AttributeStudio 8.3: Integrated Reservoir Characterization System Powered by Deep Learning Toolbox
We are thrilled to announce the general release of AttributeStudio 8.3, an integrated reservoir characterization system that includes the full workflow solutions from well correlation, seismic interpretation and reservoir property prediction. AttributeStudio version 8.3 is a major milestone in its development, including the latest algorithms and workflows of machine learning for reservoir property prediction. The easy-to-use machine learning workflows in 8.3 allow geoscientists to quantitatively integrate well data and seismic attributes by a few mouse clicks, resulting a 3-D reservoir model for optimal drilling locations.
In AttributeStudio 8.3 we are utilizing Deep Learning ToolboxTM for supervised learning, unsupervised learning and non-linear regression.
The followings are solutions provided by the Deep Learning module in AttributeStudio 8.3:
- Predict missing logs from other log curves.
- Facies classification of log facies.
- Predict reservoir facies from well log facies and seismic attributes through supervised learning.
- Avoid over-training in neural networks by selecting validation and testing samples.
- Quality control neural networks through Confusion Matrix, ROC curves, and regression plots.
- Integrate drilling data, production data and near-wellbore seismic attributes for mapping “sweet-spots” in unconventional plays;
- Understand non-linear relationships between the target reservoir property and input seismic attributes, well data, and completion data to the neural networks through Partial Dependence Plots (PDP) and Individual Conditional Expectation (ICE) curves.
By working with our clients, we have developed a new feature for advanced microseismic analysis in AttributeStudio 8.3 by generating 4-D microseismic volumes within an individual fracking stage. A 4-D visualization window is specially designed in 8.3 to visualize the in-stage microseismic events together with pump pressure data, allow users to gain more insights into hydraulic fracking process.
Attribute Studio 8.3 includes a new module of Data Analysis for estimating probability distribution, cross-correlation, variogram models. The data analysis function can be applied to most data objects in AttributeStudio project, such as training data in strata-grid, well log curves, strata-grid properties, and well top properties.
We also enhanced usability of AttributeStudio by providing several utilities, such as
- Extract property grids as log curves along well trajectories
- Stack multiple strata-grids to a large grid
- Convert strata-grids to volumes
- Display 3-D tadpoles for dip and azimuth logs or well top properties
- New interface for making 2-D grids from multiple well tops or well top properties
- Import multiple seg-y files in batch.
Geomodeling is committed to providing our users with the latest technology in an easy-to-use environment, making our users solve complex reservoir characterization problems faster.
For more information about AttributeStudio 8.3 or require a free evaluation, please contact: [email protected]
 Deep Learning Toolbox is a registered trademark of MathWorks