16.avo

25
Stanford Rock Physics Laboratory - Gary Mavko AVO 302 The Rock Physics of AVO

description

csdczs

Transcript of 16.avo

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    302

    The Rock Physics of AVO

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    303

    Water-saturated40 MPa

    Water-saturated10-40 MPa

    Gas andWater-saturated10-40 MPa

    L8

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    304

    N.1

    More than 400 sandstone data points, with porositiesranging over 4-39%, clay content 0-55%, effectivepressure 5-40 MPa - all water saturated.

    When Vp is plotted vs. Vs, they follow a remarkablynarrow trend. Variations in porosity, clay, and pressuresimply move the points up and down the trend.

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    305

    N.2

    Variations in porosity, pore pressure, and shalinessmove data along trends. Changing the pore fluidcauses the trend to change.

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    306

    Different shear-related attributes.N.2

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    307

    In an isotropic medium, a wave that is incident on aboundary will generally create two reflected waves (oneP and one S) and two transmitted waves. The total sheartraction acting on the boundary in medium 1 (due to thesummed effects of the incident an reflected waves) mustbe equal to the total shear traction acting on the boundary inmedium 2 (due to the summed effects of thetransmitted waves). Also the displacement of a point inmedium 1 at the boundary must be equal to the displace-ment of a point in medium 2 at the boundary.

    VP1, VS1, 1

    VP2, VS2, 2

    1

    1

    22

    Reflected P-wave

    Incident P-wave

    Reflected S-wave

    Transmitted P-wave

    Transmitted S-wave

    N.4

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    308

    By matching the traction and displacement boundaryconditions, Zoeppritz (1919) derived the equationsrelating the amplitudes of the P and S waves:

    sin 1( ) cos 1( ) sin 2( ) cos 2( )cos 1( ) sin 1( ) cos 2( ) sin 2( )sin 21( )

    VP1VS1

    cos 21( )2VS 22VP11VS12VP 2

    sin 22( ) 2VS 2VP1

    1VS12cos 22( )

    cos 21( ) VS1VP1

    sin 21( ) 2VP21VP1

    cos 22( ) 2VS 21VP1

    cos 22( )

    RppRpsTppTps

    =

    sin 1( )cos 1( )sin 21( )

    cos 21( )

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    309

    AVO - Shuey's Approximation

    P-wave reflectivity versus angle:

    Intercept Gradient

    R ( ) = R0 + ER0 +

    1 ( )2

    sin2 + 12

    VPVP

    tan2 sin2 [ ]

    R0 12

    VPVP

    +

    E = F 2 1 + F( )1 21

    F = VP /VPVP /VP + /

    VP = VP 2 VP1( )VS = VS 2 VS1( ) = 2 1( )

    VP = VP2 + VP1( ) / 2VS = VS2 + VS1( ) / 2 = 2 + 1( ) / 2

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    310

    P-wave reflectivity versus angle:

    Intercept Gradient

    R ( ) = R0 + 12VPVP

    2VS2

    VP2

    + 2 VSVS

    sin2

    +12

    VPVP

    tan2 sin2 [ ]

    R0 12

    VPVP

    +

    VP = VP 2 VP1( )VS = VS 2 VS1( ) = 2 1( )

    VP = VP2 + VP1( ) / 2VS = VS2 + VS1( ) / 2 = 2 + 1( ) / 2

    AVO - Aki-Richard's approximation:

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    311

    AVO Response

    P-Velocity Poisson ratio AVO response contrast contrast

    negative negative increase negative positive decrease positive negative decrease positive positive increase

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    312

    Vp-Vs Relations

    There is a wide, and sometimes confusing, variety ofpublished Vp-Vs relations and Vs prediction techniques,which at first appear to be quite distinct. However, mostreduce to the same two simple steps:1. Establish empirical relations among Vp, Vs, and porosityfor one reference pore fluid--most often water saturated ordry.2. Use Gassmanns (1951) relations to map these empiricalrelations to other pore fluid states.Although some of the effective medium models predict both Pand S velocities assuming idealized pore geometries, the factremains that the most reliable and most often used Vp-Vsrelations are empirical fits to laboratory and/or log data. Themost useful role of theoretical methods is extending theseempirical relations to different pore fluids or measurementfrequencies. Hence, the two steps listed above.

