Developments August 2021 – February 2022
This page summarizes the main improvements included in ResX versions 2021.10.0 to 2022.2.0. Our users can access more detailed information on our Support web resources
Algorithm memory use and performance
- The ResX algorithms have been adapted to support larger models with a smaller memory footprint, making it easier to work with large datasets regardless of the size of the grids, the number of wells, or the amount of production data being matched.
This has involved a significant refactoring of the Kalman update algorithm, resulting in a reduction of the memory footprint by several orders of magnitude.
Structural uncertainty user experience
ResX now provides the ability to include residual surfaces directly as model uncertainties in the Ensemble based simulation study process, greatly reducing the complexity of both the initial ensemble and the history matching workflows. It also makes it easier to recreate specific cases if required.
There is a new capability to aggregate surfaces in the Aggregate model properties process. Users can then make statistical maps and also cross sections (for structural surfaces) both for QC and to understand the updates that ResX makes when history matching the production data.
Setup of the Ensemble based simulation study
Improved ease of use in the setup of the Ensemble based simulation study process:
- Advanced filtering has been added in both the Objective function and the Localization set-up tabs.
- Enables the user to focus on specific sets of wells which form a logical group. For example, “all producer wells” or “all wells completed in Reservoir X”.
- These groupings can be based on either a “Saved search” object from the Petrel Input tree, or on a 3D discrete property from the Petrel Models tree (e.g. Zones/Segments) where ResX will identify the wells with completions within each discrete value.
Improvements to the DST features in ResX studies
After receiving substantial feedback from many studies conducted with the DST feature, significant improvement have been made. This includes better filtering of points for matching, more robust derivative computations, and some code optimization.
ResX now runs on Petrel* 2021.2
ResX in DELFI is the version of our product that supports the DELFI* E&P Cognitive Platform
* Mark of Schlumberger