Summary
Modeling complex transport processes in naturally fractured reservoirs
(NFRs) using classical continuum models may not be practically possible because
the algorithms used in this type of modeling approach for the detailed
structure of fracture/matrix systems require unreasonable computational time.
Also, fractured reservoirs are highly irregular, and finite-difference
calculations for such models often cause convergence problems. In addition, an
exact representation of a complex fracture network in classical continuum
modeling algorithms is highly difficult. An alternative is to use a
nonclassical technique known as the random-walk particle-tracking (RWPT)
algorithm.
We showed earlier (Stalgorova and Babadagli 2009) that the random-walk (RW)
technique can be adapted to model miscible flooding in a fractured porous
medium at the laboratory scale. The unknown parameters used to match the model
results were only the diffusion coefficients for oil and solvent, as the
diffusive/dispersive transport (effective in fracture and matrix) was coupled
with viscous (effective in fracture) and gravity (effective in fracture and
matrix) displacement. Advantages of this method over classical simulation
include a shorter computational time, which allows avoidance of
simplifications; the ability to model the matrix/fracture diffusion process
without any transfer function; and the representation of a complex and
irregular fracture network system.
In this paper, we modified this laboratory-scale RW model for field-scale
applications. A series of tracer-test results from the Midale field in Canada
was used to test the model. A fracture-network model was constructed on the
basis of geological data, and then we used the RWPT model to calibrate the
fracture network against tracer-test results. The results were compared to
those obtained using continuum (dual-porosity) models, and it was observed that
the connectivity and breakthrough times can be captured more correctly with the
RWPT model.
We performed a sensitivity analysis to identify the importance of different
parameters for the simulation results. The new model and observations can be
used to validate and calibrate stochastically generated fracture-network models
and to estimate the enhanced-oil-recovery (EOR) performance of NFRs.
© 2012. Society of Petroleum Engineers
View full textPDF
(
4,114 KB
)
History
- Original manuscript received:
8 June 2011
- Meeting paper published:
8 May 2011
- Revised manuscript received:
6 September 2011
- Manuscript approved:
8 November 2011
- Published online:
25 April 2012
- Version of record:
11 June 2012
View Cart
