Oil & Gas Training
and Competency Development
Competency Management system SLB NEXT

Applied Machine Learning for Reservoir Characterization, Petrophysics and Surveillance - Physics Inspired Principles, Applications and Workflows

Application of AI and machine learning (ML) is become a new addition to the traditional reservoir characterization, petrophysics and monitoring practice. Developed physics based data models are the key for applying ML techniques to solve complex problems. In other words, the machine should learn the right physical concepts using the available data to produce correct insights that we are looking for.

In addition quality data collection, physics-inspired predictive capability, and interpretability of the data-driven models should be considered.

Participants learn the elements of building machine learning-based models form initial data processing to predictive model development. Development of online and automated learning models will be discussed and demonstrated. Several case studies will be presented in the class and participants will practice developing a data-driven model during the class.

Development of the right ML-based model is started with applying intelligent data workflows to clean, quality control against the physics and generate the spatiotemporal database for the training, validation, prediction and optimization processes.

The quality of static data will be assessed. Any inconsistency and anomaly in dynamic data (pressure, rate etc) should be addressed before employing any data mining and machine learning algorithms.

Day 1
  • Review of digitalization concept in the energyindustry
  • Overview closed-loopreservoir surveillance and management
  • Outline the needs for employing machine learningand AI solutionsthat areinspired by physics for subsurface modeling and reservoir monitoring.
  • Learn the workflows for static and dynamic data inconsistency check with the physics.
  • Cover Intelligent Data Maid Workflows:


  • Data reformatting
  • Data cleaning
  • Static data processing
  • Time series signal processing
  • Anomaly detection algorithms
  • Data categorization
  • Data segmentation and partitioning
  • Spatio-temporal database generation
  • Case Studies


Day 2
  • Teach all required concepts related to statistical learning, data mining and cover Intelligent Data Miner Workflows:
  • Statistical analysis
  • Generalize linear models development
  • Advanced dimensionality reduction techniques and algorithms
  • Clustering algorithms
  • Advanced KPI identification
  • Case Studies


Day 3
  • Learn Intelligent Data Model workflows to build, test and validate the machine learning based models using different shallow and deep learning algorithms as well as hybrid learning concepts.
  • Learn to interpret the developed intelligent models and gain insight into advanced Intelligent Interpretable Data Models construction.
  • Uncertainty analysis and optimization for accurate techno-economic analysis.
  • Case Studies
Day 4

Case Studies

Applied Additional Hands On Workshop (for example permeability prediction)

Day 5

Participants Present their Workshop Results (

Learning activity mix

Any subsurface professional.

The participant will learn five building blocks of a physics-inspired machine learning model for petrophysics, reservoir surveillance and field management.

These elements are Intelligent Data Maid (TM) workflow, Intelligent Data Miner (TM) workflow, Intelligent Data Model (TM) workflow, Intelligent Interpretable Data Model (TM) and Intelligent Data Model Optimization (TM)

Upon completion of this course, participants will be able to

  • Inspect, quality control and generate spatiotemporal databases for intelligent reservoir surveillance model development.
  • Learn the correct workflows and processes for developing the right machine learning based models.
  • Learn the pitfalls of developing just machine learning models without considering the physics.
  • Learn various machine learning algorithms for data cleaning, time series signal processing, anomaly detection, statistical analysis, dimensionality reduction, clustering, KPI identification, and shallow and deep learning algorithms and Neuro-Fuzzy algorithm.
  • Learn to develop interpretable machine learning based models.
  • Learn how to quality control the trained models and select those that learned the right physical concepts.
  • Understand the online and automated machine learning models

Participants should have a basic understanding of statistics.

Filter upcoming classes by country

Set a training goal, and easily track your progress

Customize your own learning journey and track your progress when you start using a defined learning path.

In just few simple steps, you can customize your own learning journey in the discipline of your interest based on your immediate, intermediate and transitional goals. Once done, you can save it in NExTpert, the digital learning ecosystem, and track your progress.
© 2020 Schlumberger Limited. All rights reserved.