GEOPHYSICS, VOL. 69, NO. 2 (MARCH-APRIL 2004); P. 398405, 8 FIGS., 2 TABLES. 10.1190/1.

1707059

Gassmanns equation and uid-saturation effects on seismic velocities

De-hua Han and Michael L. Batzle

spots, amplitude variation with offset (AVO), and time-lapse reservoir monitoring. Despite the popularity of Gassmanns equations and their incorporation in most software packages for seismic reservoir interpretation, important aspects of these equations have not been thoroughly examined. Many of the basic assumptions are invalid for common reservoir rocks and uids. Previous efforts to understand the operation and application of Gassmanns equations (Han, 1992; Mavko and Mukerji, 1995, Mavko et al., 1998; Sengupta and Mavko, 1999) have focused mostly on individual parameter effects. Recently, Nolen-Hoeksema (2000) made a detailed effort to quantify changes in the pore-space modulus in response to changes in uid modulus. He introduced an effective uid coefcient by differentiating the bulk modulus of the uid-lled pore space K pore . He included two ratios to control the effective uid coefcient: the ratio of uid modulus to solid grain modulus and the ratio of the Biot coefcient (1941) to porosity. However, the physical meaning of the effective uid coefcient was not claried. The uid effect was in conjunction with other rock parameters. The Voigt and Reuss models were introduced to test the uid effect. We need to extend his analysis to derive both the mechanical bounds for porous media and the magnitude of the uid effect. Because the full implications of parameter interactions in Gassmanns equation are not well understood, in general practice, no constraints are placed on input parameters and there is no quality control of the results. In particular, problems arise in automated analysis in which results are usually taken at face value. In this paper, we briey list the assumptions for Gassmanns equation. We then derive mechanical bounds for the input parameters that provide stricter constraints on calculated results. The specic physical properties controlling the uid-saturation effect are then better understood. Effect of uid saturation on seismic properties The seismic response of reservoirs is directly controlled by compressional (P-wave) and shear (S-wave) velocities V p and Vs respectively along with densities. Figure 1a shows measured

ABSTRACT

Gassmanns (1951) equations commonly are used to predict velocity changes resulting from different poreuid saturations. However, the input parameters are often crudely estimated, and the resulting estimates of uid effects can be unrealistic. In rocks, parameters such as porosity, density, and velocity are not independent, and values must be kept consistent and constrained. Otherwise, estimating uid substitution can result in substantial errors. We recast the Gassmanns relations in terms of a porosity-dependent normalized modulus K n and the uid sensitivity in terms of a simplied gain function G . General Voigt-Reuss bounds and critical porosity limits constrain the equations and provide upper and lower bounds of the uid-saturation effect on bulk modulus. The D functions are simplied modulus-porosity relations that are based on empirical porosity-velocity trends. These functions are applicable to uid-substitution calculations and add important constraints on the results. More importantly, the simplied Gassmanns relations provide better physical insight into the signicance of each parameter. The estimated moduli remain physical, the calculations are more stable, and the results are more realistic.

INTRODUCTION

With improved resolution and cost efciency, seismic technologies have gained a central position in reservoir delineation and monitoring. Rock physics is an essential link connecting seismic data to the presence of in situ hydrocarbons and to reservoir characteristics. Modeling the effects of uid on rock velocity and density is a basic method used to ascertain the inuence of pore uids on seismic data. Gassmanns (1951) equations are the relations most widely used to calculate seismicvelocity changes resulting from different uid saturations in reservoirs. These equations predominate in the analysis of direct hydrocarbon indicators (DHI), such as amplitude bright

Manuscript received by the Editor September 19, 2001; revised manuscript received July 26, 2003. University of Houston, Department of Geoscience, 312 Science & Research Building, Houston, Texas 77204-5007. E-mail: [email protected]. Colorado School of Mines, Department of Geophysics, Golden, Colorado 80401-1887. E-mail: [email protected]. c 2004 Society of Exploration Geophysicists. All rights reserved. 398

Gassmanns E & Fluid Effects on V

399

dry and water-saturated P- and S-wave velocities of sandstone as a function of differential pressure. With water saturation, P-wave velocity increases slightly, whereas S-wave velocity decreases slightly. However, neither P- nor S-wave velocity is the best indicator of any uid saturation effect because of the coupling between P- and S-waves through the shear modulus and bulk density:

following Gassmanns formulation because of its clear physical meaning:

