a fully coupled hydro logical (1)

Upload: angel-ramiro-rojas-perez

Post on 07-Apr-2018

213 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    1/10

    A fully coupled method to model fracture permeability change in naturally

    fractured reservoirs

    Qingfeng Tao n, Ahmad Ghassemi, Christine A. Ehlig-Economides

    Harold Vance Department of Petroleum Engineering, Texas A&M University, College Station, TX, USA

    a r t i c l e i n f o

    Article history:

    Received 26 December 2009

    Received in revised form2 August 2010

    Accepted 27 November 2010Available online 18 December 2010

    Keywords:

    Naturally fractured reservoirs

    Fracture permeability change

    Effective stress

    Joint deformation

    Displacement discontinuity method

    a b s t r a c t

    We present a newapproachto model thefracturepermeability change in naturallyfractured reservoirsby

    combining a finite difference method (FDM) for fluid diffusivity equation in a fracture network, a fully

    coupled displacement discontinuitymethod (DDM)for theglobal relation of fracturedeformation andtheBartonBandis joint deformation model for the local relation of fracture deformation. The method is

    applied to naturally fractured reservoirs under isotropic in-situ stress conditions and anisotropic in-situ

    stress conditions, respectively. The application of the new approach shows that the fracture permeability

    decreases with the pressure depletion under isotropic in-situ stress condition. Under highly anisotropic

    stress, it is possible that shear dilation can increasefracture permeability evenas pore pressure decreases

    with production.

    Published by Elsevier Ltd.

    1. Introduction

    In many reservoirs, fractures are the main flow channels, and

    the matrix provides the main storage capacity. Somereservoirs, e.g.tight gas reservoirs, are not possible to produce without the

    existence of natural fractures (microfractures). Therefore, the

    fracture permeability is critical to hydrocarbon production.

    The dependence of formation permeability on pressure for a

    single porosity system has been investigated [112]. The pressure

    dependence of matrix permeability occurs as the porosity and

    connectivity of pores decrease with the increase in effective

    stress. But the permeability change in tight gas reservoirs mainly

    results from the closure of microcracks with the increase of

    effective stress [13].

    Generally fractures are more deformable than the matrix in a

    naturally fractured reservoir, and the permeability of fractures, not

    the matrix, dominates the flow behavior. Furthermore, fractures

    are more sensitive to pressure and stress change than the matrix,and the fracture deformation mechanism is much more compli-

    cated than matrix deformation. The effect of stress on the aperture

    and permeability of a single fracture has been well investigated in

    laboratories [1417]. Experimental data show a nonlinear relation

    between normal stress and fracture closure. Bandis et al. [16]

    presented a hyperbolic formula to represent the normal stress

    fracture closure relation. For shear deformation the experimental

    data show an approximately linear relation between shear stress

    and shear displacement before yielding, and then show a compli-

    cated relation after yielding. Shear deformation can also induce

    fracture opening as the opposed asperities of a fracture slide over

    each other and cause an increase in aperture.In naturally fractured reservoirs, there are coupled interactions

    between porous matrixand fluid, as well as between fractures. Biot

    [18,19] developed a theory of poroelasticity for porous media

    saturated with incompressible fluid to account for the coupled

    diffusiondeformation mechanism. Rice and Cleary [20] extended

    the theory for porous media saturated with compressible fluid.

    Biots theory of poroelasticity is a continuum theory for a porous

    medium consisting of an elastic matrix containing interconnected

    fluid-saturated pores. The fluid diffusion in porous media induces

    porous matrix deformation and stress redistribution, and porous

    matrix deformation also induces fluid flow and fluid pressure

    redistribution. If there is a discontinuous surface (fracture) in the

    continuum porous medium, the deformation of the fracture (open-

    ingor closing) will inducethe deformationof theporous matrixandalso pore pressure change and fluid flow.

    In addition to the interactions of fluid, porous matrix and

    fracture, there are interactions between fractures including

    mechanical deformation and fluid flow. One fracture deformation

    will cause stress change in the field and induce deformation of

    other fractures [2123]. The fluid injection or production from one

    fracture can also induce fluid pressure change in other fractures, as

    well as mechanical deformation. Curran and Carvalho [21], Cheng

    and Predeleanu [22] and Carvalho [23] developed a poroelastic

    DDM for fluid-saturated porous media with many discontinuous

    surfaces (fractures) in it. The poroelastic DDM can be applied to

    model the coupled interactions of fractures, porous matrix and

    Contents lists available at ScienceDirect

    journal homepage: www.elsevier.com/locate/ijrmms

    International Journal ofRock Mechanics & Mining Sciences

    1365-1609/$ - see front matter Published by Elsevier Ltd.

    doi:10.1016/j.ijrmms.2010.11.012

    n Corresponding author.

    E-mail address: [email protected] (Q. Tao).

    International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268

    http://-/?-http://www.elsevier.com/locate/ijrmmshttp://dx.doi.org/10.1016/j.ijrmms.2010.11.012mailto:[email protected]://dx.doi.org/10.1016/j.ijrmms.2010.11.012http://dx.doi.org/10.1016/j.ijrmms.2010.11.012mailto:[email protected]://dx.doi.org/10.1016/j.ijrmms.2010.11.012http://www.elsevier.com/locate/ijrmmshttp://-/?-
  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    2/10

    fluid in porous media with fractures. This methodhas been applied

    to simulate the hydraulic fracturing in continuum porous media

    [24]. But the poroelastic DDM has not been applied to model

    the interactions of fracture, porous matrix and fluid in fractured

    porous media. Asgian [25] applied an elastic DDM to model the

    deformable fractured reservoirs and assumed the matrix is

    impermeable, thus, ignored the fluid flow between matrix and

    fracture, which is usually the most important contribution to the

    hydrocarbon production.In this study, we present a new approach to model the fracture

    permeability change by combining a finite difference method for

    solving the coupled fluid diffusivity equation in a fracture network,

    a displacement discontinuity method (DDM) basedon Biots theory

    of poroelasticity and the nonlinear BartonBandis joint deforma-

    tion model. If there is no cooler fluid injection into the reservoir to

    maintain reservoir pressure and enhance the hydrocarbon produc-

    tion, the reservoir temperature will be constant. As a result, the

    thermal effect is not considered in this method.

