BACK-PROPAGATION AND SELF-ORGANIZING NETWORKS TO FRACTURE ANALYSIS USING THE IMAGE LOGS FMI IN ORDOVICIAN-DEVONIAN RESERVOIRS, ILLIZI BASIN, ALGERIA
Résumé
The technique of imagery used in well-logs is an effective tool for the systems of in situ recording to obtain
a high-resolution image for a better interpretation of the recorded data.This technology, widely used in
Petroleum industry, provides invaluable information on stratifications, fractures, faults, textures, and
secondary porosity.The present study deals with the Formation Micro Imager (FMI) data recorded on
Ordovician and Devonian Formations of the Illizi Basin within both Takw-1 and Take-1 wells. The aim is to
highlight their stratigraphic and structural characteristics. In most Saharan reservoirs, the measurement of
fractures represents the most important factor, especially in the field of production. In fact, the increasing
need for qualitative and quantitative analyses of the imagery to describe, in undesirable way, fractures,
faults and folds may affect the interpretation and thus the structure of the reservoirs. A study carried out on
two wells of the Illizi oilfield made it possible to obtain detailed parameters of all sedimentary and
structural events identified on electrical and sonic images using FMI. The automatic interpretation of the
seismages, revealed important information on the following characteristics: structures, stratifications,
fractures, faults, textures as well as secondary porosities. They play a significant role in the description of
the reservoir. A neural network (NN) study carried out on fractures using back-propagation (BP) and autoorganized
(SO) techniques, starting from the whole FMI data allowed to obtain a classification of fractures
and prediction of their inclination values.The results obtained showed that BP-NN can be used to predict
fracture models in other wells of the study area in order to get a broader model of dynamic fracture.