faults without any gouge filling or opposite with highly consolidated fault cores, mainly showing dilatant deformation). IV.a 2D-MODEL FOR THE ANALYSIS OF FAULTS BEHAVIOUR INCLUDING UNSTABLE PHASES The model (pad+interface = P+I) of Figure 3 represents a fault core (I) bounded by two pads (P), supposed compact and elastic, representing both the surrounding and virgin zones (cf. sect. I1). The fault core is modelled as a poroplastic interface (its elasticity being neglected), which thickness H can be neglected with regards to that of the pads h . The model is driven by the far-field displacements imposed to the pads U2,V2 (components on Ox, Oy), Sign convention : σ < 0 for compressive stresses, u > 0 for fluid pressure

2

supposed to be contractant and with a dominant shear component: (V2 (t ) < 0 and U 2 (t ) > V2 (t ) ) An external given source of pressure u ext (t ) can also interact with the fault fluid. The model also accounts for initial stress and fault fluid pressure conditions : σ yy i , σ xy i , u i .

[[V! ]] = κ! (2σ ' yy + p' c ) , [[U! ]] = κ!

x

The size of the criterion is supposed to evolve with the jump of the normal displacement V through the

[[ ]]

“consolidation

p ' c ([[V ]]) = p ' e

Using Hooke’s law the displacement-stress relationship for the pads writes:

(V2 −V1 ) i + σ yy , and: h (U − U ) σ xy = µ 2 1 + σ xyi with λ, µ, Lamé's coefficients h

σ yy = (λ + 2µ)

is an end cap model with an elliptical shape in the (σ ' yy , σ xy ) plane (Fig. 4) derived from the 3D Cam-Clay

Fault normal jump Displacement

model, M being the slope of the critical state line of the genuine Cam-Clay model (M=6Sinφ/(3-Sinφ) with φ internal friction of the interface material).

a = α/H, α being (here) the classical exponent of the consolidation laws (not to be confused with the Biot's coefficient); p ' c increases with contractance, corresponding to an improvement of fault core properties. Moreover the effective normal stress is supposed to be related to the total normal stress through Terzaghi’s law: σ yy ' = σ yy + u , with u being the fault fluid pressure, governed by the transient equation: u! + S V! / 2 = R (u ext − u ) in which the right term accounts for the communication with an external source of pressure (depending on the amplitude of the coefficient R ) and where the second left term S V! / 2 accounts for the Skempton effect within the fault. Assuming the fault filled with a soft porous material, S can be given by S~1/CflHφ (Fig.4) IVb MODEL RESPONSE - ONSET OF INSTABILITY The response (displacements, strains, stresses, fluid pressure) of the model (P+I) to any time-history of U 2 ,V2 , u ext is expressed in terms of time-derivatives after analytical calculation. Then the time evolution of the solution itself can be obtained by numerical integration of the time derivatives. The time derivatives calculation shows 4 types of situations, depending on the state of stress σ ' yy , σ xy and conditions upon the parameters of the problem:

σxy

i) elastic evolution (inactive fault) for

[[dV/dt]]

but

f < 0 or f = 0 ,

df / dt < 0 ,

ii) progressive (stable) contractant plastic evolution for f = 0 and − p ' c ≤ σ ' yy ≤ − p ' c / 2 ,

Jump of normal displacement p'c

,

[[ ]]

3 σ xy 2 2 M

Consolidation curve Critical state line Cam-Clay criterion for t=t- [[dU/dt]] Cam-Clay criterion for t=t+

p 'c ([[V ]]) , taken here as:

− a [[V ]] / 2

[[ ]]

The plastic criterion, 2

curve” i c

Fig.3 Fault model: two elastic pads (P) sheared by a Cam-Clay interface (I); antisymetric far field and pads Displacements

f (σ ' yy , σ xy , p ' c ) = σ ' yy +σ ' yy p ' c +

[[ ]]

[[ ]]

Cam-Clay poroplastic interface (I)

h

3 σ xy , M2

u2(t) v2(t)

displacement

u1(t) v1(t)

f also for the flow rule,

with [[U ]] , [[V ]] being the displacement jump components ! being the of the fault core, and κ!f = 0, κ! ≥ 0, f ≤ 0 , κ plastic multiplier. It enables the interface to be either inactive if f < 0 or active if f = 0 , with either contractant or dilatant behaviour ( V! ≤ 0 or V! ≥ 0 ) , depending on the sign of the quantity 2σ ' yy + p 'c .

y Far field Elastic pads (P) (λ,µ,σyy',σxy)

When considering

0 σ'yy

Fig.4 Cam-Clay interface characteristics; criteria for 2 values of p'c and illustration of the flow rule

iii) progressive (stable) dilatant plastic evolution for

f = 0 , − p ' c / 2 ≤ σ ' yy ≤ 0 and κ! being defined (< +∞) and positive ,

iv) the absence of solution for f = 0 , df / dt > 0 when κ! is found infinite or negative (from the consistency equation).

