Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

SPE 59045

Fractured Reservoirs: an Integrated Approach


Grard J. MASSONNAT, SPE and Marie-Odile BOCKEL-REBELLE, Elf Exploration Production

Copyright 2000, Society of Petroleum Engineers Inc.


while extrapolation is usually guided by geological concepts,
This paper was prepared for presentation at the 2000 SPE International Petroleum seismic data and production history analysis.
Conference and Exhibition held in Villa Hermosa, Mexico, 1-3 February 2000.
Assuming the extrapolation of the fracture network is
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as possible what confidence can be had in a local prediction
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
obtained with a structural model? the greatest difficulty in
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at such an approach is to relate the proposed evolution of the
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper fracture network with the evolution of the flow network
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 properties. It is known that these relations are non linear, and
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
therefore extrapolations based on them are highly uncertain.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Recent advances in fracture network modeling and flow
simulation have opened doors to new methodologies for
Abstract building 3D fractured models in the oil industry.
The understanding of the flow network organization, and
the ability to predict in 3D the distribution of fracture drains Explicit fracture models can now be built, incorporating
are hey points to enhancing well productivity, hydrocarbon most of the knowledge provided by structural geology,
recovery and therefore field productivity. geomechanics and statistics. These generic models are usually
Recent advances in fractured reservoir characterization and built around wells, constrained by cores, imaging log and
modeling have led to construct and very sophisticated seismic data.
software able to model local 3D fractured networks and run However, generating fracture networks covering areas of
one phase flow simulations on explicit fracture models. several square kilometers is still a highly uncertain process
The problem is how to take advantage of these advances and and one-phase flow simulations can be used for reducing
build more realistic 3D full field models to run multiple phase uncertainty.
flow simulations. If well test data are available for that area, they are used to
This paper proposes and integrated approach assigning calibrate the organization and properties of the fracture
reservoir of flow properties using results from pluriscale and network. Properties initially input in the model are slightly
multi-disciplinary steps. transformed until the flow simulations results honor the well
Its application on a case study largely contributed to a better test data while preserving geological consistency. The
understanding of flow behavior within reservoir. effective properties are then derived for use in the dual
porosity model. This provides very useful information for
Introduction quantifying relationships between fractures networks and flow
For years, fractured reservoirs have been modeled using networks, and significantly helps to extrapolate dual porosity
dual porosity equations from Warren and Boot1. In such parameters outside of the sampled area.
models fractured reservoirs are represented by two different This paper presents a methodology which can be applied
media: to obtain such relations between fracture and flow networks
the flow network, which represents that part of the and, through a case study, how the calibration is conduced to
fractured network which is connected and highly extrapolate effective properties to the 3D full field model.
permeable;
the matrix, which represents both the porous media and Characterization of flow network around wells
unconnected or less permeable parts of the fracture In a fractured reservoir, the evaluation of flow network
network. characteristics is very important as it can affects in the early
Equivalent parameters for these two media are usually appraisal stage as well as in mature field development stage.
obtained from interpretations of dynamic information (mud Assessing these characteristics is usually a difficult task.
loss, well tests results, production or injection logging, etc.) In the eighties2, geologists tried to solve the problem by
2 GRARD MASSONNAT and MARIE ODILE BOCKEL-REBELLE SPE 59045

