ESTIMASI PERMEABILITAS RESERVOIR DARI DATA LOG MENGGUNAKAN JARINGAN SYARAF TIRUAN PADA FORMASI MENGGALA PT CHEVRON PACIFIC INDONESIA

Abstract

A research has been conducted to apply artificial neural network in order to predict permeability of reservoir. Method of this research was analytical description. Backpropagation neural network used input layer of 4 neurons, hidden layer of 6 neurons, and output layer of 1 neuron, which was optimal architecture in this research. Result of this research showed that the biggest correlation was 0,9999 for BL#33 well and the lowest correlation was 0,9977 for BL#19 well. The rmse value of BL#19 well was 1,02%, BL#33 well was 0,21%, and was 2,42% for BL#34 well. When rmse decreased, prediction disposed approximated true value. This results indicated the solution based on backpropagation model was reasonable and feasible.