A24a PDF
A24a PDF
A24a PDF
Presented at the 60th Annual International Meeting, Society of Exploration Geophysicists. Manuscript received by the Editor August 24,
1990; revised manuscript received March 21, 1991.
∗ Formerly Department of Geology & Geophysics, The University of Calgary; presently Veritas Seismic Ltd., 200, 615-3rd Ave. SW
Calgary, Alberta, Canada T2P OG6.
‡ Department of Geology & Geophysics, The University of Calgary, Calgary, Alberta, Canada T2N ]N4.
1
Orthorhombic Physical Seismic Modeling
2
FIG. 2. The apparatus used for shear-wave splitting experiments clamps the cube of phenolic between two shear-wave transducers with mutually
parallel polarization. The cube is rotated while the transducers remain fixed. A circular protractor scale is used to determine the azimuth of
rotation.
FIG. 3. The record through Face I of a 9.6 cm cube of phenolic, FIG. 4. The record through Face 2, showing the faster S I arrival at
showing the faster S1 arrival at 1665 m/s and the slower S2 arrival 1658 m/s and the slower S, arrival at 1506 m/s. The compressional
at 1602 n-l/s. The compressional velocity in the 1-direction is velocity in the 2-direction is 3365 m/s. The polarization directions
3576 m/s. The polarization direction of the S 1 amplitude of the S1 and S2 amplitude maxima are parallel and perpendicular
maximum is parallel to the "bedding plane" of the canvas layers, respectively to the canvas layering, as in Figure 3.
while that of the S2 amplitude maximum is perpendicular to that
plane.
3
Orthorhombic Physical Seismic Modeling
through Face 2 is shown in Figure 6. This and other defined on the basis of the P-wave velocities (Figure 7).
transmission records through the phenolic show that the S1 The values quoted are group velocities based on the transit
mode generally has a greater maximum amplitude than the time measured with respect to the onset of the pulse. The
S2 arrival, indicating greater attenuation for the S2 mode. velocities are the averages of values measured through 10-
The ratios of the amplitudes of the S1 arrivals to those of and 8-cm cubes. The measured velocities for the phenolic
the S2 arrivals, measured at their maxima, have ranged from cubes were repeatable to within ±15 m/s (≈0.5 percent) for
1.1 to 1.4 for the samples tested. P-waves and ±4 m/s (≈0.25 percent) for shear waves. The
variations are likely related to small inconsistencies in the
thickness of the coupling agent used to bond the
transducers to the phenolic. Velocity variations between
different samples of phenolic ranged up to 2 percent. The
time picks used to calculate the velocities were made
directly on the digital oscilloscope for maximum accuracy.
For the following discussion, the velocities will be labeled
with 2 subscripts indicating, respectively, the directions of
propagation and polarization with respect to the three
symmetry axes (Figure 7). For example, V11 is the group
velocity for propagation and particle motion in the 1-
direction (a P wave) while V12 indicates propagation in the
1-direction with polarization in the 2-direction (an S wave).
The six shear-wave velocities measured in the principal
directions were paired as follows: V23≈V32; V31≈V13;
V12≈V21; indicating, with very small error (Table 1), only
three independent values.
FIG. 5. The record through Face 3, showing the faster S1 (1610 PRINCIPAL AXES
m/s) and slower S2 (1525 m/s) shear waves. The compressional
velocity in the 3-direction, determined separately with P-wave RAY (GROUP) VELOCITIES
transducers, is 2925 m/s. The traces for the records of Figures 3 to
5 were recorded at 5-degree intervals of rotation.
4
Orthorhombic Physical Seismic Modeling
convention (Figure 8). For the 23-plane, the direction at 45 C11 C12 C13
degrees to the 2- and 3-directions is denoted by the index 4. C22 C23
The group velocitv of the quasi-P (qP) wave in this (1)
C33
direction is thus designated V44. Polarization quasi-normal cmn =
to this 4-direction but still within the 23-plane is denoted by C44
the index 4 . Thus the quasi-SV (qSV) velocity is designated C55
V4 4 . The velocity of the corresponding SH wave, with C66
particle motion in the 1-direction, is labelled V41.
