3D Reservoir Model via Process-Aided Multiple-Point Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current geologic modeling technologies fail to precisely represent depositional stacking patterns while efficiently conditioning data, leading to costly inaccuracies in hydrocarbon reservoir simulations, particularly in decisions related to drilling and facility construction.
Innovation Solution
A method that generates a process-based model mimicking the depositional process of a hydrocarbon reservoir, extracts statistics and depositional rules, and uses object-based modeling to construct multiple unconditional geologic models, followed by a single multiple-point geostatistical simulation to create a three-dimensional reservoir model, incorporating well data and gross thickness information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If current geologic modeling technologies are used to represent depositional stacking patterns, then the model construction process is simplified, but the precision of representing spatial arrangement of discrete objects and flow barriers deteriorates
Solution Approach 1:
The patent segments the geologic modeling process into distinct phases: object-based modeling to define discrete depositional objects with precise geometries, followed by multiple-point geostatistical simulation to populate the 3D grid. This segmentation allows each phase to optimize for its specific goal - geometric accuracy in object definition and spatial distribution in the simulation phase - thereby achieving high precision without overwhelming complexity in any single step.
Solution Approach 2:
The patent applies preliminary action by first constructing object-based models that define the precise geometries and stacking patterns of depositional objects before performing the geostatistical simulation. The training images are pre-generated from these object models to encode the desired spatial arrangements and flow barrier distributions, ensuring that the final 3D model inherits the precise geometric relationships from the outset rather than attempting to achieve them through complex constraint-based simulation alone.
2Reliability
If current geologic modeling technologies are used to condition models to available data, then data integration is achieved, but the accuracy of model-based predictions for business decisions deteriorates
Solution Approach 1:
The patent merges two previously separate modeling approaches - object-based modeling and multiple-point geostatistical simulation - into a unified workflow. The object-based models provide the geometric framework and depositional rules, while the MPS simulation incorporates well data and gross thickness constraints. This merging allows the model to simultaneously achieve high predictive accuracy through the object-based geometric control and efficient data conditioning through the MPS framework, resolving the contradiction between reliability and productivity.
Solution Approach 2:
The patent introduces training images as an intermediary element that bridges object-based modeling and geostatistical simulation. These training images encode the spatial arrangements, stacking patterns, and flow barrier distributions from the object models, serving as a mediator that translates geometric concepts into simulation-ready templates. This intermediary allows well data and gross thickness information to be integrated efficiently while maintaining the geometric fidelity required for accurate predictions.
3Adaptability or versatility
If sequential simulation of individual objects is performed, then flexibility in conditioning individual well data is improved, but the ability to constrain gross thickness maps deteriorates
Solution Approach 1:
The patent transitions from sequential 2D object simulation to a unified 3D simulation approach. By constructing training images that represent the complete stacked package of objects in three dimensions, the method enables simultaneous conditioning to both individual well data and gross thickness maps. The 3D training images capture vertical and lateral relationships across the entire depositional sequence, allowing the single MPS simulation to honor both local well constraints and regional thickness patterns that were previously difficult to reconcile.
Data Source
AI summary
A method of simulating a hydrocarbon reservoir is disclosed. A process-based model is generated that mimics a depositional process of the reservoir. The process-based model is analyzed to extract statistics of geometries of a body that forms part of the reservoir, and depositional rules of the body. An object-based modeling method is applied to construct multiple unconditional geologic models of the body using the statistics and the depositional rules. Training images are constructed using the multiple unconditional geologic models. Well data and gross thickness data are assigned into a simulation grid. A single multiple-point geostatistical simulation is performed using the training images. A three-dimensional (3D) reservoir model is constructed using results of the multiple-point geostatistical simulation.