building explicative fracture models. These models, which values are included in the software. Fractures consist of a
attempt to produce more realistic fracture systems, were first body and forbidden zone. The width of the forbidden zone can
based essentially on statistics resulting from outcrop or core be either constant or variable.
studies. Seed probability grids are computed, to handle the
Recently, a new approach was presented3 for building probability that a seed appearing in the cell will survive and
models of tension fractures using a genetic process which become a fracture. Using grids filled with various parameters,
aims to reproduce the initiation, growth and interaction of geologists can set constraints concerning fracture distribution,
fractures. By comparing the resulting fracture models with intensity and orientation, which may have been derived from
the total outcrop fracture system, it is now possible to evaluate the reservoir data (wells and seismic), geochemical models or
the success of the extrapolation. However, many parameters analogous outcrop observations.
still remain uncertain and dynamic data are expected to Fracture interaction is controlled by the Impedance Factor,
provide a powerful constraint against which fracture systems which determines the ability of a fracture to impede the
can be tested. propagation of later fractures across it.
Usually, input of geological and dynamic data into Fracture Interaction can be set as an input parameter:
fractured reservoir models is always consistent. Several Number of fractures stopped
approaches for incorporating flow information have been Impedance Factor =
Total number of fractures
presented, but comparing actual well test data and numerical as well as the variations in fracture patterns in the different
simulations of flow towards the well from the 3D fracture lithologies.
network/ matrix system4 looks like best way to discriminate Many graphs showing conventional attributes of the
between the alternative geological models. fracture sets can be displayed in real time during fracture
growth. All these statistics histograms of length, spacing,
Modeling of the fracture network cumulative length, etc. can be used to calibrate the fracture
models against measured or assumed statistical distributions.
The model building procedure consists of:
Fractures growing in a reservoir model can also be
1. Input and viewing of seismic data and/ or interpreted
conditioned to cross the well at the position of real fractures
seismic surfaces and interpreted well structure data.
observed in the well. This is done by increasing the
2. Construction of near well (or interwell) structure
probability of a seeds appearing in the plane of the observed
models by extrapolating well structures including
fracture by entering a user-defined value.
sedimentary discontinuities and faults into the 3D
volume surrounding the wells. The extrapolated
surfaces can easily be processed using an interactive Well test conditioning
mechanisms until the required geometry is obtained. For most naturally fractured reservoirs, several equally
Sedimentary input might be required to complete these probable geological models can also be developed. Moreover,
models. the flow characteristics of the fractures themselves
3. Generation of large numbers of smaller scale (permeability, porosity, compressibility, etc.) are better
fractures within the near well structure calibrated to determined from sensitivity studies on well test data than
the well data and constrained as far as possible by from geological observations of well fracture aperture,
outcrop observations, geomechanical modeling and mineralisation, reactivation, roughness and other fracture
structural analysis. characteristics liable to influence fluid flow.
This is why both geometric and dynamic conditioning should
A genetic fracture modeling method has been proposed3 be used together to sort out and finally select the appropriate
using mechanical rules designed to reproduce the basic scenarios.
aspects of the fracture growth process. Tension fractures Transient pressure responses in well tests appear to be a
propagate in all directions perpendicular to the maximum powerful way of achieving this objective6. They can be
effective tensile stress5. The development of tension fractures simulated for different fracture configurations chosen to
releases the tensile stress in the rock. The stress release is represent geological end member scenarios which can have
greatest close to the sides of the fracture, usually leading to a an impact on field behavior.
decrease of the energy available for fracture growth. It creates The model simulated pressure data are compared against
a forbidden zone within which other fractures cannot real data and a fit is obtained either by changing the geometry
propagate. of the fracture model, within the constraints of the geological
With increasing stress, the number of simultaneously data, or by changing the flow properties of the fractures. By a
propagating fractures increases. process of iteration, the geological and flow data can be
Defects are represented by seeds which appear in the reconciled to produce one or more plausible models which
model. Once seeds appear, the system causes them to grow by can then be used to guide the extrapolation in the full-field
a user-defined distance, termed a step. Various models of step reservoir model. With the calibrated fracture model inferred
SPE 59045 FRACTURE MODELLING ETC. 3

