'Petroleum Analytics Learning Machine' for optimizing the Internet of Things of today's digital oil field-to-refinery petroleum system

'Petroleum Analytics Learning Machine' for optimizing the Internet of Things of today's digital oil field-to-refinery petroleum system

Pipelines

Real-time systems

Oils

Hydraulic systems

Analytical models

Support vector machines

Production

Artificial Intelligence

Big Data Analytics

IoT and the Digital Oil Field

Machine Learning

Teaching Energy Industry Professionals

The Petroleum Analytics Learning Machine (PALM) is a machine-learning-based, “brutally empirical” analysis system for managing the Internet of Things (IoT) in upstream and midstream oil and gas operations. PALM was developed in the new unconventional shale oil and gas plays of America where the simultaneous analysis of hundreds of IoT attributes from hundreds of horizontal wells with thousands of hydraulic fracture stages must be analyzed in near real-time. PALM was validated in more than 3000 shale oil and gas wells with more than 10,000 hydraulic fracture stages in the Permian Basin, TX, and the Marcellus Basin, PA. PALM comprises Machine Analytics Applications (Apps) that are big-data-centric, using computational machine learning, predictive, and prescriptive analysis techniques to maximize production of natural gas and hydrocarbon liquids while minimizing costs of operations. The PALM predictive and prescriptive technologies utilize Support Vector Machine learning, signatures, and real-time Random Forest and decision trees to steer hydraulic fractures to become high instead of low oil and gas producers as completions of horizontal shale wells progress. PALM also uses Support Vector Regression, logistic regression, Bayesian models, nearest neighbors, neural networks and deep learning networks, uniquely combined as ensemble learning tools, to weigh the importance of hundreds to thousands of geological, geophysical, and engineering attributes, both measured in the field by the IoT and computed from theoretical analyses such as reservoir simulation models and 4D seismic monitoring of production changes over time. PALM is itself an IoT system since each of these methods are written as separate apps, which are then strung together by the operator. Utilizing all oil and gas well attributes to compute Importance Weights and predicted oil, gas, and water production allows the forecasting of accurate Estimated Ultimate Recovery (EUR) over the lifetime of each well.

2017 IEEE International Conference on Big Data (Big Data)

2017-12-11

R. N. Anderson

conferencePaper

NJ2L36KC

10.1109/BigData.2017.8258496