Reservoir Management | 

Integrated Reservoir
Modeling
Reservoir Models
A” reservoir model” is a mathematical representation of a specific volume of rock
incorporating all the “characteristics” of the reservoir under study. It can be
considered as a conceptual 3D construction of a single reservoir or in some cases
of an oil/gas field.

Data derived from various sources are integrated by deterministic or

geostatistical methods, or a combination of both, to construct the model. Its
setting up, however, is a dynamic process, since a reservoir model must be
continuously up-to-dated and revised when new data become available or
inconsistencies between the predicted and real reservoir behavior are found.

The reservoir model is, therefore, the result of studies whose main objective is to
understand and describe the dynamic behavior of a hydrocarbon reservoir in
order to predict its future performance under different development and
production strategies.

From practical point of view, the integrated reservoir modeling represents now
the most valuable technical approach for estimating the oil/gas reserves and
computing the future production profiles, reducing the uncertainties always
associated with the static and dynamic reservoir descriptions.
 Workflows for integrated reservoir modeling
There are several reasons why an integrated reservoir modeling has found a
strong and rapid development:

 For a reliable evaluation of the bulk and net rock volumes, and the original
hydrocarbons in place – which are of utmost importance in
 assessing the economics of a reservoir development project
 selecting the development schemes and exploitation strategy
 selecting the basic design and size of the production facilities
 allocating equity shares with partners
 For an assessment of the minimum well number required to produce the
reservoir economically, as well as for the optimal selection of well type
(e.g. vertical, slant, horizontal, multilateral, etc.) and locations
 For an economic/technical evaluation of implementing IOR/EOR processes
to increase the final recovery
 For verifying the consistency of all static and dynamic data reducing the
uncertainties always present in a reservoir model.
The integrated reservoir modeling finds application in different stages and
phases of the reservoir life cycle.

In the case of field development it is used for:

 Estimating the HOIP

 Selecting the field development strategy
  Selecting the optimal number and locations for injector and producer
wells
 Computing the production profiles (oil, gas, and water)
 Estimating the oil and gas technical reserves
 Obtaining some basic data required by the economic evaluation
 Identifying and quantifying the key uncertainties.
In the case of producing field it is used for:

 Calibrating the geological model by matching the past production history

(fluid rates, GOR, WC, pressures, etc)
 Identifying the undrained oil/gas bearing zones
 Optimizing the production rate and the final recovery
 Keeping the right injection rate for the optimal reservoir pressure
maintenance and/or for the maximum sweep efficiency of the
displacement processes
 Locating infilling wells
 Modifying the well patterns
 Selecting the best well construction and completion design
 vertical vs. horizontal, completions
 Updating production profiles and economics.
 Dynamic reservoir model for production forecasts
In addition, the 3D integrated reservoir modeling:

 Helps the integration in a quantitative model of soft information such as

sedimentological and depositional models, faults transmissibility etc.
 Enables and promotes the joint team work of geoscientists and engineers
 Reduces the inconsistencies that can be generated by the different geo-
modeling workflows
 Allows a good and reliable volumetric evaluation of fluids initially in place
even in case of complex reservoir geometries
 Allows the most advanced 3D gridding techniques and the upscaling the
geological models
 Helps to select in a real time the optimal final well target while drilling.
Static Model
A static reservoir model is the one incorporating all the geological features (i.e.
structural, sedimentological, petrophysical, etc.) of an underground volume of
rock that can store fluids (hydrocarbons and/or water) and can allow their
movement.

In general, the static model of a reservoir is the final integrated product of the
structural, stratigraphic and lithological modeling activities, where each of these
steps is developed according to its specific workflow.

A static reservoir study typically proceeds through four main stages.

