Vous êtes sur la page 1sur 5

Integrated approach for fracture modeling of carbonate reservoir: A case study of

Baturaja formation, West Java Basin


Abdul Haris, , Agus Riyanto, , Ambar Rachmanto, and , and Adang Sukmatiawa

Citation: AIP Conference Proceedings 1729, 020082 (2016); doi: 10.1063/1.4946985


View online: http://dx.doi.org/10.1063/1.4946985
View Table of Contents: http://aip.scitation.org/toc/apc/1729/1
Published by the American Institute of Physics
Integrated Approach for Fracture Modeling of Carbonate
Reservoir: a Case Study of Baturaja Formation,
West Java Basin
Abdul Haris1, a), Agus Riyanto1, Ambar Rachmanto2 and Adang Sukmatiawa2
1
Reservoir Geophysics Program, Department of Physics, FMIPA, Universitas Indonesia, Kampus UI Depok,
Indonesia
2
PT. Pertamina EP

a)
Corresponding author: aharis@sci.ui.ac.id

Abstract. Baturaja formation has become one prospective reservoir in field A, which has potential number gas inplace
with recovery factor up to 29%. However it has a big challenge for developing because of its structural complexity. From
special core analysis data and Formation Micro Imaging (FMI) data analysis, it was shown that this reservoir is
dominated by fracture indicated by cementation exponent (m) below 2. This paper aims to perform fracture modeling
based on the simultaneous integration of geophysical, geological and engineering data to improve reservoir
characterization.These integrated data so called fracture drivers, which contain seismic attributes, curvature, porosity,
facies, acoustic impedance, elastic impedance and production data. This fracture driver will be ranked using fuzzy-logic
tool. After having rank and eliminating the less influential driver, the effect of each fracture driver on the fracturing was
analysed. The ranked drivers were used to establish the complex, non-linear relationship relating the fracture intensity to
these drivers. This process is performed by using neural-network algorithm.Our experiments show that this approach
succeed in distributing the fracture frequency, which is associated with permeability. Finally, the predicted permeability
can be useful for reservoir simulation and helps us in developing carbonate reservoir.

INTRODUCTION
Field A is located in the northern part of the West Java basin. One of the main formations in this field is Baturaja
formation which is dominated by carbonate rocks. Baturaja formation becomes a major reservoir for gas producer in
this field which has potential number of gas reserves. One of the obstacles for the development of gas reservoirs in
the carbonate rocks is the presence of fractures, since it can lead the water mobilizes easily into the wellbore, so it is
very important to know the detail of the fracture distribution. In recent exploration technology, there are several
methods for fracture identification. One of those methods is Azimuthal seismic anisotropy using a deterministic
approach with some of the approaches is not appropriate with geological conditions [1]. Moreover, this method
requires wide-azimuth seismic data, which is expensive. Another fairly new method in fracture identification is
continuous fracture modeling (CFM). This method integrates geologic, geophysical, geomechanical and reservoir
data in the fracture modeling process [2, 3]. A fracture model produced in the form of 3D grid that can be directly
used in reservoir simulation. The aim of this work is to perform 3D fracture modeling. In addition, the modeled 3D
fracture provides fracture distribution, which is very important for further development of this field. In this study,
3D PSTM seismic data, 11 well logs and 2 FMI data were used to build 3D fracture model. In order to verify the 3D
fracture model, historical water production data was used.

International Symposium on Current Progress in Mathematics and Sciences 2015 (ISCPMS 2015)
AIP Conf. Proc. 1729, 020082-1–020082-4; doi: 10.1063/1.4946985
Published by AIP Publishing. 978-0-7354-1376-4/$30.00

020082-1
INTEGRATED APPROACH FOR FRACTURE MODELING
The fracture modeling was performed by following sequential steps: (a) fracture intensity interpretation, (b)
fracture drivers classification and (c) training-validation of fracture model. The first step is doing interpretation to
FMI data to identify fracture intensity. Fracture intensity is quantity of the fractures presents around the well. In this
case, 2 FMI data from well A and well B which are located away each other were used. This means, this data is
representative enough for fracture distribution of the field.
Interpretation was focused on the depth range of 1900-2100 m, which was suspected as target zone of carbonate
reservoir. From FMI data interpretation results, two parameters of fracture, dip azimuth and dip angle were obtained.
Then they were counted as fracture intensity based on the window range with interval of 5 m. Figure 1 shows the
fracture interpretation result of well A, both dip azimuth and dip angle are in the first column, the second column
shows the gamma ray which indicates the lithology, and the third column shows the fracture density. In addition,
this calculated fracture intensity was then distributed to the 3D structural framework from 3D seismic data to
produce 3D fracture model.
The second step is fracture driver classification. This step is intended to find the most correlated fracture drivers
against the fracture intensity. Fracture drivers correlation process performed by a non-linear correlation method of
overall existing fracture drivers to fracture intensity, which had been up-scaled. It is based on the concept that there
is a complex relationship between the fracture drivers with fracture intensity. The correlation value of the entire
fracture drivers is shown in Table 1. Eight fracture drivers are selected as the most influential drivers in the case of
Field A. Those drivers are porosity, shale volume, maximum curvature, fault distance, acoustic impedance, variance,
ant-track, and iso-frequency.
The third step is training and validation, which was performed by Neural Network algorithm. This was done in
order to find close relationship between fracture intensity, which was extracted from FMI data and Fracture drivers,
which derived from seismic (seismic attributes) data [4]. In this Neural Network algorithm, fracture intensity was
used as a target for introducing the weight function to predict the intensity fracture.
In this study, 50% of the data was used, which was randomly selected. The selected data was then used as a
training data, while the remaining 50% was used as the validation process. The relation between predicted fracture
intensity and targeted fracture intensity is illustrated in Fig. 2(a), while its validation process is shown in Fig. 2(b). It
could be understood that predicted fracture intensity shows a good match with correlation coefficient around 0.75.
The result of fracture modeling is illustrated in Fig. 3(a). This figure was extracted based on the horizon slice of
top Baturaja formation, which was obtained from 10 times realization. The colors in Fig. 3(a) represent the high
(light color) to low (dark color) fracture intensity. In order to avoid the not-prospect area, which is associated with
higher fracture intensity and higher water saturation, Fig. 3(b) was proposed. Roughly speaking, it can be concluded
that the eastern part is more prospect than western part.