    2. Fully coupled displacement discontinuity method

    Based on Biots theory of poroelasticity, Carvalho [23] gave the

    fundamental solutions of induced stress and pore pressure for afinite thin fracture segment with a fluid injection/production

    source in an infinite two-dimensional homogeneous and isotropic

    porous medium saturated with a compressible single-phase fluid.

    The induced stress and pore pressure by a single long fracture or

    many fractures with fluid injection/production can be obtained by

    discretizing the fracture or fractures into Nfracture segments and

    summing the influences of all N fracture segments.

    2.1. Constitutive equations of a porous medium saturated with a

    compressible single-phase fluid

    The relation of stress to strain and pore pressure for a linear

    isotropic poroelastic medium is given by Biots theory of poro-

    elasticity [18]

    sij 2G eij diju

    12u ekkh i

    dijap 1

    where s and e are stress and strain, respectively, G and u are shearmodulus and Poissons ratio, ekk is the volumetric strain, i and j are

    either x, y, or z and dij is the Kronecker delta. Combining a static

    force balance equation with Eq. (1) and ignoring the body force,

    yields the Navier equation of poroelasticity

    Gui,kk G

    12nuk,kiap,i 0 2

    The total volumetric deformation (ekk) consists of pore space

    change (zp) and the solid grain deformation (zs). The solid grain

    deformation is dueto fluidpressure andeffective stress loading: (i) the

    effect of fluid pressure (the compression stress or strain is negative)

    Bs1 p

    Ks1f 3

    (ii) the effect of effective stress loading

    Bs2 sukk3Ks

    4

    where Ks is the bulk modulus of the solid grains and f is the

    porosity. The average effective stress (skk0/3) has the followingrelation with the volumetric strain and pore pressure [23]

    sukk3

    suxx suxx suxx3

    Kmekk KmKs

    p 5

    where Km is the bulk modulus of the whole system including fluid

    andsolidgrains.Combining Eqs. (3)and (4), andsubstituting Eq.(5)

    yield the solid grain deformation

    Bs KmKs

    ekk p

    Ks

    KmKs

    1f !

    6

    Thepore space change is obtained by subtracting thesolid grain

    deformation from the total volumetric strain and using the defini-

    tion of Biots coefficient (a 1Km=Ks)

    Bpaekk

    af

    p

    Ks 7

    2.2. Fluid diffusion equation in the porous media

    Thefluidmass balance equation gives that thenet fluid flowrate

    is equal to the sum of the increase of fluid mass in the pore space

    and injected/produced fluid

    rfqm,i @rfVf

    @trfqs 8

    where rfis the fluid density, qm is the fluid flow rate in matrix, Vf ispore volume, qs is the production or injection rate and tis the time.

    The fluid is compressible and the fluid density is pressure depen-

    dent

    @rf@p

    cfrf 9

    where cfis thefluid compressibility. In a unit volumeporous media,

    the pore space is f, and the pore space change is zp, and Eq. (8) is

    rewritten as

    rfqm,i f@rf@t

    rf@Bp@t

    rfqs 10

    Darcys flow is assumed

    qm kmp,i 11

    where k is the matrix permeability and m is the fluid viscosity.Substituting Eqs. (7), (9) and (11) into Eq. (10), neglecting the term

    with @p=@xi2

    [26] and assuming small change in the pore volumeyield

    p,ii

    m

    kM

    @p

    @t am

    k

    @ekk@t

    mqsk

    12

    where 1=Mfcf af=Ks.

    2.3. Fundamental solutionsfor a single fracture segment in an infinite

    two-dimensional porous medium

    For a plane strain condition, there is a constant discontinuity in

    the media and also constant flow rate (injection or production)

    along a thinfracturewitha lengthof 2a from t0 (Fig.1). Theinitialconditions aredefined in Eq.(13) andthe inner andouterboundary

    conditions are defined in Eqs. (14) and (15). Since only the induced

    solutions for changes in stress, displacement and pore pressure are

    needed, the initial values of stress, displacement and pore pressure

    are set as zero. The initial conditions are given by

    at t 0, forall x,y : p 0, ux uy 0, sxx syy sxy 013

    The boundary conditions at the inner boundary are given by

    at y 0, for xj jra : uxx,0uxx,0 Ds, uyx,0uyx,0 Dn, qs 2aq0 14

    where q01 m3/s. The outer boundary condition is

    as ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffix2 y2q -1 : ux uy 0, sxx syy sxy p 0 15

    Q. Tao et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268260

  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    3/10

    Using the initial and boundary conditions, Eqs. (2) and (12) can

    be solved for separate inner boundary conditionsconstant

    volume flow rate injection/production (ux(x, 0)ux(x, 0+)0,uy(x, 0

    )uy(x, 0+)0, qs 2aq0) at the inner boundary andconstant displacement discontinuity (DD)(ux(x, 0

    )ux(x, 0+)Ds,uy(x, 0

    )uy(x, 0+)Dn, qs0) at the inner boundary [23]. Theinduced displacement, pore pressure and stress at any point (x, y)

    andtime tby the constant volume injection/production rate and by

    the displacement discontinuities including normal and shear

    displacement discontinuities through the fracture segment are

    given in Appendix A and Appendix B, respectively [23]. The final

    fundamental solutions for poroelastic DDM are obtained by

    combining the solutions of the constant volume rate fluid injec-

    tion/production and the constant displacement discontinuities in

    the fracture segment.