1. Structural modeling
 Reconstruction of the geometrical and structural properties of the
reservoir, by defining a map of its structural top and the set of faults
running through it. This stage of the work is carried out by integrating
interpretations of the geophysical surveys with the available well data.
 A  structural model showing faults and layering
2. Stratigraphic modelling
 Definition of a stratigraphic scheme using well data, which form the
basis for well to well correlations. The data consist of electrical,
acoustic and radioactive wireline logs, and of results of core
analysis, integrated where possible with information from specialist
studies and production data.
 Example of a 3D Stratigraphic Model – L. Cosentino
3. Lithological modeling
 Definition of the lithological types (basic facies), which are
characterized on the basis of lithology, sedimentology, and
petrophysics. This classification into facies is a convenient way of
representing the geological characteristics of a reservoir, especially
for the purposes of subsequent three-dimensional modelling.

Example of a stochastic model of facies –  L. Cosentino

4. Petrophysical modeling
 A quantitative interpretation of well logs to determine some of the
main petrophysical characteristics of the reservoir rock, (porosity,
water saturation, and permeability). Core data represent the
essential basis for the calibration of interpretative processes.
 A petrophysical model showing porosity distribution and values
The results of these different stages are integrated in a two (2D) or three-
dimensional (3D) context, to build an integrated geological model of the
reservoir.

This model represents the reference frame for calculating the quantity of
hydrocarbons in place, and on the other, forms the basis for the initialization of
the dynamic model.

Dynamic model
The dynamic model combines the static model, pressure- and saturation-
dependent properties, well locations and geometries, as well as the facilities
layout to calculate the pressure/saturation distribution into the reservoir, and the
production profiles vs. time.
A dynamic model can be used to simulate several times the entire life of a
reservoir, considering different exploitation schemes and operating conditions to
optimize its depletion plan.
 Integrated reservoir modeling and simulation

Reservoir simulation is a branch of petroleum engineering developed for

predicting reservoir performance using computer programs that through
sophisticated algorithms numerically solve the equations governing the complex
physical processes occurring during the production of an oil/gas reservoir.

Basically, a reservoir simulation study involves five steps:

1. Setting objectives
2. Selecting the model and approach
3. Gathering, collecting and preparing the input data
4. Planning the computer runs, in terms of history matching and/or
performance prediction
5. Analyzing, interpreting and reporting the results.
Factors to consider in selecting the simulation model are

 The recovery process of the reservoir

 the model must be able to reproduce the main reservoir drive
mechanisms
 Quality and type of the available information- which influence the level of
detail to use in the modeI
 Type of answer required
  The desired accuracy of the expected results will influence the
design of the simulation model
 Available resources
 human, economic and technological resources
Different types of simulators are used to represent the drive mechanisms of
different types of reservoirs, and the selection depends on the type and behavior
of the original reservoir fluids and on the predominant process controlling the
reservoir production and hydrocarbon recovery:

 Black-oil model
 It assumes that the thermodynamic behavior of the reservoir
hydrocarbon system can be well represented only by two
components: the ”stock-tank oil” and the “separator gas”. The
classical PVT studies supply all the data required by this approach.
 Compositional model
 It assumes that the reservoir hydrocarbon system can be well
represented only by a number of components and
pseudocomponents (C1+N2, C2, …, C7+, …). The thermodynamic
behavior of such system is described by the use of an EOS
(equation of state) that is usually calibrated with the data of PVT
studies.
 Thermal model
 It is used in case of reservoirs where an EOR process based on
thermal recovery techniques is applied. This is the case of heavy oil,
extra heavy oil, and bitumen reservoirs, in which the oil viscosity is
so high that does not allow any primary production or the
implementation of any conventional injection process (cold water,
gas).
 The thermal EOR processes that can be simulated include: SAGD,
cyclic steam injection, steam flooding, hot and cold water injection,
and in situ combustion.
Upscaling of the geological models is key issue in the reservoir simulation. It is
basically a process by which a very heterogenous region of the reservoir rock
described with a huge amount of “fine grid cells” is replaced by an equivalent less
heterogeneous region made up of a number of single coarse-grid cells. The
“upscaled geological model” must, however, maintain the same storage and
transport properties of the reservoir rock described with detail by the “fine
geological model”. The upscaling process, therefore, is essentially an averaging
procedure in which the static and dynamic characteristics of a fine-scale model
are approximated by those of a coarse-scale model.
 Conceptual illustration of the upscaling process
In a numerical simulation study historical production/injection data (oil, gas, and
water rates) must be supplied to the mathematical model. Of course, good
quality production/injection data are essential for a reliable simulation study, in
terms of direct input data and reference data to evaluate the accuracy of the
history match phase.
Past history matching is the most practical method for testing a reservoir model’s
validity and for calibrating the geological model. Basically history matching is a
process of reservoir parameter adjustment in such a way that the simulated
reservoir behavior reproduces the actual reservoir behavior.
History matching process should also help to identify possible points of
weaknesses in the initial reservoir model, and should help to find and evaluate
the most efficient ways to overcome them.

