Oil and Gas

Participants will learn to:

1. Build confidence in core machine learning and data science concepts, tailored specifically to Exploration, Production, and other Upstream Oil & Gas domains.
2. Evaluate and select the most appropriate analytical workflows and machine learning algorithms for addressing domain-specific challenges.
3. Gain hands-on experience in implementing end-to-end solutions using modern open-source tools and Python libraries relevant to Upstream applications.

Day 1

Lectures

• Introduction to Data Science in the Oil & Gas Industry.
• Introduction to Python:
• Python basics.
• Data Wrangling (using Pandas and SQL).
• Data visualisation.

Daily Learning Outcomes

• Solve upstream challenges with Python – Confidently apply Python and leading machine learning libraries to tackle problems in reservoir engineering, production analysis, and field optimisation.
• Build scalable data workflows – Develop reusable pipelines for analysing production data from both local files and databases, applicable to small and large oil & gas fields.
• Prepare complex reservoir data efficiently – Rapidly transform production and pressure data from multi-compartment reservoirs for material balance analysis, formatted for PETEX MBAL.
• Accelerate reservoir simulation analysis – Streamline the review of large numbers of simulation results to quickly assess history match quality and forecast reliability.
• Create impactful visualisations – Produce high-quality plots of production, pressure, and well log data suitable for technical reviews, presentations, and reporting.

Case Studies and Exercises

• Production data analysis – Norne Field, The North Sea.
• Production data analysis and visualisation.
• Data preparation for material balance calculation (PETEX MBAL).
• Reservoir simulation model QC.

Day 2

Lectures

• Numerical optimisation.
• Exploratory Data Analysis.
• Uncertainty Evaluation and Decision Making.

Daily Learning Outcomes

• Apply numerical optimisation to complex tasks - Use a range of optimisation techniques to fit rate-time data and pressure gradients, improving reservoir depletion analysis and enabling consistent initialisation of reservoir simulation models.
• Upscale static models intelligently - Convert fine-grid geological models into simulation-ready coarse grids, with full control over upscaling parameters to balance dimensionality reduction and geological details preservation.
• Estimate volume-in-place - Perform probabilistic volume-in-place assessments that incorporate input uncertainty, enabling quick volumetric estimates without building a full geological model.
• Optimise water and gas injection allocation - Determine the most efficient distribution of water and gas injection volumes across injection wells to maximise oil recovery, using a minimal number of targeted simulation runs using experimental design and surrogate models.

Case Studies and Exercises

• Decline Curve Analysis (DCA).
• PVT data preparation for reservoir simulation.
• Probabilistic volume-in-place estimation.
• Static model upscaling.
• Waterflood optimization.

Day 3

Lectures

• Introduction to Machine Learning.
• Clustering methods.
• Anomaly detection methods.
• Dimensionality reduction methods.

Daily Learning Outcomes

• Apply ML concepts in practice - Confidently use machine learning terminology and translate business and technical requirements into an actionable project scope.
• Select fit-for-purpose ML techniques - Choose the most appropriate algorithms based on problem type, data availability, data quality, and performance requirements.
• Streamline reservoir model evaluation - Screen static model scenarios to simplify history matching, reduce simulation runs, and quantify the impact of geological uncertainty.
• Optimise electrofacies definition - Determine the optimal number of electrofacies for realistic property distribution in reservoir models.
• Automate production well monitoring - Develop fully automated workflows to detect abnormal behaviours in production wells, helping prevent production losses and reduce environmental risks.

Case Studies and Exercises

• Electrofacies identification based on well log data.
• Screening of static model realisations.
• Detection of abnormal behaviours in production and injection wells.

Day 4

Lectures

• Core machine learning concepts.
• Regression methods.
• Machine learning model turning.

Daily Learning Outcomes

• Design robust ML studies - Structure and execute machine learning projects with high reproducibility, clarity, and model quality.
• Apply essential ML concepts - Master train/test split, cross-validation, objective functions, bias–variance trade-off, and hyperparameter tuning through hands-on exercises.
• Predict and optimise well performance - Forecast new well productivity and optimise completion design in unconventional reservoirs without physics-based simulation.
• Develop data-driven fluid property models - Build accurate ML models to predict saturation pressure using available field data, even when full PVT experiments are lacking.
• Automate ML model hyperparameter tuning - Use modern tools to efficiently identify optimal model configurations and reduce manual experimentation time.

Case Studies and Exercises

• Production forecasting for unconventional reservoirs.
• Saturation pressure prediction.

Day 5

Lectures

• Classification methods.
• Neural networks and Deep learning.
• Advanced machine learning topics:
• Imbalanced datasets.
• ML model interpretability.

Daily Learning Outcomes

• Communicate results effectively – Explain machine learning outputs clearly to both technical and business audiences to support QA/QC and decision-making.
• Classify lithofacies from well logs – Develop robust ML-based lithofacies classification models to extend interpretation to wells without core data.

Case Studies and Exercises

• Lithofacies identification.
• Production forecasting for unconventional reservoirs.

A reservoir engineer, geologist, petrophysicist, or production engineer with 5+ years of experience who wants to build strong foundational and practical skills in scientific programming, data science, and machine learning.