Zeitschrift für Geographie und Naturkatastrophen
Offener Zugang

Abstrakt

Prediction of Hydraulic Flow in the New Zohr Carbonate Reservoir: Eastern Mediterranean using Artificial Neural Networks

Amir Maher Sayed Lala

A new gas reservoir includes the carbonates of Upper-Cretaceous Formation in the Zohr oilfield of eastern Mediterranean Sea in Egypt. The main aim of this study is to assess the new carbonate reservoir by thin section study and estimate hydraulic flow units HFUs by smart system. This carbonate formation is now considered the most important gas reservoir in northern Egypt. In this paper five microfacies were identified based on microscope petrographic analysis. The examined rocks were formed in lagoon, shoal and open marine depositional environments. The relationships between microfacies and flow units are further evaluated in this study. The determination of such relationships has proven to be challenging due to petrographic complications arising from diagenetic processes. The correlation behind pore space percentage and permeability is important to recognize hydraulic flow in the reservoir under consideration in this study. Flow Zone Indicator (FZI) approach was applied to estimate flow zones from borehole core data. To attain the goal of this work the Artificial Neural Network (ANN) technique was implemented to predict HFUs in un-cored wells. The entries parameters compensated neutron porosity (NPHI), sonic transient time (DT), Spectral Gamma Ray (SGR), and total porosity (PHIT), and formation density (RHOB) in the entry window, five cells in the obscure window and a cell as result window were used for ANN. The rock samples and logging information from wellbores Aa and Bb were used for ANN model construction.

After validating the obtained data from well (Cc) as a model input, it was executed with well (Dd) that just well logging data were available. Correlation of the results obtained from the ANN model with real data proved the reliability of the smart technique for inferring of HFUs in un-cored intervals across the field. Accordingly, the neural network technique creates the real relationship among hydraulic zones and the logging information in un-cored well.