Ks = Kd + Kd =

Kd (2) (3)

K 0 (1 K d / K 0 )2 , 1 Kd /K0 + K0/K f

Vp = Vs =

K + 4/3 = ,

M , (1)

s = d ,

where K is bulk modulus, is shear modulus, M is the compressional modulus, and is the bulk density. In contrast, if we plot bulk and shear moduli as functions of pressure (Figure 1b), the water-saturation effect becomes evident: (1) the bulk modulus increases about 50%, whereas (2) shear modulus remains constant. Clearly, bulk modulus is more sensitive to water saturation. The bulk-volume deformation produced by a passing seismic wave results in a pore-volume change and causes a pressure increase in pore uid (water). This pressure increase stiffens the rock frame and causes an increase in bulk modulus. Shear deformation, however, does not produce a pore-volume change, and consequently different uids do not affect shear modulus. Therefore, any uid-saturation effect should correlate mainly to a change in bulk modulus. Gassmanns equation Gassmanns equations provide a simple model for estimating the uid-saturation effect on bulk modulus. We prefer the

where K 0 , K f , K d , K s , are the bulk moduli of the mineral grain, uid, dry rock, and saturated rock frame, respectively; is porosity; and s and d are the saturated and dry-rock shear moduli. K d is an increment of bulk modulus as a result of uid saturation of dry rock. These equations indicate that uid in pores will affect bulk modulus but not shear modulus, which is consistent with our earlier discussion. As Berryman (1999) pointed out, a shear modulus that is independent of uid saturation results directly from the assumptions used to derive Gassmanns equation. Numerous assumptions are involved in the derivation and application of Gassmanns equation: 1. the porous material is isotropic, elastic, monomineralic, and homogeneous; 2. the pore space is well connected and in pressure equilibrium (zero-frequency limit); 3. the medium is a closed system with no pore-uid movement across boundaries; 4. there is no chemical interaction between uids and rock frame (shear modulus remains constant). Many of these assumptions may not be valid for hydrocarbon reservoirs and depend on rock and uid properties and the in-situ conditions. For example, most rocks are anisotropic to some degree, invalidating assumption (1). The work of Brown and Korringa (1975) provides an explicit form for an

Figure 1. (a) Measured P- and S-wave velocities on a sandstone sample at dry and water-saturated states, as a function of pressure. (b) Bulk and shear moduli at dry and water-saturated states, as a function of pressure.

400

Han and Batzle

anisotropic uid substitution. Also, in seismic applications, it is normally assumed that Gassmanns equation works best for seismic data at frequencies less than 100 Hz (Mavko et al., 1998). Recently published laboratory data (Batzle et al., 2001) show that acoustic waves may be dispersive in rocks within the typical seismic band, invalidating assumption (2). In such cases, seismic frequencies may still be too high for the application of Gassmanns equation. Pore pressures may not have enough time to reach equilibrium, and the rock remains unrelaxed or only partially relaxed. A Gassmann-type calculation provides an estimate of relaxed velocity at zero frequency, which is a lower bound of the uid-saturation effect. Other similar violations of the assumptions often lead to the misapplication of Gassmanns equation. We will discuss this issue in detail in a separate paper.
SIMPLIFIED GASSMANNS EQUATION

where (1 K n ( )) is the Biot parameter (Biot, 1941). The Biot parameter is a relative measure of the difference between the mineral and dry-frame moduli. Furthermore, if we apply the Voigt bound for K n , and because usually K 0 K f , it is reasonable to assume that

0 1 K n ( )

K0/K f

(7)

for sedimentary rocks with high porosities ( > 15%). Therefore, the uid-saturation effect of the Gassmann equation can be simplied as

