Documente Academic
Documente Profesional
Documente Cultură
the human expert reasoning process much more realistically Zone of Interest
than do the conventional systems15. Fuzzy set theory is an The focus of this study was the area limited by the
efficient tool for modeling the kind of uncertainty associated rectangle (x=820000 y=130000, x=828000, y=140000), in the
with vagueness, imprecision and/or a lack of information northern part of Hassi Mesaoud field (Figure1). Basically this
regarding a particular element of the problem at the hand. area includes the northern part of zone 13, the southern part of
This study aims to apply the above discussed techniques to zone 8 and the interzone in between them. Some parts of zones
characterize naturally fractured reservoirs and Hassi Messaoud 6, 9 and 10 were also taken into account (Figure 2)
field Algeria is used as a case study. The interpretation of recent image logs recorded from
some vertical and horizontal wells, core analysis and pressure
Literature Review buildup tests indicate that this area is naturally fractured.The
The application of Artificial Intelligence tools such as zone of interest; especially the interzone mentioned above; is
Fuzzy Logic and Neural Network is evolving as an oilfield now an object of a development plan by drilling new
technology. During the last few years several studies have horizontal wells. The study presented here aims to obtain an
been conducted in the field of petroleum engineering by image of the most probable zones to be naturally fractured.
applying artificial intelligence. The major applications are The first step consists of building a stochastic model of
seismic data processing and interpretation, well logging, permeability, porosity and shale volume (Vsh).
reservoir mapping and engineering.
Recent applications of artificial intelligence tend to focus Data Collection
on reservoir characterization. Big efforts have been devoted to Geologically this area is located on the top of the reservoir.
the area of naturally fractured reservoirs. Balch14 used the well Basically, it includes three layers R2, ID, and D1. The ID was
data set consisting of traditional logs and tuned-to-core eroded and was not observed in several wells, for such reason
estimates of secondary porosity from FMI image to train an geologists treat the ID and D1 as a single layer Dh.
artificial neural system to predict the secondary porosity. Data were collected from the two layers together. Core
Neural network has also been used to provide porosity and permeabilities and porosities were collected from 36 vertical
permeability estimates across a study area by training on wells with sampling frequency of four cores per meter.
seismic amplitude information. Sadiq and Nashawi12 presented Gamma ray data were taken from 20 vertical wells and two
a neural network model to predict fracture gradient pressure; horizontal wells lying in the interzone 8-13 (OMPZ333 and
the results indicated that neural network is not only feasible OMPZ34).
but yields quite accurate results. We also gathered CMR results from these two horizontal
Ouenes et. al6,7,8,9,10,11 introduced a collection of artificial wells in terms of permeability and porosity. Theses two
Intelligence tools to model the interwell region of fractured horizontal wells helped in constructing horizontal variograms
reservoirs. Assuming that fractures can be represented by well and selecting the appropriate model type to be used later in the
performance at grid block scale, Ouenes used AI tools to map geostatistical software.
between different geological and geomechanical drivers and
the well performance. Eduardo13 proposed a Fuzzy algorithm Statistical Analysis
for fracture detection using conventional well logs and drilling The statistical analysis carried out in the present
information. He also suggested the use of Fuzzy Curvature study showed a great heterogeneity in reservoir properties in
Index for application to naturally fractured reservoirs. terms of permeability and shale volume. The coefficient of
Ouenes and Lee9 presented an integrated fractured variation, Cv, for the case of permeability varies between 1.2
reservoir model using both discrete and continuum approach. and 3 and the average mean value varies from few millidarcies
Gauthier19 applied successfully such Integrated Fractured up to more than 100mD. Vsh exhibit less heterogeneity with Cv
Reservoir Characterization to a case study in North Africa range between 0.4 to 0.8 whereas porosity does not change too
much with Cv between 0.22 and 0.55 and a mean value of
Hassi Messaoud Field description about 8%.
