|
June/July 2003 |
| Technical Focus | |
|
Fig.
14. Sketch of a recumbent slump fold, adapted from Reading (1986). Note
that the down slope dip is opposite to that of the fold limbs.
Fig.
15. Example of a dip stereoplot (upper hemisphere projection) displaying
typical deep marine feature classified sedimentary dips.Fig.
16. Example of the construction of an azimuthal vector walkout plot using
dip azimuth data.
Fig.
16. Example of the construction of an azimuthal vector walkout plot using
dip azimuth data.
Fig. 17. Example of an optimum televiewer image resolving thin (3-6 cm) sandstone beds and lenses for net sand determination, from Lawrence (2002).
Fig.
18. Interpreted geological relationships through one sand body imaged
three times by a RAB tool.
Fig.
19. Illustration of the application of bed top and base dip geometry to
calculate distance to bed feather edge.
Fig.
20. Illustration of the thinning rate differences between channel sands
and sheet sands. |
Finding
Yourself In Deep Water (Part II) Introduction Sediment
Dispersal Inferences in Deep Marine Sediments One of the most reliable indicators of local slope orientation is the measurement of slumped beds. These are typically large enough for accurate dip measurement from one or several fold limbs (Figure 14). However, comparisons between slumps seen on images, and regional palaeoslope orientations determined from high resolution 3D seismic, reveals that the local slump direction does not necessarily relate to regional slope. It is, however, quite common for slumps to be oriented with the axis of a channel, as in Figure 15. However, slump trends orthogonal to channel complex axes are also commonly seen. Such bedding orientation inferences must be analysed in the context of the local conceptual depositional model. Another
fruitful source of more subtle sediment dispersal indicators comes from
the naturally low angle bedding occurring in deep marine sediments. As
in the example presented in Figure 15, structural dip is determined from
thick claystone intervals, assumed to represent the palaeo-horizontal.
Once structural dip has been rotated back to horizontal, any low angle
residual dip in well-bedded sequences may represent lateral accretion
on subtle topographic features, such as erosion surfaces, slump scars
or channel levees. Alternatively, subtle dip trends may reflect differential
compaction between individual sedimentary bodies, however, such features
tend also to exhibit other vertical dip magnitude trends. A useful method
for the identification of subtle dip changes in shallow dipping sediments
is the vector walkout plot. Such plots are generated by cumulatively adding
orientation vector data, usually either azimuth or inclination (Figure
16). These plots are generated by the head-to-tail stacking of unit vectors
sequentially up through a given succession. Each azimuth is drawn from
the population of derived dip data. They provide a visual running average
of mean azimuth for a studied succession. Sharp deviations in azimuth
direction recognised from these plots can sometimes flag significant stratigraphic
breaks. Analysis
of Thin Beds Using Image Data But what do we mean by thin beds? In the log analysis sense, thin beds are individual beds that cannot be individually resolved by conventional open hole resistivity logs. The effect of this log resolution limitation, quoted in literature, ranges from 10-30% underestimation in net to gross. It is being increasingly recognised that borehole images can provide a basis for extending net sand calculation well beyond the resolution of conventional wireline logs. Lawrence (2002) illustrates that even an acoustic televiewer, logged in good hole conditions in moderately consolidated sediments, can provide sufficient acoustic bedding contrast to extract sandstone bedding data for detailed net to gross studies (Figure 17). Net sand information derived from borehole images can also be used as the basis for bed thickness statistical analysis, which can be a powerful tool in the study of many deep marine sequences. While beyond the scope of this article, there is a growing school of log analysis that uses the high resolution of borehole image logs to generate a square bed 'earth model' of the thin sandstone and mudstone units in a stacked sequence. This earth model is used to convolve conventional resistivity logs, correct for shoulder bed effects and derive a more representative solution for bed saturation at bed boundaries. The bed model can also be used to generate synthetic resistivity logs, forward modelled using tool response equations. Applications