Ks = Kd +

K d K d + G ( ) K f ,

(8)

where G ( ) is the simplied gain function dened as

G ( ) =

Gassmanns formulation is straightforward, and the simple input parameters typically can be directly measured or assumed. This simplicity is a primary reason for its wide application in geophysical techniques. However, derivation of the rock and uid input parameters frequently remains ambiguous. As a consequence, we frequently have little control over the validity of uid-substitution calculations. We have made efforts to clarify how the uid-saturation effects are controlled by rock parameters (Han, 1992). A simple graphic construction developed by Mavko and Mukerji (1995) introduces an intercept porosity R for a dry rock that is based on bulk porosity, drybulk modulus, and solid-frame modulus. However, the physical meaning of the intercept porosity is not clear. Additionally, multiple-parameter effects on Gassmanns calculation remain ambiguous. As a consequence, for application of Gassmanns calculation, generally we have not been able either to constrain input parameters or to obtain quality control on the results. In this section, we regroup Gassmanns equation with combined rock parameters. Under certain conditions, we can simplify this equation further and clearly dene the controlling parameters for uid-saturation effects. The primary measure of a rocks velocity sensitivity to uid saturation is its normalized modulus K n , the ratio of dry bulk modulus to that of the mineral:

[1 K n ( )]2 2 = .

(9)

Kn = Kd /K0.

(4)

Equation (8) is a simplied form of Gassmanns equation, with clear physical meaning: uid-saturation effects on the bulk modulus are proportional to a simplied gain function G ( ) and the uid modulus K f . The G ( ) in turn depends directly on dry-rock properties: the normalized modulus and porosity. In general, G ( ) is independent of uid properties (ignoring interactions between rock frame and pore uid). Equation (9) also shows that the normalized modulus or the Biot parameter must be compatible with porosity. Otherwise, G ( ) can be unstable, particularly at small porosities. We need to know both the simplied gain function of the dry rock frame and the poreuid modulus to evaluate the uid-saturation effect on seismic properties. Figure 2 shows that for sandstone samples (Han, 1986) at high differential pressure (>20 MPa), the K s of water-saturated sands calculated using the simplied form is overestimated by 3% for porosities greater than 15%. These errors will decrease signicantly with a low uid modulus (gas and light-oil saturation). For low-porosity sands with high clay content, the simplied Gassmanns equation substantially overestimates watersaturation effects. This simplied Gassmanns equation has a clear physical meaning with systematic errors. This formulation can guide us in applying Gassmanns equation and in assessing the validity of such calculations.

This function can be complicated and depends on rock texture (porosity, clay content, pore geometry, grain size, grain contact, cementation, mineral composition, and so on) and reservoir conditions (pressure and temperature). This K n can be determined either empirically or theoretically. For relatively clean sandstone at high differential pressure (>20 MPa), the complex dependence of K n (x , y , z , . . .) can be simplied as a function of porosity:

K n (x , y , z , . . .) = K n ( ).
From equation (2), the bulk-modulus increment equal to

(5)
K d is then Figure 2. The simplied Gassmanns formulation overestimates the water-saturation effect on bulk modulus of sandstone samples (1986) at a differential pressure greater than 20 MPa. For porous sands (porosity >15%), the error is around 3%.

K 0 [1 K n ( )]2 Kd = 1 K n ( ) + K 0 / K f

(6)

Gassmanns E & Fluid Effects on V CONSTRAINTS ON GASSMANNS EQUATION

401

frame is at this lower bound:

The assumptions contained in Gassmanns equation do not constrain basic rock parameters. In equations (2), there are ve parameters, and usually the only applied constraint is that the parameters are physically meaningful (>0). When one applies Gassmanns equation, one generally handles input parameters as being completely independent. Values for K 0 and K f are estimated or assumed. K s or K d are calculated from V p and Vs , and density either comes from log data or is estimated along with . Incompatible or mismatched data often generate wrong or even unphysical results, such as a negative modulus. Unrealistic results can be hidden when one is performing a uid modulus substitution without examining K d . In reality, only K 0 and K f are completely independent. K s , K d , and are actually closely correlated. Bounds on K d as a function of , for example, constrain the bounds of K s . The Voigt bound Assuming the porous medium is a Voigt material, which is a high bound for K d ,

K d max =

K0 = K R. 1 + K0/K f

(17)

Note that the modulus increment K from dry to uid saturation is equal to the Reuss bound (equation 15):

Ks = Kd +

K d max = K R .

(18)

K d = K 0 (1 ).

