15.01.2015 Views

Microseismic Monitoring and Geomechanical Modelling of CO2 - bris

Microseismic Monitoring and Geomechanical Modelling of CO2 - bris

Microseismic Monitoring and Geomechanical Modelling of CO2 - bris

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

CHAPTER 6.<br />

GENERATING ANISOTROPIC SEISMIC MODELS BASED ON GEOMECHANICAL SIMULATION<br />

mica content, <strong>and</strong> show a 0 greater than the global trend. Aspect ratio also appears to increase with<br />

increasing clay <strong>and</strong> mica content from 1784 to 1788. This would appear to confirm that increasing<br />

clay <strong>and</strong> mica content correlates with larger values <strong>of</strong> a 0 .<br />

In order to underst<strong>and</strong> this relationship, it is helpful to consider the micro-structure for clean<br />

<strong>and</strong> clay-rich samples. Figure 6.13 shows back-scatter electron micrograph (BSEM) pictures for the<br />

clean s<strong>and</strong>stones 1909 <strong>and</strong> 1950, <strong>and</strong> for the clay-rich sample 1784. The clean samples show r<strong>and</strong>om<br />

orientation <strong>of</strong> quartz <strong>and</strong> feldspar, with a r<strong>and</strong>om orientation <strong>of</strong> the diagenetic calcite (Valcke et al.,<br />

2006). The clay-rich sample shows a preferred orientation <strong>of</strong> the mica <strong>and</strong> clay grains with vertically<br />

aligned rotational symmetry due to compaction. Figure 6.13 highlights the dominance <strong>of</strong> the mica <strong>and</strong><br />

clay, which is also noted in the ultrasonic data, where a strong VTI symmetry is observed. It is not<br />

clear whether this strong lithological anisotropy has a significant influence on aspect ratio estimates.<br />

For instance, does this initial VTI skew the inversion estimates, or does the presence <strong>of</strong> significant<br />

amounts <strong>of</strong> mica <strong>and</strong> clay lead to an inherent micro-structural bias <strong>of</strong> larger aspect ratios It is<br />

difficult with such limited data to conclude with any certainty that it is the presence <strong>of</strong> clay particles<br />

alone that causes aspect ratios to increase. More velocity-stress data for shales <strong>and</strong> shaley (clay rich)<br />

s<strong>and</strong>stones, with accompanying petrophysical analyses, are necessary.<br />

6.5 Comparison <strong>of</strong> rock physics models<br />

So far I have outlined two rock physics models, the third-order elasticity approach <strong>of</strong> Prioul et al.<br />

(2004), <strong>and</strong> the approach developed in this thesis <strong>and</strong> Verdon et al. (2008). Other models available but<br />

only touched on briefly in this chapter are the R-factor (Hatchell <strong>and</strong> Bourne, 2005), <strong>and</strong> models based<br />

entirely on empirical calibration (e.g., Mink<strong>of</strong>f et al., 2004). Each model has particular advantages <strong>and</strong><br />

issues, <strong>and</strong> each requires its own assumptions. Therefore each has its role, or niche, in the geophysicists<br />

toolbox.<br />

The simplest method for dealing with stress sensitive velocities is to use an empirically defined<br />

relationship. Cores samples are taken from the reservoir, <strong>and</strong> velocities measured at the stresses <strong>of</strong><br />

interest. The advantage <strong>of</strong> such an approach is that no assumptions need to be made about the<br />

physics controlling stress-sensitive seismic velocities. However, such an approach will be limited in its<br />

validity to the parameter space tested in experiments. In a real scenario, the triaxial stress tensor<br />

will vary continuously across a reservoir <strong>and</strong> overburden, <strong>and</strong> will vary through time as a result <strong>of</strong><br />

production. Therefore it is unfeasible to conduct experiments for every stress state. Models are needed<br />

to extrapolate from experiments to the stress condition at each point in the reservoir.<br />

One approach is to use a linear interpolation between velocities measured at hydrostatic stress<br />

conditions (e.g., Mink<strong>of</strong>f et al., 2004). Such an approach requires no assumptions about the response<br />

<strong>of</strong> seismic velocities to stress. However, such an approach is extremely limited, because in reality rocks<br />

around the reservoir are not at hydrostatic stress. An approach is needed that can map triaxial stress<br />

changes to anisotropic variations in seismic properties. Furthermore, experimental observations show<br />

130

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!