Predicting Shear Wave Velocity Using Petrophysical Logs and Deep Learning Algorithms in One of the Hydrocarbon Fields in Iran

Authors

1 School of Mining Engineering, College of Engineering, University of Tehran, after jalal Al Ahmad St., North Kargar St., Tehran, Iran

2 School Mining Engineering, College of Engineering, University of Tehran, after jalal Al Ahmad St., North Kargar St., Tehran, Iran

10.22055/aag.2024.45698.2426

Abstract

Abstract
Shear wave velocity is one of the most important parameters in petrophysical and geomechanical modeling. Many experimental models have been introduced to determine shear wave velocity, each of which is specific to a specific region. One of the recently used methods is intelligent methods. In this study, shear wave velocity has been predicted using deep learning algorithms in one of the hydrocarbon fields in southern Iran. In this article, Pearson's correlation coefficient was used to select the features, and in the following, multilayer perceptron neural network (MLP), recurrent neural network + multilayer perceptron neural network (LSTM+MLP), convolutional neural network + multilayer perceptron neural network (CNN+MLP) was used the shear wave velocity estimation and the error value and coefficient of determination (R2) were calculated for the training and test data. Also, in order to ensure the results of the algorithms, a part of the data was separated as blind data, and the error and coefficient of determination were calculated for the blind data, and the coefficient of determination was RMLP^2=0.7989، R(LSTM+MLP)^2=0.8984، R(CNN+MLP)^2=0.9032. The results show that the error is higher and the coefficient of determination is lower for the MLP network compared to the LSTM+MLP and CNN+MLP networks. According to the obtained results, deep learning methods can be used to predict shear wave velocity as a suitable and low-cost method.
Keywords: Shear Wave Velocity, Petrophysical Data, DSI Log, Asmari Reservoir, Deep Learning.

Keywords

Main Subjects