(10)

Substituting equation (10) into Gassmanns equation (2) gives

K d min = K f ,
and

(11)

Again, Gassmanns equation is consistent with the dry and uid-saturated Reuss bounds. The K d max is the maximum uid-saturation effect predicted from Gassmanns equation (Figure 3). Physically, with the weakest frame, uids have a maximum effect. It is interesting that for the Voigt bound, the uid-saturation effect on the modulus increases with increasing porosity. This is opposite of the uid-saturation effect on the modulus for the Reuss bound, which decreases with increasing porosity. At a porosity of 100%, both the Voigt and Reuss bounds in the Gassmanns calculation show that the uid effect on modulus equals the uid modulus. In many reported applications, these general bound values have often been ignored. For example, a K s calculated directly from log data (V p , Vs , and bulk density) may be lower than the Reuss bound. This results in a negative value for K d . Such bounds of the uid effect on bulk modulus provide constraints for the input and output parameters of Gassmanns calculation.

Ks = Kd +

K d min = K 0 (1 ) + K f .

Critical porosity

(12)

Because this Voigt bound is the stiffest upper limit, the uidsaturation effect on bulk modulus here ( K d min ) will be a minimum (Figure 3). This is the rst constraint derived from the Gassmanns equation: The minimum of bulk modulus increment resulting from uid saturation is proportional to the porosity of the rock (the simplied gain function G ( ) = ) and the modulus of the pore uid. K0. We know that uid modulus is a positive value, but K f Therefore, from equations (10) and (11),

K0 Ks Kd .
The Reuss bound

(13)

The low modulus bound for porous media is the Reuss bound:

(1 ) 1 = + , KR K0 Kf K0 K f KR = . (1 ) K f + K 0

(14) (15)

Reservoir rocks in general are far from the Voigt and Reuss bounds, as Figure 4 shows for sandstones. Dolomite with vuggy pores may approach the Voigt bound, and highly fractured rocks may approach the Reuss bound. However, there is a great difference between these idealized bounds and most real rocks: The bounds shown in Figure 4 do not limit most observed, naturally occurring porosity. The vast majority of rocks have an upper limit to their porosity, usually termed critical porosity, c (Yin, 1992; Nur et al., 1995). At this highporosity limit, we reach the threshold of grain contacts (Han, 1986). This c modies the Voigt model (Figure 3) to provide tighter constraints for dry and uid-saturated bulk moduli for sands. This triangle physically correlates with both the Voigt and Reuss bounds. The uid-saturation effect on modulus is consistent with the Voigt triangle: It increase with increasing porosity and is limited by the Reuss bound at the critical porosity. The modied Voigt triangle provides a linear formulation and a graphic procedure for Gassmanns calculation: the uid saturation effect on bulk modulus is proportional to normalized porosity and the maximum uid saturation effect on bulk modulus (Reuss bound) at the critical porosity (Figure 3):

For completely empty (dry) rocks, the uid modulus K f is equal to zero. Thus, both the Reuss bound and the normalized modulus ( K n R ) for a dry rock in this case equals zero (for nonzero porosity):

K d = /c K Rc .

(19)

K n R ( ) = K d / K 0 = 0.

(16)

This is consistent with the earlier work done by Mavko and Mukerji (1995), with a slightly different physical formulation. For typical sandstones, the critical porosity c is around 40%. Thus, we can also generate a simplied numerical formula of the normalized modulus K n for the modied Voigt model:

Substituting equation (16) into the Gassmanns equation (2), we nd the uid-saturation effect on bulk modulus when the

K n ( ) = 1 /c = 1 2.5 .

(20)

402

Han and Batzle

Using this in Gassmanns equation (6) yields the uidsaturation effect

Kd =

6.25 K 0 < 6.25 K f . 1.5 + K 0 / K f

(21)

at high differential-pressure (>20 MPa) conditions, D ranges from about 1.45 to slightly more than 2.0, depending on consolidation of the rock. By inserting this D function into equation (9) we nd

EMPIRICAL MODEL FOR NORMALIZED MODULUS Kn

G ( ) = D 2 (2 D )2 .

(26)

Extending our empirical approach to a rst order, both Pand S-wave velocities can correlate linearly with porosity at high differential pressure ( Pd = 40 MPa). For dry, clean sands (Han, 1986),

Figure 6 shows the modied Voigt model based on critical porosity and the D function model, with D = 2. This simple calculation provides a useful quality-control tool to check any Gassmann-type calculation for sandstones.
SOLID MINERAL BULK MODULUS

V p = 5.97 7.85 km/s, Vs = 4.03 5.85 km/s,

assuming that the density of these sands is equal to

(22)

d = 2.65 (1 ) g/cm3 .

