INTEGRATED RESERVOIR

CHARACTERIZATION AND MODELING

Mickaele Le Ravalec
Brigitte Doligez
Olivier Lerat Introduction

ISBN: 2-901638-15-5 Book DOI: 10.2516/ifpen/2014001

EAN: 9782901638155 Introduction DOI: 10.2516/ifpen/2014001.c000
 We briefly recap in the following pages what a reservoir model is, which kinds of data can
be considered, and how data integration can be performed. The main point is to
emphasize the need for implementing integrated workflows.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 1
 As soon as we are interested in underground geological objects, we have to face a major
difficulty: the measurements performed to better characterize these objects are very
expensive. As such, they are rather limited.
In such conditions, numerical simulation is a useful and complementary tool. We then
refer to computer models to represent the objects under consideration. These models are
given as input to flow simulators, which provide numerical responses. This helps
understand fluid flows.
According to Wikipedia:
- reservoir modeling involves the construction of a computer model of a petroleum
reservoir, for the purposes of improving estimation of reserves and making decisions
regarding the development of the field, and
- reservoir characterization refers to reservoir modeling activities up to the point when a
simulation model is ready to simulate the flow of fluids.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 2
 Therefore, a central goal in reservoir characterization and modeling is to build reliable
reservoir models, which help decision makers optimize field development. The more
data-consistent the model, the sounder the predictions. Thus, the key point is to integrate
and reconcile all available data - geological, geophysical and production data - into
reservoir models.
Creating asset teams with geologists, geophysicists and reservoir engineers is a first step
towards integration, but it is not enough. There is a clear need:
1) for integrated modeling workflows focusing on a single objet, i.e., the reservoir model,
and
2) for advanced techniques to adjust the reservoir model at any level of the workflows
and strengthen its consistency with data.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 3
 Oil and gas reservoirs are geological formations with fluids within the pore space. These
formations lie deep underground and have very complex configurations and properties.
The pore space comprises the spaces that are unoccupied by the solid rock material:
e.g., spaces between grains and fractures.
Usually, formations are heterogeneous, compressible, while the fluid is multiphase, multi-
component, and with interactions between phases and components.
A reservoir is mathematically an infinite-dimensional system, i.e., an infinite number of
parameters is required to describe a reservoir fully, because both the rock formation and
the fluid in the pore space are continuous and have very complicated and variable
properties.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 4
 The physical system that describes reservoirs is governed by mainly three fundamental
laws and all mathematical techniques are simply an application of these fundamental
equations.
- Mass conservation law: the total mass is conservative, i.e., the difference between the
mass inflow and outflow in a cube equals the sum of mass accumulation.
- Darcy’s law: derived empirically, this law indicates a linear relationship between the fluid
velocity relative to the solid rock and the pressure head gradient. Its theoretical basis was
provided by, e.g., Whitaker (1966).
- Equation Of State (EOS): this is an algebraic expression to represent PVT properties of
phases. Numerous EOS have been proposed to represent the phase behavior of pure
substances and mixtures in the gas and liquid states since Van der Waals introduced his
expression in 1873. These equations were generally developed for pure fluids and then
extended to mixtures through the use of mixing rules.
Other equations are also considered:
- Energy conservation law: the energy is conservative (to be considered when reservoirs
are produced using for instance steam injection).
- Relative permeability and capillary pressure relationships: these are dynamic
relationships in multiphase flows. The relative permeability kr of a phase is a function of
water saturation Sw while water saturation is related to capillary pressure Pc.
We combine these fundamental laws and the initial and boundary conditions so that the
reservoir is represented as a mathematical model.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 5
 As solving the continuous problem is not straightforward, it is approximated by a discrete
system, i.e., an array of discrete cells, delineated by a grid, which may be regular or
irregular. This array is usually three dimensional, even though 1D and 2D models are
sometimes used. It often comprises several millions of grid blocks. This numerical object
is the reservoir model.
In short, a 3D reservoir model consists of the following elements: 1) upper and lower
boundaries that correspond to the top and base horizons (they are determined from logs
and seismic interpretation), 2) faults that also contribute to the definition of the structural
model, 3) facies and petrophysical properties such as porosity, permeability, saturations,
net-to-gross (the value of each property is implicitly deemed to apply uniformly throughout
the volume of the cell), and 4) properties describing the fluids in place with kr-Pc curves,
a PVT model describing formation volume factors and solution gas ratios, Water-Oil
Contact (WOC) and Gas-Oil Contact (GOC), coefficients describing the activity of
aquifers...
All of these elements are interconnected. They are highly dependent on one another. For
instance, water saturation can be a function of the height above contact and the height
itself depends on the position of the reservoir, the contacts, porosity and permeability.
The net-to-gross ratio, denoted N/G or NTG, is the proportion of rock formed by reservoir
rock (range is 0 to 1).