Similarly, we use the indices 5 and 6 to denote propagation of nine independent coefficients (e.g. Nye, 1985). Using
in the 31 - and 12-planes, respectively, at 45 degrees. The the elastic equations of motion the stiffnesses Cmn may be
P-, SV-, and SH-wave group velocities are thus labeled V55, estimated from the observed body-wave velocities and the
density of the phenolic (see the Appendix). The computed
V5 5 , and V52, in the 31-plane, and V66, V , and V63, in the
66 stiffnesses are summarized in Table 1. Along the principal
12-plane (Figure 8), in each instance only for the special axes the phase and group velocities are equal and the
cases of rays at 45 degrees to the symmetry directions. stiffnesses were computed directly using equations (A-45)
Each of the velocities along the diagonal raypaths is the and (A-46). Along the diagonal raypaths, the direction of
average of two measurements (between the two pairs of the wavefront normal (i.e. the slowness direction) is not, in
opposing edges of the cube) which had equivalent raypaths general, the same as the 45-degree direction of the raypath
relative to the principal axes within each of the three (i.e., of energy transport). The procedure used to compute
principal planes. The two traveltimes for each of the the slowness directions, the phase velocities, and the related
diagonal raypath pairs were virtually identical, differing by stiffnesses for the diagonal raypaths is described in the
two sample points (100 ms) or less in all cases. Four Appendix.
measurements were also recorded for raypaths from corner Nine independent velocity values, are required to enable
to corner of the cube, with similarly small differences in the complete determination of the stiffness matrix for the case
velocities. This symmetry confirmed that the presumed of orthorhombic anisotropy. These could include the three
principal planes, chosen to correspond to the planar P-wave velocities along the principal axes, three shear-
layering of the canvas fabric and the orthogonal weave of wave velocities (one of each pair, or their average) also
fibers in the phenolic, are indeed the seismic anisotropic along the principal axes, and three P-wave or SV-wave
symmetry planes. velocities, each for a raypath perpendicular to one and at 45
45° AXES degrees to the other two principal axes. In principal,
RAY (GROUP) VELOCITIES measurements at other orientations could be used but these
would require considerably more complex solutions.
Since we actually observe more than nine velocities, the
internal consistency of the orthorhombic symmetry model
can be checked. In addition to observations in this context
already mentioned above, equation (A-44) was used with
the shear-wave velocities observed along the principal axes
(Figure 7) to calculate additional independent values for the
SH-wave velocities along the diagonal raypaths (Figure 8).
For example,
(
V41′ = 2V13V12 / V132 + V122 )
12
(2)
= 1633m / s,
where V41, the observed value, is 1636 m/s, differing by
0.18 percent from V’41 the calculated value. Similarly, V’52
FIG. 8. The results of transmission measurements between
opposing edges of the phenolic cube are summarized. The =1583m/s, equal toV52, and V’63 = 1559m/s, differing by
propagation directions were at 45 degrees to two of the principal 0.l9 percent from the V63 value of 1556 m/s. Clearly, the
axes and perpendicular to the third. SH-mode velocities observed along the diagonal raypaths
conform very well to the assumed orthorhombic symmetry
ORTHORHOMBIC ANISOTROPY model.
For the orthorhombic symmetry system. the 3 x 3 x 3 x 3 The off-diagonal stiffnesses have been computed (Table 1)
stiffness tensor Cijkl (see the Appendix) may be reduced to a using the velocities measured on diagonal raypaths
6 x 6 symmetric matrix, namely: [equations (A-47)] both for P (giving Cij) and for SV
(giving cij ). The differences between pairs (Cij, cij ) seem
5
Orthorhombic Physical Seismic Modeling
rather large at first; however, the stiffness values are quite The results of these experiments indicate that the
sensitive to small velocity changes. For each wave type we orthorhombic symmetry system is appropriate to describe
compare the two velocities (one measured and one the anisotropy of this material. In particular, three distinct
calculated) corresponding to a stiffness pair (Cij, cij ). The compressional velocities were measured in mutually
orthogonal directions consistent with the visible structural
following are the relative deviations from the mean of these symmetry of the material; the observed shear-wave
velocity pairs, with the corresponding deviation for the velocities along the principal axes are equal in pairs (Vij =
stiffness in square brackets (all given as percentages) -21- Vji); the observed diagonal SH velocities agree well with
plane: ±0.3 (P), ±1.0 (SV), [±2.5]; 13-plane: ±0.6 (P), ±2.4 those calculated independently [equation (2); Table 1]; and
(SV),[ ±5.7]; 23-plane: ±0.5 (P), ±2.0 (SV), [±4.4]. the two independent determinations of each off-diagonal
stiffness (C23, C31, C12) , from a qP and a qS measurement,
are in good agreement (Table 1).
