Machine Learning for Petroleum Engineers: Maximizing Asset Value with ML

---

Disciplines: Data Science and Engineering Analytics

Course Description

This immersive 2-day course introduces participants to Machine Learning methods aimed at maximizing production, while integrating business concepts from the best-selling book “The Great Game of Business”. Through dynamic role-playing, the class simulates real world challenges, fostering innovative thinking and strategic problem-solving. Participants will explore various machine learning methods, understanding their benefits and potential challenges. 
Designed to provide a comprehensive understanding of Machine Learning and its applications in the petroleum industry, the course focuses on production and reservoir engineering. Participants will form multidisciplinary teams to represent asset teams within an operating company. These teams will tackle the challenge of adding value while keeping costs low.  Navigating limited resources and poor data quality, teams will apply machine learning concepts and devise strategies to enhance asset value.

Key Concepts:

• Data-Driven Decisions: Utilize machine learning to make informed decisions despite the challenges of poor data quality.
• Value Creation: Focus on strategies that enhance the asset's value through reserve additions and production optimization.
• Cross-Disciplinary Collaboration: Highlight the need for effective communication and collaboration among geologists, engineers, and data scientists.
• The Great Game of Business Concepts:
• Open-Book Management: Teams have access to the same set of data, and performance metrics to foster transparency and informed decision-making.
• Mini-Games: Specific challenges within the role-play scenario that target critical aspects such as production optimization, and reserve addition.
• Scoreboards: Real-time tracking of each team's performance against key metrics.

Role-Play Scenario:
There are several competing teams operating in the Golden Basin: "Petro Best", “Petro Alpha”, “Petro Omega”, etc. Each team is tasked with increasing production from their prized acreage position while managing costs effectively. The Golden Basin has complex geology, aging infrastructure, and inconsistent data quality. Teams must navigate these challenges while also collaborating with each other to overcome them. Each multidisciplinary team will learn and apply machine learning techniques to increase value from their prized acreage position.  The team that finds the most value through the application of these techniques wins!

Day 1: Fundamentals of Machine Learning and EDA 

• Welcome and course objectives.
• History and fundamental concepts of machine learning: features, training, testing, key algorithms, supervised and unsupervised learning.
• Exploratory data analysis (EDA): Familiarization with methods to identify, collect, manipulate, transform, normalize, clean, and validate data.
• Strategies to overcome data gaps: Data imputation, ensemble models, active learning.
• Algorithms and fundamental EDA libraries in Python.
• Teams are introduced to the challenge and provided with the history of the Golden Basin, geological, operations data, and a list of available resources.
• Practice: Exploratory Data Analysis (EDA) in Python and Colab.
• Group discussion: Insights from the data, key issues, and opportunities.
• Mini-game 1: Each team presents their preliminary analysis and strategy. The team that gains the most insights and opportunities from their EDA data wins!
• Wrap-up and summary of key points from Day 1.

Day 2: Machine Learning Methods and Algorithms

• Algorithms and fundamental bases of Machine Learning: classification, regression, and decision trees.
• Basic requirements for a multivariate model: e.g; homoscedasticity, linearity, multivariate normality, etc.
• Measures of uncertainty and reliability in model predictions: Confidence and prediction intervals, variance, and standard deviation.
• Metrics used to evaluate model performance.
• ‘Curse of dimensionality’, ‘model overfitting’, ‘bias’.
• Benchmarking metrics and model optimization: Cross-validation, grid search, random search, ensemble methods, Bayesian optimization.
• Machine learning in Python: Fundamental methods and libraries.
• Practice: Regression techniques for predicting well performance, optimization, and model selection.
• Mini-game 2: Showcasing the value added to the Golden Basin. Teams present their ML model(s) and production-enhancing strategy from the modeling exercise. The team with the most accretive strategy wins!
• Wrap-up and summary of key points from Day 2.

Learning Objectives:

• Machine Learning methods with emphasis on petroleum applications including production and reservoir areas.
• Exploratory data analysis to identify and overcome data gaps.
• Statistical techniques (e.g.; filtering, clustering, classification, and decision trees).
• Requirements of a multivariate model: homoscedasticity, linearity, normality, etc.
• Familiarization with model metrics, and optimization. Explaining bias, the ‘curse of dimensionality’ and ‘model overfitting’.

Learning Level

Intermediate

Course Length

2-days

Why Attend

This course is your gateway to mastering Machine Learning methods while integrating business concepts from the best-selling book, “The Great Game of Business.” Here’s why you won’t want to miss out:

Real-World Challenges: Through dynamic role-playing, you’ll tackle complex scenarios that mirror the challenges faced by asset teams in operating companies. From limited resources to poor data quality, you’ll learn to navigate obstacles and devise innovative strategies.
Maximize Production: Discover how Machine Learning can supercharge production optimization. Explore statistical techniques, multivariate models, and model metrics. Understand bias, the ‘curse of dimensionality,’ and model overfitting.
Cross-Disciplinary Collaboration: Effective communication and collaboration are key. Geologists, engineers, and data scientists unite! Learn to make data-driven decisions that enhance asset value.
The Great Game of Business: Dive into open-book management, mini-games, and real-time scoreboards. Your team’s performance matters—track it against key metrics!
Role-Play Scenario: Picture yourself in the Golden Basin, competing with other teams (“Petro Best,” “Petro Alpha,” “Petro Omega”) to increase production from prized acreage positions. Complex geology, aging infrastructure, and inconsistent data quality—can your multidisciplinary team apply machine learning techniques to win?
 

Who Attends

This is an intermediate-level course designed for technical professionals in the petroleum engineering sector, including data analysts, reservoir and operations engineers.

Special Requirements

Participants are expected to have intermediate to advanced experience in the oil industry and understand the fundamentals of reservoir and/or production engineering. A laptop computer and basic programming skills, particularly in Python, are also required.

CEUs

1.6 CEUs (Continuing Education Units) are awarded for this 2-day course.

Cancellation Policy

All cancellations must be received no later than 14 days prior to the course start date. Cancellations made after the 14-day window will not be refunded. Refunds will not be given due to no show situations.

Training sessions attached to SPE conferences and workshops follow the cancellation policies stated on the event information page. Please check that page for specific cancellation information.

SPE reserves the right to cancel or re-schedule courses at will. Notification of changes will be made as quickly as possible; please keep this in mind when arranging travel, as SPE is not responsible for any fees charged for cancelling or changing travel arrangements.

We reserve the right to substitute course instructors as necessary.

Instructor