Summary
The application of geomechanics in reservoir flow simulation has increased
substantially since it was recognized that the modeling of geomechanical
effects was necessary to predict important phenomena such as compaction,
subsidence, wellbore failure. However, its application is strongly limited due
to the use of a single-grid system for both reservoir flow and geomechanics
deformation. In the case of a large field-scale simulation, the use of a
single-grid system gives rise to an extremely large number of gridblocks. On
one hand, for an accurate modeling of fluid flow, the gridblocks need to be
reasonably small around wells and sharp fronts. Yet, these small gridblocks may
not be essential for geomechanics computations. On the other hand, accurate
geomechanics calculations may require many gridblocks in the overburden,
underburden, and sideburden (country rock) that are not necessary for fluid
flow. In this work, a dual-grid technique is combined with an iterative
coupling method to resolve the problem. In this dual-grid technique, the
reservoir flow grid and the geomechanics grid are distinct in order to model
efficiently both fluid flow phenomena and geomechanics deformations. A method
to couple the two grid systems is described. The use of this grid coupling
approach reduces the simulation run time substantially with results that are
very close to the single-grid method. A series of examples illustrating the
application of this dual-grid concept and the corresponding run-time reduction
are described.
Introduction
The coupling between a reservoir (fluid and heat flow) simulator and a
geomechanics (constitutive stress-strain) simulator is still an active area of
research. Among the different time/implicitness coupling approaches (i.e.,
fully coupling, iterative coupling, explicit coupling, and pseudocoupling)
(Tran et al. 2004), the iterative coupling approach is still considered the
most practical technique in field applications (Chin et al. 2002; Coombe et al.
2001; Settari and Walters 2001). Compared to a fully coupled approach,
iterative coupling is easier to maintain and yet gives comparable results.
Nevertheless, irrespective of the time coupling method being used, problems
related to large computer memory requirements and long computer running time
still exist. This is because a geomechanics simulator normally solves a much
larger number of unknowns per gridblock than a reservoir simulator does. If the
same (coincident) grid is used for both simulators, a full-field coupled
problem requires significantly more computer time and memory than the run
without coupled geomechanics calculations, which makes the coupled runs
unattractive.
To overcome this challenge, an improved gridding technique is introduced
whereby the reservoir and geomechanics grids are not required to be coincident.
A reservoir grid can cover a subregion of a geomechanics grid, or vice versa.
With this approach, overburden, underburden, and sideburden blocks included in
a geomechanics grid can be eliminated from the reservoir grid if there is no
fluid or heat flow in those regions. In addition, the two grids can be refined
or coarsened in different regions independently according to the scale of the
various physical processes of interest. For typical thermal recovery processes
with fronts, the number of geomechanics gridblocks can be much smaller than the
number of reservoir gridblocks, resulting in a much reduced computer time and
memory requirement for a coupled run.
© 2009. Society of Petroleum Engineers
View full textPDF
(
1,351 KB
)
History
- Original manuscript received:
7 July 2008
- Meeting paper published:
21 September 2008
- Revised manuscript received:
10 April 2009
- Manuscript approved:
18 April 2009
- Published online:
5 November 2009
- Version of record:
12 March 2010
View Cart