Hassi Messaoud (HMD) field is situated in the
Algerian part of Sahara desert and was discovered in 1956. Establishing PDF’s and CDF’s
With its reserves (OOIP) of several billion m3 in place, the Using 7090 samples of permeability and porosity and
HMD field is considered a giant one. Mainly solution-gas 8000 sample of Vsh, PDF’s and CDF’s were built for each of
driven, the productive formation is Cambrian sandstones the three parameters, (Figure 3, 4 and 5). Examining the three
3400m deep with an average thickness of 300m and 3 figures, we can notice that:
productive layers (R2, Ra, and Ri from bottom to top). • Vsh, calculated from GR log, shows a negatively skewed
The R2 and Ri layers are characterized by poor distribution where most of data lie to the low end of the
petrophysical properties. The important layer is Ra and is data range.
characterized as heterogeneous with an average porosity of 8% • Core porosities follow very well a normal distribution. In
and a permeability of 5 mD. The large extension of the field fact after taking a cut off of 25 values from the amount of
and the presence of fault barriers and the facies lateral 7090 available data, all the remaining 7065 lie on the range
variation have confirmed the nonuniform depletion throughout of 0 and 0.20 giving a nice Gaussian shaped distribution
the field, which has led to its zonation, Figure 1. • Permeability, however, follows a log normal distribution.
The corresponding lognormal histogram shows basically
SPE 84870 3
two clusters with almost the same percentage suggesting Mesaoud sandstone reservoir generated using conditional
the presence of two groups of data. These two clusters simulation and spherical areal variograms. The reddish color
correspond to the layers R2 and Dh respectively. indicates good characteristics (high k, high φ and low Vsh),
whereas the bluish color indicates poor characteristics (low k,
Areal and Vertical Variability (Variograms Construction
low φ or high Vsh). All of the three maps represent the
and Analysis)
While parameters of vertical semivariograms are distribution of the properties in a succession of good sand
usually available from the well data, the lateral parameters bodies distributed among a poor sandstone medium, into
must be established from the normally sparse well data. It is which are the dispersed discontinuous siltstones which have
important that uncertainties in these parameters be low porosity and permeability. They also show the distinction
acknowledged in the modeling schemes. of the inter zone13-8 of low petrophysical characteristics
which confirm the actual zonation in this area.
Areal Variability
The purpose of presenting areal semivariograms driven Geomechanical drivers
from horizontal well data in this study is to diagnose the In the previous section, the geological model was
lateral variability and sketch out the appropriate model type built for the reservoir properties of permeability, porosity and
(i.e., Spherical, Gaussian or Exponential).This model will be shale volume. These parameters will serve as a set of inputs
used later on to fit the sparse well data that characterize the for the upcoming neural network. In this section we will deal
lateral variability in four different directions. with the calculation and analysis of structure shape (slopes and
Areal variability is available from horizontal well logs curvatures), distance to faults and, bed thickness (Figure 13).
interpretation in the northwest to southeast direction.
Horizontal semivariograms were calculated using Slopes and Curvatures
permeabilities and porosities obtained from the results of Stress in rocks depends on the curvature. As such,
CMR interpretation and Vsh obtained from the GR logs. the shape of the formation structure top can have a significant
Figures 6-8 show horizontal semivariograms of permeability, role on fracture development. Curvature analysis is inherent
porosity, and shale content respectively. They all illustrate a from Maurry’s model that relates fracture porosity and
combination of exponential and spherical variability. They permeability to structural curvatures.
also reveal some periodicities represented by a succession of ∂ 2z ∂2z
2
the rectangle is filled with black color. Four classes have been Input Fuzzification
distinguished A, B, C and D, where class A presents the best This fuzzy inference system has been built on 8 rules, each
characteristics and class D presents the poorest ones. rule depends on resolving the inputs into a number of different
This study covered 24 wells where sufficient data were fuzzy linguistic sets: skin is low, fractured length is high, class
available. The descriptive information has been converted into indicator is low and so on. Before the rules can be evaluated
digital data presented in Table 12.4. It was noted that class D the inputs must be fuzzified according to each of the linguistic
was absent almost in all the wells. For such reason, those very sets. As an example, the Figure 15 shows that to what extent
rare fractures of class D were added to class C. Open fractures the skin is low. In that manner all the inputs are fuzzified and
represent 10 to 96 % among the whole fracture system with an overall the qualifying membership functions required by
average of 65%. the rules.
Fracture Index model (FFI). We notice that the major drivers presented to the network the error is large. This problem
for the first case are those related to lithology (permeability occurs mostly in case of large networks with only few
and porosity) and then come the structural drivers whereas for available data.
the second case it is those geomechanics related drivers which There are a number of ways to avoid overfitting3.
come first followed by lithology drivers. However, with just fixing the error and the number of epochs
to an adequate level (not too low , not to high) and dividing
Neural Network Design the data into two sets; training and testing; one can ovoid such
After trying several architectures with different problem by making several realizations and selecting the best
number of layers, neurons and transfer functions, the best of them.
results have fallen on the one described in Figure 17. This
network has two hidden layers with logarithmic transfer Training Design
functions implemented in all neurons. The first hidden layer For both models 75% of the available data were used
has eight neurons. Each neuron has a bias and is fully to train the neural network. The other 25% were left aside for
connected to all inputs, which means 8×8+8 =72 connections. testing purpose. Prior to any modeling all data were scaled to
The second hidden layer has also eight neurons where each the range [0 1]. Three training algorithms were tested:
one has a bias. The connection between this layer and the conjugate gradient, scaled conjugate gradient and Levenberg
previous one is ensured by neuron to neuron connection (each Marquadt. We have found, however, that the second and the
neuron from the second hidden layer is connected to one and third ones are more suitable for our case. Training stop criteria
only one neuron from the first hidden layer and vise versa). was chosen to a maximum number of epochs of 150 and a
This layer has then 8+8 =16 connections. All the neurons of mean squared error of 0.005.
the second hidden layer are connected to the output layer. The
output layer has only one neuron without bias which means 8 Testing
more connections. Finally we have 96 connections to be Several realizations were conducted where
adjusted by a proper training algorithm. sensitivities were made on the training algorithm parameters:
the training algorithm itself, the stopping criteria and the wells
Inputs Selection used in the training and testing sets. Once the training process
Selecting the appropriate number of inputs is crucial, yet converges, the testing data set must be presented to the
difficult task. However, we have found that taking the first network. If the testing present good agreement between the
eight inputs among the available thirteen leads to good results. actual and the predicted fracture index, the bias and weight
The number of eight inputs has neither been chosen arbitrarily matrices must be saved and kept aside. If not, the realization
nor under certain precise rules. It is the result of the is cancelled. Such process has been repeated several times
following observations: until a satisfied number of realizations with good testing
• We only have few data available for the training. results were achieved.
Taking all the thirteen inputs with a similar architecture results
in 221 connections and thus very little chances to ovoid Predicting
the overtraining. Once the training leads to satisfactory results, the
• At least the first layer of the network should have a network is considered to be trained and generalized well. It is
full connection with the inputs. then ready to predict the fracture index on the other grid
• The more inputs we introduce, the better modeling blocks. From those best realizations obtained in the previous
we could achieve. two sections, the one that produce a fracture map
• To better model the problem we should have inputs corresponding to intervals where we do have high confidence,
from various sources is selected
• Some inputs have a strong relationship with the
fracture index than others do. Neural Network Results: PI model
Taking these five remarks into account and the fuzzy For this model, productivity index was taken as a
ranking results of the drivers, we have chosen to take the first fracture index assuming that well performance is linearly
eight inputs. related to the fractures. Only early PI’s were selected, these
PI’s range from 0.001 to 1.68 m3/h/kg/cm². Since the output
Neural Network Modeling transfer function in the Neural Network model is sigmoid
A key element in using Artificial Neural Network in which produces an output in the range of [0 1], we must
this study is that the fracture distribution is a complex process provide our data by fracture indexes in the same range [0 1]. If
to be modeled by classical regression. A neural network we take FI = 0.5 × PI, we get our fracture index in the range of
modeling process comprises three basic steps, training, testing [0 0.84] leaving a degree of freedom of 0.16 for the network to
and predicting. predict higher values (if they could be). From the 46 available
wells, 10 were chosen from different locations and left aside
Training: The Overfitting Problem for testing purpose. The training/testing and predicting process
One of the most common problems in the training process has been conducted using a Matlab program based on BPNN.