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    313

    N.5

    0

    1

    2

    3

    4

    5

    6

    7

    0 0.5 1 1.5 2 2.5 3 3.5 4

    SandstonesWater Saturated

    Vp (k

    m/s

    )

    Vs (km/s)

    mudrockV s = .8621Vp- 1 . 1 7 2 4

    Castagna et al. (1993)V s = .8042Vp- . 8 5 5 9

    Han (1986)V s = .7936Vp- . 7 8 6 8

    (after Castagna et al., 1993)

    0

    1

    2

    3

    4

    5

    6

    7

    0 0.5 1 1.5 2 2.5 3 3.5 4

    ShalesWater Saturated

    Vp (k

    m/s

    )

    Vs (km/s)

    mudrockV s = .8621 Vp- 1.1724

    Castagna et al. (1993)V s = .8042Vp- . 8 5 5 9

    Han (1986)V s = .7936Vp- . 7 8 6 8

    (after Castagna et al., 1993)

    N.5

    MudrockVs = .86 Vp - 1.17

    Han (1986)Vs = .79 Vp - 0.79

    Castagna et al. (1993)Vs = .80 Vp - 0.86

    MudrockVs = .86 Vp - 1.17

    Han (1986)Vs = .79 Vp - 0.79

    Castagna et al. (1993)Vs = .80 Vp - 0.86

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    314

    012345678

    0 0.5 1 1.5 2 2.5 3 3.5 4

    LimestonesWater Saturated

    Vp (k

    m/s

    )

    Vs (km/s)

    V s = Vp/ 1 . 9Pickett (1963)

    Castagna et al. (1993)V s = -.05508 VP2 + 1.0168 Vp - 1.0305

    water (after Castagna et al.,1993)

    012345678

    0 0.5 1 1.5 2 2.5 3 3.5 4

    DolomiteWater Saturated

    Vp (k

    m/s

    )

    Vs (km/s)

    Castagna et al. (1993)V s = .5832Vp - . 0 7 7 7 6

    Pickett (1963)V s = Vp/ 1 . 8

    (after Castagna et al., 1993)

    N.6

    Pickett(1963)Vs = Vp / 1.9

    Castagna et al. (1993)Vs = -.055 Vp2 + 1.02 Vp - 1.03

    Pickett(1963)Vs = Vp / 1.8

    Castagna et al. (1993)Vs = .58 Vp - 0.078

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    315

    0

    1

    2

    3

    4

    5

    6

    0 0.5 1 1.5 2 2.5 3 3.5 4

    Shaly SandstonesWater Saturated

    Vp-sat c>.25Vp-sat c 25 % Vs= .8423Vp-1 .099

    clay < 25 %Vs= .7535Vp- .6566

    mudrock Vs= .8621Vp-1 .1724

    (Data from Han, 1986)

    0

    1

    2

    3

    4

    5

    6

    0 0.5 1 1.5 2 2.5 3 3.5 4

    Shaly SandstonesWater Saturated (hf)

    Vp-sat Phi>.15Vp-sat Phi 15 % Vs = .7563Vp-.6620

    porosity < 15 %Vs = .8533Vp-1.1374

    mudrock Vs = .8621Vp-1.1724

    (Data from Han, 1986)

    N.7

    MudrockVs = .86 Vp - 1.17

    Clay < 25%Vs = .75 Vp - 0.66

    Clay > 25%Vs = .84Vp-1.10

    porosity > 15%Vs = .76Vp - 0.66

    porosity < 15%Vs = .85 Vp - 1.14

    MudrockVs = .86 Vp - 1.17

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    316

    Dry Poissons Ratio Assumption

    02468

    10121416

    0 5 10 15 20 25 30 35

    clay < 10%clay > 10%

    Vs2

    dry

    Vp2 dry

    = 0.01 = 0.1

    = 0.2

    = 0.3

    = 0.4

    Shaly Sandstones - Dry

    The modified Voigt Average Predicts linear moduli-porosity relations.This is a convenient relation for use with the critical porosity model.

