Seismic inversion for reservoir characterization Petroleum

Sérgio Soares Chicungo

July 2012
Abstract

Characterized a reservoir is predict its behavior to ensure that we will have an optimized
production, and in the case of deviations, the plan development is adjusted by the updating of
which is based on studies.

The aim of this study is to evaluate a model of petrophysical properties (porosity) characterized
only by information from wells and compares it with a model characterized by wells and seismic
data, a cube of acoustic impedances, based on geostatistical simulation methodologies and co-
stochastic simulation.

Keywords

Petroleum Reservoir

Geostatistics

Acoustic Impedance

Porosity

Permeability

Uncertainty

Direct Sequential Simulation (DSS)

seismic data and production, and geological

information, aiming to build a 3D model of
1.1. Introduction porosity, permeability and saturation. This
reservoir model is essential for the planning
Characterize a reservoir is predict its of the process of development of the field.
behavior to ensure that its behavior is
consistent with expected, and in the case of In many cases the only source of
deviations, the plan development is information available for the construction of
adjusted by the updating of which is based reservoirs model is based solely on
on studies. petrophysical readings, leading to
unreliable models with uncertainty, high.
For this it is essential to gather data that are
acquired from different sources, methods Thus, the modeling of petroleum reservoirs
and scales, such as laboratory data, wells, or systems aims at developing and

1
implementing mathematical and stochastic • Permeability to allow fluid flow through the
models for the characterization of reservoir to the producing wells.
petroleum reservoirs. It involves the
construction of an oil of a reservoir for the Thus the porosity and the permeability is a
purpose of improving the estimation of
reserves decisions on the development of key feature of a petroleum reservoir, for the
the field production and accumulation of
hydrocarbons.
Of the wide variety of heterogeneous
methods for characterizing the reservoir,
the stochastic simulation algorithms, in 1.1.2. Porosity
general, are of great importance for the
The porosity is the amount of voids or pores
estimation procedures, since they allow to:
in the rock, which control the volume of fluid
• Play more faithfully the basic statistics, that may contain the rock. The amount of
histograms and the experimental fluid existing in reservoir rock is a function
variograms, and; of pore volume thereof.

• Knowing the uncertainty of the variable Is defined mathematically as percentage

associated with each location is sampled porosity, as the relationship between pore
volume (or air voids) and the total volume of
The reservoir and its rock.
characteristics
Vp
A reservoir is formed of one (or more) of
 (%)   100 1
Vt
subsurface formations containing fluids and
/ or gaseous hydrocarbons, of sedimentary
Where: - Porosity (%) - Volume of pores; -
origin (or not, in some exceptions). The
reservoir rock is porous and permeable, total volume of the rock
and its structure is limited by the
impermeable barriers that trap the Note that the porosity does not contain
hydrocarbons. information about the morphological
parameters of the pores, such as surface
area and / or the degree of connectivity. To
describe the morphology of the pores, it is a
function which takes the value 1 for the
points (x, y, z) located in the pores, or 0 to
the point in the rock matrix. Using this
function will be possible to represent the
morphological characteristics of the porous
medium. The porosity could be obtained
Figure 1.1 - Example of structural traps from integral, represented by:

A reservoir must have two fundamental

properties of commercial interest to    ( x, y, z )dV 2
V
present:

The porosity is distinguished in two forms,

• Porosity to contain fluids;
the effective and total. As regards the

2
disposal of the sediment can be classified
as primary and secondary
2.1.1. Basic geoestatistcs
1.1.3. Permeability concepts

The permeability of the rock is defined as

2.1.2. Exploratory data analysis
its conductivity to the fluid, ie the ability to
allow fluid to pass through. Depends on the In exploratory data analysis, we
way the pores are interconnected, ie, the characterize the statistical behaviour of the
effective porosity. univariate and multivariate set of existing
information. This step aims to extract all
2.1. The geostatistic on the possible information about the data we
reservoir characterization have, so they are characterized and
quantified the relationship between different
The geostatistic provides two different
types of data and information (from the
methods of calculation which usually boil
seismic logs).
down into two categories: i. is., estimation
and simulation. Both techniques use the 2.1.3. Univariate analysis of data
variogram and correlogram as measures of
spatial continuity. It is the analysis on a single variable, the
main objective is to present the feature or
The common way of measuring the spatial the tendency of a data variable.
representation is the variogram and
correlogram. The arithmetic mean is a first example of
one. It is usually represented by:
The seismic inversion Geostatistics has

  xi 3
been a technique commonly used to embed 1 n
mx
information in stochastic models of the Grid. n i1
Basically, the inversion methods
The dispersion of the sample is usually
geostatistics perform a sequential
approximation in two steps: acoustic calculated by the variance, which

impedance values are simulated for each represents the mean of the squared

detail (a column of a gid and 3D) data differences between the observed values

based on the well and of spatial patterns as and the average of the variable. Is given by:

shown by the variogram; then in a second 

) 2 4
1 n
step, the impedance values are processed S2  
n i1
( xi  x
by a convolution with a known amplitude
estimate for a given seismogram which can 2.2. Bivariate analysis
be compared with the real seismic.
It aims to study simultaneously two (or
more) variables. Establish relationships
between variables and determine whether
the differences between the distributions of