    Induced pore pressure

    px,y,t pdnx,y,t Dn pdsx,y,t Ds pqx,y,t qint 16

    Induced displacement

    uxx,y,t udnx x,y,tDn udsx x,y,tDs uqxx,y,tqintuyx,y,t udny x,y,tDn udsy x,y,tDs uqyx,y,tqint 17

    Induced stress

    sxxx,y,t sdnxx x,y,tDn sdsxxx,y,tDs sqxxx,y,tqintsyyx,y,t sdnyy x,y,tDn sdsyyx,y,tDs sqyyx,y,tqint

    sxyx,y,t

    sdn

    xy x,y,t

    D

    n sds

    xyx,y,t

    D

    s sq

    xyx,y,t

    q

    int 18

    where Dn and Ds are the normal and shear displacement disconti-

    nuity sources, and qint is the fluid source term in a fracture

    (interface flow rate between fracture and matrix), and the super-

    scripts dn, ds and q denote normal displacement discontinuity

    source, shear displacement discontinuity source and fluid source,

    respectively. The induced pore pressure, pq, displacement in x

    direction, uxq and in y direction, uy

    q, stress components, sxxq , syy

    q and

    sxyq by the constant rate fluid injection/production from a fracture

    segment are listed in Appendix A. The induced pore pressure, pdn

    and pds, displacement in x direction, uxdn and ux

    ds, and in y direction,

    uydn and uy

    ds, stress components, sxxdn, syy

    dn, sdnxy , sxxds, syy

    ds and sxyds by the

    constant normal and shear discontinuous displacement of a

    fracture segment are listed in Appendix B.

    3. Joint deformation model

    In this study, we only consider those fractures with two rough

    surfaces to contact each other and the fluid flow can flow through

    the void spaces between them as illustrated in Fig. 2. The fracture

    can deform normally and laterally to the fracture surface as the

    stress acting on it changes.

    Bandis et al. [16] presented a hyperbolic model for the normal

    deformation of fracture based on a large body of experimental data

    sun Dn

    aabDn19

    wheres0n is theeffective normalstress and Dn is thenormal closureof fracture; aa and b are constants and related with the experi-

    mentally determined parameters initial normal stiffness (Kni) and

    the maximum possible closure (Dnmax) as aa1/Kni and aa/bDnmax. Eq. (19) can be rewritten by substituting Kni and Dnmaxfor aa and b

    sun KniDn1Dn=Dnmax

    20

    The normal stiffness (Kn) is therefore derived as follows as a

    function of Dn or s0n

    Kn @sun@Dn

    Kni1Dn=Dnmax 2 21

    or

    Kn Kni1sun= KniDnmax sun 2 22

    The change of shear stress has a linear relationship with the

    change of shear displacement before yielding

    Dss KsDDs 23where Ks is the shear stiffness. The fracture slips when the shear

    stress exceeds the shear strength (sc) of the fracture. The tworough surfaces slide each other and cause an increase in apertures,

    which is known as shear dilation [17]. We used an approximate

    linear relation for the aperture increase (DDn-dilation) due to shear

    movement

    DDndilation DDstanfd 24where fd is the dilation angle.

    4. Fluid flow in the fracture network

    The aperture of real fracture varies in space (Fig. 2) and the fluid

    flow inside is also very complicated due to the rough surfaces. But

    Darcys law is still valid and the rough fracture can be represented

    by a fracture with the average fracture aperture, which has been

    verified by Witherspoon et al. [27]. The fluid balance equation in

    x

    y

    Dn+

    Ds+

    qs

    Fig. 1. A thin line fracture in an infinite two-dimensional elastic porous medium,

    and the line fracture starts from (a,0) and ends at (a,0).

    Fig. 2. A fluid filled fracture subject to normal and shear stress.

    Q. Tao et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268 261

  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    4/10

    the fracture includes theflow from theconnected fractures andthe

    interface flow from the connected matrices

    rfqf,i @ rfwfDL

    @tDLrfqintrfqs 25

    where qf is the flow rate in the fracture per unit formation

    thickness, qint is the interface flow rate per fracture length per

    unit formationthickness,DL is thelength of fracture segment andqs

    is theproductionrate per unit formationthickness. Theflow rate inthe fracture can be obtained by using Darcys law

    qf kf wfm

    rp 26

    where kf is the fracture permeability and calculated from the

    fracture aperture using the Cubic law and wf is the fracture

    aperture. Combining Eqs. (9) and (26), and ignoring the second

    order of the small term p,i [26], the net fracture flow rate is

    rfqf,i rfkfwf

    mp

    ,ii 27

    The fluid mass change in the fracture includes two parts, one is

    due to fracture volume change and another one is due to fluid

    density change. The fracture volume change is mainly from thefracture aperture change

    @Vf@t

    DL @wf@t

    28

    Eq. (28) can be rewritten by substituting Dn for wf

    @Vf@t

    DL @Dn@t

    29

    The fluid mass change due to fluid density change is

    @m

    @t wfDL

    @rf@t

    30

    Substituting Eq. (9) into Eq. (30) yields

    @m

    @t cfrfwfDL

    @p

    @t 31

    Combing Eqs. (25)(31) yields the fluid diffusion equation in

    fracture network

    kfwfm

    p,ii wfDLcf

    @p

    @tDL @Dn

    @tDLqintqs 32

    5. Numerical implementation

    The fractured reservoir is treated as a fracture network in a

    porous medium saturated with a compressible single-phase fluid.