from the proposed discrete fracture modeling approach, permeability measured on the plug) An effective permeability
important flow properties of the fracture network can be (matrix + fractures) is obtained in a pseudo permanent regime
derived and these are then used as input parameters in the multiplied by the height (KH test), as well as values of and
reservoir simulations. , respectively.
The seismic data analysis helped us define a possible relation
Applications between fracturing and fold geometry.
The method used is described below through a case study
on a field that have been producing for several years. Modeling
The different analyses are then rendered consistent in
Structural setting order to generate a realistic 3D fracture model. The method
The structure is an elongated anticline, located in the consists in reproducing the fracture network, in as much
folded foreland which is part of a very recent mountain range detail as possible, in a volume around the well (fig. 4). The
(essentially Miocene to Recent). Furthermore, the fold volume should be approximately equivalent to the
deformation related to this recent compression followed the investigation volume of the well test.
development of inherited faults trending North-South and sub The software used3 allows to integrate all types of data
East-West. These faults play a part in the current structural concerning fracturing , from the seismic interpretation to core
scheme, in which they delimit domains of more or less data. The various stratigraphic surfaces and the major faults
intense folding (fig.1) were input in the software using a deterministic method: a
At smaller scale, the anticline was affected by two older combination of markers resulting from the seismic
directions which mark the shape of the fold as well as the interpretation and indicators defined at the wells.
various fluid contacts which have been observed: the field The zone selected for the fracture modeling is located in the
comprises structural compartments (fig.2). vicinity of a N-S fault at the top of the anticline. The
following parameters were selected as a function of the
Sedimentological setting
The depositional environments of the various Cretaceous analytical data:
fractures parallel to the fold axis are presented over the
units correspond to channels trending perpendicularly to the
paleocoastline. The channels range between 2 and 300 meters entire structure. Their growth was essentially confined to
large on average, for several kilometers in length. They are within the layer.
N-S fractures are associated with the nearby fault.
separated by very fine carbonate beds.
E-W fractures are not present in this area (too far from
Fracturing: faults with the same trend).
Fracture data reveal several different origins and scales for Moreover, the uncertainty on other parameters such as
the fractures. persistence of fractures associated with the neighboring fault,
Core analyses on cores from 5 wells (4 vertical wells and 1 or interaction between two fracture sets led us to propose
slanted well) located in the different structural compartments several equiprobable fracture scenarios (fig. 5).
of the fields. The analysis was calibrated with imaging logs. It Phenomenological simulation for determining dual
revealed the following points about the field: porosity
Evident fracturing in three different directions. The main For each of the two geological models constructed on this
direction is parallel to the anticline axis. This set of basis, well test simulations were conducted using the finite
fractures is probably related to folding. element code previously described6. The first permeability
Two secondary directions, N0 and N80 were observed values were assigned to the matrix (Kplug) and to different
around the N-S and E-W faults, respectively. It is quite fault sets (estimated values). The code used simulates flow in
likely that both these sets of joints started forming either fractures as well as fluid exchanges between the matrix and
immediately before or the major fold. the fractures.
the fracture set defined here is usually confined to the The dynamic fracture properties and geometry are
most carbonate-rich facies. The aperture is often small (in adjusted, as far as the geological constraining permits, to fit
the main direction) and partially filled with calcite. pressure data.
Stratiform stylolites are quite frequent. Microfractures The results of these simulations make it possible to
(less than 5 cm in length) are often associated with the determine a fracture scheme which is consistent with dynamic
stylolites: they are usually open and filled with insoluble data. Equivalent properties (fracture permeability, block size,
residue at their ends. etc.) were computed in the following step.
Well tests from 2 wells were interpreted. One well was located The selected fracture model provided a better fit between
at the top of the structure and the other in the vicinity of a N- dynamic and geological data (fig. 5). It also favored generally
S fault. The dual porosity behavior (fig.3) is obvious intra-bed fracturing, with predominance of the fracture set
(permeability obtained with the test is much higher than the related to the fault (N-S).
4 GRARD MASSONNAT and MARIE ODILE BOCKEL-REBELLE SPE 59045