3
these variables are statistically significant, for the characterization and study of the
with the purpose of searching for wells.
influences, causalities or coincidences.
The data, although synthetic, addressing a
2.2.1. Spatial continuity real case, proving to be sufficiently reliable
(variography) to illustrate how to apply the methods to
assess the impact geostatisticos qualitative
The study of variography or spatial
and / or quantitative integration of seismic
continuity aims at knowledge of the spatial
data in the construction of an oil reservoir
dispersion of the experimental data and the
petrophysical model.
degree of anisotropy between the variables
in question. These data are given in an oil field is
containing 10 wells, whose dimension (in
The variogram is the instrument used for
the directions X, Y, Z) is 50m 50m 61.
this purpose it is intended to represent
quantitatively the variation of a regionalized 3.2. Univariate statistics
phenomenon in space (Huijbregts, 1975). (histograms)

2.2.3. Simulation and Direct 3.2.1. Acoustic impedance

Sequential Co-simulation (SSD-
The acoustic impedance is defined as the
SSD and CO)
product of the density (ρ) and the speed v
The direct sequential simulation (DSS) is a of the medium traversed by the seismic
simulation method which uses the original wave. The acoustic impedance can be
variable does not require any processing, clearly linked to many properties of the
which is clearly an advantage over SSG reservoir, such as the lithology and porosity
SSI or will to co-simulation or simulation of constituents and some of the pores.

  v ) 5
continuous variables (Smith, 2006 .)
(I 

3.1. Case study and analysis of

results

3.1.1. Variables in study

The initial information that serves as the

subject of this paper consists of
petrophysical data containing readings of
values of acoustic impedance, porosity,
permeability and acoustic impedance
respective hub which will serve as
secondary and supplementary information

4
 Figure 3.1 - Location and distribution of the
acoustic impedance in the wells and the
histogram data of acoustic impedance and its
distribution function

From the histogram we can get an idea of Figure 3.2 - Location and distribution of the porosity in
the wells and the histogram data of porosity and its
how the variable is distributed, we note that
distribution function
the average distribution of the sample is in
the order of 7083.51, the variance can be The image shows the location of the wells, the
considered high in the order of 2,460,129.06. wells have a low acoustic impedance, have very
You can also check that the probability high values of porosity, this phenomenon can be
distribution function which can be attributed explained because of the negative correlation
seeing the histogram, it will be much like a between these two properties.
normal distribution.
The histogram of this variable shows that the
Can also be considered a nearly symmetrical average porosity is 0.31 and the variance is 0.01.
histogram in which frequency is higher near The distribution function which can be attributed to
the center, with a slight decrease markedly at this normal histogram is asymmetric with extreme
the left and right of center. The mean and values, where the frequency is higher for high
median can be considered similar and close values of porosity.
to the center.

3.2.3. Permeability
3.2.2. Porosity

5
 3.3.1. Acoustic impedance

Figure 3.3 - Location and distribution of the

permeability in the wells and the histogram data of
permeability and its distribution function

The image clearly shows that the wells

show low values of permeability; this detail
can indicate that the container has a good
storing ability pear hydrocarbons.

The histogram of this variable shows that

this variable is the mean and variance
1086.96 is 97583.79. The median is
practically equal to the average and are
located in the center of the histogram, and
the probability distribution function is similar
to a normal distribution, be it a symmetrical
histogram in which frequency is higher in
the center or near this, and gradually
decreases towards both sides Figure 3.4 - Variogram adjusted acoustic
impedance
symmetrically.

3.3. Spatial continuity

(variography) 3.3.2. Porosity
Experimental variograms were calculated
for different directions, but I will mention
only those of greatest interest, especially
those representing the principal directions,
because they give the idea of variable
distribution without regard to a particular
direction.

It is important to cite all the variograms

were fitted using the exponential model is
the one that best fits the variogram.

The direction is the main (90, 0), a minor (0,

0), minir 2 (0, 90) and the omni-directional
(0, 0).

6
 Initially it was made considering the SSD
only information data from the wells and
then co-SSD regarded as the cube of
acoustic impedance as a secondary image.
30 simulations were performed for the
impedance, porosity and permeability and
calculated the mean and variance of each
set of simulation to see how the data vary
Figure 3.5 - Variogram adjusted porosity from simulation to simulation and to
compare and quantify the uncertainty in
each simulation method.

3.3.3. Permeability No need for a more detailed analysis of the

behavior of each of the 30 simulated
images, we performed the calculation of the
variograms, histograms and basic statistics
of the different representations and
compared with the sample data from wells.

The following is a series of images

corresponding to each set of simulations,
as well as the mean and variance of each
of the simulations 30 and their histograms. I
have only one image for each simulation,
since they have very similar visual aspects.

4.1. Acoustic Impedance

The DSS of this property of the reservoir

(acoustic impedance) was based only on
data from 10 wells, with the aim of
improving the characterization of
heterogeneity and incorporate the
uncertainty in the estimate.

Marked by the low values of acoustic

impedance, it can be seen (in the figure
below) that this property distribution along
the reservoir. At the top, continuous layers
Figure 3.6 - Variogram adjusted permeability
can be identified with low values of acoustic
4. Direct Sequential impedance, which may correspond to the
main carrier range of hydrocarbons. The
Simulation
lower layers have continuities less
This work presents the application of the pronounced, with average to high acoustic
SSD algorithm on the readings of acoustic impedance.
impedance and porosity, performed in 10
wells in a cube of dimensions (50m 50m
61m) (x, y, z), later to compare with the
results of Direct Sequential Co-Simulation
which will be integrated into the data cube
acoustic impedances as auxiliary
information.

7
 Figure 4.2 - Histogram of the simulation of
acoustic impedance and histogram of the
original data of acoustic impedance

Figure 4.1 - Picture of the 1st simulation of the

acoustic impedance plane (x, y); perspective

As might be expected, the simultaneous

analysis of these results to the images
generated histograms, lets say that for the
variable acoustic impedance, comparing
the histograms of simulated values and the
sampling of the wells, it appears in both the
first class histograms values as a group
most frequented, there is still an almost
uniform distribution for the intermediate
classes and a roughly similar mean and
variance. However it is observed in both
cases a weak values representative of the Figure 4.3 - Variograms of the simulated
acoustic impedance values from the DSS, and
extreme right the average coefficient of
the theoretical model fitted to experimental
variation and also low.
variograms

Regarding the variogram with the main

directions, the results are also
extraordinary. The theoretical model of the
actual values of the variogram fits well to
the variogram of the simulated values.

8
4.1.2. Porosity are similar. However histograms have the
same distribution of the samples, the
distribution law which can be attributed to
these histograms is asymmetric with normal
extreme values, where the frequency is
higher for high values of porosity

This similarity might suggest that this

simulation is variable in accordance with
the expected, ie, that the reflected
simulation equiprobable set of images with
a very close spatial variability of the sample
data.

Figure 4.4 - Image of a simulation of the

porosity in the plane (x, y); perspective

Figure 4.6 - Variograms of the simulated values

of porosity from the DSS, and the theoretical
model fitted to experimental variograms.

The variogram simulation analysis shows

that the simulation of the variogram
presentam much resemblance to the
original data, so that the exponential model
was chosen to adjust the variogram, such
as the sample data set of the porosity.
Figure 4.5 - Histogram of the simulation
histogram of porosity and porosity of the original
data

The histograms show that the variance is

less than the simulation data for Pit
although the average porosity in both cases

9
4.2. Direct Sequential Co-Simulation

On similarity of the DSS, there were 30 co-

simulations (presented only in the first figure
below) about 570 readings of porosity in 10 wells.
The coefficient of correlation between porosity
and AI is 0.84, which shows a strong
interdependence between the spatial variables of
interest, and validating the application of the Figure 4.8 - Variograms of porosity values co-
simulated from the DSS, and the theorical model
methodology.
fitted to experimental variograms.

5. Conclusive Analysis

Having carried out 30 simulations for each

of the variables and made the calculation of
means and variances in each case, one
can consider that the results were
satisfactory. The simulations were
conducted using the sequential simulation
right in which I had in mind only the
information of the wells.

The achievements obtained as a result of

direct sequential simulation there were
areas of high variability of the porosity and
to a lesser extent to acoustic impedance.
However, outputs that are obtained by
direct co-simulation sequential co-located
with local correlation, and for which used
the image impedance of the hub as
secondary information, and show less
variability lowest uncertainty as to the
previous method. Correlations and
Figure 4.7 - Image of a co-simulation of the
information from univariate and bivariate
porosity in the plane (x, y); perspective
analysis were of great importance for the
analysis and interpretation of images
obtained after the simulations.

BIBLIOGRAPHY
Gomes J. S; Alves F. B. (2007) - Universo
da indústria petrolífera, FCC

HUIJBREGTS, C.J. Regionalized variables

and quantitative analysis of spatial data. In:
Davis, J.C. & McCullagh, M.J. (ed) Display
and analysis of spatial data.

10