Geoscience Reference
In-Depth Information
Fig. 6.25 The relationship
between permeability
architecture (anisotropy
expressed in terms of an
amalgamation ratio) and
key underlying controls
on confinement (Based
on discussions with D.
Stanbrook and E. Stephens)
small
large
small
large
Size of container
the seismically-observed reservoir can be consid-
erable. This is particularly the case in unconfined
systems. The challenge is therefore to see beyond
the seismic, particularly if the reservoir has a
stratigraphic trap with no direct means of defin-
ing the field extent.
For reservoir modelling, the possibility of
unseen HCIIP requires a concept, and ideas on
confinement (and hence likely connectivity) can
be tested against information from material bal-
ance. There is therefore greater emphasis than
usual on starting the modelling exercise with
guidance from dynamic data, as this can inform
the first conceptual sketches of the reservoir.
The example shown in Fig. 6.26 is from a gas
field in the UK Central North Sea modelled after
5 years of production (Bentley and Hartung
2001 ). The structure is a small field in which
gravity flows were observed to onlap a palaeo-
topographic high above a salt dome. Models
limited to observations of sand intervals in the
wells could be history matched but simple
matches could only be achieved if additional
volumes were present
uncertainty, it was therefore valid to consider
scenarios including additional, unobserved reser-
voir sands, an interpretation which would be
consistent with additional confined flows
depositing around the palaeo high. 4D seismic
data from this field subsequently supported this
hypothesis.
6.5.3 Thin Beds
Incorrect representation of thin beds is prevalent
in deep water stratigraphic traps where signifi-
cant undrilled sub-seismic HCIIP may be pres-
ent. Even when penetrated by wells, the thin beds
may be unresolved on logs and reservoir pay
intervals will typically be underestimated.
The impact of the log sampling problem in a
reservoir modelling workflow is illustrated in
Fig. 6.27 . Modellers will usually check blocked
property (porosity) logs against raw log data as a
QC step, but this is only worthwhile if the raw
data points are valid in the first place. If not, the
subsequent cross-plotting of incorrect blocked
log data with permeability data is invalid.
in beds not
initially
identified from seismic.
In the analysis of
Search WWH ::




Custom Search