structural geology, propagation mechanics and - Stanford School of ...
structural geology, propagation mechanics and - Stanford School of ...
structural geology, propagation mechanics and - Stanford School of ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
5.3. Finite difference/finite element method<br />
For the purposes <strong>of</strong> computational efficiency, a coupled, two-step finite<br />
difference/finite element numerical method was used to compute k * . In order to<br />
adequately resolve centimeter-thick DBs within meter-scale patterns, finely discretized<br />
grids a few hundred cells on a side were generally necessary, with each cell being<br />
assigned one <strong>of</strong> two isotropic permeability values: kmatrix or kb<strong>and</strong> (e.g. Figure 6.6). Again<br />
for the sake <strong>of</strong> computational efficiency, two coarsening steps were used to compute the<br />
final, upscaled permeability tensor. For example, to compute k * for a pattern meshed at<br />
400 × 400 cells, the first coarsening step might be to a 20 x 20 cell system (yielding 400<br />
intermediate permeability tensors) <strong>and</strong> then to the final k * in the second step. An efficient<br />
finite difference method was used to make the first coarsening step. The second step to k *<br />
required a finite element procedure that is less efficient, but capable <strong>of</strong> h<strong>and</strong>ling the<br />
matrix <strong>of</strong> full permeability tensors produced during the first step. Repeat testing revealed<br />
that the final k * computed is relatively insensitive to the size <strong>of</strong> the coarsening steps used,<br />
varying by less than 10% for the patterns studied.<br />
6. Effective permeability results<br />
The governing equations, boundary conditions <strong>and</strong> numerical methods detailed above<br />
were used to calculate 2-D effective principal permeabilities for each <strong>of</strong> the three<br />
characteristic DB patterns described for the Aztec. For the simpler, more regular parallel<br />
<strong>and</strong> cross-hatch examples, we used idealized patterns to assess how effective<br />
permeability varies as a function <strong>of</strong> the relevant physical variables—b<strong>and</strong> thickness,<br />
spacing <strong>and</strong> intersection angle—<strong>and</strong> present analytical solutions for comparison where<br />
possible. For the more generally irregular anastomosing pattern, which exhibits<br />
substantial spatial variability that defies analytical treatment, we assess effective<br />
permeability for a representative outcrop pattern. In order to emphasize reductions in<br />
permeability attributable to DBs, all permeability values reported below <strong>and</strong> in the<br />
figures are normalized by that <strong>of</strong> the undeformed s<strong>and</strong>stone matrix (~ 0.1 to 1.0 darcys)<br />
to yield dimensionless ratios <strong>of</strong> relative permeability. Thus in the analyses that follow,<br />
DBs are assigned isotropic internal permeability values <strong>of</strong> 10 -2 <strong>and</strong> 10 -3 , representing the<br />
mid range <strong>of</strong> values for DBs as reported by other workers <strong>and</strong> summarized above.<br />
157