(23)

Because the modulus is the product of the density and the square of the velocity, we obtain an equation that is cubic in terms of porosity. The bulk modulus can be derived as

K d = (1 A + B 2 C 3 ) K 0 ,

(24)

where A = 3.206; B = 3.349; C = 1.143. Equation (24) can be further simplied if porosity is not too high ( < 30%):

As we mentioned previously, the normalized modulus K n controls the uid saturation effect, rather than K 0 or K d individually. The mineral modulus K 0 is as important as is K d . The K 0 is a mineral property. However, in most applications of Gassmanns equation, only K d is measured. Properties of the mineral modulus K 0 are often poorly understood and oversimplied. K 0 is the modulus of the solid material that includes grains, cements, and pore llings. If clays or other minerals are present with complicated distributions and structures, K 0 can vary over a wide range. Unfortunately, few measurements of K 0 have been made for sedimentary rocks (Coyner, 1984), and the moduli of clays are a particular problem (Wang et al., 1998; Katahara, 1996). Measured data (Coyner, 1984) show that at

K d = (1 D )2 K 0 ,

(25)

where D for clean sandstone is equal to 1.52. This includes an empirical expression of the normalized modulus as a direct dependence on porosity and the D parameter. Table 1 and Figure 5 show empirical relations generated from dry velocity data of relatively clean rocks. The parameter D has been derived based on approximate linear velocity-porosity relations. D represents, in the rst order, the correlation of porosity to bulk modulus for relatively clean sandstones and clastic sediments. It is related to rock texture, pressure, and uid saturation and should be calibrated for local reservoir conditions. For shaly sands, the clay effect on the modulus should be corrected before the D parameter can be derived. For consolidated rock

Figure 3. Illustration of both the Voigt high and Reuss low bounds for dry rock, uid-saturated rock, and bounds of predicted uid-saturation effect by the Gassmanns equation.

Figure 4. A typical velocity distribution for clean and shaly sands, weakly cemented sands, fractured rocks, and suspensions, in comparison with the Voigt and Reuss bounds and critical porosity (modied from Marion, 1990).

Gassmanns E & Fluid Effects on V

403

high pressures (>20 MPa), K 0 for sandstone samples ranges from 33 to 39 MPa. K 0 is not a constant and can increase by more than 10% with increasing effective pressure. Clearly, mineral modulus K 0 can vary across a wide range, depending on mineral composition, distribution, and in situ conditions.

Figure 7 shows the inuence of K 0 on Gassmanns calculation. This case uses a dry bulk modulus calculated with a mineral modulus of 40 GPa and D = 2 in the D function. The water-saturation effect was calculated for three mineral moduli, of 65, 40, and 32 GPa, and a water modulus of 2.8 GPa.

Figure 5. Normalized modulus based on Voigt and Reuss bounds and empirical D functions for different rocks (See Table 1).

Figure 7. Different mineral-frame bulk-modulus effects on calculated bulk modulus with water saturation (brine modulus: K f = 2.8 GPa) and D = 2 function model.

Figure 6. Comparison of the Gassmann and simplied Gassmann calculation for brine-saturation effect (brine modulus K f = 2.8 GPa) for the modied Voigt bound, with c = 40% and D = 2 function model. Table 1.

Figure 8. Fluid-saturation effect on calculated bulk modulus, with typical water, oil, and gas saturation and D = 2 function model.

Compiled empirical relations and relative D-por. models for different rocks. V -emp. relation Pe = 40 MPa V p = 5.41 6.35 Vs = 3.57 4.57 V p = 5.97 7.85 Vs = 4.03 5.85 V p = 5.81 9.42 Vs = 3.89 7.07 V p = 6.47 5.84 Vs = 3.39 3.03 V p = 6.19 9.80 Vs = 3.20 4.90 V p = 6.78 9.80 Vs = 3.72 5.20 K d = (1 A + B 2 C 3 ) K 0 A 3.053 3.206 3.283 2.815 4.244 3.578 B 3.070 3.349 3.284 2.639 5.820 4.020 C 1.016 1.143 1.014 0.824 2.605 1.358 K d = (1 D )2 K 0 D 1.450 ( K 0 = 32.5 GPa) 1.523 ( K 0 = 37.0 GPa) 1.584 ( K 0 = 36.0 GPa) 1.340 ( K 0 = 71.9 GPa) 1.970 ( K 0 = 66.8 GPa) 1.705 ( K 0 = 94.4 GPa)

Rock type Dry shaly sandstone Dry clean sandstone Silicate clastic (Castagna et al., 1985) Dry vuggy limestone Dry limestone Dry dolomite

* K d units are V -emp. relation.

