Prédiction Des Paramètres Pétrophysiques À L’aide Des Méthodes De L’intelligence Artificielle, Cas Du Réservoir Tagi Champs De Chebet El-nakhla Bassin D’oued M’ya, Sud-est De L’algérie.

Résumé: The objective of this thesis focused on the estimation of the missing porosity and permeability using several techniques of artificial intelligence based on well-logging data of the heterogeneous reservoir of the Lower Triassic shaly Sandstone case of the Chebet Nekhla oil field basin of Oued Mya southeast of Algeria. The Integration of GA-MLR, ANN, SVM, was done to determine porosity from well-logging data. The results show that the ANN results give the best correlation coefficient between real porosity and simulated porosity (R2=0.9373). However, the SVM method shows the performance of this technique in comparison to other methods, according to that SVM can be recommended to predict porosity in all other wells in this oil field with high reliability. For permeability the GA-MLR technique is more reliable in comparison to the other methods, gives the best correlation coefficient between real permeability and simulated permeability (R2=0.75014). Despite, it is low value and that due to the heterogeneity of our reservoir. It does not give a high performance. The Results of this thesis confirm the efficiency of artificial intelligence tools for predicting petrophysical parameters (porosity, permeability) even they are heterogeneous and they show high performances. The used techniques allow gain of time on the one hand, and to minimize the time of laboratory measurements on the other hand. «Gain time gain money »

Publié dans la revue: