Image logs hold important information about the subsurface sequences and they provide information about bedding and fault/fracture spatial distribution and characteristics. They can supply insight on the rock texture, textural organization and porosity types and distribution. To reduce the subjectivity of the interpretation and cut the interpretation time we developed and tested a new semi-automatic process for image log interpretation using a new software. This process led to the development of an expert system (called I2AM) that exploits image processing algorithms and clustering techniques, to analyze and classify borehole images. This system extrapolates the maximum amount of information from the image logs by considering not only the surfaces that cut the borehole but also the textural features of the images. Once the image log are analysed the application of clustering techniques to the values extracted from the borehole images supply a consistent classification of the images and the propagation of this classification along the logged section. In this way, we can automatically extract rock properties information with two main advantages: (i) avoid the subjectivity of the interpretation, (ii) reduce the interpretation time. The final results of this process is a set of “image facies” identified along the image log obtained by a largely automated log interpretation, although some level of human interaction and correction is still necessary. We define the clustering application as semi-automatic because the interpreter can decide, based on his geological background and on the geological characteristics of the logged section, to keep the clusters/classes proposed by the system or modify the number of clusters/classes. The clustering process and the propagation of the classes along the logged section is very fast (30 seconds) allowing an interactive approach, producing several scenarios with different number of classes and/or allowing a quick update of the image log interpretation once more data/knowledge is acquired. This approach was tested on 5 wells from north Africa where a previous image log interpretation was performed. The new interpretation based on this system made 3 years later (with more data and information) produced more refined results in very short time.
The main advantages to use the integration between geology and artificial intelligence techniques to Interpret Image Logs. And Example from Algeria.
FERRARETTI, Denis;GAMBERONI, Giacomo;
2010
Abstract
Image logs hold important information about the subsurface sequences and they provide information about bedding and fault/fracture spatial distribution and characteristics. They can supply insight on the rock texture, textural organization and porosity types and distribution. To reduce the subjectivity of the interpretation and cut the interpretation time we developed and tested a new semi-automatic process for image log interpretation using a new software. This process led to the development of an expert system (called I2AM) that exploits image processing algorithms and clustering techniques, to analyze and classify borehole images. This system extrapolates the maximum amount of information from the image logs by considering not only the surfaces that cut the borehole but also the textural features of the images. Once the image log are analysed the application of clustering techniques to the values extracted from the borehole images supply a consistent classification of the images and the propagation of this classification along the logged section. In this way, we can automatically extract rock properties information with two main advantages: (i) avoid the subjectivity of the interpretation, (ii) reduce the interpretation time. The final results of this process is a set of “image facies” identified along the image log obtained by a largely automated log interpretation, although some level of human interaction and correction is still necessary. We define the clustering application as semi-automatic because the interpreter can decide, based on his geological background and on the geological characteristics of the logged section, to keep the clusters/classes proposed by the system or modify the number of clusters/classes. The clustering process and the propagation of the classes along the logged section is very fast (30 seconds) allowing an interactive approach, producing several scenarios with different number of classes and/or allowing a quick update of the image log interpretation once more data/knowledge is acquired. This approach was tested on 5 wells from north Africa where a previous image log interpretation was performed. The new interpretation based on this system made 3 years later (with more data and information) produced more refined results in very short time.I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