404

Han and Batzle

Results show that for the same K d and K f , the bulk-modulus increment K d resulting from uid saturation increases with increasing mineral modulus K 0 . Errors resulting from uncertainty of K 0 can be signicant for low-porosity rocks. Because of a lack of measurements on mineral moduli, often we must use measured velocity-porosity-clay-content relationships for shaly sandstone (Han, 1986) to estimate the mineral modulus. Assuming zero porosity and a grain density of 2.65 g/cm3 , we can derive mineral bulk and shear moduli from measured P- and S-wave velocities. The results are shown in Table 2. For relatively clean sandstones (with clay content of a few percent), the mineral bulk modulus ( K 0 ) typically is 39 GPa, which is a stable value for differential pressures higher than 20 MPa. Mineral shear modulus (0 ) is around 33 GPa, which is signicantly less than the 44 GPa for a pure-quartz aggregate. Shear modulus is more sensitive to differential pressure and to clay content. For shaly sandstones, the mineral bulk modulus decreases about 1.7 GPa per 10% increment of clay content. Derived mineral bulk modulus can be used for the Gassmanns calculation, if there are no directly measured data or reliable models for calculation. Lithology detection is often a goal of seismic interpretation. In modeling the seismic response, we often face the challenge of how to separate the uids inuence from the lithology effect. To estimate this effect, we need to perform lithology substitution by using different values for the simplied gain function in Gassmanns equation:

Fluid substitution is a primary application of Gassmanns equation. With a change of uid saturation from uid 1 to uid 2, the bulk modulus increment is equal to

K 21 () G ( ) ( K f 2 K f 1 ),

(28)

where K f 1 and K f 2 are the moduli of uids 1 and 2, respectively, and K 21 represents the change in increment caused by substituting uid 2 for uid 1. Equation (28) uses the fact that the simplied gain function G ( ) of the dry-rock frame remains constant as the uid modulus changes (this may not be true for some real rocks). The uid-substitution effect on bulk modulus is simply proportional to the difference in uid bulk moduli. This form of uid substitution is similar to that derived by Mavko et al. (1998):

K f1 K s2 K s1 = K 0 K s1 (K0 K f 1) K 0 K s2 K f2 Kd = . (K0 K f 2) K0 Kd

(29)

Note that the uid-substitution effect calculated through equation (29) is based on dry-frame properties. Deriving the dry modulus often helps us to examine validity of input parameters and output results of Gassmanns calculation. If we know the simplied gain function for a rock formation, we can estimate the uid-substitution effect without knowing shear modulus:
2 2 2 V p 2 1 V p 1 + G ( ) ( K f 2 K f 1 ),

K 21 () [G 2 ( ) G 1 ( )] K f .

(27)

(30)

We should incorporate this factor into the uid-substitution scheme. However, this process is not commonly performed and is still poorly understood.
FLUID MODULUS AND FLUID-SATURATION EFFECTS

The uid modulus is another key independent parameter in Gassmanns equation. Because identication of uid types is often the primary goal of a seismic program, uid properties are of critical signicance. However, uid properties are often oversimplied in seismic applications. Although complex, uid properties are systematic. Oil properties depend on density (or API gravity), gas-oil ratio (GOR), gas gravity, pressure, and temperature conditions. Under different conditions, the uid phase and its seismic properties can vary dramatically (Han and Batzle, 2000a, b; Batzle and Wang, 1992). As we mentioned previously, Gassmanns prediction is approximately proportional to the uid bulk modulus K f , with different gain functions. Thus, the stiffer the pore uid, the higher the bulk modulus. Table 2. Grain bulk and shear modulus for shaly sands; mineral modulus derived from empirical velocities relation of shaly sandstones. C is the fractional clay content. C =0 Pd (MPa) 40 30 20 10 K0 (GPa) 39.03 39.08 39.27 38.74 0 (GPa) 32.83 31.91 30.45 26.46 C = 0.1 K0 (GPa) 37.27 37.26 37.30 36.72 0 (GPa) 29.40 28.56 27.29 25.73 C = 0.2 K0 (GPa) 35.51 35.44 35.35 34.72 0 (GPa) 26.16 25.40 24.30 22.94

