fajrulhq

Classification of Rock Facies from Well Logs with Comparison Supervised Machine-Learning :

An Innovative Method to Understand Complexities Behind Reservoir Characterization.

Introduction

Background

The growth of both data volume and data types in the development of a modern oilfield is stretching the capabilities of traditional manual workflows. Moreover, the complex physics behind the reservoir characterization is only partially understood through a deterministic, model-based analysis. The oil and gas industry are turning to new analysis tools to overcome these challenges through a more data driven approach considering the stochastic nature of resource plays. These tools are known as machine learning.

These methods can help to enhance the productivity of geologists, geophysicists, and petroleum engineers by automating tasks and performing most time-consuming analysis. Reservoir characterization project is working on applications of new machine learning technologies to solve a wide range of exploration and development problems. With the recent development in algorithms, computational power, and availability of enormous amounts of data, the implementation of machine learning has spurred the interest in the oil and gas industry and brought into the forefront of future energy.

Machine Learning

Machine learning is a subset of Artificial Intelligence that focuses on learning and analyzing data. Many of these approaches involve statistics, and some often refer to this field as statistical learning. Machine learning uses a variety of approaches to learn underlying rules that govern data or systems to represent the reality it models accurately. These methods allow us to make sense of large volumes of data with many variables while avoiding the biases that humans can bring to such analysis.

Study Area

The location of the study area is in the Hugoton Panoma Field, South of Kansas. There is natural gas production throughout the years since 1922 (Martin K. Dubois & Doveton (2006).

The names of all the wells are :

Shrimplin

Shankle

Luke G U

Cross H Cattle

Nolan

Recruit F9

Shrimplin

Churchman Bible

Dataset

Download CSV Notebook

Amount of Data

Result

Correlation Matrix

Accuracy Score

62%
Decision Tree
73%
Random Forest
71%
K-Nearest Neighbors
60%
Logistic Regression
64%
SVM

Reference

Abbas, I., (2017). Comparison Algorithm Kernels on Support Vector Machine (SVM) To Compare The Trend Curves with Curves Online Forex Trading. Jurnal Informatika Upgris, 2(2).
Antoine Guitton, Hua Wang, & Trainor-Guitton, Whitney. 2017. Statistical identification of faults in 3D seismic volumes using a machine learning approach. CWP.
Brownlee, Jason. 2016. Gentle Introduction to the Bias-Variance Trade-Off in Machine Learning. Machine Learning Mastery.
Cao, D. Z., Pang, S. L., & Bai, Y. H. (2005, August). Forecasting exchange rate using support vector machines. In 2005 International Conference on Machine Learning and Cybernetics (Vol. 6, pp. 3448-3452). IEEE.
Dwihusna, N. 2020. “Seismic And Well Log Based Machine Learning facies Classification in the Panoma- Hugoton Field, South Kansas.” Department of Geophysics. Colorado School of Mines
Galarnyk, Michael. 2017. PCA using Python (scikit-learn).Towards Data Science.
Hall, B. 2016. Facies classification using machine learning. The Leading Edge, 35(10), 906-909.
Srivastava, Tavish. 2018. Introduction to k-Nearest Neighbors: A powerful Machine Learning Algorithm (with implementation in Python R). Analytics Vidhya.
Shaikh, Raheel. 2015. Feature Selection Techniques in Machine Learning with Python.Towards Data Science.
Mitchell, Tom. 1997. Machine Learning. McGraw-Hill Science, Engineering, and Math.

Our Team

Ilham Diaz Rahmat N.

Geophysics UGM

M Fajrul Haqqi

Geophysics UGM

Fadjar Novel

Mechanical Engineering UGM

Gadjah Mada Oil & Gas Club

SMiLe Team