    As in dual-porosity models, the fracture network provides the main

    flow channels and the porous media provides the main storage

    media. On production, the fluid flows from matrixto fractures, then

    in fractures to the well. The fracture network is discretized intosmall fracture segments and the change of aperture and perme-

    ability of every fracturesegment withproduction aredeterminedby

    combining the BartonBandis joint deformation model, the DDM,

    and a FDM solving the diffusivity fluidequation in fracturenetwork.

    5.1. Local relation between stress and displacement to fracture

    deformation

    Forany fracture in thefracturenetwork (Fig.3), the deformation

    must comply with the fracture deformation model. The relation

    between effective normal stress change Dsn0 and normal displace-

    ment DDn of the ith fracture segment is

    Dsnui

    Ki

    nDDni

    33

    The normal stiffness Kn is a coefficient that is dependent on the

    fracture. The effective stress (tension is treated as positive) is

    defined as

    snu sn ap 34where a 1Km=Ks. For a fracture, when the bulk modulus ofsystem Km is much less than the solid bulk modulus Ks, the Biot

    coefficient becomes unity, and the effective stress is given by

    snu sn p 35

    Substituting Eq. (35) for effective stress in Eq. (33) yields (for

    each fracture segment)

    Dsi

    n Dpi KnDDi

    n 36The relation of shear stress change Dss and shear displacement

    DDs is

    Dsi

    s Ki

    sDDi

    s 37The shear stiffness is a constant before yielding and reduces to

    zero after yielding. The normal deformationDDn-dilationdueto shear

    dilation is

    DD

    i

    ndilation DDi

    s tanfd 38Eq. (36) must be rewritten when the normal deformation

    induced by shear dilation is considered

    Dsi

    n Dpi Ki

    n DDi

    n DDi

    stanfd

    39

    5.2. Global relation between stress and displacement to fracture

    deformation

    In the fracture network with m fracture segments, there are

    interactions among fractures. The stress change of the ith fracture

    segment is influenced by the deformation of all the fracture

    segments in the system. The dependence of change of normaland shear stresses ofith fracture segment on the normal and shear

    deformation,and theinterface flowrateof allfracturesegmentsare

    derived from Eqs. (16) and (18) [28]

    Dsi

    n Xm

    j 1A

    ij

    DDj

    n Xm

    j 1Bij

    DDj

    s Xm

    j 1Cij

    qj

    int

    Dsi

    s Xm

    j 1Eij

    DDj

    n Xm

    j 1Fij

    DDj

    s Xm

    j 1Kij

    qj

    int

    Dpi

    Xmj 1

    Lij

    DDj

    n Xm

    j 1Hij

    DDj

    s Xm

    j 1Nij

    qj

    int 40

    whereAij

    , Bij

    , Cij

    , Eij

    , Fij

    , Kij

    , Lij

    , Hij

    , and Nij

    are the influence coefficients ofjth

    fracture element on the ith fracture element.

    n

    s

    Fig. 3. Local relation of fracture deformation.

    Q. Tao et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268262

  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    5/10

    The DDM solutions are both space and time dependent, and the

    fundamental solutions are based on constant displacement dis-

    continuities and constant interface or source flow rates. However,

    for practical applications, the displacement discontinuities and

    interface flow rates are timedependent. The timemarchingscheme

    shown in Fig. 4 is used to allow source strengths (the displacement

    discontinuities and interface flow rate) to change with time.

    Starting each boundary integration from an initial homogeneous

    status avoids the needfor volumetricintegration [23]. Therefore,allthe previous increments of source strengths must be included

    while numerically integrating the effect of source strengths at each

    time step. The induced stress and pore pressure on the ith fracture

    segment by the increments of source strengths are

    Dsi

    nt Xm

    j 1A

    ij

    ttxDDjx

    n Xm

    j 1Bij

    ttxDDjx

    s Xm

    j 1Cij

    ttxDqjx

    int

    Xx1

    h 0

    Xmj 1

    Aij

    tthDDjh

    n Xx1

    h 0

    Xmj 1

    Bij

    tthDDjh

    s Xx1

    h 0

    Xmj 1

    Cij

    tthDqjh

    int

    Dsi

    st Xm

    j 1Eij

    ttxDDjx

    n Xm

    j 1Fij

    ttxDDjx

    s Xm

    j 1Kij

    ttxDqjx

    int

    Xx1

    h 0 Xm

    j 1Eij

    t

    th

    DD

    jh

    n

    Xx1

    h 0 Xm

    j 1Fij

    t

    th

    DD

    jh

    s

    Xx1

    h 0 Xm

    j 1Kij

    t

    th

    Dq

    jh

    int

    Dpi t

    Xmj 1

    Lij

    ttxDDjx

    n Xm

    j 1Hij

    ttxDDjx

    s Xm

    j 1Nij

    ttxDqjx

    int

    Xx1

    h 0

    Xmj 1

    Lij

    tthDDjh

    n Xx1

    h 0

    Xmj 1

    Hij

    tthDDjh

    s Xx1

    h 0

    Xmj 1

    Nij

    tthDqjh

    int

    41where DD

    jx

    n, DDjx

    s and Dqjx

    int are the source strength increments for

    the jth fracture segment at the current time step, x; DDjh

    n, DDjh

    s and

    Dqjh

    int are the previous source strength increments of for the jth

    fracture segment at time step h, which indexed from 1 to x1.

    Aij

    t

    th

    , B

    ij

    t

    th

    , C

    ij

    t

    th

    , E

    ij

    t

    th

    , F

    ij

    t

    th

    , K

    ij

    t

    th

    , L

    ij

    t

    th

    , H

    ij

    t

    th

    and N

    ij

    tth arethe influence coefficients ofjth fracture element onthe ith fracture element at time step h.