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 6
 We focus on the identification of petrophysical properties because
- they are related, directly on indirectly, to all type of data - geological, geophysical and
production data;
- they (especially permeability k) are often the main properties influencing fluid flows.

The spatial distributions of petrophysical properties are very difficult to characterize for at
least the reasons listed below.
First, petrophysical properties can be very heterogeneous (Dutton et al., 2003; Eaton,
2006), meaning that they can vary a lot with space.
Second, there is a deep lack of quantitative data. Except for seismic, most reservoir data
are determined at well points. These points account for less than 1% of the volume of the
whole reservoir even for fully developed and brown fields.
Third, measurements are subject to errors.
This contributes to induce high uncertainty.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 7
 Referring to Wikipedia, homogeneity and heterogeneity are defined as follows.
- “Homogeneity is the state of being homogeneous. Pertaining to the sciences, it is a
substance where all the constituents are of the same nature; consisting of similar parts,
or of elements of the like nature.”
- “Heterogeneity is the state of being heterogeneous. It is the nature of opposition, or
contrariety of qualities. It is diverse in kind or nature; composed of diverse parts, or
resulting from differing causes. In general, a heterogeneous entity is composed of
dissimilar parts, hence the constituents are of a different kind.”
In short, a given property is homogeneous when it is the same everywhere. When it
varies with space, it is said to be heterogeneous.
Reservoirs are inherently heterogeneous because of the variety and the complexity of
geological formation processes. This results in spatial variations in porosity, permeability,
saturation, facies distribution, fault and fracture distributions, orientations...
The following slides point out the occurrence of heterogeneities at all scales, focusing
mainly on 4 of them (Schulze-Riegert and Ghedan, 2007): micro, macro, mega and giga
scales.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 8
 Reservoir heterogeneity exists at multiple scales and is related to complex and intricate
geological formation processes such as erosion, transport and deposition of sediments.
Thus, channels, lobes, bars or meanders can be observed at large scale. Heterogeneity
can be also evidenced inside these geological objects. For instance, lateral accretion and
meter-scale cross-bedding are often observed in channels. Besides, changes in the
nature of contacts, grain size or fracture density are commonly detected at the much
smaller scale of cores.
Reservoir heterogeneity produces variations in petrophysical properties. For instance,
permeability of the Culebra Dolomite aquifer, approximately 450 m above the repository
horizon of the Waste Isolation Pilot Plant in southeastern New Mexico, USA, was shown
to vary by five orders of magnitude (Holt, 1997). On the other hand, Henriette et al. (1989)
made detailed permeability measurements of porous blocks of sandstone and limestone
measuring 15 × 15 × 50 cm3. Each block was cut into three hundred small plugs and
permeability was measured for each of them. Permeability variations were shown to
approximately span two orders of magnitude.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 9
 Microscopic heterogeneities occur at the level of pores and grains. They are evidenced
from laboratory measurements and can be expressed in terms of sizes of pores and pore
throats, shapes and sizes of grains, openings of throats, types of minerals, roughness of
pores, etc. These heterogeneities have to be related to formation processes: deposition,
subsequent deformation phenomena with compaction, cementation, and dissolution.
These microscopic heterogeneities tend to create preferential flow paths at the pore
scale: the displacing fluids flow while some residual or trapped oil is left behind.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 10
 Macroscopic heterogeneities are detected at the level of cores. Again, they are identified
from laboratory measurements. Experiments are performed to estimate porosity,
permeability, saturation, capillary pressure, wettability, etc. They can vary from one core
to another one even though extracted from a very close location.
The laboratory experiments are usually used to calibrate the logs and well tests. They are
provided as inputs to the flow simulators.
Macroscopic heterogeneities influence the shape of the displacing fluid front. They impact
the sweep efficiency and as a consequence the amount of bypassed oil.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 11
 Megascopic heterogeneities correspond to the interwell spacing scale. They can be
detected from well test analysis, log correlations, or from the comparison of reservoir
simulations with production data.
Megascopic heterogeneities include lateral discontinuities of individual strata, pinchouts,
fluid contacts, trends, reservoir compartmentalization…
They basically impact fluid flow the same way as macroscopic heterogeneities, but at a
larger scale. In short, they can speed up the displacing fluids to the producers leaving
behind large quantities of bypassed oil.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 12
 Last, the gigascopic heterogeneities occur at the basin scale. This is the delineation
reservoir scale.
In other words, gigascopic heterogeneities correspond to variations in reservoir
architecture. They can be caused by the original depositional phenomena or subsequent
structural deformations and modifications due to the tectonic activity.
Understanding how they are spatially distributed is essential. Some of the oil saturated
areas may remain uncontacted if gigascopic heterogeneities are poorly described.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 13
 As previously emphasized, reservoir heterogeneities may affect fluid flow at any scale.
They are mainly related to deposition processes, diagenesis and/or the structural imprint
on the reservoir.
The sedimentary structures (figure 1), facies associations, depositional environments and
geometry of sedimentary bodies produce heterogeneities at various scales (Eschard and
Doligez, 1993). Their spatial distributions are controlled by the stacking pattern of genetic
units (figure 2) - sequences defined at the reservoir unit scale.
Besides, the mechanical or chemical transformations that affect sediments after
deposition (i.e., diagenesis) are another source of heterogeneity, which can be
superimposed to the previous ones (figure 3). As an example, dolomitization (white zone
in figure 3) can affect the reservoir, modifying its initial properties. This type of diagenesis
can be associated to faults.
Faults can disconnect or connect distinct regions of the reservoir (figure 4). They have a
major impact on reservoir production. When sealed, they act as barriers to flow. On the
other hand, they can also contribute to fluid flow even more than the rock matrix all
around.
Methodologies have been proposed (Van de Graff et al., 1992) to evaluate how much
heterogeneities impact fluid flow referring to the analysis of parameters describing facies
reservoir properties (mainly N/G, porosity and permeability) as well as parameters
describing the geometric features of the reservoir (sand bodies and faults).

