Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Earth science
Reexamination Certificate
2000-08-08
2002-04-30
McElheny, Jr., Donald E. (Department: 2862)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Earth science
C702S014000
Reexamination Certificate
active
06381543
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method for forming a model of an underground zone, constrained both by dynamic data and static data obtained by means of measurements or observations.
One of the main current research means for improving location techniques relative to zones likely to contain hydrocarbons consists in giving better descriptions of the inner architecture of reservoirs. Engineers therefore essentially have two types of data: dynamic data (reservoir) and static data (seismic).
I) BACKGROUND OF THE INVENTION
An inversion technique is commonly used for constructing really representative reservoir models, starting from a geologic model given a priori which is gradually modified in order to reduce the differences between the values of the parameters or physical quantities calculated by the model and values of these quantities that can be observed or measured at certain points of the medium, by best minimizing a cost function J.
The model can be optimized by inversion of static data that can be deduced from seismogram interpretations or that have been obtained by measurements in exploration wells. The most commonly used parameter for inversion is the acoustic impedance of the medium.
The model can also be optimized by inversion of dynamic data: production data spread over a period of time or resulting from well tests. However, the data used concern very localized zones in the immediate neighbourhood of production wells. The parameters used for inversion are here the total permeability of the medium and the porosity thereof In both cases, the inversion algorithm is the algorithm shown in FIG.
1
.
The models obtained by inversion of static and dynamic data are achieved separately by specialized reservoir engineering and seismic interpretation teams. The solutions which specialists in these two fields come to are often multiple and generally very different according to whether static data or dynamic data are used.
I-1 Flow Model
Consider a direct one-dimensional model relative to the flow of fluids between an injection well and a production well in a reservoir diagrammatically represented (
FIG. 2
) by a cylinder of length L extending between the two wells. Knowing the pressure Po of the fluid at an initial time, its evolution within a time interval [
0
,T] is governed by the following equations defining the direct model:
p
(
x
,
0
)
=
p
0
(
x
)
∀
x
∈
[
0
,
L
]
-
k
(
0
)
∂
p
∂
x
(
0
,
t
)
=
q
0
(
t
)
∀
t
∈
]
0
,
T
]
∂
p
∂
t
+
∂
∂
x
(
-
k
(
x
)
∂
p
∂
x
)
=
0
∀
x
∈
]
0
,
L
[
and
∀
t
∈
]
0
,
T
]
p
(
L
,
t
)
=
p
L
∀
t
∈
]
0
,
T
]
where:
p(x,t) is the pressure of the fluid at the time t and at the position x,
k(x) is the permeability of the rock at the position x,
q
o
(t) is the rate of inflow at the time t,
P
L
is the pressure at the distance L from the injection well, that are solved according to a finite-difference numerical scheme. A description of the permeability at the interfaces inspired by the finite-volume schemes allows to take account of the stream continuity at the interfaces. It is assumed that the reservoir is an << assembly of sections >> of homogeneous lithologies and of different permeabilities to be identified. As pressure measurements have been performed at the injection well, at x=0, the following objective function is constructed:
J(k)=½∫
0
T
(p(0,t)−{circumflex over (p)}(t))
2
dt
and inversion of the parameter or of the physical quantity k is thus achieved by means of conventional optimization methods, such as the adjoint state method which allows accurate calculation of the gradient of the cost function in relation to the parameter.
I-2 Seismic Model
Consider a sedimentary basin (
FIG. 3
) consisting of several geologic layers, having each their own acoustic impedance, explored by seismic means. Seismograms showing the response of the medium (waves reflected by the subsoil discontinuities) in response to the emission of seismic waves by a source are available. The wave equation is a hyperbolic equation where the unknown y(t,z) is the subsoil vibration depending on the time t and on the depth z, with the acoustic impedance of the medium &sgr;(z) as the parameter. The same type of finte-difference scheme as that used for the flow equation is used to solve this equation, i.e. a scheme using a description of the impedances at the interfaces inspired by the finite-volume schemes. Because of the initial conditions, this is here an explicit scheme. The geologic interfaces located by seismic reflection means are the boundary layers between the media having different acoustic impedances.
The medium being quiescent at the initial time t=0, and knowing the pressure impulse g(t) imparted to the medium at the surface at each time t, vibration y(z,t) is the solution to the direct model given by the following modelling equation system (z being the depth in two-way time):
(
P
s
)
{
σ
(
z
)
∂
2
y
∂
t
2
-
∂
∂
z
(
σ
(
z
)
∂
y
∂
z
)
=
0
for
z
>
0
and
for
t
>
0
y
(
z
,
0
)
=
∂
y
∂
t
(
z
,
0
)
=
0
for
z
≥
0
-
σ
(
0
)
∂
y
∂
z
(
0
,
t
)
=
g
(
t
)
for
t
>
0
y
(
L
,
t
)
=
0
for
t
>
0
As above, the medium is considered to be an assembly of sections sliced vertically and having each an impedance value. Measurements ŷ being known, a cost function is constructed:
J(&sgr;)=½∫
0
T
(y(0, t)−ŷ(t))
2
dt
which allows to perform the acoustic impedance inversion.
I-3 Problem Linked with Lack of Data
The success of an inversion procedure greatly depends, as it is well-known, on the quality and the quantity of observations available. The more observations, the better the quality of the inversion result. A lack of data leads to averaged information on the parameter to be inverted.
An inverse problem performed on two meshes including information on parameters p
1
, p
2
with an insufficient number of data only allows to obtain the mean value of the two parameters. As for reservoirs for example, it is the harmonic mean {overscore (p)} of the two parameters that will be inverted as the result:
1
p
_
=
1
2
(
1
p
1
+
1
p
2
)
A great number of possible parameter doublets p
1
and p
2
and consequently of different models for which the (reservoir or seismic) simulations will adjust to the observations corresponds to the same harmonic mean {overscore (p)}. This phenomenon can be verified with the following example of 1D simulation of a flow performed in a reservoir consisting of about twenty meshes, with a permeability k
i
per mesh (at the centre of the mesh). It is a trimodal medium representing a permeable medium (sandstone for example) crossed by an impermeable barrier (clay for example). One tries to know if the model inversion allows to find the permeability gradient. The data are the pressures recorded at the injection well in the course of time.
As shown in
FIG. 4A
, the distribution of the permeabilities obtained after inversion (graph k
2
) with an exact gradient calculated by means of the adjoint state method is very far from the distribution used for creating the injection well data (graph k
0
), whereas the responses of the direct model (
FIG. 4B
) applied to this distribution (graph P
2
) perfectly match the data (graph P
0
). This shows that several models, sometimes very far from the real model, can solve the inverse problem.
II) THE METHOD ACCORDING TO THE INVENTION
The method according to the invention allows to form, by means of an inversion technique, a representation of the variatio
Guerillot Dominique
Pianelo Laurent
Institut Francais du Pe'trole
McElheny Jr. Donald E.
LandOfFree
Method for forming a model of a geologic formation,... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for forming a model of a geologic formation,..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for forming a model of a geologic formation,... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2860492