Quantitative Interpretation

Predict elastic and petrophysical properties from seismic angle stacks using a new quantitative interpretation (QI) machine learning tool!

A new tool has been added to the quantitative interpretation (QI) module. The QI machine learning reservoir characterization tool uses machine learning (ML) to predict elastic and petrophysical properties (porosity, volume of shale, etc) from seismic angle stacks.

Horizon Clean-up

A new version of the horizon clean-up process is now available! Improve your structural models by removing poor-quality data.

This method automates the prediction of high and low-quality seismic horizon data in the areas near to faults, using a combination of distance and algorithm-based filters. The resulting filter attributes can be used to remove poor—and retain good quality—data, enhancing the model.

EDFM

Embedded discrete fracture modeling (EDFM) is now available for use in hydraulic induced fracture scenarios and large-scale systems in naturally fractured reservoirs!

EDFM is now available in Petrel software and the Intersect™ high-resolution reservoir simulator, to allow fractures to be represented as discrete objects during reservoir modeling and simulation. The integrated embedded discrete fracture modeling to simulation workflow is supported in the Intersect simulator from 2024.1.

Perceptual color

Perceptual color tables now available in Petrel subsurface software for improved visualization!

Perceptual interpolation occurs in a color space known as Oklab. The Oklab color space is a perceptually uniform color model that accurately represents how humans perceive color. It ensures a visually consistent and balanced transition, taking into account factors such as lightness, chroma, and hue. 

Machine learning (ML) and inversion in Petrel Quantitative Interpretation

The Petrel Rock Physics and Inversion plug-in provides a set of robust, interactive tools to explore the relationship between seismic amplitudes and reservoir properties in 3D and 4D. Leading-edge seismic AVO inversion techniques allow solving jointly for lithofacies and elastic properties. With petrophysical joint inversion, multiphysics measurement can be integrated for a better estimation of reservoir properties.

This functionality is now included as part of Petrel Quantitative Interpretation.

Machine Learning based prediction of log and reservoir properties has also been added to Petrel Quantitative Interpretation. Allowing users, a faster turn around time for reservoir characterization workflows or integrated reservoir characterization for accelerated and more informed decisions.

Machine learning for property modeling

Machine learning for property modeling combines tried-and-tested geostatistical methods with machine learning to change the paradigm of reservoir property modeling. This new process dramatically reduces the time spent on geostatistical data analysis and parameterization tasks. These time savings enable you to spend more time analyzing output property model realizations, and their associated uncertainties, to rapidly deliver reservoir models with increased efficiency and confidence to make better decisions.

Traditional geostatistical reservoir modeling is complex and time consuming. Detailed knowledge of geostatistics is generally required and the de-trending of input data, alongside stationarity assumptions, needs to be done per zone, per facies and per region. These extensive data preparation tasks are not needed in Machine Learning for Property Modeling as the geostatistical models are embedded into machine learning.

This has several beneficial implications; firstly, an unlimited number of variables can now be used to provide additional conditioning of the model. Secondly, the time normally spent on detailed parameterization can instead be spent on better understanding the geological relationships between the input and output data and gaining a more comprehensive understand of the subsurface.

Traditional property modeling efforts produce a distribution of the target property with additional effort required to gain an understanding of property uncertainty. Machine learning for property modeling produces an integrated, consistent estimation of the property distribution at every cell. This allows geomodelers to obtain robust and unbiased estimates of the property, the associated uncertainty and sweet spots. These additional outputs enable you to quickly gain a better understanding of where reservoir-grade conditions are likely to be present, and where more detailed analysis is required.

Generate Visage ensembles from Petrel

In Petrel 2021.1. you can generate ensembles of Visage finite-element geomechanics simulations from within Petrel uncertainty and optimization (U&O) process. Apply variables and vary properties to obtain a quantitative assessment of geomechanical uncertainty.