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 14
 Reservoir models are inputs to flow simulators. They can be used to evaluate fluid
displacements: the reliability of the simulated responses obviously depends on the
reliability of the models themselves. A model is considered as reliable provided it
respects the available data.
Data are split into two main groups: static and dynamic data.
Static data do not vary with time. They include:
- measurements on core samples extracted from wells: they provide very high-resolution
information;
- downhole logging tools: they are used to describe the electrofacies and the
petrophysical variations along wells. Bedding, fracture features, faults and stratigraphic
features can be identified and quantified by borehole imaging tools and core analysis.
Such data are said direct as they provide values directly characterizing the studied
property.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 15
 However, such direct measurements of geological and petrophysical properties are very
sparse and sample only a small reservoir volume. They are supplemented by indirect
measurements: 3-D seismic. 3-D seismic data are indirect as they do not directly
characterize the target property.
Due to its usual low resolution, 3-D seismic is routinely used for defining only the
structural model.
However, unlike laboratory and log data, 3-D seismic provides information over large
areas, which makes it an invaluable candidate for better characterizing geological and
petrophysical properties. In the case of high resolution seismic, we can use inverted
seismic data instead of focusing on raw seismic data (Doligez et al., 2002; Dubrule, 2003;
Lerat et al., 2007). Inverted seismic data basically comprise velocities and impedances.
Seismic inversion refers to the inverse modeling of reservoir properties from raw seismic
data. Inverted seismic data can help capture reservoir property variability between wells.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 16
 The second group of data includes dynamic data, i.e., data which vary with time because
they depend on fluid flows. They mainly comprise production data, i.e., data measured at
wells such as bottom hole pressures, oil production rates, gas-oil ratios, tracer
concentrations, etc.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 17
 Since the late nineties, dynamic data also consist of data derived from 3-D seismic
surveys repeated at successive times, that is 4-D seismic data (Eastwood et al., 1994;
Benson and Davis, 2000; Arts et al., 2002; Behrens et al., 2002; Guderian et al., 2003;
Roggero et al., 2012). Ideally, a base 3-D seismic survey is acquired before starting
production. Then, monitor surveys are successively acquired after a few years of
production. The differences between the successive seismic responses are caused by
fluid movements, pressure changes, temperature changes, fluid property or
compositional changes, or rock changes. They help monitor fluid fronts and pressure
domains between and beyond wells.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 18
 Each of the data types mentioned above is associated to its own scale of measurement,
its own level of precision, and a given method of measurement with its own physical
principles.
This makes data integration even more challenging.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 19
 Data help understand reservoir heterogeneity.
Thus, well log and core data are used to identify facies as well as petrophysical
properties (e.g., porosity, permeability...). In addition, they make it possible to assess
diagenesis.
On the other hand, well tests and production data provide information about permeability
and connectivity in between wells. They contribute to improve our understanding of fluid
flows and sweep efficiency.
Last, seismic data are used to infer the structure of reservoirs with mainly horizons (top
and base) and faults. They permit to determine the boundaries of the major stratigraphic
units, which impacts fluid flows and sweep efficiency.
In case of high resolution data, they may also help capture reservoir heterogeneity.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 20
 The distinct measurements are used to identify and characterize different groups or types
of rocks.
In geology, a sedimentary facies is a distinctive rock, which forms under certain
conditions of sedimentation, reflecting a particular process or environment (Reading,
1996). It is widely used in connection with sedimentary rocks, but it is not restricted to
them.
The term lithofacies is used to refer to a facies characterized by particular lithologic
features (e.g., grain size, mineralogy, bedding). A lithofacies may be a single bed a few
millimeters thick or a group of beds hundreds of meters thick.
A lithotype is a lithofacies or group of lithofacies defined on the basis of certain selected
petrophysical properties.
Log types or electrofacies are also identified from the classification of wireline log data.
In this case, a classe is associated to a group of data with similar electrical / radioactive /
acoustic / ... characters. To be meaningful, these classes have to be calibrated to core
information: facies, plug data, etc. It is worth mentioning that they may depend on fluid
properties.
A rock type is lithotype or log type defined from a fluid-flow point of view. It is an interval
of rock with unique pore geometry, determined mineralogical composition and is related
to certain specific fluid flow features (saturation, relative permeability, capillary pressure,
SCAL - Special Core Analysis - measurements...).