P V 22 3365 C 22 15.445
P V 11 3576 C 11 17.443
S V 12 1665 (avg)
S V 13 1602 (avg)
S V 23 1506 (avg)
∗
See equation (2)
6
Orthorhombic Physical Seismic Modeling
8.9 percent higher than V21. In the 32 plane, V is 1569 interpretation that the anisotropy conforms very closely to a
44 system of orthorhombic symmetry.
m/s, 3.5 percent higher than V32. SV anisotropy in the 31-
plane is lowest (0.9 percent) despite this plane exhibiting Physical modeling is currently proceeding with the
the strongest P and SH anisotropy. Although some of these phenolic and involves the recording of shot gathers as well
observations may seem surprising intuitively, they are all as simulated VSP and crosswell experiments. Observations
quite reasonable since the material has nine independent of the effect of orthorhombic anisotropy on moveout
stiffnesses. In the 23-plane for instance, P and SH velocities are being reported by the present authors (Brown
anisotropies depend on C22 and C33 (P) and C55 and C66 et al., 1991) and tomographic reconstruction will be
(SH) whereas the SV anisotropy depends not only on C22, described in future publications. Physical model data using
C33, and C44, but also on C23. phenolic laminate is proving to be a valuable adjunct to
numerical studies of the increasingly important topic of
Origin of the anisotropy seismic anisotropy.
The cause of the anisotropy in the phenolic laminate
appears to be related to the layering and the weave of the ACKNOWLEDGMENTS
canvas fabric. The material behaves like a stack of nets set This research is supported by funding provided by the
in a gel, with different fiber densities and orientations in the corporate sponsors of the CREWES Project. The authors
directions of the three principal axes. The many causes of wish to acknowledge the technical contributions of Mr.
anisotropy in rocks range from the microscopic to the Malcolm Bertram of the Department of Geology and
7
Orthorhombic Physical Seismic Modeling
Geophysics and Mr. Eric Gallant of the CREWES Project, Liu, E., Crampin, S., and Booth, D. C., 1989, Shear-wave
both at The University of Calgary. Finally, we would like splitting in cross-hole surveys: Modeling: Geophysics, 54,
to thank Dan Ebrom and Linda Zimmerman for their 57-65.
knowledgeable reviews of this paper and for their
Musgrave, M. J. P., 1970, Crystal acoustics: Holden-Day.
constructive suggestions, which have led to an improved
final version. Nikitin, L. V., and Chesnokov. E. M., 1984. Wave
propagation in elastic media with stress-induced
REFERENCES anisotropy: Geophys. J. Roy. Astr,. Soc., 76, 129-133.
Aki, K., and Richards, P. G., 1980, Quantitative Nye, J. F., 1985, Physical properties of crystals: Oxford
seismology: Theory and methods: W. H. Freeman and Co. University Press.
Banik, N. C., 1984, Velocity anisotropy of shales and depth Rathore, J. S., Fjaer E., Holt, R. M., and Renlie, L., 1990.
estimation in the North Sea basin: Geophysics, 49, 1411- Acoustic anisotropy of synthetics with controlled crack
1419. geometries: Presented at the 4th Internat. Workshop on
Brown, R. J., Lawton, D. C., and Cheadle, S. P., 1991, Seismic Anisotropy.
Scaled physical modelling of anisotropic wave propagation: Sayers, C. M., 1988, Stress-induced ultrasonic; wave
multioffset profiles over an orthorhombic medium: velocity anisotropy in fracture rock: Ultrasonics, 26, 311-
Geophys. J. Internat., in press. 317.
Bullen, K. E., 1963, An introduction to the theory of Tatham, R. H., Matthews, M. D., Sekharan, K. K., Wade,
seismology: Cambridge University Press. C. J., and Liro, L. M., 1987, A physical model study of
Crampin, S., 1981, A review of wave motion in anisotropic shear-wave splitting and fracture intensity: 57th Ann.
and cracked elastic-media: Wave Motion, 3, 343-391. Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
-- 1984, An introduction to wave propagation in anisotropic 642-645.
media: Geophys. J. Roy. Astr. Soc., 76, 17-28. Thomsen, L., 1986, Weak elastic anisotropy: Geophysics,
-- 1985, Evaluation of anisotropy by shear-wave splitting: 51, 1954-1966.