of LWD Images Our experiences in the interpretation of LWD images has shown that the immediate impact on geological description is in fine scale structural evaluation. Most of our studies find that the notion of uniform structural tilt, usually defined from near vertical well sections, is wrong. Sections are commonly found to be very gently folded, with fold amplitudes of 2 to 10 m associated with wavelengths of 100 m to 500 m. Many of these folds are not detected on seismic. Fluid drainage within thin reservoir sand bodies may become an issue in such circumstances. LWD images have also proved valuable in fault and fracture detection, and the passage of a horizontal well through a fault damage zone allows estimates to be made of fault throw. A major advantage of horizontal wells through channelised turbiditic sections is the ability to repeatedly intersect a channel margin (Figure 18). From such data, each surface of intersection can be orientated and their position and strike used to constrain channel margin directions (McGarva et al, 1999). If a large number of channel margin contacts have been sampled and orientated within a channel complex, it is possible to statistically model likely height-width ratios for the sampled channel sand bodies. Horizontal wells are commonly drilled with the intention to run parallel or near parallel to one particular bed for substantial distances. In addition, the passage of horizontal wells through a gently folded succession allows a sedimentologist to trace lateral variations in lithology and facies type across a field for a given bed. Both of these situations allow horizontal changes in sandbody architecture and facies type to be assessed. For instance, we have observed rapid lateral facies changes from well-constrained, clean and visually massive sheeted sandstones to more argillaceous and well-bedded sections over distances less than 220 m. In contrast, other instances showed very little internal change within correlated channel elements displaying progressive abandonment. Reservoir
Modelling Using Image and Log Data Images and core allow detailed vertical sections of facies data to be derived, often over hundreds of metres. This data is depth-based and can generate very detailed bed thickness information. Is it possible to derive likely sand bed dimensions within a basin using thickness data? The short answer is no (Malinverno, 1997), unless we have prior knowledge of the shape of the deposits (Luthi, 2001) when a scaling relationship can be derived. Derived values of reservoir volumes depend strongly on the assumed shape of the deposits. Possession of orientation data for the top and base of individual turbidite beds allows the extension of these bedding planes into the surrounding formation. For suitable bed tops and bases it is possible to determine where these beds will thin to a feather-edge, and the strike of the intersection (Figure 19). Calculated thinning rates are a proxy for bed continuity. Although bedding surfaces are rarely exactly planar, this technique allows relative estimates of thinning rates to be made between different sections. A further larger-scale approach to bed thinning rates can be made from wells and proximal sidetracks. Individual beds, bed packages and layers are traced between wells. A numerical thinning rate is then calculated for these sections. We have found that basin floor sheet sandstones have different thinning rates compared to channel fill successions and have successfully discriminated between the two where seismic data is very poor (Figure 20). Examination of a series of LWD images from horizontal wells within a turbiditic field showed a dominance of oversteep sandstone-shale contacts. These were originally interpreted as faulted and/or erosive margins. However, analysis of channel element geometries using ellipses as a model indicates that vertical wells into typical sandstone bodies will tend to underestimate mean dips, whilst horizontal wells will overestimate mean dips. Horizontal wells preferentially cut the oversteep parts of the sandstone-shale contact and this results in an overestimate of mean contact dip. It is not because the dips are steep by deformation (e.g. faulting), they are steep by default of a sampling bias. Results from the modelling replicate the general form of the data sets from wells. The inverse problem, given the mean dip, what is the geometry of a typical sandstone body, has also been considered and has been used to estimate thickness:width ratio for sandstone channel bodies, and the likely shape and extent of carbonate nodules. Borehole images deliver spatial content to a reservoir description. The development of new methodologies and techniques will give us the ability to fully use this detail in future reservoir models. Conclusions Image logs will, in most situations, provide key sedimentary detail to enhance understanding of the reservoir and associated sequences. High-resolution images are particularly useful as a source for sedimentary descriptive data from deep marine sedimentary systems. However, a systematic approach to interpretation is essential. While accurate, feature classified dip information can be acquired from deep water sediments, the interpretation of sediment dispersal directions in such sequences can be challenging and must be conducted in the context of a conceptual depositional model. LWD logs, while of lower resolution, are becoming a recognised source of lateral data regarding the length and width of facies tracts and the nature of lateral contacts. This is an exiting new area of image log application with much potential for the future. These image logs can also be used to provide accurate bedding statistics, which can be used to generate useful thickness trend statistics, and are being increasingly used for other forms of bed modelling and advanced log analysis techniques. Acknowledgements References
|
