Significant stress changes are generated when producing reservoirs compact due to large reductions in the reservoir pore-pressure. These stress changes are not confined to the reservoir. The stress and strain is redistributed to the surrounding formations, modifying both velocity and thicknesses in these formations. These changes often manifest themselves as significant timelapse time differences on migrated 4D images.
Various authors (Hatchell et al, 2003 and Barkved et al, 2005) have used geomechanical modelling to explain these 4D timeshifts, thereby gaining valuable insight into the behaviour of the whole subsurface around some compacting reservoirs. This has so far been accomplished by assuming a simple relationship between thickness and velocity changes. The modelling is presumably repeated using various updated relationships until a match is obtained with the observed 4D timeshifts.
We present an approach in which the 4D time differences are measured on prestack data. Without relying on any assumed relationship between velocity and thickness changes, we use the additional non-zero offset information combined with raytracing and linear least squares inversion techniques to derive the thickness and velocity changes. These resulting velocity and thickness changes combined with density and pore-pressure well data can then be converted to stress and strain changes. The technique should therefore help to close the loop between seismic 4D time differences and geomechanical stress and strain changes.
Egozi, Uzi (Veritas DGC) | Yates, Matt (Veritas DGC) | Omana, Jose (Veritas DGC) | West, Bruce Ver (Veritas DGC) | Burke, Nick (BP) | Mesa, Mario (BP) | Moreno, Eduardo (BP) | Checa, Jamie (BP) | Martinez, Monica (BP) | Alfonso, Hector (Ecopetrol) | Calderon, Jose E. (Ecopetrol)
We propose a new workflow for anisotropic depth imaging. We developed a new technology, a tilted transverse isotropy (TTI) well mis-tie (WMT) tomography to determine the anisotropy. It was successfully incorporated into a workflow with multi azimuth reflection tomography and TTI Kirchhoff prestack depth migration (PSDM) to produce a good velocity model. The image produced with this velocity model is reducing the risk for future well drilling decisions and has already made impact on some of those decisions.
The structural complexity of this Colombian land survey and the resulting challenge of the velocity model building required leading edge technologies. Incorporating the well data from this producing field, we designed a workflow to meet the main processing objectives:
(1) Build a velocity model that will produce a better depth image and good fault termination without fault shadows.
(2) Determine the anisotropic velocity model parameters, V0, e, d, θ and φ, that will lead to flat depth migrated gathers after TTI Kirchhoff PSDM.
(3) Use reflection tomography for the velocity inversion.
(4) Tie all the well-tops with errors of 1% or less.
The Cusiana Cupiagua Sur project is a merged land dataset. The Cupiagua 3D program was acquired in 1995 and 1996 with 4 cables along SE-NW direction. The natural bin size was 20x20 meter, nominal fold was 20 and far offset along the cables was 3160 meters. The Cusiana 3D program was acquired 900 to the Cupiagua program, in SW-NE direction. It was acquired in 1997 and 1998 with 6 cables. The natural bin size was 15x15 meter, nominal fold was 12, and far offset along the cables was 3585 meters. There are 165 exploration and production wells, some of which have sonic logs and checkshots; many have gamma ray logs and most have well tops.
Since most of the anisotropy is a result of the sediment bedding (Uhrig and Van Melle, 1955), the anisotropy was assumed to be TTI above the Yopal fault and VTI below it. The velocity model area was then divided into four major parts: the area right below the topography, the area above the Yopal fault, the area below the Yopal fault and above the Cusiana fault, and the prospect zone.
Initial Velocity Model
We derived a refraction tomography velocity-static solution by using the first break picks with initial two-layer static model. The velocity from the refraction tomography solution was extracted. The long wavelength part of the static solution was removed, so only the short wavelength static was applied to the prestack data. The velocity model was then trimmed deeper than 200 meter below sea level and smoothed. The PSTM RMS velocity field was converted to interval velocity in depth using Dix (1955) formula. The refraction tomography velocity model was merged together with the Dix converted RMS velocity model by replacing the converted RMS velocity with the shallow refraction tomography velocity model. The velocity model was then vertically calibrated.
