Page 8 - RFP_2013_Cilona

correspondence of the host rock and minimum values at
the intersection between ZBs (Fig. 8 a). It is interesting issue, and investigating the internal architecture of both
to notice how the interplay between the presence of
open slip surfaces and the low-porosity fault rocks leads
Figure 7: Schematic flowchart showing the
workflow we developed to export the fracture
distribution from a standard DFN model into a
fluid flow simulation software. The bifurcation
point represents the choice of the best
combination of geo-cellular volume and cell
sizes. The boxes around each step have
different color to represent software operation
(violet) and manual operation (red). The dashed
arrow represent a step that is theoretically
possible but has not been done for this model.
to an intermediate porosity value of the fault zones (Fig. ZBs (Fig. 8 b). Due to its higher accuracy the second
8 a).
From the map (Fig. 8 a) it is not possible to detect
the effect that CBSs have on porosity. As already
mentioned this is due to the size of the CSBs compared
to the geo-cellular volume cells. In order to address this
faults and ZBs, we performed another porosity
calculation (same workflow). For this latter calculation
we kept unvaried the same structure array (Fig. 6) but
we changed the volume. The volume used had a top
face area measuring one ninth of the previous one (50 x
312 m; Fig. 8). Moreover the top face of the cells area
was reduced to 0.09 m2 (0.3 x 0.3 m).
As expected the result of this second calculation
mirrored the previous one. However, since the cells
were smaller in size (Fig. 8 b), the
compartmentalization effects of CSBs are now
detectable. Thanks to this smaller volume, it is possible
to visualize the anastomosing architecture of faults and
volume has been used for the fluid flow simulations.
Fluid flow tests
As explained in the previous paragraph, the fluid
flow simulation were run at the scale of the second
volume (Fig. 8 b). The data from MOVETM format have
been exported into generic ascii format to be loaded
into MODFLOW 2005. At this stage it was necessary to
replace normalized porosity with real permeability.
Since the map (Fig. 8 b) perfectly match the DFN
model, establishing a correlation between the color
codes and the single structures was straight forward.
Despite the literature lacking of systematic dataset on
the permeability of deformation bands and faults in
porous carbonates, very complete datasets are available
for the same structures in sandstones. The hydraulic
behavior of deformation bands and faults in porous
carbonates (Tondi, 2007; Rath et al, 2011) is analogous
to that one of the same structures in porous sandstones
(e.g. Antonellini & Aydin, 1994; Sternlof et al., 2006;
Ahmadov et al., 2007). Thus for our model we used
typical values from the sandstones literature. The
selected permeability “Kxx” are listed in Table 1
(expressed in orders of magnitude).
Since MODFLOW only deals with hydraulic
conductivity, the permeability values had to be
converted (Tab. 2). The aforementioned hydraulic
conductivity values are representative of one direction
(Kxx), the other two principal directions (Kyy and Kzz)
of the permeability tensor have been assigned according
to the following assumption: Kxx=Kyy < Kzz (where
Kzz = 10 Kxx). This assumption is justified by the fact
that within fault zones and ZBs (Fig. 2 b, c), lenses of
less-deformed rock are comprised between low-
permeability structures (i.e. CSBs and fault rocks).
These lenses are elongated in the direction of 2 (in
agreement with Fossen & Bale, 2007).
The model imported in MODFLOW is shown in
Figure 8 (cell size 0.5 x 0.5 m). Figure 9 (a) shows the
Stanford Rock Fracture Project Vol. 24, 2013 E-8