An Artificial Neural Network for Predicting Rate of Penetration in AL- Khasib Formation – Ahdeb Oil Field

Abstract

The main objective of this study is to develop a rate of penetration (ROP) model for Khasib formation in Ahdab oil field and determine the drilling parameters controlling the prediction of ROP values by using artificial neural network (ANN).

     An Interactive Petrophysical software was used to convert the raw dataset of transit time (LAS Readings) from parts of meter-to-meter reading with depth. The IBM SPSS statistics software version 22 was used to create an interconnection between the drilling variables and the rate of penetration, detection of outliers of input parameters, and regression modeling. While a JMP Version 11 software from SAS Institute Inc. was used for artificial neural modeling.

     The proposed artificial neural network method depends on obtaining the input data from drilling mud logging data and wireline logging data. The data then analyzes it to create an interconnection between the drilling variables and the rate of penetration.

     The proposed ANN model consists of an input layer, hidden layer and outputs layer, while it applies the tangent function (TanH) as a learning and training algorithm in the hidden layer. Finally, the predicted values of ROP are compared with the measured values. The proposed ANN model is more efficient than the multiple regression analysis in predicting ROP. The obtained coefficient of determination (R²) values using the ANN technique are 0.93 and 0.91 for training and validation sets, respectively. This study presents a new model for predicting ROP values in comparison with other conventional drilling measurements.