During June 2004 well A-1 was drilled in the onshore Coastal Swamp area of Nigeria (Figure 1). The target of the drilling had been planned on seismic, which had rudimentary AVO processing applied. Other wells in the area had shown good correlation between the seismic and hydrocarbon finds. However on well A-1 although hydrocarbons were detected they did not tie with the AVO anomalies seen in the seismic data. Two more wells were about to be drilled from the same location into adjacent fault blocks so it was necessary to be sure that the seismic was telling the full picture. A volume of data around the well site was extracted and reprocessed through a sequence that included high resolution velocity analysis and surface consistent scaling. Intercept and Gradient data were extracted to calculate a product gradient volume. Although anomalies could be seen on the product gradient it was not until the attributes were inverted and the calibrated Lambda-Rho volumes calculated that the full potential of the seismic was revealed.
Summary Test results indicate that single sensor three component receivers provide better data than conventional groups of geophones in rough terrain. The images from the 3C data have better resolution and better imaging of dipping reflectors. Arguably, the benefits of the 3C data are much more significant than the small increase in random noise that we observe on the single sensor data. Introduction Imaging targets under rough terrain like the mid continent overthrust belt in the USA is a challenge. Recent advances in acquisition technology include single sensors and multicomponent systems.
A method is presented for performing fluid substitution without knowledge of the S-wave velocity that is based on the Biot-Gassmann relationship. In order to perform this, the Biot-Gassmann relationship is rearranged in terms of the measurable properties of P-wave velocity and density. The user must supply the porosity, the Biot coefficient, the bulk modulus of the fluid and the bulk modulus of the solid. All but the Biot coefficient can be estimated or calculated using established techniques. Thus, the key parameter containing the non-uniqueness in the problem is the Biot coefficient. The Biot coefficient describes pore space stiffness and hence how the rock responds to changes in fluid, porosity, and pore type. Several methods to estimate the Biot coefficient are discussed. Further, the Biot coefficient has well-defined upper and lower bounds, limiting the range of allowable solutions.
In addition the critical angle also plays an important role for heavy oil plays in northeastern Alberta. Reflections at the Paleozoic often become critical due to the large velocity contrast between the overlying clastics and underlying carbonates. This becomes a problem because the large supercritical reflection coefficients obscure the overlying zone of interest. The traditional way of dealing with this is to limit the range of angles used in the AVO inversion. This is problematic since the critical angle may be as low as 25 degrees, again limiting reliability of the estimates of the AVO inversion. Limiting the angles used in the AVO inversion is due to the common and mistaken belief that the Aki and Richards (1980) linearized approximation (equation 5.44) of the Zoeppritz equation is only valid for subcritical angles of incidence. For example, de Nicolao et al. (1993) state that the linearized approximation is only valid for precritical angles. Aki and Richards (1980) state as a precondition for using the linearized approximation (equation 5.44) that the angles of incidence and transmission must be less than 90 degrees, thus precluding the critical angle.
Much interest in time-lapse results hinges on time-lapse AVO. This paper therefore investigates the sensitivity of prestack repeatability to positioning issues. The normalised rms of the difference data (nrms), commonly used as a measure of repeatability, requires some understanding, and accordingly the first part of this work develops a model of its behaviour as the data signal to noise ratio varies, and time shifts and amplitude variations occur between datasets.
Analysis of a near offset volume from a North Sea survey indicates that most variability in nrms measured at the target level in a given bin, comes from changes in source-receiver azimuth. Changes in the inline or crossline position of both sources and receivers are much less significant. However, a large part of the non-repeatability of the seismic data appears to be unrelated to positioning, instead being a function of the signal to noise ratio and other factors related to both acquisition and geology.