RESULTS AND DISCUSSION


The fracture modeling in the carbonate reservoir of Baturaja formation has been successfully conducted. All
sequence steps has been carried out properly starting from FMI interpretation, selection of fracture driver and
prediction of fracture intensity model by using Neural Network algorithm. In general, the resulted fracture model
(Fig. 3(a)) shows that mostly high fracture intensity can be identified in the area where is close to fault zone. This
indicates that the fault becomes a dominant factor in the process of fracture formation in Baturaja formations.
Furthermore, it can be concluded that the attributes of the structure has a dominant portion in the deployment
process of fracture, which is indicated by higher coefficient correlation of the structural drivers.
Our prediction on the fracture model shows a good match, which is verified by well production data. It can be
seen on the map of the fracture intensity to be overlaid by water saturation property (Fig. 3(b)). This illustration aims
to give an idea where is the prospect and the not-prospect area in term of the presence of fracture. The presence of
fracture in the area occupied by fluid (gas zone) is identified as prospect area. This area should be concerned to
maintain the production well since this situation could accelerate the water breakthrough. It could be observed from
Fig. 3(b) that there is high fracture intensity in well A-05, but it is adjacent to the water contact. It causes a rapid
water breakthrough and proven by performance of well that rapidly rising water production. This situation is shown
in Fig. 4.

020082-2
TABLE 1. Correlation value between all
fracture drivers and intensity fracture
No. Driver Correlation
1 Mean curvature 0.770
2 Max curvature 0.756
3 Distance 0.749
4 Gaussian curvature 0.732
5 Min curvature 0.725
6 Azimuth of max curvature 0.522
7 Azimuth 0.346
8 Porosity 0.330
9 Vshale 0.289
10 Acoustic impedance 0.246
11 Iso frequency 0.227
12 Most positif curvature 0.204
13 Variance 0.194
14 Ant tracking 0.164
15 Maximum curvature 0.155
16 Cosin phase 0.136
17 Seismic 0.122
18 Attenuation 0.080
19 Frequency 0.071
20 Chaos 0.050
FIGURE 1. Fracture Interpretation and Intensity Fracture
Log in well A-1010

(a) (b)

FIGURE 2. (a). Plot of fracture intensity between data and training result using 50% data, correlation 0.9925,
0.9925
(b). Plot of 50% remaining fracture intensity data in validation process
process, correlation 0.747

Clearer observation of water breakthrough cocondition is illustrated in Fig. 5. The dark grey indicates high fracture
fra
intensity with high water saturation values. The light grey represents low fracture intensity and low water saturation
values. This situation gives more possibility that primary porosity is dominant than the secondary porosity (fracture).
The medium grey indicates the area with high fracture intensity and low water saturation, the area is more
prospective. However, we must be careful since this area is in close contact blue color that indicates water content,
which has a high risk of water breakthroug
breakthrough.

020082-3
(a) (b)
FIGURE 3. (a) Cumulative probability model of fracture intensity with the cut-off 4. (b) Map of sum up
of fracture intensity overlaid with water saturation

Water Rate Production at well A

FIGURE 4. Water rate production in well A-05 FIGURE 5. Geobody extraction from crossplot probability
fracture and water saturation

CONCLUSIONS
The resulted fracture model is very useful and has a good agreement with the well production data. Intensity
fracture model is dominantly controlled by the following fracture driver; porosity, shale volume, maximum
curvature, the distance from the fault, acoustic impedance, variance, ant-track, and iso-frequency, where the
structural attributes are more dominant than the stratigraphic attributes. The good agreement of the modeled fracture
of Baturaja formation is illustrated on the overlay of physical property between water saturation and fracture
intensity as well as geobody modeling, which introduce an area that is still possible to be developed as well as areas
that are prone to the emergence of water breakthrough.

REFERENCES
1. A. Ouenes and J. L. Hartley, in Proceeding of Invited Papers, SPE Annual Technical Conference and
Exhibition, (Society of Petroleum Engineers, Texas, 2000), 1.
2. C. Jenkins, A. Ouenes, A. Zellou, and J. Wingard, AAPG Bull. 93, 1597 (2009).
3. A. Ouenes, Comput. Geosci. 26, 953 (2000).
4. A. Ouenes, A. M. Zellou, G. Robinson, D. Balogh, and U. Araktingi, in Proceeding of Invited Papers, SPE
International Petroleum Conference, (Society of Petroleum Engineers, Texas, 2004), 1.

020082-4

Vous aimerez peut-être aussi