Basic HTML Version
OH – ET – VA - LL: Analysis of Dynamic Data in Shale Gas Reservoirs – Part 1 – Version 2 (December 2010)
p 22/24
13 – What we do not take (yet) into account
The numerical NL model is the best reference as it is the only one taking both geometry and
nonlinearities into account. However even this model is an over-simplification of the actual
physics behind shale gas production from fractured horizontal wells. In this section, we have
selected a few of the key physical processes that we believe to be missing – and that we may
develop in a coming consortium...
Complex fracture networks
Today, we only consider regular, planar hydraulic fracture systems, as described in section 6.
Although our fractured horizontal well options account for interferences between fractures, and
honor 3D effects around the wellbore or fractures, the considered geometry remains very
simple compared to the high complexity of actual flow paths. In general, fracture jobs create a
hydraulic (primary) fracture system similar to our model, but also activate a pre-existing
natural (secondary) fracture network. The secondary network’s stimulation level and its
connection to the primary system have a decisive influence on the well productivity, and need
to be properly addressed.
With our current model, one would account for the secondary network by artificially increasing
the length and/or conductivity of the hydraulic fractures. This classically leads to a reservoir
model with (1) a low-permeability matrix and (2) a system of high conductivity parallel
fractures accounting for both the primary and secondary networks interactions.
Another possible approach could consist in artificially increasing the permeability of the matrix
in the vicinity of the well, in order to account for an effective matrix plus secondary network
medium. This seems less relevant - especially at early time - if the secondary network is
conductive. Moreover, this requires the use of a composite zone around the fractured
horizontal well, with model input parameters relatively hard to interpret. Note that it is also
possible to assign double-porosity to the reservoir, in order to simulate the interaction between
the matrix and a more permeable fissure system. This, however, remains an approximate,
non-practical model. In particular, this double-porosity behavior must also be restricted to the
vicinity of the fractured horizontal well.
It is far from certain that a coupled primary-secondary fracture network behaves as a single
effective, primary-like system. Hence, a major improvement would be the ability to grid and
simulate the secondary network, using either an orthogonal, regular pattern (orthogonal to the
primary network), or a more complex, random systems. With this option, we would be able to
understand and quantify the impact of the interactions between the primary and secondary
fracture networks.
Naturally this improved geometry brings more complexity to the problem definition, and
although we will not address such issues here it calls for advanced data integration – the best
we can think of is microseismics – and further modifications in the analysis workflow.