Two possible approach to History Matching

Once calibrated, the simulation models are then used to compute the production
forecasts considering various hypotheses for the reservoir exploitation.

In simple cases, this prediction phase can be performed in a few days, while the
in more complex cases it can take several months depending on:

 the size (i.e. number of cells), the type (i.e. black oil, compositional) and
geological features of the model,
 the complexity of the wellbore system and of the surface facilities layout
 the number of predictions to be run.
A general sequence for running the prediction phase can be summarized as
follows:

 Input Data for predictions – definition of the cases to be run

 Setting guidelines and constraints – to simulate the future production
performance of a field
 Inflow and outflow well performance
 Running the prediction cases
 Uncertainty assessment.
 Schematic of a 3D uncertainty workflow.
It shows the most common inputs, stages in the 3D modelling process and
corresponding outputs
#12 Hoffmann Plots – Theoretical Foundation
/ Tech Talks / By admin / June 9, 2020

PVTsim users often ask why we do not assign any importance to Hoffmann plots
in the QC module of PVTsim Nova.

The idea of a Hoffmann plot is to validate whether a gas and a liquid separator
sample are in equilibrium at separator conditions. 
The below table shows a recombined reservoir fluid composition for a volatile oil
and the separator gas and liquid compositions at 69 bar and 65 oC.  The
recombined reservoir fluid composition will only be representative of the
reservoir fluid if the separator gas and separator liquid compositions are in
thermodynamic equilibrium at the separator conditions. It would be useful to
have a method to clarify if that is the case.

Recombined Reservoir fluid Separator Gas Separator Liquid

Component
mol % Mol% Mol%

N2 1.775 2.320 0.226

CO2 2.929 3.390 1.620

H2S 11.450 10.669 13.667

C1 61.314 76.443 18.340

C2 2.707 2.873 2.235

C3 2.145 1.786 3.166

iC4 0.772 0.498 1.550

nC4 1.521 0.851 3.426

iC5 0.872 0.332 2.407

 nC5 0.872 0.289 2.527

C6 1.470 0.268 4.885

C7 1.666 0.134 5.871

C8 1.783 0.083 6.529

C9 1.520 0.040 5.609

C10+ 7.204 0.025 27.944

Recombined reservoir fluid and separator gas and liquid compositions at 69 bar
and 65  oC. The recombined fluid has a C7+  molecular weight of 181.5 and a
C7+  density of 0.824 g/cm3 
The method of Hoffmann et al. was  published in 1953 in a paper entitled
“Equilibrium Constants for a Gas-Condensate System” (Petroleum Transactions,
AIME). 
We have dug into the theory behind Hoffmann plots and found it is based on
simple and approximate correlations that are otherwise no longer used in oil
industry.
The Hoffmann plot assumes that a linear relation exists for

where i is a component index and 

K             Equilibrium constant (ratio of component mole fractions in gas and

liquid)
P             Pressure
Pc            Critical pressure
T              Absolute temperature
TB            Boiling temperature at atmospheric pressure
Tc            Critical temperature 
A Hoffmann plot for the separator gas and liquid in the above table is shown
below. A perfect Hoffmann plot would have all the blue dots on the orange line,
which criterion is not completely fulfilled on the below plot

Hoffmann plot for volatile oil separated at 69 bar and 65 oC.