    5.3. Solution method

    Discretizing the Eqs. (37) and (39) for the local relation of

    fracture deformation in time, and combining the local and global

    relations yields

    pi t

    Xmj 1

    Aij

    ttxDDjx

    n Ki

    nDDix

    n Xm

    j 1Bij

    ttxDDjx

    s Ki

    ntanfdDDix

    s

    X

    m

    j

    1

    Cij

    ttxqjx

    int Xx1

    h

    0 X

    m

    j

    1

    Aij

    tthDDjh

    nXx1

    h

    0 X

    m

    j

    1

    Bij

    tthDDjh

    s

    Xx1

    h 0

    Xmj 1

    Cij

    tthqjh

    intKi

    n

    Xx1h 0

    DDih

    n tanfdXx1

    h 0DD

    ih

    s

    pi 0

    Xmj 1

    Eij

    ttxDDjx

    n Xm

    j 1Fij

    ttxDDjx

    sKi

    sDDix

    s Xm

    j 1Kij

    ttxqjx

    int

    Xx1

    h 0 Xm

    j 1

    Eij

    tthDDjh

    n Xx1

    h 0 Xm

    j 1

    Fij

    tthDDjh

    s Xx1

    h 0 Xm

    j 1

    Kij

    tthqjh

    int

    Ki

    s

    Xx1h 0

    DDih

    s

    pi t Xm

    j 1Lij

    ttxDDjx

    n Xm

    j 1Hij

    ttxDDjx

    s Xm

    j 1Nij

    ttxqjx

    int

    pi 0Xx1

    h 0

    Xmj 1

    Lij

    tthDDjh

    nXx1

    h 0

    Xmj 1

    Hij

    tthDDjh

    s

    Xx1

    h 0

    Xmj 1

    Nij

    tthqjh

    int 42

    The diffusivity equation, Eq. (32), is discretized in space and

    time for a given fracture network using an implicit finite differencemethod. For the ith fracture segment at the time step, x

    Xmj 1

    Cpij

    pj tDLDD

    ix

    n DLDqix

    int wfDLcfpi txDL

    Xx1h 0

    Dqih

    intXx

    h 0qih

    s

    43where Cp

    ij

    is the fluid coefficient matrix [28]. The production rate

    from ith fracture segment qih

    s is also discretized in time in Eq. (43).

    All left terms in Eqs. (42) and (43) are unknown and all right terms

    areknown. When the production rate andinitial reservoirpressure

    are given, the normal and shear fracture displacement, interface

    flow rate, and fluid pressure can be obtained by solving the linear

    equation Eqs. (42) and (43).

    6. Applications

    6.1. Fracture permeability change under isotropic in-situ stress

    conditions

    In this section the fracture permeability change during produc-

    tion and its effect on transient wellbore pressure are investigated

    for a well with constant production rate (2 m3/day) from a unit

    reservoir thickness (1 m) in formation with a fracture network

    consisting of two sets of orthogonal vertical fractures surrounded

    by an effectively infinite porous medium as in Fig. 5. Only two-

    dimensional flow and deformation are considered, and change in

    the vertical direction is ignored. The in-situ stress field before

    production is assumed to be isotropic with compression set to

    21 MPa. To better illustrate the geomechanic effects during pro-

    duction, the reservoir pressure is set very close to the in-situ stress

    at 20 MPa. The two joint parameters, initial normal stiffness and

    maximum closure, characterizing the normal deformation of

    fracture are 0.5 GPa/m and 0.8 mm, respectively. The nonlinear

    relationship between effective normal stress under compression

    and fracture closure is shown in Fig. 6. The fracture aperture at the

    initial condition (zero effective normal stress) is assumed as

    0.8 mm. Thefractureapertureunder theinitial in-situ stress before

    production is assumed as 0.229 mm for all fractures. Other para-

    meters are listed in Table 1. Because the fracture permeability

    dominates the reservoir permeability, changes in matrix perme-

    ability are neglected and assumed as constant during production.

    (

    xj,yj,

    )

    (xj, yj, 0)

    (xj, yj, )

    0 1 t

    (xj, yj, 1)

    Fig. 4. Time marching scheme, w represents Dn, Ds or qint [23].

    Q. Tao et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268 263

  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    6/10

    Fig. 7 shows the reservoir pressure distribution after 360 days

    on production. In this case, thelowest pressure is 14.2 MPa, andthe

    highest pressure is 18.6 MPa. Fig. 8 shows that the fracture

    permeability declines with production, and changes from 4428

    to 280 D at the well, and from 4428 to 1920 D at the boundary. The

    slope ofthe trend in Fig. 6 gives the normal fracture stiffness, which

    changes with the effective stress, from a small value at small

    effective stress to a rapidly increasing value at high effective stress.

    As such, the fracture is more deformable when the reservoir

    pressure is close to the in-situ stress than when there is a large

    contrast between them. To study the influence of a higher stiffness,

    considerthe sameinitialfracture aperture of 0.229 mm and fracture

    permeability of 4428 D before production, but set the initial in-situ

    stress to a value that increases the effective stress while all other

    properties remain same. Fig. 9 shows the fracture permeabilitychange at the well for different effective in-situ stress conditions.

    The influence of production on the fracture permeability change

    strongly depends on the initial effective stress condition, and

    decreases rapidly with increase in the effective in-situ stress. The

    fracture permeabilityonly reduces 3.5%of the initial permeabilityof

    4428 D for the case with an effective in-situ stress of 10 MPa.

    However fracturepermeability loss for the case with an effective in-

    situ stress of 1 MPa is 93.7% of the initial permeability.

    The weakness of fracture is critical to the influence of production

    on fracture permeability change. If all other conditions are the same,

    the weaker the fracture, the more fracture permeability reduction.