    These are equivalent to the dry rock Vs/Vp relation and the dry rockPoissons ratio equal to their values for pure mineral.

    The plot below illustrates the approximately constant dry rockPoissons ratio observed for a large set of ultrasonic sandstonevelocities (from Han, 1986) over a large rance of effective pressures (5< Peff < 40 MPa) and clay contents (0 < C < 55% by volume).

    N.8

    Kdry = K0 1c

    dry = 0 1

    c

    VSVP

    dry rock

    VSVP

    mineral

    dry rock mineral

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    317

    Kriefs Relation (1990)The model combines the same two elements:1. An empirical Vp-Vs- relation for water-saturatedrocks, which is approximately the same as the criticalporosity model.2. Gassmanns relation to extend the empiricalrelation to other pore fluids.Dry rock Vp-Vs- relation:

    where is Biots coefficient. This is equivalent to:

    where

    and Kf are two equivalent descriptions of the pore spacestiffness. Determining vs. or K vs determines therock bulk modulus Kdry vs .

    Krief et al. (1990) used the data of Raymer et al. (1980) toempirically find a relation for vs :

    Kdry = Kmineral 1 ( )

    1Kdry

    =1K0

    +K

    1K

    =1v p

    dv pd

    PP = constant

    ; =dv pdV

    PP = constant

    =KdryK

    1 ( ) = 1 ( )m ( ) where m ( ) = 3 / 1( )

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    318

    Assuming dry rock Poissons ratio is equal to themineral Poissons ratio gives

    N.9

    Kdry = K0 1 ( )m ( )

    dry = 0 1( )m ( )

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    319

    Expressions for any other pore fluids are obtained fromGassmanns equations. While these are nonlinear, theysuggest a simple approximation:

    where VP-sat, VP0, and Vfl are the P-wave velocities of thesaturated rock, the mineral, and the pore fluid; and VS-satand VS0 are the S-wave velocities in the saturated rockand mineral. Rewriting slightly gives

    where VR is the velocity of a suspension of minerals in afluid, given by the Reuss average at the critical porosity.This modified form of Kriefs expression is exactlyequivalent to the linear (modified Voigt) K vs and vs relations in the critical porosity model, with the fluid effectsgiven by Gassmann.

    which is a straight line (in velocity-squared) connecting themineral point ( ) and the fluid point ( ). A moreaccurate (and nearly identical) model is to recognize thatvelocities tend toward those of a suspension at highporosity, rather than toward a fluid, which yields themodified form

    VP02 , VS02 Vfl2 , 0

    VP sat2 Vfl2

    VS sat2=VP 02 Vfl2

    VS02

    VP sat2 VR

    2

    VS sat2=VP 0

    2 VR2

    VS02

    VP sat2 = Vfl

    2 + VS sat2 VP 0

    2 Vfl2

    VS 02

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    320

    0

    5

    10

    15

    20

    25

    30

    35

    0 5 10 15 20

    Vp-Vs Relation in Dry andSaturated Rocks

    V p2

    (km

    /s)2

    Vs2 (km/s)2

    saturated

    dry

    Sandstones mineral point

    fluid point

    N.10

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    321

    N.11

    0

    10

    20

    30

    40

    50

    0 4 8 12 16

    Vp-Vs Relation in Sandstoneand Dolomite

    V p2

    (km

    /s)2

    Vs2 (km/s)2

    Sandstone

    Dolomite

    mineralpoints

    fluid points

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    322

    VS = 12 Xii = 1

    LaijVPjj = 0

    Ni+ Xii = 1

    LaijVPjj = 0

    Ni 1 1

    Greenberg and Castagna (1992) have given empiricalrelations for estimating Vs from Vp in multimineralic, brine-saturated rocks based on empirical, polynomial Vp-Vsrelations in pure monomineralic lithologies (Castagna etal., 1992). The shear wave velocity in brine-saturatedcomposite lithologies is approximated by a simple averageof the arithmetic and harmonic means of the constituentpure lithology shear velocities:

