Detecting Faults by Integration of Seismic Attributes Using Neural Networks in One of Iranian Oil Fields

The main object of this study is to detect faults in one of Iranian oil fields with using the seismic attributes and neural networks. The faults play an important role in creating areas of high porosity and permeability in reservoir rocks. Therefore accurate detection of fault areas has special importance in maximizing production from petroleum traps. Faults mapping are applied in detection of faults based on combination of individual seismic attributes using neural network system to create new attributes, which give the optimal view of the fault zones. Firstly, in this study, different individual attributes like energy, average frequency squared,, dip variance, similarity were calculated on the seismic section. Then the set of attributes and the representative points of the fault and non fault locations were introduced as input to artificial neural network system to train. A neural network was trained on attributes extracted at the example points set in each fault and non-fault class to generate new attributes. Finally, fault probability cube was obtained which trained neural network to apply the entire data set. High continuity across the faults and great contrast between faults and fractures with surrounding provides more accurate tracking of faults.