Geophysics, 50, 142-152.
-- 1989, Suggestions for a consistent terminology for ----- 1988, Reflection seismology over azimuthally
seismic anisotropy: Geophys. Prosp., 37, 753-770. anisotropic media: Geophysics, 53, 304-313.
Ebrom, D. A., Tatham, R. ll., Sekharan. K. K., McDonald, Vlaar, N. J., 1968. Ray theory for an anisotropic
J. A., and Gardner, G. H. F., 1990. Hyperbolic traveltime inhomogeneous elastic medium: Bull. Seis. Soc. Am., 58,
analysis of first arrivals in azimuthally anisotropic rnedium: 2053-2072.
A physical modeling study: Geophysics, 55, 185-191 Winterstein, D. F., 1990, Velocity anisotropy terminology
Ensley, R. A., 1989, Analysis of compressional- and shear- for geophysicists: Geophysics, 55, 1070-1088.
wave seismic data from the Prudhoe Bav Field: The leading Yale, D. P., and Sprunt, E. S., 1989, Prediction of fracture
Edge, 8, no. 11, l@13. direction using shear acoustic anisotropy: The Log Analyst,
Fedorov, F. I., 1968, Theory of elastic waves in crystals: 30, 65-70.
Plenum Press.
APPENDIX
Helbig. K., 1983, Elliptical anisotropy - Its significance and
meaning: Geophysics, 48, 825-832. RELATIONSHIPS AMONG VARIOUS ELASTIC-
WAVE PARAMETERS IN AN ANISOTROPIC
Keith, C. M., and Crampin, S., 1977, Seismic body waves MEDIUM OF ORTHORHOMBIC SYMMETRY
in anisotropic media: Reflection and refraction at a plane
interface: Geophys. J. Roy. Astr. Soc., 49, 181-208. Basic theory and the Kelvin-Christoffel equations
The equations of motion governing wave propagation in a
Kendall, J.M., and Thomson, C. J., 1989, A comment on
generally isotropic elastic medium are given by many
the form of the geometrical spreading equations, with some
authors (e.g. Bullen, 1963; Fedorov, 1968; Musgrave,
numerical examples of seismic ray tracing in
1970- Aki and Richards, 1980; Crampin, 1981, 1984; Nye,
inhomogeneous, anisotropic media: Geophys. J. Internat.,
1985) For infinitesimal displacements ui, Cartesian
99, 401-413.
coordinates xi, density ρ, stress tensor σij and body forces
Lewis, C., Davis, T. L., Vuillermoz, C., Gurch, M., per unit mass gi:
Iverson, W., and Schipperijn, A. P., 1989,Three-
dimensional imaging of reservoir heterogeneity, Silo Field, ρu&&i = σ ij , j + ρg i (A-1)
Wyoming: 59th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, 763-765.
8
Orthorhombic Physical Seismic Modeling
where ",j" denotes the partial derivative with respect to xj will be real. (Throughout this appendix vertical bars denote
and where the Einstein summation convention (for repeated
determinant.)
indices) applies.
A further consequence of the symmetries embodied in (A-
The stress tensor, in terms of the strain tensor εkl and the 8) is that there are only 21 independent stiffnesses, Cijkl.
stiffness tensor Cijkl, is given in accordance with Hooke’s Following, e.g., Musgrave (1970), Nye (1985), and
law by: Thomsen (1986), the fourth-order stiffness tensor may be
σ ij = C ijkl + ε kl , (A-2) written as a second-order (6 x 6) symmetric matrix:
where C ijkl → C mn
ε kl = 1
2
(u l ,k + u k ,l ) . (A-3) where
These equations of motion, and their solution for By introducing the so called Kelvin-Christoffel stiffnesses,
monochromatic plane-wave motion, are considered by given by Musgrave (1970) as:
many authors (e.g., Fedorov, 1968; Musgrave, 1970; Keith Γ ik = C ijkl n j n l (A-11)
and Crampin, 1977; Aki and Richards, 1980; Crampin,
1981, 1984) but here we follow Musgrave’s treatment most
closely. We assume harmonic plane-wave displacement,
expressed as equations (A-7) and (A-9) may be rewritten as:
u k = Ap k exp [i ω (s r x r − t )] , (A-5)
Γ11 − ρv 2
Γ12 Γ13 p1
(A-12)
Γ21 Γ22 − ρv 2 Γ23 p 2 = 0
where A is the amplitude factor, pk is the unit polarization Γ31 Γ32 2
Γ33 − ρv p 3
(or particle displacement) vector, ω is angular frequency, sr
is the slowness vector, and in this equation only, i = − 1 . and
The slowness vector gives the direction of the wavefront Γ11 − ρ v 2
Γ12 Γ13
normal and may further be written: (A-13)
Γ21 Γ22 − ρ v 2 Γ23 = 0
s r = v −1 n r , (A-6) Γ31
Γ32 Γ33 − ρ v 2
where v is phase velocity and nr is the unit slowness (or Equations (A-12) and (A-13) are known as the Kelvin-
wavefront-normal) vector. From equation (A-4), (A-5), Christoffel equations.