    The initial normal stiffness and maximum closure characterize the

    normal fracture deformation. For the above reservoir and production

    conditions, the fracture permeability change is simulated for threedifferentfractures withdifferent kniand Dnmax (weakest: kni0.8 mm,Dnmax0.5 GPa/m; medium: kni0.357 mm, Dnmax5 GPa/m; stron-gest: kni0.247 mm, Dnmax50 GPa/m). For the three kinds offractures, all fracture apertures at the normal effective stress of

    1 MPa are 0.229 mm. As a result, they have the same initial fracture

    permeability of 4428 D. The fracture aperture reduces with the

    increase of the effective normal stress. The weakest fracture has

    the maximum reduction and the strongest one has the minimum

    reduction (Fig. 10). Fig. 11 shows the comparison of the fracture

    permeability change with time for the three different kinds of

    fracture. For the weakest fracture, the fracture permeability of the

    fracture at the well decreases by 93.7% of the initial permeability,

    while the fracture permeability for the strongest fracture only

    decreases by 21% of the initial permeability.

    6.2. Fracture permeability change under high anisotropic in-situ

    stress conditions

    We consider anisotropic in-situ stress conditions in this case.

    The shear deformation of a fracture is approximately linear before

    yielding and is treated as linear here, as is characteristic of a

    constant shear stiffness value. The shear stiffness is abruptly

    reduced to zero after yielding. The yielding stress can be calculated

    using the following formula:

    tpeak snu tanfi 44

    where fi is the internal friction angle. For reservoirs already at

    the critical stress conditions the fractures are already yielded.

    Well

    Fig. 5. Well located atthe center ofa fracturedfield, whichis surrounded by matrix

    rock of effectively infinite extent.

    0

    2

    4

    6

    8

    10

    0

    Fracture closure (mm)

    Effectivenormalstress(MPa)

    0.2 0.4 0.6 0.8

    Fig. 6. Nonlinear fracture normal deformation.

    Table 1

    Rock and fracture parameters in the modeling.

    Area (m2) 1000 1000Shear modulus G (GPa) 5.9

    Possoins ratio u 0.16

    Undrained Possoins ratio uu 0.31

    Matrix permeability (md) 0.8Matrix porosity f 0.2

    Biots coefficient a 0.83Fluid viscosity m (cp) 1Fluid compressibility (/MPa) 6.8 104

    Fig. 7. Pore pressure distribution after 360 days production.

    Q. Tao et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268264

  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    7/10

    Therefore, the fractures are week and the shear stress disturbance

    can result in large shear deformation. The shear deformation will

    induce some normal deformation by dilation.

    In Fig. 12 a fractured reservoir with high anisotropic in-situ

    stress (s130 MPa, s323 MPa)has are two sets of fractures withanangleof601. Theshearstiffness before yielding is 100 GPa/m,the

    internal friction angle is 301, the dilation angle is 51, and all other

    parameters are the same as those in the isotropic case listed in

    Table 1. All fractures are already yielded before production and the

    production with a constant rate of 2 m3/day induces not only the

    normal deformation but also large shear deformation.

    Fig. 13 shows the direction and magnitude of the shear displace-

    ment after 360 days production. If the shear dilation induces moreopenness of the fracture than the closure induced by the increase of

    the effective normal stress, the fracture permeability will increase

    with production instead of reduction. Fig. 14 shows the fracture

    permeability distribution after 360 days production. There is still

    reduction of fracture permeability for those fractures in dark

    blue. But the fracture permeability for other fractures increases

    compared with the initial fracture permeability of 4428 D. The

    fracture permeability and shear displacement are compared and

    show consistent change (Fig. 14). Fig. 15 shows that the fracture

    permeability increases with production both for the fracture inter-

    sected by the well and for a fracture at the boundary with the

    maximum enhancement at early time. But the permeability of the

    fracture intersected by the welldecreaseslater as the effective stress

    keeps increasing.And thepermeabilityof thefracture at thetop rightcorner increases until very latetime and changes the trend whenthe

    effect of compression exceeds the effect of dilation. Therefore, under

    highly anisotropic stress conditions production may enhance the

    fracture permeability.

    7. Conclusions

    Production in naturally fractured reservoirs will cause reservoir

    pressure change, thereby changing the stress. The stress change

    will change the fracture aperture and permeability, thereby

    influencing the production. In this study, we developed a new

    approach to model the fracture permeability change in naturally

    fractured reservoirs by combining a finite difference method

    0

    1000

    2000

    3000

    4000

    5000

    0

    Time (hr)

    Fractureaperture(mm)

    fracture at boundary

    fracture intersected with Well

    1000 2000 3000 4000 5000 6000 7000 8000 9000

    Fig. 8. Fracture permeability declines with time.

    0

    1000

    2000

    3000

    4000

    5000

    0

    Time (hr)

    Fracturepermea

    bility(darcy)

    Effective in situ stress = 1 MPa

    2 MPa

    5 MPa

    10 MPa

    1000 2000 3000 4000 5000 6000 7000 8000 9000

    Fig. 9. Effect of initial effective in-situ stress on the fracture permeability change.

    Fig. 10. Relation of fracture aperture with the effective normal stress for differentfractures.

    Q. Tao et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268 265

  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    8/10

    solving the coupled fluid diffusivity equation in fracture network,

    the nonlinear BartonBandis joint deformation model, and the

    poroelastic displacement discontinuity method (DDM).

    Fig. 11. The influence of weakness of fracture on the fracture permeability change with the fluid production.

    Fig. 12. Well located at the center of a fractured field under anisotropic stress field

    and the fractured network are surrounded by matrix rock.

    Fig. 13. Shear displacement distribution after 360 days production for the case

    fractures are already yielded before production. The arrow represents the shear

    direction.

