Geoscience Reference
In-Depth Information
for alternative well designs to be tested against
likely productivity.
It is generally preferable to design the well
paths in reservoir models which capture all factors
likely to impact a fairly costly investment deci-
sion. Most geoscience software packages have
good well design functionality allowing for accu-
rate well-path definition in a high resolution static
model. Figure 1.4 shows example model for a
proposed horizontal well, the trajectory of which
has been optimised to access oil volumes (HCIIP)
by careful geo-steering with reference to expected
stratigraphic and structural surfaces.
Some thought is required around the
determinism-probability issue referred to in the
prologue and explored further in Chap. 2 , because
while there are many possible statistical
simulations of a reservoir there will only be one
final well path. It is therefore only reasonable to
target the wells at more deterministic features in
the model - features that are placed in 3D by the
modeller and determined by the conceptual geo-
logical model. These typically include fault blocks,
key stratigraphic rock units, and high porosity
features which are well determined, such as chan-
nel belts or seismic amplitude 'sweet spots.' It is
wrong to target wells at highly stochastic model
features, such as a simulated random channel,
stochastic porosity highs or small-scale probabilis-
tic bodies (Fig. 1.5 ). The dictum is that wells
should only target highly probable features; this
means well prognoses (and geosteering plans) can
only be confidently conducted on models designed
to be largely deterministic.
1.4
Models as a Front End
to Simulation
The majority of reservoir models are built for
input to flow simulators. To be successful, such
models have to capture the essential permeability
heterogeneity which will impact on reservoir
performance. If the static models fail to capture
this, the subsequent simulation forecasts may be
useless. This is a crucial issue and will be
discussed further at several points.
The requirement for capturing connected per-
meability usually means finer scale modelling is
required because permeability is a non-additive
property. Unlike models for volumetrics, the
scope for simple averaging of detailed heteroge-
neity is limited. Issues of grid geometry and cell
shape are also more pressing for flow models
(Fig. 1.3 ); strategies for dealing with this are
discussed in Chap. 4 .
At this point it is sufficient to simply appreci-
ate that taking a static geological model through
to simulation automatically requires additional
design, with a focus on permeability architecture.
1.5
Models for Well Planning
If the purpose of the modelling exercise is to assist
well planning and geosteering, the model may
require no more than a top structure map, nearby
well ties and seismic attribute maps. Wells may
also be planned using simulation models, allowing
Fig. 1.3 Rock model (
a
) and property model (
b
) designed for reservoir simulation for development planning (
c
)
Search WWH ::




Custom Search