and (A-6) one obtains:
In the case of orthorhombic symmetry only nine of the, in
(C ijkl n j nl − ρv δ ik p k = 0
2
) . (A-7) general, 21 indepent stiffnesses, Cmn, are nonzero. The six
independent Kelvin-Christoffel stiffnesses are then
Thus, the determination of the details of the wave motion Γ 11 = n 12 C 11 + n 22 C 66 + n 32 C 55
has been cast as an eigenvalue problem in which, having Γ 22 = n 12 C 66 + n 22 C 22 + n 32 C 44
specified Cijkl (the stiffnesses of the medium) and nr (the (A-14)
direction of phase propagation), one can solve for pk (the Γ 33 = n 12 C 55 + n 22 C 44 + n 32 C 33
9
Orthorhombic Physical Seismic Modeling
and
Ak = Γkk (no summation) . (A-27)
10
Orthorhombic Physical Seismic Modeling
ρ vV 3 = p 22 n 3 C 44 + p 32 n 3− 1 (ρ v 2 − n 22 C 44 ) .
Further, from equation (A-12) for this case we get:
p2 n n (C + C 44 ) ρv 2 − n 22 C 44 − n32 C 33
= 2 2 3 2 23 =
p 3 ρv − n2 C 22 − n32 C 44 n2 n3 (C 23 + C 44 )
(A-36)
from which
p 22 ρ v 2 − n 32 C 33 − n 22 C 44
= . (A-37)
p 32 ρ v 2 − n 22 C 22 − n 32 C 44
since the raypath or group-velocity direction is at 45
degrees to the 2- and 3-axes, V2 = V3 and the two right-hand
sides of equations (A-35) are equal. From this and equation
(A-37) one can eliminate p2 and p3 and obtain:
11
Orthorhombic Physical Seismic Modeling
( ) [ (
V444 (n 2 + n3 ) n32 − n 22 + 2V442 (n 2 + n3 ) n 23 V222 + V232 ,
2
) Expression for stiffnesses in terms of group velocities
( )] (
− n33 V332 + V232 + 4 n34V332V232 − n 24V222 V232 = 0 ) For completeness, expressions for the nine stiffnesses, for
the case of orthorhombic symmetry, are here summarized.
(A-40) These equations follows directly from (A-21) to A-23), (A-
36), and (A-39), as well as their cyclic variant.
in which all of the Vij have been measured experimentally.
Now, since n 2
2
+ n32 = 1 , equation (A-40) can, in C 11 = ρ V 112
principle, be solved for n2 and n3. In practice, we determine C 22 = ρ V 222 (A-45)
n2 and n3 by iterative substitution. Knowing n2 and n3 and C 33 = ρ V 332
v44 from equation (A-39), equation (A-36) and (A-37) can
be solved for C23 and the polarization, p2/p3. Similarly,
C 44 = ρ V 232 = ρ V 322
using V44 (qSV ) in (A-40), we will get different values
C 55 = ρ V 312 = ρ V 132 (A-46)
(in general) for n2, n3 and v 44 ; but assuming the C 66 = ρ V 122 = ρ V 212
orthorhombic model to be a reasonable one, we should get
the same result for C23.
C 23 =
ρ
n 2 n3
{[
1
2
(n2 + n3 )2 V442 − n32V232 − n32V332 ] (A-47a)
The SH wave. – Choosing equation (A-34) instead of (A-
33) we have, from (A-14) and (A-31):
[1
2
(n2 + n3 )2 V442 − n22V222 − n32V232 ]}
12
− ρ V232
ρv 2
= n C 66 + n C 55 = Γ11
2 2
(A-41)
ρ
{[ (n ]
41 2 3
12