    Fig. 14. Distribution of fracture permeability and shear displacement (shown with

    arrows) after 360 days production for the case fractures are already yielded before

    production.

    2000

    4000

    6000

    8000

    10000

    12000

    14000

    0

    Time (hr)

    Fractureperm

    eability(darcy)

    Fracture at the top right corner

    Fracture intersected by well

    2000 4000 6000 8000 10000

    Fig. 15. Fracturepermeability changeswith production forthe case thefracture are

    already yielded before production.

    Q. Tao et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268266

  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    9/10

    Fracture aperture and permeability decreases with pressure

    reduction caused by the production in naturally fractured reser-

    voirs under isotropic stress conditions. But the magnitude of the

    change is dependent on the initial effective in-situ stress. For low

    initial effective in-situ stress (the reservoirpressure is very close to

    the magnitude of stress), the normal stiffness is small if the initial

    normal stiffness is small, i.e., weak fractures. The small change of

    reservoir pressure and effective stress can induce large fracture

    closure andpermeability loss. Butfor hard rock (high initial normalstiffness) or high effective in-situ stress, the normal stiffness is

    large. The change of fracture aperture and permeability is small

    even for large reservoir pressure change. Therefore, whether the

    reservoir is stress sensitive can be decided by laboratory tests on

    the properties of fractures and field tests of the in-situ stress. For

    stress sensitive fractured reservoirs, the method developed in this

    study can be applied to evaluate the change of fracture perme-

    ability during production and its influence on production. How-

    ever, underhighly anisotropic in-situ stress condition, the fractures

    can be at the critical stress condition, and a small change of the

    shear stress can induce large shear displacement. The fracture

    aperture and permeability can be enhanced due to shear dilation

    while the reservoir pressure is decreasing.

    Appendix A. Fundamental solutions for fluid source

    Induced pore pressure pq, displacement uq and stress sq bycontinuous unit fluid source along a line fracture segment

    r2 xxu2 y2 A:1where x0 varies from a to +a. Recall that

    E1x Z1

    x

    eu

    udu A:2

    The induced pore pressure is given by

    pq

    m

    4pkZ

    a

    aEi

    x

    2

    dxu

    A:3

    where

    x r2ffiffiffiffi

    ctp A:4

    The induced displacements are

    uqx am12u

    16pkG1u 2ctex2 r

    2E1x22

    2ct lnr2 E1x2 " #a

    aA:5

    uqy am12u

    16pkG 1u 4ctarctanxxu

    y

    !aa

    &

    yZa

    aE1x2dxu4cty

    Zaa

    ex2

    r2dxu

    )

    A:6

    The induced stresses are

    sqxx am12u8pk1u xx

    1

    x2 e

    x2

    x2E1x2

    " #aa

    2Za

    aE1x2dxu

    ( )

    A:7

    sqyy am12u8pk1u xxu

    1

    x2 e

    x2

    x2

    E1x2 " #a

    aA:8

    sqxy am 12u 8pk 1

    u

    y 1x

    2 e

    x2

    x2

    E1x2 " #a

    a

    A:9

    Appendix B. Fundamental solutions for displacement

    discontinuities source

    1. Induced pore pressure, displacement and stress by the

    continuous unit normal displacement discontinuity along a line

    fracture segment.

    Induced pore pressure

    pdn

    G

    uu

    u

    2par212u1uu 2

    x

    xu

    r2 1ex2

    !a

    a B:1Induced displacement

    udnx 1

    4p 1u 12uln9r9uuu1uu

    lnr E1x

    22

    1ex2

    2x2

    " #(

    y2

    r21 uuu

    1uu

    1 1

    x2

    ex2

    x2

    " #)aa

    B:2

    udny 1

    4p1u 21uarctanxxu

    y

    ln9r9

    &

    xxuyr2

    1 uuu1uu 1

    1

    x2

    ex2

    x2 " #)

    a

    a

    B:3

    Induced stress

    sdnxx G

    2p1uxxu3xxuy2

    r4 uuu

    1uu

    (

    xxu3xxuy2

    r4 3xxuy

    2xxu3r4

    1ex2

    x2

    2xxuy2ex

    2

    r4

    " #)aa

    B:4

    sdnyy G

    2p1uxxu3 3xxuy2

    r4 uuu

    1uu

    (

    xxu3 3xxuy2

    r4 xxu

    33xxuy2r4

    1ex2

    x2 2xxu

    3ex2

    r4

    " #)aa

    B:5sdnxy

    G

    2p1uxxu2yy3

    r4 uuu

    1uu

    (

    xxu2yy3

    r4 3xxu

    2yy3r4

    1ex2 x2

    2xxu2yex

    2

    r4

    " #)aa

    B:6

    where vu is the undrained Poissons ratio.

    2. Induced pore pressure, displacement and stress by the

    continuous unit shear displacement discontinuity along a line

    fracture segment.

    Induced pore pressure

    pds

    Guuu

    2par2

    12u1uu

    2y

    r2

    1

    ex

    2

    !a

    a B:7

    Induced displacement

    udsx 1

    4p1u

    (21uarctan xxu

    y

    ln rj j

    xxuyr2

    1 uuu1uu

    1 1

    x2

    ex2

    x2

    " #)aa

    B:8

    udsy 1

    4p 1u 12u ln rj juuu1uu

    lnr E1x

    22

    1ex2

    2x2

    " #(

    y2

    r21 uuu

    1uu 1

    1

    x

    2 e

    x2

    x

    2 " #)a

    a

    B:9

    Q. Tao et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268 267

  • 8/3/2019 A Fully Coupled Hydro Logical (1)

    10/10

    Induced stress

    sdsxx G

    2p 1u

    ( 3xxu

    2yy3r4

    uuu1uu

    3xxu2yy3

    r4 3xxu

    2yy3r4

    1ex2 x

    2 2y

    3ex2

    r4

    " #)aa

    B:10

    sdsyy G

    2p 1u

    (xxu2yy3

    r4 uuu

    1uu

    xxu2yy3

    r43xxu

    2yy3r4

    1ex2 x

    2 2xxu

    2yex2

    r4

    " #)aa

    B:11

    sdsxy G

    2p1uxxu3xxuy2

    r4 uuu

    1uu

    (

    xxu3xxuy2

    r4 3xxuy

    2xxu3r4

    1ex2 x2

    2xxuy2ex

    2

    r4

    " #)aa

    B:12References

    [1] Gray DH, Fatt I, Bergamini G. The effect of stress on permeability of sandstonecores. SPE J 1963;3:95100.

