Zhengyun Zhou

A comparison between methods that discriminate fluid content in unconsolidated sandstone reservoirs

Three seismic attributes commonly used to predict pore fluid and lithology are the fluid factor ( ΔF ) , Poisson impedance (PI), and lambda-rho ( λρ ) . We evaluated the pore-fluid sensitivity of these attributes with both well-log and seismic data in Tertiary unconsolidated sediments from the Gulf of Mexico where sand and shale are the only expected lithologies. While the sensitivity of one attribute versus another to discriminate pore fluid is often debated in the literature, the sensitivities of the three attributes are not independent but can be traced back to the fluid factor, which is a function of the P- and S-wave normal-incident reflection coefficients. Interestingly, the fluid factor, which is a reflectivity attribute, at the top of a hydrocarbon-saturated reservoir, is basically independent of the shale properties above the reservoir. It is a function of the brine and hydrocarbon impedances of the reservoir. The next attribute, Poisson impedance, is thenequal to the fluid factor times the sum o...

Water-saturation estimation from seismic and rock-property trends

Summary The ability to estimate water saturation for a thin-bed reservoir from seismic is greatly enhanced if two rockproperty transforms are employed. One transform linearly relates the wet normal-incident reflectivity [NI(wet)] to the hydrocarbon NI. The other transform relates the far-trace amplitude to NI for each saturation state. These transforms are derived from rock-property trends that are local to the prospect. With these two transforms and the AVO gathers at the prospect and at the down-dip water-equivalent reservoir, a test statistic can be developed that differentiates economic gas from fizz saturation. The methodology doesn’t require a calibration well that ties the seismic unless the bed thickness is desired.

Is there a basis for all AVO attributes

Well-log data indicate strong linear relationships exist between acoustic impedance (AI) and shear impedance (SI) for different lithologies and pore-fluid saturations. In fact, linear relationships also exist between normal-incident reflection coefficients for P wave (NIP) and for S wave (NIS). However, the basic underlying attribute that permits the discrimination can be traced back to the fluid factor discussed by Smith and Gidlow (1987).

Pore-fluid Quantification: Unconsolidated Vs. Consolidated Sediments

The fluid factor (ΔF), Poisson Impedance (PI) and lambda-rho (λρ) are three seismic attributes commonly used for pore-fluid discrimination. While it is debated that one attribute is a better pore-fluid discriminator than the other, their sensitivity to pore fluid is not significantly different because the sensitivity of the three attributes are directly related to PI. In addition, the sensitivity of the three attributes is increased if attribute calibration is finalized in the horizon domain. The effectiveness of each attribute was tested for unconsolidated sediments in the Gulf of Mexico and consolidated sediments in the Southern Gas Basin of the North Sea. However, in the Paleozoic consolidated sediments of SGB, the extra sensitivity of the impedance attributes (PI andλρ) appear to be needed for a robust estimate of pore fluid. Introduction Pore-fluid and lithology prediction are two objectives in amplitude-versus-offset (AVO) analyses. Smith and Gidlow (1987) defined the fluid factor as the weighted difference between the reflectivities of Pand S-wave velocity. Goodway et al. (1997) introduced the pore-fluid attribute λρ along with the lithology attribute μρ, where λρ is related to the acoustic impedance AI and shear impedance SI as λρ = (AI) – 2(SI). Hilterman (2001) and Hedlin (2000) related λρ to Gassmann’s pore-space modulus and Russell et al. (2003) noted that λρ would be a better pore-fluid discriminator if the factor 2 in its equation relating AI and SI was varied depending on the local rock properties. Quakenbush et al. (2006) showed that the axes of the crossplot between SI vs. AI can be rotated (“pore-fluid projection”) and the new attribute defined by the rotated abscissa is a good pore-fluid discriminator, which he called Poisson impedance. Our objective is to quantify the pore-fluid sensitivity of these three AVO attributes. Two Desirable Pore-Fluid Measurements The linear approximation of Zoeppritz’s equation can be expressed as a function of the incident angle θ as RC(θ) ≈ NIVEL/cos(θ) + NIDEN + NIRIGsin(θ) (1) where RC is the reflection coefficient, NIVEL and NIDEN are reflectivities associated with the velocity and density variations across the interface. NIRIG = -2Δμ sin(θ)/ρα, where Δμ is the shear rigidity difference between the lower and upper media and ρ and α are the average velocity and density of the upper and lower media. Figure 1: Zoeppritz responses for a shale over sand interface when the sand is saturated with brine, fizz (low gas saturation) and gas. In Figure 1, the downshift of the fizz response from the wet response is primarily due to the NIVEL term in equation 1 when gas is introduced. The third term with NIRIG is approximately the same for all three pore fluids. Additional gas lowers the AVO Zoeppritz response to the “gas” curve. The difference between the economic gas and fizz curves is mainly caused by the changes in NIDEN. The density contribution is the same at all angles during the transition from fizz to gas saturation. The discrimination of pore fluid is mainly in the measurement of (NIGAS-NIWET) or (NIFIZZ-NIWET) and not the slope. The second desired measurement is the pore-space modulus of Gassmann’s equation, which is related to λρ. Our task is to develop seismic attributes that relate to (NIGAS-NIWET) and the pore-space modulus. These are investigate with well-log data first. Well Measurements in Unconsolidated Sediments In the northern portion of the Gulf of Mexico (GOM), velocity and density values for wet sand and shale were extracted from 150 wells at 60-m intervals. From 27003300-m depth, the rock properties of 183 brine-saturated reservoirs and the encasing shale were selected. These are mainly unconsolidated Pliocene-Miocene sediments. Fluid-substitution rock properties were generated for all 183 reservoirs (estimated S-wave velocity). The normalincident P-wave (NIP) and S-wave (NIS) reflection coefficients for brine-, oiland gas-saturations are illustrated in Figure 2a. In Figure 2b, the fluid factor is based on a rotation of the axes (ΔF=NIP-0.72NIS) while in Figure 2c the fluid factor has an additional translation term (ΔF=NIP-0.72NIS+0.03), which has an expected value of zero for the fluid factor from brine-saturated -0.3 -0.2 -0.1 0 0 10 20 30 40 50 Incident Angle (Deg) Am pl itu de 93% NIVEL

Stringent assumptions necessary for pore-fluid estimation

Summary We investigated the sensitivity of several assumptions associated with a new method for predicting pore-fluid type and saturation using seismic amplitudes and rock-property transforms. The method derives regression coefficients for rock-property transforms from well-log curves and then applies the transforms to the near- and far-angle seismic stacks to yield estimates of the normal-incident reflection coefficient (NI). The NI values are then related to watersaturation (Sw). The sensitivity of two assumptions associated with the method is evaluated. The first is that the porosity of the prospect and down-dip reservoir is the same. The second is that the ratio of the far-angle amplitude is scaled properly to the near-angle amplitude during processing. A field example from South Marsh Island (SMI) demonstrates the method and the sensitivity study.