is the overfitting. This happens when the error on the training Figure 18 shows the training testing results. It may seem that
set is driven to a very small value, but when new data is the predicted values are somehow different from the actual
values, but they can be considered in a good agreement for a
SPE 84870 7
Nomenclature 18. Weiss, W.W. and Balach, R.S.: “How Artificial Intelligence
Rfi : Borehole corrected short normal resistivitly, Ω-m. Methods can Forecast Oil Production”, SPE 75143
Rfo : Borehole corrected log normal resistivity, Ω-m. 19. Gauthier et al.: “Integrated Fractured Reservoir
Rmf : Mud filtrate resistivity, Ω-m. Characterization: a Case Study in a North Africa Field.”, SPE
Rma : Matrix resistivity, Ω-m and, 65118 .
Rw : Water resistivity,Ω-m.
20. Aissaoui, K.: “Contribution of Horizontal Well Logs and
Pressure Data in Stochastic Modeling.” MS Thesis, Norman,
References Oklahoma 2001.
1. Haykin, S.: “Neural Networks,” McMaster University Hamilton,
Ontario, Canada. Prentice Hall International, Inc. Table 1 Horizontal wells performance
2. Hagan, M.T.: Howard B. Demuth, Mark Beal,: “Neural Network
Design”, PWS Publishing Company.
WELL PI
Skin kx ky kz
NAME m3/h/kg/cm2 mD mD
mD
3. Demuth, H. and Beale, M: “Neural Network Toolbox for use
with MATLAB”, User’s Guide, Version 4, MATLAB 6.0. OMPZ34 1.14 -2.7 1.9 200 82
OMPZ111 3.4 1.7 17 35 0.7
4. Demuth, H. and Beale, M.: “Fuzzy Logic Toolbox for use with
MATLAB”, User’s Guide, Version 2, MATLAB 6.0. OMPZ331 0.412 -1.9 1.75 6.12 1.98
Well Name
Characterization and Performance Forecasting Using
Geomechanics and Artificial Intelligence,” paper SPE 30572 Class A Class B Class C
Open
Close
Open
Close
Open
Close
7. Zellou, A.M., Ouenes, A. and Banik, A.K: “Improved Fractured
Reservoir Characterization Using Neural Networks,
Geomechanics and 3-D Seismic,” paper SPE 30722
# 01 2.84 0 3.78 0.95 0 0
8. Ouenes, A., Zellou, A.M., Nova, T., Basinski, P.M. and Head, # 02 0 0 10.51 2.97 0 0
C.F.: “Practical Use of Neural Networks in Tight Gas Fractured
Reservoirs: Application to the San Juan Basin.” Paper SPE # 03 0 0 15.49 1.46 0 0
39965 # 04 0 0 10.00 7.50 0 0
9. Ouenes, A. and Hartley, L.: “Integrated Fractured Reservoir # 05 0 0 9.53 6.32 0 2.08
Modeling Using Both Discrete and Continuum Approaches.” # 06 0 0 5.71 6.86 0 0
Paper SPE 62939
# 07 0 0 23.56 0.61 0 0
10. Ouenes, A.: “Practical application of Fuzzy Logic and Neural # 08 0 0 13.09 0.15 0 0
Networks to Fractured Reservoir Characterization”, Computers
and Geosiences, Vol.26, pp. 953-962, 2000. # 09 0 0 6.75 27.58 0 0
11. Barman, A., Ouenes, A. and Wang, M.: “Fractured Reservoir # 10 0 0 8.85 0.77 0 0
Characterization Using Streamline-Based Inverse Modeling and # 11 1.08 0 28.85 9.19 0 0
Artificial Intelligence Tools.” Paper SPE 63067. # 12 0 0 11.54 13.08 0 0
12. Sadiq, T. and Nashawi, I.S.: “Using Neural Networks for # 13 1.88 0 24.38 7.15 0 0
Prediction of Formation Fracture Gradient.” Paper SPE 65463.