σxy 40 (MPa)

Dilatance (m)

0.6

B(p'c-) P, σ(t -)

30 A(p'c-)

0.4

20 B(p'c+) = σ(t+)

0.2

10 Jump of normal displacement

0 -80

-20 -40 -60 Stress jump from σ(t -) to B(p'c+) Curve of points A(p'c) Curve of points B(p'c)

0

σ'yy (MPa)

Fig.5 Illustration of the stress jump in case of instability The first situation is that of an inactive fault (with

! =V ! = 0 , since the fault elasticity is neglected) U 1 1 when the effective state of stress evolves inside the (current) plastic domain criterion or comes back to it. The second one corresponds to the case of an active fault during interseismic period, submitted to a large (negative) σ ' yy with regards to the shear stress. Then the contractant behaviour of the fault would result from a dominant crushing effect of the core material. The third situation would also correspond to the case of an active fault during interseismic period, but under high shear stress compared to the effective normal stress. In that case the kinematic behaviour of the fault can be explained as for the Mohr-Coulomb criterion, by dominant dilatant mechanisms associated to shear displacements. The fourth case is directly linked to the occurrence of unstable evolution of the structure (P+I). This case only happens when some conditions upon the model parameters are satisfied. The calculation shows that this case is linked to the consistency equation which writes Dκ! = N with : & V! # 6µ N = (2σ ' + p' )$ R(u − u) + (λ + 2µ ) 2 ! + σ U! yy c $ ext h ! M 2 h xy 2 % "

λ +2µ

D = (2σ'yy+p'c )2 (

h

+ S) +

18µ 2 σxy +a(2σ'yy+p'c )σ'yy p'c M4h

Since N is positive for the far field displacement considered here, the problem arises for D being equal to 0 or negative. Actually the condition D ≤ 0 plays the role of a criterion for the onset of instability. It can be shown that the consideration of inertial forces within the pads with the generation of P and S seismic waves makes it possible to find a solution again to the model (P+I) during that phase. For a given value of p ' c such as :

p' c ≥ p' c lim =

and for

6µ ahM 2

& # 2 $1 + M (λ + 2µ + Sh) !! $ 3µ % "

S , such as : ), # µ ,& ahM 2 *$$ p' c −1!! / 2M 2 − 2 − λ / µ ' h *% 6µ ' " 2

S ≤ S lim =

+

(

the condition D ≤ 0 delimits an arch sector A(p'c), B(p'c) along the dilatant side of the Cam-Clay fault criterion (Fig.5). When p ' c decreases until p ' c lim the sets of points A(p'c), B(p'c) draw two convergent curved lines in the (σ ' yy , σ xy ) plane (Fig.5). For p ' c < p ' c lim or S > S lim , the existence of unstable evolutions disappears. Now coming back to the effect of the fluid on the onset of instability, this one is contained in the conditions S ≤ S lim or S > S lim observing that the first one (possibility of unstable evolutions) will rather be obtained for fluids with high compressibility and conversely for the second one (stable evolutions only). It also shows that instabilities may arise even when all the data (stresses, fluid pressure, model parameters) are fixed, except the fluid compressibility due to changes of its composition. This is illustrated on the Figure 6 which shows how the curved domains A(p'c), B(p'c) evolve with increasing values of Cfl (from Cfl1 to Cfl4, points A1 to A4) and can reach an effective state of stress σ (Point P) already located on the fault criterion. These conditions can be interpreted through the unstabilizing role of low Skempton effect within the fault, during very rapid evolution. A1

A2

A3

B

P •

Shear stress

A4

Increasing fluid compressibility

Effective normal stress

Fig.6 Increasing instability domains BA for increasing fluid compressibilities (BA1 for Cfl1 , etc. e.g. increasing gas content). For a state of stress such as P, instability occurs for compressibilities between Cfl3 , and Cfl4 (A3 , A4). The model also enables to predict the new stable situation reached at the end of an unstable evolution, characterized by shear and dilatant displacement jumps, a relaxation of the shear stress, an increase of the effective normal stress and a decrease of the fault fluid pressure (Fig.5). V CONCLUSION AND PERSPECTIVES a) Three main parameters appear as fundamental in progressive or unstable deformation/rupture initiation of faults: the link with an external source of pressurized fluids,