    Castagna et al. (1992) gave representative polynomialregression coefficients for pure monomineralic lithologies:

    Regression coefficients for pure lithologies with Vp and Vsin km/s:

    Xi = 1i = 1

    L

    VS = ai2VP2 + ai1VP + ai0 (Castagna et al. 1992)

    whereL number of monomineralic lithologic constituent

    Xi volume fractions of lithological constituentsaij empirical regression coefficientsNi order of polynomial for constituent iVp, Vs P and S wave velocities (km/s) in composite brine-

    saturated, multimineralic rock

    Lithology a i2 a i1 a i0S andstone 0 0.80416 -0.85588Limestone -0.05508 1.01677 -1.03049Dolomite 0 0.58321 -0.07775Shale 0 0.76969 -0.86735

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    323

    1

    2

    3

    4

    5

    6

    7

    0 1 2 3 4 5

    SandstoneL imestoneDo l om i t eS h a l e

    Vp (

    km/s

    )

    Vs (km/s)

    Note that the above relation is for 100% brine-saturated rocks.To estimate Vs from measured Vp for other fluid saturations,Gassmanns equation has to be used in an iterative manner.In the following, the subscript b denotes velocities at 100%brine saturation and the subscript f denotes velocities at anyother fluid saturation (e.g. this could be oil or a mixture of oil,brine, and gas). The method consists of iteratively finding a(Vp, Vs) point on the brine relation that transforms, withGassmanns relation, to the measured Vp and the unknown Vsfor the new fluid saturation. the steps are as follows:

    1. Start with an initial guess for VPb.2. Calculate VSb corresponding to VPb from the empiricalregression.3. Do fluid substitution using VPb and VSb in the Gassmannequation to get VSf.4. With the calculated VSf and the measured VPf, use theGassmann relation to get a new estimate of VPb. Check withprevious value of VPb for convergence. If convergencecriterion is met, stop; if not go back to step 2 and continue.

    N.12

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    324

    Coefficients for the equation b = aVp2 + bVp + c

    L ithology a b c Vp Range (Km/s)

    Shale - .0261 .373 1.458 1.5-5.0Sandstone - .0115 .261 1.515 1.5-6.0Limestone - .0296 .461 0.963 3.5-6.4Dolomite - .0235 .390 1.242 4.5-7.1Anhydrite - .0203 .321 1.732 4.6-7.4

    Coefficients for the equation b = dVpf

    L ithology d f Vp Range(Km/s)

    Shale 1.75 .265 1.5-5.0Sandstone 1.66 .261 1.5-6.0Limestone 1.50 .225 3.5-6.4Dolomite 1.74 .252 4.5-7.1Anhydrite 2.19 .160 4.6-7.4

    Both forms of Gardners relations applied to log and lab shale data, aspresented by Castagna et al. (1993)

    Polynomial and powerlaw forms of the Gardner et al. (1974) velocity-densityrelationships presented by Castagna et al. (1993). Units are km/s andg/cm3.

    N.15

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    325

    Both forms of Gardners relations applied tolaboratory dolomite data.

    N.16

  • Stanford Rock Physics Laboratory - Gary Mavko

    AVO

    326

    Both forms of Gardners relations applied to laboratory limestone data. Notethat the published powerlaw form does not fit as well as the polynomial. wealso show a powerlaw form fit to these data, which agrees very well with thepolynomial.

    Both forms of Gardners relations applied to log and lab sandstone data, aspresented by Castagna et al. (1993).

    N.17