where 1 , 2 , V p1 and V p2 are the densities and velocities of rock saturated, respectively, with uid 1 and uid 2. Both equations (28) and (30) are direct results from simplied Gassmanns equation (equation 8). Mavko et al. (1995) have suggested a similar method. In Figure 8, we show the typical uid-modulus effect on the saturated bulk modulus K s . Even at a modest porosity of 15%, changes can be substantial. At in situ conditions, pore uids are often multiphase mixtures. A dynamic uid modulus may also depend on uid mobility, uid distribution, rock compressibility, and seismic wavelength. Another approach is to use the P-wave modulus ( M ) to replace bulk modulus in Gassmanns equation. This works reasonably well for sandstones. The validity of this simplication results from the approximate equivalence of the ratio of dryframe bulk and shear modulus ( K d /d ) to the ratio of mineral (quartz) bulk and shear modulus ( K 0 /0 ).
ESTIMATING DRY BULK MODULUS

To estimate saturation effects on rocks practically, we need to obtain the dry bulk modulus. From Gassmanns equation, we derive K d from uid-saturated K s and uid-saturation effect Ks :

Kd = Ks

Ks = Ks

K 0 (1 K s / K 0 )2 , K0/K f + Ks /K0 1 (31)

where the uid-saturation effect K s is based on measured uid-saturated modulus K s and results in a slightly different formulation for K d in equation (2). If we reformulate K s

Gassmanns E & Fluid Effects on V

405 ACKNOWLEDGMENTS

with the Reuss bound K R of the rock with saturated uid,

Ks =

2 K0

K R (K0 Ks ) K R. + Ks K R 2 K0 K R
2

(32)

Again, the Reuss bound provides the maximum uidsaturation effect.