Extrapolation of these parameters to the entire field Uncertainty on multi-scenario fracture networks can be
This delicate step is essential for understanding the reduced by constraining the models to single phase flows in
dynamics of the entire field. It is unavoidable for locating the well-test investigation radius. Both the geometry of the
production or injection wells, as well as for choosing the right system and the flow characteristics of the fractures can be
strategy for secondary recovery. adjusted to fit actual data.
Extrapolation can be carried out using a geological model Extrapolation is model driven; it results in an integrated
and phenomenological model constrained by flow data, as approach in which geological concepts and field dynamics are
described above, and by taking into account geological reconciled.
concepts and the behavior of the different wells. This approach must be performed using multiscale
In this case study, a geological concept can be defined as analyses in order to provide consistency to the extrapolation.
one of the three points below: On a case study, both the observations of field
compartmentalization by faults and open fractures in the
understanding of the regional tectonic setting: the field is
same direction as sealed faults provided a good illustration of
located in a region considered to be a folded foreland
the dual behavior of the fracture system:
(fig. 1) in which folding shows the following specific
the fault itself acts as a barrier
characteristics: highly elongated (several tens of
the related damaged zone plays a significant role as a
kilometers long) narrow (several kilometers) asymmetrical
drain.
anticlines, generally affected by reverse faulting on their
This dual behavior should be reproduced around seismic -
outer limbs. The presence of major oblique directions with
illustrated faults in the 3D full-field model in order to propose
respect to the axis of the mountain range can be explained
a realistic permeability field for multi-phase flow simulations.
by inherited structures. These major boundaries delimit
individual structural sectors which reacted differently to
References
the recent compressional deformation.
understanding of field compartmentalization: which both
Warren, J.E., and Root, P.J: The Behavior of Naturally Fractured
horizontal (no connectivity between the various reservoirs) Reservoirs, Society of Petroleum Engineers Journal
and vertical (fluid contacts are very different on either side (September 1963) pp 245-255
of the inherited N-S and sub-E-W faults). This Long, J.C.S.: A Model for Steady State Flow in Random Three-
compartmentalization suggests that the faults have good Dimensional Networks of Disc-Shaped Fractures, Water
seal efficiency and the barriers are not fractured. Resources Research (1985) 21 pp 1105 - 1115
understanding of the fracture network: the fit described Swaby, P.A. and Rawnsley K.D. :An interactive 3D Fracture-
earlier of the phenomenological fracture model to the Modeling Environment, SPE Computer Applications, June
dynamic data allow us to explain the possible interaction 1996
between the various fracture sets, as well as their Miller I., Lee G., Dershowitz W. and Sharp G. :MAFIC -
Matrix/Fracture Interaction Code with Solute Transport - User
distribution. N-S and sub E-W fractures are located only
Documentation, Golder Associates Inc. Redmond, Washington
in the vicinity of faults in the same direction. Fractures (1994).
parallel to the fold axis exhibit a distribution which is Pollard, D.D. and Aydin, A.L.: Progress in understanding jointing
related to fold geometry. Moreover, very few fractures over the past century, Bulletin Geological Society of America
apparently run across the entire reservoir series. We can (1988) 100, pp. 1181-1204
therefore propose general interpretation of the fracture Wei L., Hadwin J., Chaput E., Rawnsley K.D., Swaby P.A.:
network for the entire field. Discriminating Fracture Patterns in Fractured Reservoirs by
Pressure Transient Tests, SPE 49233, 1998
The behavior of the various wells in this field (PI,
production history, etc.) agrees with the general interpretation
of the fracture network.

Conclusions
Recent advances in the understanding of the growth of
tension fractures has made it possible to propose a system for
modeling fractures in a reservoir. This method of fracture
generation, fracture growth and fracture control has given the
system a great deal of flexibility and made it possible to
produce realistic tension fracture patterns.
SPE 59045 FRACTURE MODELLING ETC. 5

0 100 km

Main thrusts
Studied
area reverse faults, minor thrust

strike slip faults

Main structural boundaries

FOLDING ZONATIONS

?
folds having very high wavelength (15 20 Km)
folds having high wavelength (10 15 Km)
folds having medium wavelength (5 10 Km)
folds having small wavelength (2 5 Km)

folded foreland

trends derived from the gravimetric interpretation

trends derived from the magnetic interpretation

Fig.1 Regional structural environment of the studied field

Map at top reservoir

North

+ +

+ LEGEND

Probableinheritedaccidents

Locationofthewell

Well with welltestdata

Fig.2 Compartmentalisation of the field according to the inherited accidents


6 GRARD MASSONNAT and MARIE ODILE BOCKEL-REBELLE SPE 59045

1983/02/02-0614 : OIL

ENDWBS
DP + DERIVATIVE (PSI/STB/D)
10 -1

PD=1/2
10 -2
10 -3

10 -3 10 -2 10 -1 10 0 10 1
DT (HR)

Fig.3 Well test data and interpretation

Fig.4 3D view of the field and location of the fracture model


SPE 59045 FRACTURE MODELLING ETC. 7

Fig.5 Selected fracture model

You might also like