Geoscience Reference
In-Depth Information
sealing properties. In reservoir modelling studies
there are two main activities:
1. Representing fault geometry as accurately
as possible based on the best available seismic
data.
2. Representing the fault flow properties, using
various methods for fault seal analysis.
s
1
a
60°
s
3
s
2
6.7.1.3 Geometry
Estimating fault throw is a key uncertainty, as
seismic image quality tends to deteriorate close
to faults. Fault connections in the 3D network
are a particular issue as fault intersections are
rarely resolved accurately from seismic. It is
therefore typically necessary to edit raw fault
interpretations from seismic to produce a net-
work which is structurally plausible (Fig.
6.46
).
Judging whether the fault network interpreted
from seismic is indeed plausible and reasonable is
assisted by the knowledge that fault systems - unlike
joint systems - are fractal in nature (Scholz and
Aviles
1986
; Walsh et al.
1991
) so fault networks
show size and property distributions which usually
follow a power law. Walsh and Watterson (
1988
)
showed that for many real fault datasets the length of
a fault, L, is correlated with the maximum displace-
ment on the fault, D, such that D
s
3
b
30°
s
1
s
2
s
2
c
30°
s
3
s
1
Fig. 6.44
Anderson theory of faulting relating faults to
the principal stress directions: (
a
) normal faults, (
b
) thrust
faults, and (
c
) strike-slip faults
L
2
/P (where P is
a rock property factor). A 10 km-long fault would
typically have a maximum displacement of around
100 m. Similar relationships between fault thickness
and displacement have also been established by
Hull (
1988
)andEvans(
1990
).
¼
We therefore need to translate structural geolog-
ical features into their flow properties, and this is
not an easy task. Faults often give rise to 'tales of
the unexpected' in reservoir modelling studies
because:
They are relatively narrow features, hard to
sample in well and core data and usually pres-
ent on a sub-seismic scale;
They generally have very low permeability
and high capillary entry pressure;
They are very heterogeneous, both in the
plane of the fault zone and perpendicular to
that plane;
They introduce new layer connections due
to fault offsets.
To have any chance of anticipating the
potential effects of faults on flow behaviour in
a reservoir, we need some appreciation of the
mechanics of faults and the nature of their
6.7.1.4 Sealing Properties
Figure
6.47
shows an example fault where a few
metres of displacement have created a fault with
a thickness of a few centimetres. Also clearly
seen in this example is the drag of a shale layer
along the fault surface creating a baffle or seal
between juxtaposed sandstone layers - the for-
mation of a 'fault gouge' (Yielding et al.
1997
;
Fisher and Knipe
1998
).
Empirical data from fault systems has led to a
set of quantitative methods for predicting the
sealing properties of faults. The most widely
used method is the shale gouge ratio, SGR, pro-
posed by Yielding et al. (
1997
) who showed that
the cumulative shale bed thickness in a faulted
siliciclastic reservoir sequence could be used to
