Akademik Veri Yönetim Sistemi
21th International Petroleum and Natural Gas Congress and Exhibition of Turkey, Ankara, Türkiye, 27 Ekim - 29 Kasım 2023, ss.131-132
Traditional description of unconventional reservoir like
any natural fracture reservoirs relies heavily on quality
and analysis of image logs and core data at discrete well
locations. These data provide high vertical resolution
information but soon becomes challenging to use and
propagate away from the borehole. The presented
modeling process is generally based on pure stochastic
workfl ow that tries to achieve realistic 3D model of
fractured reservoir by matching the well information,
in-situ stress and capture inter-well heterogeneity
using 3D seismic data. This workfl ow has been
tested on unconventional natural fracture reservoir in
Diyarbakir basin where late Silurian – lower Devonian
age Dadas-1 organic rich shale member exhibits both
low porosity and extremely low matrix permeability.
Acting as one of the main source rocks in SE Turkey,
it represents a self-sourced unconventional play. The
interpretation of image logs and core samples reveals
clusters of fractures suggesting a naturally fracture
reservoir type 1. This study aims at using an integrated
approach spanning from seismic interpretation to image
log data analysis and 3D geomechanics to develop a
discrete fracture network model (DFN) and to provide
new insights into distribution of hydrocarbons since
fractures is solely responsible for making this reservoir
producible(Figure 1). In addition, a derived DFN model
off ers an opportunity to improve a reservoir modeling
(static and dynamic), to provide the basis for design of
an optimum well placement, stimulation, completion,
production and could serve as a guide on how to
improve the seismic acquisition/resolution to highlight
valuable fracture zones.
METHODS, PROCEDURES, PROCESS
This integrated study uses borehole image log
data acquired in the Dadas-1 member interval of 3
exploration wells: A, B and C, 3D seismic volume
and 3D geomechanical model. Fracture modeling was
performed using a fracture modeling software that
calculates fracture permeability, porosity and matrix
block size on 3D reservoir grids by constructing the
DFN model. The main steps of the workfl ow were: 1).
to use previously interpreted image log data and classify
the natural fractures by fracture sets using dip azimuth
distribution. The wellbore fracture data is dominantly
striking in E-W and N-S directions with a large number
having a high (> 60 degree) dip angle. Important to
note is that the statistical likelihood of intersecting high
(>50 degree) angled fractures is reduced by drilling
vertical wells, suggesting that the vertical wells used
in this study may be underestimating the true fracture
density; 2) evaluate possible fracture drivers, defined
as any 3D properties that can be sensitive to or capture
directly fracture intensity information in the interwell
space. The fracture drivers used in our workfl ow
are various post-stack seismic attributes: Variance,
Curvature, Chaos, Ant-Tracking, Sweetness and etc.
All tested seismic attributes are used as input to a multiscale
statistical correlation analysis where previously
derived 1D intensity logs using image logs were
matched. Because seismic domain comes with inherent
resolution limitations compared to borehole data the
fracture intensity logs were generated with diff erent
window filter size to evaluate the best scaling factor for
optimizing correlation with the seismic domain. The
best fracture drivers for the total intensity log before
splitting on sets appear to be 3D Curvature (≈0.66
correlation factor), Chaos (≈0.43 correlation factor)
and fl atness (≈0.55). These 3 seismic cubes supervised
by interpreted intensity values at well locations were
used to derive 3D fracture intensity cube (or 3D
fracture driver). The QC crossplot between interpreted
and predicted intensity logs shows correlation value
0.88; 3) to review this analysis in the context of the
seismic structural interpretation and regional tectonic
framework. Since the fracture distribution and density
relate to the tectonic features (faults), Ant-tracking
3D seismic attribute was used to derive seismically
resolved faults/fractures. All captured by Ant-tracking
discontinuities were split into two tectonic sets based
in the strike direction and relationship to tectonic
events: Set-1 fault polygons in black that have West-
East orientation and formed during 1st tectonic event
and Set-2 fault polygons in red that striking in North-
South and formed during 2nd tectonic period (Figure
2); 4) to employ 3D geomechanical model to determine
the likelihood of natural fractures undergoing tensile
reactivation. This model does not consider variables
such as fracture plane roughness, cementation, pressure
variation, or the possibility of crystal bridging, which
has been shown to enable fractures to remain open and
permeable albeit not preferentially aligned with σ1.
Importantly, all fracture types (conductive, partially
conductive and resistive) were used for the modeling.
Natural fractures in shale, as weak planes of mechanical
heterogeneity, can reactive and widen the treatment
zone, aff ect propagation and intensity of artificial
fractures. Even the cemented fractures, i.e. mineral
veins, can considerably contribute to eff icient hydraulic
fracture treatment, because of the weak chemical bond
between the fracture-filling minerals and their wall
rocks that can be easily broken apart.