Rock Type Classification Using Machine Learning
Winner MEA 2018 Award
Incorporated in the Emerson SeisEarth Integrated Canvas, Rock Type Classification uses a brand new algorithm to generate rock type volumes calibrated to facies logs. With a streamlined workflow, the tool is available to any interpreter directly within the interpretation platform, with no learning curve needed.
Rock Type Classification is a supervised Machine Learning solution based on a Democratic Neural Network Association (DNNA). The goal of this method is to predict from the seismic attributes, away from the wellbore, facies, specifically lithofacies, or rock types determined from logs. It employs an “ensemble” of naïve neural networks running in parallel that simultaneously learn from an identical training dataset, each neural network with a different activation function. This architecture tends to minimize the possibility of biasing. It also includes a secondary training stage where new seismic data, away from the well bore, is introduced and “voted” on for training set inclusion, to stabilize network training while preventing overlearning.
The outcome of this process is a reliable and realistic description of the reservoir, for both conventional and unconventional plays. It predicts the most likely facies distribution and the probability relative to each facies. By addressing many of the limitations of traditional seismic inversion and classification methods for building a spatial rock type or facies model, DNNA provides a practical approach to “invert” directly for the desired model facies resolution and heterogeneity, including fluid overprint. The different studies where this technology has been used demonstrate that the results obtained show higher resolution than conventional approaches.
One of the most significant activities performed by exploration and development geoscientists is the creation of a geologic facies model. Spatial determination of the lateral and vertical facies heterogeneities of a reservoir has a direct impact on the field development plan, as an inaccurate or incomplete determination of facies distribution will lead to unrealistic reservoir behavior. This approach results in less guesswork when quantifying uncertainty in rock type distribution. Deliverables are interactively generated for in-depth analysis and are reservoir simulation ready, which is critical for reservoir geologists and engineers to better understand reservoir behavior.
Benefits of Rock Type Classification
- Enables better risk management and well planning optimization
- Brings new potential about seismic data reliability for predicting reservoir facies away from wells, especially when referring to prestack data, which carry more information with any type of seismic attributes.
- Provides faster images of the subsurface while still maintaining accuracy, thus helping to accelerate the decision-making process in the drilling location determination.
Rock Type Classification, the unique classification technology from Emerson E&P Solutions, has won Hart Energy’s 2018 Meritorious Award for Engineering Innovation (MEA) in Exploration/ Geoscience.
The sheer volume of well and seismic data that needs to be analyzed has made Machine Learning an effective approach to the transformation and analysis of subsurface data, producing outputs in minutes or hours rather than months or years. Since Machine Learning integrates data of different resolutions (core, wireline and seismic) and different domains, it enables geologists, reservoir engineers, and geophysicists to work together to ensure that disparate data is calibrated, and results validated. It is ideal for Cloud implementation.