    [2] Vairogs J, Hearn CL, Dareing DW, Rhoades VW. Effect of rock stress on gasproduction from low-permeability reservoirs. J Pet Tech 1971;23:11617.

    [3] Thomas RD, Ward DC. Effect of overburden pressure and water saturation ongas permeability of tight sandstone cores. J Pet Tech 1972;24:1204.

    [4] Raghavan R, Scorer JDT, Miller FG. An investigation by numerical methods ofthe effect of pressure-dependent rock and fluid properties on well flow tests.SPE J 1972;12:26775.

    [5] Vairogs J, Rhoades VW. Pressure transient test in formations having stress-sensitive permeability. J Pet Tech 1973;25:96570.

    [6] Samaniego VF, Brigham WE, Miller FG. Performance-prediction procedure fortransient flow of fluidsthroughpressure-sensitive formations. In: Proceedingsof the 51st annual technical conference and exhibition, New Orleans, 36 Oct1976, paper SPE 6051.

    [7] Samaniego-V F, Brigham WE, Miller FG. An investigation of transient flow ofreservoir fluidsconsidering pressure-dependentrock andfluid properties. SPEJ1977;17:14050.

    [8] Jones FO, Owens WW. A laboratory study of low-permeability gas sands. J Pet

    Tech 1980;32:163140.[9] Samaniego F, Cinco-Ley H. On the determination of the pressure-dependent

    characteristics of a reservoir through transient pressure testing. In: Proceed-

    ings of the annual technical conference and exhibition, San Antonio, 811 Oct

    1989, paper SPE 19774.[10] Buchsteiner H, Warpinski NR, Economides MJ. Stress-induced permeability

    reduction in fissured reservoirs. In: Proceedings of the annual technical

    conference and exhibition, Houston, 36 Oct 1993, paper SPE 26513.[11] Chin, LY, Raghavan R, Thomas LK. Fully coupled analysis of well responses in

    stress-sensitive reservoirs. In: Proceedings of the annual technical conferenceand exhibition, New Orleans, 2730 Sept 1998, paper SPE 48967.[12] Davies JP, Davies DK. Stress-dependent permeability: characterization and

    modeling. In: Proceedings of 1999 SPE annual technical conference and

    exhibition, Houston, Texas, 36 October 1999, paper SPE 56813.[13] Ostensen RW. The effect of stress-dependent permeability on gas production

    and well testing. SPE Form Eval 1986;1:22735.[14] Iwai K. Fundamental studies of fluid flow through a single fracture, PhD

    dissertation, Univ Calif, Berkeley, 1976.[15] GoodmanRE. Methodsof Geological Engineeringin DiscontinuousRocks. New

    York: West Pub; 1976.[16] BandisSC, LumsdenAC, BartonNR. Fundamentalsof rockjointdeformation.Int

    J Rock Mech Min Sci Geomech Abstr 1983;20:24968.[17] Barton NR, Bandis SC, Bakhtar K. Strength, deformation and conductivity

    coupling of rock joints. Int J Rock Mech Min Sci Geomech Abstr 1985;22:

    12140.[18] Biot MA. General theory of three-dimensional consolidation. J Appl Phys

    1941;12:15564.[19] BiotMA. General solutions of theequations of elasticity andconsolidation fora

    porous material. J Appl Mech 1956;78:916.[20] Rice JR, Cleary MP. Some basic stress diffusion solutions for fluid-saturated

    elastic porous media with compressible constituents. Rev Geophys 1976;14:

    22741.[21] Curran,JH, CarvalhoJL. A displacementdiscontinuitymodelfor fluid-saturated

    porous media. In: Proceedings of the sixth international congress on rock

    mechanics, Montreal, 1987, p. 738.[22] ChengAHD, Predeleanu M. Transient boundaryelementformulation for linear

    poroelasticity. Appl Math Modell 1987;11:28590.[23] Carvalho JL. Poroelastic effects and influence of material interfaces on

    hydraulic fracturing. PhD dissertation, Univ Toronto, Toronto, 1990.[24] Vandamme LM,RoegiersJC. Poroelasticity in hydraulicfracturingsimulators. J

    Pet Tech 1990;42:1199203.[25] Asgian M. A numerical-model of fluid-flow in deformable naturally fractured

    rock masses. Int J Rock Mech Min Sci 1989;26:31728.[26] Lee J, Rollins J, Spivey J. Pressure Transient Testing. Richardson, Texas: Soc

    Petrol Eng; 2003.[27] WitherspoonPA,Wang JSY,IwaiK, Gale JE. Validityof cubiclawfor fluid flowin

    a deformable rock fracture. Water Resour Res 1980;16:101624.[28] Tao Q. Numerical modeling of fracture permeability change in naturally

    fractured reservoirs using a fully coupled displacementdiscontinuity method.

    PhD Dissertation, Texas A&M Univ, College Station, Texas, 2010.

    Q. Tao et al. / International Journal of Rock Mechanics & Mining Sciences 48 (2011) 259268268