To arrive at the linearity in Equation (1), the pure component vapor pressure of
component i is assumed to follow the below simplified Antoine equation (1888)

where ai and bi are constants specific for component i.

At the normal boiling point (TB) the vapor pressure equals atmospheric pressure
(1 atm), giving
This allows the Antoine equation to be rewritten to

At the critical point the vapor pressure equals Pc and the temperature is Tc. This
gives the following expression for bi

For an ideal liquid mixture and an ideal gas, the K-factor of component i can be
calculated from Raoult’s law (1887)

which means

An ideal mixture is one, where there is no difference between the interaction

between two molecules of the same chemical species and two molecules of
different chemical species.
Combining Equation (8) and Equation (5) gives

This is a special case of the Hoffmann relation where the straight line assumed in
the Hoffmann method follows the equation Y=X. The right-hand side is
independent of pressure, and a pressure correction is needed to arrive at the final
Hoffmann relation.
Poynting (~1900) introduced a pressure correction through what he called a
modified saturation pressure as defined below. It was later known as the pure
component fugacity

is a so-called fugacity coefficient. Replacing  in Equation (8)

by 
assume

where Con_A and Con_B are constants independent of component index.

Combined with Equation (5)  that gives

which reproduces Equation (1). 

Several questionable approximations have been made, which makes the
Hoffmann relation highly uncertain:

 The pure component vapor pressure does not in general follow the
Antoine equation
 A separator gas is not an ideal gas
 A separator oil is not an ideal mixture.
 The Poynting correction is an approximation and component dependent.
Looking at the above Hoffmann plot, one may get the impression that the
assumed linear relation is close to being fulfilled, but that is not quite true. The
below table shows the composition of the sampled separator gas and of the gas
composition you would get with the Hoffman relation, i.e. if all the blue dots on
the above Hoffman plot were moved to be on the orange line. The Hoffmann gas
composition deviates significantly from the measured one.

Component Separator GasMol% Hoffmann GasMol% %Dev

N2 2.32 1.735 -25.2

CO2 3.39 3.015 -11.1

H2S 10.669 10.670 0.0

C1 76.443 78.383 2.5

C2 2.873 2.642 -8.0

C3 1.786 1.497 -16.2

 iC4 0.498 0.387 -22.3

nC4 0.851 0.698 -18.0

iC5 0.332 0.256 -22.8

nC5 0.289 0.221 -23.5

C6 0.268 0.212 -20.9

Measured separator gas composition at 69 bar and 65 oC and the one that obeys
the Hoffmann relation.

The Hoffmann method was good practice at a time when computers and cubic equations
of state (EoS) had not yet made their way into oil industry. Cubic equations were known
at the time Hoffmann presented his method, but phase equilibrium calculations using
cubic equations were out of reach until computers in the seventies became a widely used
engineering tool in oil industry.

Today, more refined EoS-based QC techniques exist. Within seconds, it is possible to

perform EoS-based flash and phase equilibrium calculations, which are much more
accurate than the correlations forming the basis of the Hoffmann plot.

The below plot shows phase envelopes of the separator gas and separator liquid
simulated using a cubic EoS. To be in equilibrium at the separator conditions the phase
envelopes must have a point of intersection at the separator temperature and pressure
(65oC and 69 bar), which is seen to be the case.

QC check for separator conditions. The phase envelopes of a gas and liquid in
equilibrium at the separator conditions must intersect at the separator T and P.
With a cubic equation, it is straight forward to perform a PT flash calculation of
the recombined reservoir fluid composition at separator conditions and compare
the simulated K-factors with those derived from the sampled separator
composition. A plot of experimental K-factors versus simulated K-factors should
give a straight line with  Y=X as illustrated in the below figure.

Experimental versus simulated K-factors at separator conditions.

The two latter plots are examples of the QC checks performed by the QC module
in PVTsim Nova.