Machine Learning in Oil & Gas Industry: A Novel Application of Clustering for Oilfield Advanced Process Control

Abstract

Data Analytics is an emerging area that involves using advanced statistical and machine learning algorithms to discover information & relationsips present in different types of data. The work described in this paper illustrates the application of machine learning techniques to an Oilfield Advanced Process Control (APC) project involving deployment of APC at a large onshore conventional oilfield in Saudi Aramco. APC implementation enables better control and optimization of the production from hundreds of oilwells. APC rollout at the large oilfield involved APC deployment on 300+ oil wells. Using conventional APC implementation methodology, the rollout would be very difficult to manage and would have taken about 3 man years which was not practical. Use of innovative data analytics techniques was essential to ensuring the timely deployment of such a large scale APC project. A machine learning algorithm used to cluster similarly behaving wells, enabled significant (80%) reduction in the engineering effort and operator involvement in developing the models for each well. This allowed the implementation to be completed one year in advance thus realizing the APC benefits earlier than planned.

Performance Analysis of Deep Learning Implementation in Operational Condition Forecasting of a Gas Transmission Pipeline Network

“…For example, forecasting of gas shale or wells production can be done using classical statistical methods such as Support Vector Machine (SVM), linear regression, or autoregressive (AR) [7] or the contemporary method such as neural network [8]. Even though some more advanced techniques of deep learning have developed, traditional machine learning techniques such as a nearest neighbor or random forest are still largely used, such as for beam pump dynamometer data classification [9], prediction of dead oil viscosity correlation [10], offshore structure predictive modeling [11], oilfield process control [12], drifting behavior detection [13], and crude oil contamination detection [14].…”