Reservoir Model Types - Reservoir Model Design - page 216

Fig. 6.53 Example fault-plane mesh (circa 5 km long
and 100 m high) showing hanging-wall to footwall
grid connections with estimated fault transmissibility
multipliers: hot colours representing higher transmissibility

at fault margins ( left )and cold colours representing lower
transmissibility closer to the centre of the fault ( right )
(Statoil image archive, Statoil ASA, reproduced with

#

permission)

This example illustrates, once again, the
importance of the conceptual geological model
as a foundation for reservoir modelling. In this
case, failure of the initial conceptual model to
include the effects of low-density fracture
systems was the source of the failure to match
the dynamic field data. The change in the
geological interpretation was prompted by an
accumulation of production data which was
inconsistent with other interpretations. Once
sense-checked against core data, outcrop data
and structural geological theory, a new model
emerged, which was not only geologically plau-
sible but also led to significantly improved inter-
pretation of rather complex subsurface flow
behaviour. This case also highlights the impor-
tance of understanding faults as heterogeneous
3D zones, rather than 2D planes of offset.

Although it is critical to maintain alternative
reservoir concepts through the life cycle of a reser-
voir development, there is often a reluctance to
expend the additional effort required to build alter-
native structural models. This is unwise, especially
as fault uncertainties are always significant - they
are never perfectly imaged on seismic data
and their fractal nature means sub-seismic fault
populations will always exist. The question is
how important are the effects of faults compared
with other factors? These uncertainties are best
handled by establishing fault-model workflows
(e.g. Fig. 6.53 ) and testing alternative models
within the uncertainty span. If deemed significant,
fault-related uncertainties should be evaluated
either using a relatively simple sensitivity analysis
(e.g. Brandsæter et al. 2001b ), a practical
modelling work-around such as the Douglas
Field case (Bentley and Elliott 2008 ), or using
an integrated experimental design scheme for
assessing the impact of different factors on reser-
voir performance metrics (e.g. Lescoffit and
Townsend 2005 ; Manzocchi et al. 2008a ).

6.7.1.7 Fault-Related Uncertainties
The Douglas Field Example highlights the limi-
tation of becoming locked into a best-guess or
base-scale conceptual geological model.