Relationships among elastic-wave values (R pp, R ps, R ss, Vp, Vs, ρ, σ, κ)

Transcription

1 Relationships among elastic-wave values (R pp, R ps, R ss, Vp, Vs, ρ, σ, κ) Robert R. Stewart, Qi Zhang, Felix Guthoff * ABSTRACT The average Vp/Vs value of a set of layers is a weighted sum of the interval velocity ratios. The average value is also bounded by the maximum minimum interval values. The average value will change according to changes in the target layer. The thicker the layer, the greater its influence on the average value. The approximate equations for converted-wave reflectivity R ps pure-shear reflectivity R ss can be combined to give a simple relationship between the two: R ps ~ 4sin(θ)R ss, where θ is the reflection angle of the shear wave. The normalized elastic parameters ( α α,, ρ ρ ) are estimated from summed P-P P-S reflection coefficients using a linear inversion method. ormalized Poisson s ratio ( σ σ ) Lamé parameters ( λ λ, κ κ ) can then be computed from the estimated velocity density changes. AVERAGE Vp/Vs VALUE OF MULTIPLE LAYERS Often in seismic analysis we extract a low-resolution or macroscopic parameter, such as average velocity, which is dependent on higher resolution values such as interval velocities. We may thus be interested in understing how the micro-values effect the macro-parameters. In this case, how do interval P- S- velocity ratios affect the average velocity ratio. This is interesting for several reasons. One is when picking events isochrons on P S sections, we often take several cycles between picked events (Miller et al., 1995). This means that a series of layers are entering into the isochrons, isochron ratios thus overal Vp/Vs calculation. The question is how does the overall or average Vp/Vs value relate to the interval Vp/Vs values? Average Vp/Vs calculation Suppose that we have a layered medium (with layers, ) having P-wave S- wave interval velocities (α i, i ). Each layer has thickness zi a set of transit times: s for one-way P waves t i for one-way S waves (Figure 1). t ip What is the average velocity ratio for the whole section? Let s first define an average Vp/Vs value as trhe ratio of average velocities (after Sheriff, 1984):. * Institute für Geophysik, Münster, Germany CREWES Research Report Volume 7 (1995) 10-1

2 Stewart, Zhang Guthoff γ Z T Tpp, (1) Z Ts where Z is the total depth travelled, Tp is the one-way P-wave travel time to depth Z, Ts is the one-way S traveltime from Z to the surface, then γ = Ts T s T p. (2) But t i s = γ i t ip, γ = s t i Σ T p = Σ γ i t i p T p (3) γ = Σ γ i r i (4) p where r i =t i Tp T p or the fractional transit time. t i p t i s γ 1 = α 1 1 α 1, 1 α 2, 2 α j, j Z α k, k γ = α α, Fig. 1. Plane-layer elastic medium with layers. Thus, the average Vp/Vs value is the transit-time weighted sum of the interval velocity ratios. Furthermore, γ will be bounded by the minimum maximum interval ratios ( γ i ) as shown below: 10-2 CREWES Research Report Volume 7 (1995)

3 γ = γ i r i min(γ i )r i = min(γ i ) r i = min(γ i ) (5) γ = γ i r i max(γ i )r i = max(γ i ) r i = max(γ i ) (6) Thus, min (γ i ) γ max (γ i ). Examples We can take several examples to show the effect of a variable velocity layer on the average Vp/Vs value. The medium s velocities are given in Table 1, Figure 2 shows the results. We see that for an isochron ratio or average Vp/Vs determination across a thick stack of layers, say 130 m, with only a 10 m target interval, there is little impact of that layer. On the other h, a 100 m target layer has a large influence on the final Vp/Vs value. Table 1. Five layer elastic model to compute the average Vp/Vs value. Layer Thickness (m) Vp (m/s) Vs (m/s) Vp/Vs Two more examples, directly related to current field cases are shown. We observe the effects of altering the reservoir thicknesses Vp/Vs values for the Lousana isku case Blackfoot s channel example. Table 2. Average Vp/Vs values for Lousana isku case Layer Thickness (m) Vp (m/s) Vs (m/s) Vp/Vs Wabamun salt Calmar shale isku anhydrite isku porous dolomite isku tight dolomite A thick porous dolomite influences the average Vp/Vs value significantly (Figure 3). A thin porous region will have an effect that is probably lost in the noise of real data. CREWES Research Report Volume 7 (1995) 10-3

4 Stewart, Zhang Guthoff Table 3. Average Vp/Vs values for Blackfoot s channel Layer Thickness (m) Vp (m/s) Vs (m/s) Vp/Vs Mannville Glauconitic channel Basal quartz Results from the model of Table 3 are shown in Figure 4. If we assume that we can pick real variations in Vp/Vs down to about 0.05, then a Glauconitic s with thickness above about 10 m should produce an anomalous Vp/Vs value z3= z3= z3=50 Average Vp/Vs z3=30 z3= Vp / Vs of third layer FIG. 2. Variation of the average Vp/Vs value over the 5 layer model (Table 1) with changes in the third layer Vp/Vs value CREWES Research Report Volume 7 (1995)

5 Average Vp /Vs z3=5 z3=10 z3=20 z3=30 z3= Vp /Vs of third layer FIG. 3. Variation of the average Vp/Vs value from the Lousana isku model (Table 2) Average Vp / Vs z3=5 z3=15 z3=25 z3= z3= Vp / Vs of third layer FIG. 4 Variation of the average Vp/Vs value across the Blackfoot s channel model (Table 3). CREWES Research Report Volume 7 (1995) 10-5

