Microseismic Monitoring and Geomechanical Modelling of CO2 - bris

CHAPTER 4. A COMPARISON OF MICROSEISMIC MONITORING OF FRACTURE STIMULATION DUE TO WATER
VERSUS CO 2 INJECTION
ξ 1 α 1 γ δ
Water 0.1 120 ◦ 0.04 0.1
CO 2 0.01 141 ◦ 0.038 0.0
Table 4.3: Splitting inversion results for the events during water and CO 2 injection.
4.6.2 Interpretation of datasets
The results of the inversion for SWS during water injection are plotted in Figure 4.24 and listed in
Table 4.3. I note that as anticipated from the inversions with synthetic data, the fracture strike and
sedimentary fabric are well imaged (with α=120 ◦ and γ=0.04) while the fracture density is not well
constrained. As an independent measure of fracture strike, the event locations (Section 4.2) indicate
the formation of fractures trending at approximately 120 ◦ from the injection well. The match between
fracture strikes estimated from event locations and from SWS demonstrates the success of the SWS
inversion. I plot the splitting predicted by the best fit model in Figure 4.12, and note a good match
between my model and the observed splitting.
The inversion results for during CO 2 injection are shown in Figure 4.25 and listed in Table 4.3.
The best fit model parameters are α=141 ◦ and γ=0.038. The fracture strike appears to be poorly
constrained. This is because, at very low values of fracture density, the fracture strike parameter
becomes unimportant (if there are no fractures, it doesn’t matter what direction ‘no fractures’ are
striking). The range within the 90% confidence interval, for higher fracture densities, does match the
fracture orientation imaged by the event locations during CO 2 injection. The fracture density given
by the inversion for during CO 2 injection is less than that for during water injection. This might
suggest that the CO 2 injection has indeed caused a smaller amount of fracturing. However, note
that the 90% confidence surfaces from water and CO 2 injection overlap between ξ∼0–0.06, hence this
conclusion cannot be supported given the limitations that the source-receiver geometry impose on
the ability to constrain fracture density. This limitation has been identified a priori using synthetic
forward modelling.
The match between γ for both stages is also encouraging. Kendall et al. (2007) note that the
strength of VTI fabric (given by γ) often correlates with reservoir quality, as the presence of clay
particles both reduces reservoir quality and introduces VTI symmetry. Although specific information
about the lithologies at either depth is not available, the rocks at both depths are believed to be similar.
Therefore I anticipate that γ should be similar for both depths, and this is indeed the case, a further
indication of the success of this inversion method. Although I have no way to independently verify
γ by any other method, the values found are well within the range expected for typical sedimentary
rocks (Thomsen, 1986).
78