Environmental Engineering Reference
In-Depth Information
be utilized to simulate tracer tests. The numerical model for tracer analysis assumes
that the migration of tracer species (particles) is dictated by a flow-dependent prob-
ability in each grid. The probability of tracer migration for a given orientation is ob-
tained by dividing the flow rate in the direction with the total flow rate for the grid.
Based on the field experience and the development of numerical simulation
code, we have elaborated the flow chart for the procedure of FRACSIM-3D shown
in Fig.
6.11
. Based on field data such as fracture density (number), orientation, and
fractal dimension of fracture size, natural fracture networks are generated within a
cubic fracture generation volume assuming its spatially random distribution. Field
information such as fracture density and orientation can be obtained from the obser-
vation of cores and well logs.
It is crucially important to make a good estimate of the initial fracture aperture
and shear dilation angle for the fractures having different dimensions in the gener-
ated fracture network. A predictive model for fracture aperture and shear dilation
angle has been incorporated into FRACSIM-3D. The model predicts the initial frac-
ture aperture based on initial permeability data to be measured in a borehole, and
the shear dilation angle of each fracture is estimated numerically on the basis of the
fractal nature of the fracture surface.
Because the distribution of natural fractures is a stochastic characteristic, it is not
technically feasible to determine
a priori
size and spatial location of each natural
fracture in the fracture network generated using the fractal model. Thus, we have
proposed a method for determining the natural fracture distribution by comparing
numerical predictions with field observations. Specifically, a number of natural
fracture networks are generated using random seeds (say 50 realizations in the ex-
ample to be described below).
First, based on the natural fracture networks generated, we carry out a hydrau-
lic stimulation analysis to compare the numerical outputs with field observations.
The field data to be used for this comparison are rock to fluid ratio (RFR), and
shape and dimension of the created reservoir. The RFR parameter is a measure
of the amount of fracture void space created during hydraulic stimulation, which
can be computed from the rock volume stimulated and the injected fluid volume.
The stimulated rock volume, and the shape and dimension of the created reser-
voir may be estimated using field date obtained from microseismic methods. This
evaluation step is labeled as Level I. If preliminary and short-term water injec-
tion test results such as impedance and tracer responses are available for the site,
short-term circulation analyses may be further performed. Flow rates and tracer
responses between injection and production wells may be used to compare the
numerical outputs with field observations. This assessment stage is denoted as
Level II.
This process is repeated for all natural fracture networks prepared, and the frac-
ture model that matches best with the field observations is then selected. An analy-
sis of long-term water circulation is subsequently carried out to predict the thermal
extraction performance of the created reservoir, based on the selected best fracture
network model for a pair of injection and production wells. Thermal drawdown
(time history of water temperature at the production well) can be calculated for the