6 Stewart, Zhang Guthoff A APPROXIMATE RELATIOSHIP BETWEE R ps AD R ss We may have a converted-wave reflectivity section or a pure shear section, depending on what survey was conducted. Because both surveys relate to shear-wave properties it is natural to ask several questons: Which is better? Cheaper? How do they relate to each other? The first question is the subject of much current interest study. More on it soon. The second question has an answer it generally will be the P-S survey, as it is basically a stard P-P survey with just a different geophone. What the relationship is between R ps R ss can be shown as below. The equations from Aki Richards (1980) that approximate the converted-wave reflectivity R ps pure-s reflectivity R ss are given as: R ps θ (4 2 p 2 = pα 2cos θ [(1 22 p cos ψ α 4 2 cos ψ α cos θ ) ρ ρ cos θ ), (7) ] R ss θ = 2 1 (1 42 p 2 ) ρ ρ ( 1 2cos 2 θ 42 p 2 ) (8) Suppose that ψ, thus θ p are small then R ss ~ 1 2 ( ρ ρ + ) (9) R ps ~ pα 2 [(1 + 2 α ) ρ ρ +4 α ] = pα 2 [4 α ρ ρ +4 α 4 α +(1 2 α ) ρ ρ ] = pα 2 [ 8Rss α +(1 2 α ) ρ ρ ] ow as α 1 2, the second term in the equation is very small. Thus R ps θ ~ 4sin θ R ss θ (10) So the converted-wave reflectivity is approximately related to the pure-s reflectivity CREWES Research Report Volume 7 (1995)

7 COUPLED P-P AD P-S IVERSIO Suppose that we are trying to estimate α α from the Aki Richards (1980) equations. We could first exp the equations in powers of the ray parameter p: R pp ~ R o pp p 2 (2 2 ρ ρ α2 4 α α +42 ) (11) where Then where R o pp = 1 2 ( ρ ρ + α α ), R ps ~ pα p2 ( ) [(1 + 2 α ) ρ ρ + 4 α ]. (12) R pp +R ps = C o +C 1 p + C 2 p 2 + C 3 p , (13) C o =R o pp, C 1 = α 2 [(1 + 2 α ) ρ ρ + 4 α ], C 2 = (2 2 ρ ρ α2 4 α α +42 ), C 3 = α2 4 [(1 + 2 α ) ρ ρ + 4 α ]. This could be analysed now by a polynomial line fit up to the third power of the variation of R pp + R ps with offset (p). Once we have estimates of C 0,...C 3, then we can pose the problem as a matrix inverse. ote that C 3 is directly related to C 1 will not constrain the problem. So: C 0 C 1 C 2 = A 0 B CDE ρ ρ α α (14) C = GP (15) where A, B, C, D, E are functions of α,,ρ. We could estimate the scaled elastic changes by using a damped (with ε 2 ) leastsquare solution, for example: CREWES Research Report Volume 7 (1995) 10-7

8 Stewart, Zhang Guthoff P =(G T G) G T G+ε 2 1 G T C (16) Once we have the normalized density velocity changes from equation (16), we may be interested in finding other elastic parameter changes such as Poisson s ratio the Lamé parameters. POISSO S RATIO AD LAMÉ PARAMETER VARIATIO Poisson s ratio σ can be written as a function of the P- S-wave velocities as: where σ= α (α 2 2 ) = γ2 2 2(γ 2 1) γ = α (17) How do variations in α effect σ? Suppose that changes in the velocities α are small then: σ = σ σ α + (18) α Using equations (17) (18), the relative variation of σ is: σ σ = 2α 2 2 (α 2 2 )(α ) ( α α ). (19) We note that if α =c, where c is a constant, then α α = changes in α produce no change in Poisson s ratio ( γ is constant). If, however α =c+d, where c d are constants, then: α α = α α d c (20) σ σ changes accordingly. We recall that α 2 =(λ+2µ)/ρ, (21) 2 = µ / ρ, (22) where µ λ are the Lamé parameters. ow, it may be useful to isolate λ for petrophysical analysis as λ may depend more directly on pore fill CREWES Research Report Volume 7 (1995)

9 λ = ρα 2 2ρ 2 (23) Exping equation (23) for small velocity density changes gives Again, a λ λ changes. λ λ =( 2 ( α α 2 2)[α2 α ) 22 ( )] + ρ ρ (24) section might have less influence of lithology highlight pore-fill For completeness, we can also estimate the incompressibility (bulk modulus) changes: κ =λ µ = ρα2 4 3 ρ2, (25) κ κ =( 2 ( α α 2 2)[α2 α ( )] + ρ ρ (26) COCLUSIOS The average Vp/Vs value is a weighted sum of the interval velocity ratios. The average value is also bounded by the maximum minimum interval values. It will change according to changes in the target layer. The thicker the layer, the greater its influence on the average value. The shear reflectivities (R ps R ss ) are approximately related by a simple sine function. It should be possible to construct one section given the other. A method is presented to calculate normalized velocity density changes from P-P P-S reflectivity coefficients. Using these changes we can also calculate changes in Poisson s ratio, incompressibility, the Lamé parameters. ACKOWLEDGEMET Dr. Clint W. Frasier of Chevron Petroleum Technology Corporation derived a similar result to equation (10) independently in an internal Chevron memo dated May 10, REFERECES Aki, K. Richards, P.G., 1980, Quantitative seismology: Theory methods: W.H. Freeman Co., v.1. Miller, S.L.M., Lawton, C.D., Stewart, R.R., Szata, K., 1995, Interpreting P-S seismic data from Lousana: CREWES Research Report, ch. 10, v. 7. Sheriff, R.E., 1994, 2nd ed., Encycolpedic dictionary of exploration geophysics: Soc. Expl. Geophys. CREWES Research Report Volume 7 (1995) 10-9