the presence/appearance of gas, the contractant/dilatant behaviour of the fault core, b) according to the state of stress, as illustrated by the Cam-Clay plasticity criterion, the fault core may exhibit contractant or dilatant behaviour at the contact with the failure criterion (plastic deformation), c) when contractant, the fault core compacts progressively, its geomechanical properties being improved, as illustrated by the enlargement of the consolidation pressure, d) when dilatant, the deformation can be progressive (stable) or unstable (rupture): - outside two limits given by the sandwiched model (points A and B, Fig.5), the deformation is progressive, - inside these two limits, the instability occurs (rupture), with a jump in shear stress, effective normal stress, displacement and generation of seismic waves, e) for inactive faults, rapid link with overpressurized fluid source can initiate progressive deformation or unstable slip (rupture), f) for inactive faults or active during interseismic periods, when fluids are invaded by gas, the stress path resulting from tectonic stresses is modified due to the minimized Skempton's effect, inducing larger effective normal stresses than in the absence of gas, allowing the fault core to support larger shear stress, therefore delaying temporarily the occurrence of the plastic deformation (progressive or unstable), g) - in case of liquid saturated fluids, the Skempton effect tends to decrease the pore pressure during rapid dilatant evolution resulting in an increase of effective normal stress, finally prone to stabilize faults and rupture evolution, - in case of fluid containing gas, there is no Skempton effect, no pore pressure decrease and no stabilizing effect on the fault, h) the model shows an interesting feature: all stress, pore pressure and geomechanical properties being constant, a fault in condition of dilatant progressive deformation can switch to instability (rupture) when pore fluids are invaded by gas. Tables of Skempton and mechanical coefficient values for different rock types, a detailed description of the CamClay interface model and case studies using it are beyond the scope of the present contribution, but are the subjects of follow-up papers. The presented model has been easily implemented for numerical simulations. Then interestingly such calculations show that using realistic data (attached for instance to some given rock depth and pad size) the model also predicts realistic results, in terms for instance of fault shear jump displacement during instable events. Such numerical results and comparison with field situations will be also presented in follow-up papers. The observations or not of gas at the surface is a problem of structural geology and gas reservoir engineering. Leakage of gas happens above and around gas reservoirs. Case histories show that gas may migrate underground over very large distances, redissolve, disappear or be stacked in unexpected geological reservoirs. The observation of gas at the surface around an active fault may therefore not be systematical before an earthquake. Nevertheless when gas

appears at the surface it is surely an indicator that: (i) it may have induced very detrimental effects on the stress path, (ii) rupture may be triggered by a very small additional tectonic loading or an external overpressure, (iii) the shear stress values may then be great. REFERENCES Miller S.A. "Properties of large ruptures and the dynamical influence of fluids on earthquakes and faulting", JGR, Vol. 107, N° B9, 2182, [doi : 10.1029/2000JB000032] 2002 King C., Zhang W. and Zhang Z., Earthquake-induced Groundwater and gas Changes, PAGeo 163(2006) 633-645 Hartmann J., Levy, JK Hydrogeological and gas geochemical earthquake precursors - A review for application Nat Hazards 34 (3): 279-304 Mar 2005 Heicheng Earthquake Study Delegation (1977) Eos Trans. AGU, 58, pp 236-272 Wakita H. "Geochemical challenge to earthquake prediction", PNAS, USA, Vol.93, pp. 3781-3786, April 1996 Kuo, MCT., Fan, K. & al. A mechanism for anomalous decline in radon precursory to an earthquake GROUND WATER 44 (5): 642-647 SEP-OCT 2006 Tanaka H., Fujimoto K., Ohtani T., & Ito H., 2001 Structural and chemical characterization of shear zones in the freshly activated Nojima fault, Awaji Island, southwest Japan JGR 106, 8789-8810 Fitzenz D. D. & Miller S. A. "Fault compaction and overpressured faults : results from a 3-D model of a ductile zone" Geophys. Int. J. (2003) 155, pp 111-125 Gratier J.-P. Fault Permeability and Strength Evolution Related to Fracturing and Healing Episodic Processes (Years to Millennia): the Role of Pressure Solution, Oil & Gas Science and Technology – Rev. IFP Energies nouvelles 2010, DOI: 10.2516/ogst/2010014 Boutareaud S., Wibberley A.J. C., Fabbri O., & Shimamoto T., Permeability structure and co-seismic thermal pressurization on fault branches: insight from the Usukidani fault, Japan, Geol. Soc. London Spec. Publ. 2008; v.299; p. 341-361 Faulkner D. R., Mitchell T. M., Rutter E. H. & Cembrano J. On the structure and mechanical properties of large strikeslip faults, Geol. Soc. London, Sp. Pub. 2008; v. 299; p. 139-150 doi:10.1144/SP299.9 William D, McCain JR., The Properties of Petroleum Fluids, 2nd Edition, PennWell Rice, J. R., & M. P. Cleary, Some basic stress-diffusion solutions for fluid-saturated elastic porous media with compressible constituents, Reviews of Geophysics and Space Physics, vol. 14, pp. 227-241, 1976. Cappa F., Rutqvist J., Yamamoto K., 2009. Modelling crustal deformation and rupture processes related to unwilling of deep CO2-rich fluids during the 1965-1967 Matsushiro Earthquake Swarm in Japan. J Geophys Res, 114, B10304, doi:10.1029/2009JB006398 Maury V. & Zurdo C. "Drilling-induced lateral shifts along pre-existing fractures : a common cause of drilling problems » SPE Drilling and Completion March 1996 pp 17-23, presented as paper N°27492 at the IADC/SPE Conf. Dallas, 1993 (first publication at the ISRM/SPE Symposium Rock at Great Depths Pau 1989, Balkema Pub