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 21
 As previously stated, there are not enough data and there are also uncertainties in the
available data due to errors in the measurements or errors in interpretation.
In the case of geophysical data, this entails uncertainties in the determination of reservoir
structure (locations of horizons, of faults), hence in the volume of hydrocarbon initially in
place (HCIIP) and eventually in production forecasts.
Similarly, the uncertainties in geological data involve uncertainties in sedimentary and
petrophysical models. Again, this means uncertainties in the volume of hydrocarbon
initially in place and in production forecasts.
On the other hand, the uncertainties in dynamic reservoir data strongly impact the flow
simulation results, hence the accuracy of the production forecasts.
Likewise, the uncertainties in fluid data can significantly influence the production
forecasts, more especially in Enhanced Oil Recovery (EOR) contexts.
The various sources of uncertainties (Vincent et al., 1999; Corre et al., 2000; Schulze-
Riegert and Ghedan, 2007) are briefly recapped hereafter.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 22
 There are many sources of uncertainties in seismic data that can be identified during the
acquisition, processing and interpretation phases.
They are listed above.
For illustration purposes, various experiments were performed to reveal and quantify
uncertainty in seismic interpretation (Rankey and Mitchell, 2003; Bond et al., 2007). They
showed that you also have to account for conceptual uncertainty. This actually explains
why you may have a large range of interpretations even though based upon a single data
set.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 23
 Geological interpretations are clearly affected by the uncertainties in geological data,
themselves related to the data sets, data types or the density of the data points. However,
the most prominent uncertainty source is that geological interpretations or geological
modeling have a high degree of subjectivity. This is crucial when fixing the boundaries of
the main geological objects.
For instance, the interpretation of a given layer seen in a well has an uncertainty
depending on the quality of the data collected in this well. Geological interpretation makes
it possible to envision what is going on in between the wells.
However, due to the high degree of subjectivity, the quantification of geological
uncertainties is very difficult.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 24
 A typical example is depicted above. Assume that there are two wells drilled in a
reservoir. Two facies are identified along these wells: thin beds of a tight facies, called
Facies 1, and a facies with good reservoir properties, called Facies 2. Given this
information, the problem consists in defining a geological concept. What is in between the
wells? Do Facies 1 heterogeneities correspond to short, intermediate or extended
barriers? How can we quantify the uncertainty associated to a given geological scenario?