# 14 0 0 5.43 27.32 0 0
13. Eduardo, Quintero, J., Martinez, L.P. and Gupta, A: # 15 0.60 0.2 18.10 5.40 0 0
“Characterization of Naturally Fractured Reservoirs Using
Artificial Intelligence.” paper SPE 67286. # 16 26.60 6.60 22.98 24.04 0 0
14. Balch, R., Weiss, W. and Wo, S.: “Correlation of Traditional # 17 0 0 2.23 2.05 0 0
Wire-Line Logs to FMI Estimated Secondary Porosity”, For: # 18 4.41 2.06 2.94 2.65 0 0
Yates Petroleum Corporation –April 1998. # 19 0 1.79 0.22 0.11 0 0
15. Soto, R. B., Garcia, J.C., Torres, F. and Perez, G.S.: # 20 0 0 6.09 3.85 13.40 0
“Permeability Prediction Using Hydraulic Flow Units and
Hybrid Soft Computing Systems,” SPE 71455. # 21 0 0 24.29 5.71 0 0
# 22 0 0 5.56 4.44 0 0
16. Weiss, W.W., Wo, S., Weiss, J.W. and Weber, J.: “Data Mining
at a Regulatory Agency to Forecast Waterflood Recovery” SPE # 23 2.01 5.97 7.78 0.42 0 0
71057 # 24 0 0 22.83 0.83 0 0
17. Balch, R.S., Broadhead, R. F., Weiss, W.W. and Hart, D. M.:
“Regyonal Data Analysis to Better Predict Drilling Success:
Brushy Canyon Formation”, SPE 75145
SPE 84870 9
(fraction)
High_K
Skin
NPI
(%)
FFI
FL
CI
Model
Model
Model
FFI
FFI
800 100%
PI
PI
600 75%
Frequency
% Sum
Permeability 100 57.70 1 4
2
88.52 23.29 2 8 400 50%
Porosity
Vsh 53.52 36.30 10 5 200 25%
Slope SN 59.98 28.72 8 7
0 0%
Slope WE 44.74 19.12 12 11
0.04
0.12
0.2
0.28
0.36
0.44
0.52
0.6
0.68
0.76
0.84
0.92
1000 125%
Frequency
% Sum 25
800 100%
20
600 75%
Frequency
% Sum
15
Variance
400 50%
10
200 25%
5
0 0%
1 3 5 7 9 11 13 15 18 More 0
0 50 100 150 200 250 300 350 400 450 500
Class
Lag distance
Figure 4. Histogram & cumulative distribution function
"CDF" of porosity.
800 125% Figure 8. Average horizontal semivariogram of Vsh.
Frequency
% Sum
100%
600
75%
Frequency
% Sum
400
50%
200
25%
0 0%
-2 -1 0 1 2 3 4 5 6 7 8
Class
4000
3000
Variance
2000
1000
0
0 25 50 75 100 125 150 175 200
Lag distance
6
Variance
0
0 25 50 75 100 125 150 175 200
Lag distance
Figure 12. Map showing the structure top over which the
directional derivatives were calculated
Bed Thickness
Slopes and
Curvatures
x
Distance to
Faults
1.0
Training
Testing
Predicted Fracture Index
0.8
0.5
0.3
0.0
0 0.2 0.4 0.6 0.8 1
Figure 15. Input fuzzification. Actual Fracture Index
1
Trainig
Testing
Predicted Fracture Index
0.8
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
5
x 10
1.4
1.39
1.38
High
1.37
1.36
1.35
1.34
1.33
1.32
Low
1.31
1.3
8.2 8.21 8.22 8.23 8.24 8.25 8.26 8.27 8.28
5
x 10
Figure 19. Fracture intensity map based on the productivity Figure 22. Fracture Network Map (PI model).
index (PI model).
5
x 10
1.4
1.39
1.38
High
1.37
1.36
1.35
1.34
1.33
1.32
Low
1.31
1.3
8.2 8.21 8.22 8.23 8.24 8.25 8.26 8.27 8.28
5
x 10