CONCLUSIONS

We thank the industry sponsors of Phase III Fluid Consortium for guidance and nancial support. We also thank the editor and reviewers for their comments and suggestions.
REFERENCES Batzle, M. L., and Wang, Z.-J., 1992, Seismic properties of pore uids: Geophysics, 57, 13961408. Batzle, M. L., Han, D., and Castagna, J. P., 2001, Fluids and frequency dependent seismic velocities of rocks: The Leading Edge, 20, 168 171. Berryman, J. G., 1999, Origin of Gassmanns equations: Geophysics, 64, 16271629. Biot, M. A., 1941, General theory of three-dimensional consolidation: Journal of Applied Physics, 12, 155164. Brown, J. S., and Korringa, J., 1975, On the elastic properties of a porous rock on the compressibility of the pore uid: Geophysics, 40, 608 616. Gassmann, F., 1951, Uber die elastizitat poroser medien: Vierteljahrsschrift der Naturforschenden Gesellschaft in Zurich, 96, 123. Castagna, J. P., Batzle, M. L., and Eastwood, R. L., 1985, Relationship between compressional-wave and shear-wave velocities in clastic silicate rocks: Geophysics, 50, 571581. Coyner, K. B., 1984, Effects of stress, pore pressure, and pore uids on bulk strain, velocity and permeability in rocks: Ph.D. dissertation, Massachusetts Institute of Technology. Hamilton, E. L., 1956, Low sound velocities in high porosity sediments: Journal of the Acoustical Society of America, 28, 1619. Han, D., 1986, Effects of porosity and clay content on acoustic properties of sandstones and unconsolidated sediments: Ph.D. dissertation, Stanford University. Han, D., 1992, Fluid saturation effect on rock velocities in seismic frequencies: presented at SEG Summer Research Workshop, Big Sky, Montana. Han, D., and Batzle, M., 2000a, Velocity, density and modulus of hydrocarbon uidsData measurement: 70th Annual International Meeting, Society of Exploration Geophysicists, Expanded Abstracts, paper RPB 7.2. Han, D., and Batzle, M., 2000b, Velocity, density and modulus of hydrocarbon uidsEmpirical modeling: 70th Annual International Meeting, Society of Exploration Geophysicists, Expanded Abstracts, paper RPB 7.3. Katahara, K., 1996, Clay mineral elastic properties, Expanded abstract, 66th Annual International Meeting, Society of Exploration Geophysicists, Expanded Abstracts, paper RP 1.4. Marion, D. P., 1990, Acoustic, mechanical, and transport properties of sediments and granular materials: Ph.D. dissertation, Stanford University. Mavko, G., and Mukerji, T., 1995, Pore space compressibility and Gassmanns relation, Geophysics: 60, 17431749. Mavko, G., Chan, C., and Mukerji, T., 1995, Fluid substitution: Estimating changes in V p without knowing Vs : Geophysics, 60, 1750 1755. Mavko, G., Mukerji, T., and Dvorkin, J., 1998, Rock physics handbook: Tools for seismic analysis in porous media: Cambridge University Press. Nolen-Hoeksema, R. C., 2000, Modulus-porosity relation, Gassmanns equations, and the low frequency elastic-wave response to uids: Geophysics, 65, 13551363. Nur, A., Mavko, G., Dvorkin, J., and Gal, D., 1995, Critical porosity: The key to relating geophysical properties to porosity in rocks: 65th Annual International Meeting, Society of Exploration Geophysicists, Expanded Abstracts, 878. Sengupta, M., and Mavko, G., 1999, Sensitivity analysis of seismic uid detection: 69th Annual International Meeting, Society of Exploration Geophysicists, Expanded Abstracts, Paper BHRP 7.2. Wang, Z.-J., Wang, H., and Cates, M. E., 1998, Elastic properties of solid clays: 68th Annual International Meeting, Society of Exploration Geophysicists, Expanded Abstracts, Paper RP 3.2. Yin, H., 1988, Velocity in sand-clay mixture: Stanford Rock Physics Data. Yin, H.-Z., 1992, Acoustic velocity and attenuation of rocks: isotropy, intrinsic anisotropy, and stress-induced anisotropy: Ph.D. dissertation, Stanford University.

Gassmanns equations are used widely to calculate uidsubstitution effects. Unfortunately, the underlying assumptions are often violated, and the validity of the resulting calculations is unknown. Several factors can be incorporated into the analysis to make the results more physically meaningful and reliable. 1) A simplied form of Gassmanns equation claries the physical control of uid-saturation effects on rock bulk modulus, through the simplied gain function of dry rock (dependent on the normalized dry bulk modulus and porosity) and uid modulus. Rock parameters, such as normalized dry bulk modulus and porosity, are strongly correlated, which effectively reduces the number of free parameters. 2) The dry and uid-saturated Voigt-Reuss bounds of bulk modulus provide physical limitations on Gassmanns equation. The minimum increment of bulk modulus resulting from uid saturation, consistent with the Voigt bound, is proportional to porosity and uid modulus. The maximum increment of bulk modulus with changing saturation, consistent with the Reuss bound, is equivalent to the Reuss bound of uid-saturated rock itself. This is because the Reuss bound of dry rock is zero. 3) The normalized bulk modulus can be a complicated function of rock textures and in situ conditions, which may lead to wide variations in the results of applying Gassmanns equation. However, simplied modulusporosity trends can be incorporated, resulting in simple polynomial dependence on porosity in sandstones. Fluidsubstitution effects are then a straightforward function of porosity and the difference in uid moduli. 4) The calculated dry-frame modulus or mineral modulus can be in substantial error if rock properties are not consistent or assumptions are violated. Although these relationships have been applied for many years, considerable basic research still needs to be done on many of the controlling factors. Fundamental components, such as the mineral moduli (particularly for clays), are rarely measured. The exact character of the pore-volume modulus (Brown and Korringa, 1975) is unknown, and the validity of replacing it with the mineral modulus is ambiguous. The inuence of mixed uid phases and uid mobility is also not incorporated. Additionally, theory applies strictly to the low-frequency range, thereby permitting no frequency dependence. With increasing application of seismic data to extracting and predicting reservoir and uid properties, we will need more constrained and tested forms of Gassmanns relations and other porous-media theories.