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 25
 The geological or reservoir models can be provided as inputs to flow simulators to
estimate how production evolves with time.
Some of the parameters describing these geological/reservoir models, such as fault
sealing capacity and aquifer size are unimportant for the static point of view, but have
strong implications for the dynamic behavior of the reservoir, that is production. These
parameters are also subject to uncertainties: they comprise horizontal permeabilities
(Kh), permeability anisotropy, ratio of vertical to horizontal permeabilities (Kv/Kh), relative
permeabilities (kr) and capillary pressures (Pc), aquifer properties, fault properties,
extension of barriers, well properties...
The sources of uncertainties are related to the possible errors in measurements, but also
to the non stationarity of these properties and to the lack of data.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 26
 In upfront reservoir engineering studies, it is essential to pay attention to the uncertainties
in reservoir properties (geological and petrophysical data). Likewise, the uncertainties in
fluid properties have to be accounted for as they strongly impact the accuracy of
production forecasts.
These uncertainties are mainly related to the lack of representative samples extracted
from the reservoir and to the spatial variations in fluid properties within the reservoir.
In addition, if focusing on the data available, a source of uncertainty is related to the way
the properties of interest are measured in the laboratory. For instance, accurate
measurements are quite difficult to obtain for heavy oil. Significant variations have been
pointed out for heavy oil viscosity measured from different laboratories using different
measuring techniques. Similarly, you have uncertainties in compositional analysis,
volumetric measurements in the PVT laboratory.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 27
 The lack of data and the complexity of the formation processes make reservoir models
and production forecasts uncertain. This uncertainty cannot be removed. Therefore,
reservoir engineers resort to alternate techniques to account for it. They can use industry-
recognized correlations between given properties. They can also investigate hydrocarbon
production for probable scenarios.
An alternative is to build reservoir models respecting the available data. This calls for
geostatistics and history-matching.
Geostatistics is a branch of statistics that studies spatial or spatiotemporal data. It is
currently applied in petroleum geology, but also in hydrology, meteorology,
oceanography, geochemistry, forestry, epidemiology...
On the other hand, history-matching consists of adjusting a reservoir model until it
reproduces (as well as possible) the production data collected in the past, that is the
historical production. Broadly speaking, this problem is an inverse problem. Similar
problems are also considered in many other disciplines such as hydrology, meteorology,
oceanography...
In the chapters hereafter, we focus on the data integration problem referring both to
geostatistics and history-matching.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 28
 Until recently, data integration was considered as a sequential two-step process.
1- The first step, performed by geologists and geophysicists, was based upon the building
of a geological model using geostatistical techniques. The result was a geological model
populated by petrophysical properties respecting geological and geophysical data, but not
production data (more generally speaking, dynamic data).
2- Then, this model was passed along to the reservoir engineer.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 29
 2- The reservoir engineer fed a flow simulator with this model and obtained numerical
production responses. Most of the time, a significant discrepancy was observed between
the responses and the data.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 30
 2- Therefore, the reservoir engineer made the required changes to compel this model to
respect the production data. This history-matching process is an inverse problem. It is
usually solved referring to optimization. However, at this stage, the main concern of the
reservoir engineer was the production data solely: he somewhat disregarded the static