New post-processing tools in Petrel 2021.1 enable easily visualization and interpretation the output ensembles, so geoscientists can derive insights. Compatibility with the charting window enables geoscientists to present geomechanics summary findings for a selected region, individual simulations, or numerous examples in an ensemble.

The mean and standard deviation for various simulation results can be calculated and visualized in 3D using a new stress calculator tool (see picture above). The calculation enables geoscientists to determine the geomechanical uncertainty at any point in the modeled subsurface, providing a better understanding of subsurface stress and strain to enable better well placement and field development decisions

Faster depogrid simulation exports

When exporting depogrid cases for simulation, Petrel 2021.1 optimizes the data footprint and runtime. For example, the export time for a 2.26 million cell depogrid, with a 25 realization Uncertainty and Optimization (U&O) workflow where properties are modified, reduces from 17 minutes (14.7 GB data) to 4 minutes (0.8 GB data).

Data will only be exported from depogrid case files if there has been a relevant change since the last export. This optimization of the data footprint means that re-export time will be substantially reduced. A further reduction in the data footprint and export runtime is achieved by sharing common grid files, case data, and properties across U&O cases.

The combined optimization of the export data footprint offers a significant reduction in export runtime–a 75% reduction for a 2.26 million cell depogrid with a 25 realization U&O workflow where properties are modified–by eliminating the need to re-export unchanged and common data. Runtime savings will vary depending on the size and complexity of the model.

Identify critically stressed faults

Identifying severely stressed faults in the subsurface is an important component of defining future field development plans, especially for EOR, carbon capture and storage, and geothermal workflows. The new distance-to-failure algorithm, which uses Mohr-Coulomb failure criteria, enables geoscientists to make an initial assessment of the fault model to identify areas of faults that are oriented favorably for slip and therefore more likely to conduct fluids up the fault.

Distance-to-failure analysis is used to determine the likelihood of sliding on individual faces of the fault surface. Fault faces that are oriented favorably for slip are considered to be critically stressed and are more likely to conduct fluids up the fault. A detailed understanding of which faults are likely to be leakage pathways is critical not only for oil and gas operations, but also for long term CO₂ storage integrity and geothermal operations.

The new distance-to-failure operation in Petrel 2021.1 employs Mohr-Coulomb failure criteria and enables geoscientists to perform an initial assessment of the fault model so they can identify potential areas that may require further investigation.

The distance-to-failure operation, which is available for structural frameworks, depogrids, pillar grids, discrete fractures, and points sets, is located in the Structural analysis tab of each object's Settings dialog.

Automate over 90% of work steps in Petrel

The workflow editor is one of the most powerful tools available in Petrel subsurface software. Use it to automate work steps from everyday housekeeping to assessing reservoir uncertainty and share them across your organization using the new export/import functionality.

By automating laborious and time-consuming procedures in the workflow editor, you can focus on analysis and delivering value.

The Petrel 2021's workflow editor now supports:

• Conversion of triangular meshes to editable triangle meshes (and vice versa)
• Deletion of fracture sets
• New work steps to create and edit point and polygon attributes enable geoscientists to create and edit string attributes and manipulate attribute spreadsheets
• Conversion of structural framework model horizons to surfaces
• Generation of Visage finite-element geomechanics simulator ensembles

Easily manage Petrel performance

The new Petrel subsurface software health monitor tool helps users monitor and manage the performance of Petrel, and provides hints and tips to better manage performance so you can work uninterrupted.

The Petrel health monitor interface displays real-time memory consumption, dedicated Graphic Processing Unit (GPU), User, and Graphics Device Interface (GDI) handles, as well as traffic light indicators that indicate when system resources are being over-stretched. Hover over hints and tips to enable you to proactively manage performance and assist in preventing performance issues.

The new Petrel health monitor can be accessed in the Home tab on the ribbon, or via the GDI, User, RAM and GPU indicators in the bottom right of the Petrel subsurface software interface.