data. At the end, the model, named reservoir model, could more or less reproduce the
production data, but was likely to be inconsistent with respect to the geological and
geophysical data. In this regard, the production profiles predicted from such a model are
questionable.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 31
 Reservoir models typically fall into two categories: geological models or reservoir models.
Geological models are created by geologists and geophysicists and aim to provide a
static description of the reservoir, prior to production.
Reservoir models (or reservoir simulation models) are created by reservoir engineers and
provided to fluid flow simulators to simulate the flows of fluids within the reservoir, over its
production lifetime.
Sometimes, a single "shared earth model" is used for both purposes.
More commonly, a geological model is constructed at a relatively high (fine) resolution. A
coarser grid for the reservoir simulation model is constructed, with perhaps two orders of
magnitude fewer cells. Effective petrophysical properties are then derived from the
petrophysical properties populating the geological model by a process of "upscaling“ and
these effective properties are assigned to the coarser reservoir model.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 32
 Reservoirs are very complex porous media due to the complexity of geological formation
processes. Models are just approximate representations that make it possible to
understand the spatial distribution of hydrocarbons and to identify the best production
design to extract these hydrocarbons.
The finely gridded geological model is populated with petrophysical properties (facies,
porosity Phi, permeability k, fluid saturations sat and net-to-gross N/G). It is used to
evaluate the volume of hydrocarbons initially in place (HCIIP) and its spatial distribution
within the reservoir.
The reservoir model is coarser for practical purposes. It is given as an input to the fluid
flow simulator to predict fluid flows. Therefore, it encompasses less, but coarser grid
blocks to make flow simulation feasible in a reasonable amount of time. The reservoir
model has to be populated with additional petrophysical properties such as relative
permeabilities kr and capillary pressures Pc. This model is used to help decision-makers
to identify production designs optimizing the production of hydrocarbons.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 33
 The integration of data provided by various disciplines raises multiple issues.
An integration process based upon two independent steps (geology and geophysics first,
then reservoir) is not feasible in practice as it does not allow to preserve the consistency
of the model with the geological, the geophysical and the production data together. An
integrated modeling workflow has to be envisioned.
An additional difficulty is that data are associated to different scales meaning that their
integration has to be performed at different scales. This calls for complicated workflows
with appropriate scale changes to make it possible to account for data at the right scale.
Reservoir or geological models consist of millions of grid blocks to be populated with
petrophysical properties. In other words, there are millions of unknown petrophysical
property values and probably not enough data to characterize their spatial distributions
with no uncertainty.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 34
 The traditional approach to data integration was linear or sequential with each discipline
passing its model to the next one with no looking back for consistency checking.
The new approach is dynamic. It is an iterative process including a sequence of modeling
activities (Le Ravalec et al., 2012) with:
- first, stochastic simulation tools to create the initial geological model,
- then, upscaling tools to get the corresponding coarse reservoir model,
- a flow simulator to simulate the flow responses to be compared to the available
production data, and
- a petro-elastic model to simulate the seismic responses to be compared to the available
production and 4D-seismic data.
This provides the framework for the model to be updated when any modeling parameter
is modified. Iteration is a choice and can happen anywhere in the workflow.

Integrated reservoir characterization

and modeling - DOI:
10.2516/ifpen/2014001.c000 35