Stratigraphic Forward Model Calibration Using Escape Clauses
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Stratigraphic Forward Modelling (SFM) tools face challenges in calibration due to their tendency to converge to a single minimum, especially when dealing with complex and multimodal error functions, leading to inefficiencies and potential trapping in local minima, which complicates the inversion of geological data.
Innovation Solution
A method using a conditional sampling algorithm with explicit escape clauses to identify and prevent oversampling of local minima and admissible solutions, allowing the algorithm to explore other regions of the solution space by defining error thresholds and neighborhoods, thereby promoting a multi-modal inversion approach.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a conventional Neighborhood Algorithm is used for SFM calibration, then the algorithm can find optimal solutions, but it tends to converge to a single minimum and get trapped in local minima
Solution Approach 1:
The algorithm dynamically adjusts the sampling process by introducing escape clauses that activate when local minima are detected. The sampling strategy transitions from intensive local sampling to broader exploration when stagnation is detected, making the search process adaptive rather than static
Solution Approach 2:
The invention changes the parameter sampling strategy by introducing explicit escape clauses that modify the sampling distribution. When a local minimum is detected through the local minima escape clause, the algorithm changes parameters to sample from broader regions, thereby escaping the trapped state and continuing the search for admissible solutions
2Ease of operation
If manual trial and error calibration is used, then the process is simple to understand, but it is time consuming and prone to subjective bias
Solution Approach 1:
The algorithm performs self-calibration by automatically detecting local minima and triggering escape clauses without human intervention. The system serves itself by monitoring its own performance and autonomously adjusting the sampling strategy to escape stagnation and continue searching for admissible solutions
Solution Approach 2:
The invention implements feedback mechanisms where the algorithm continuously monitors the error function and sampling distribution. When feedback indicates convergence to a local minimum (through the local minima escape clause conditions), the system adjusts its behavior by changing the sampling strategy to explore other regions of the solution space
3Power
If deterministic gradient descent is used for inversion, then the approach is computationally efficient, but it converges to a single global minimum and cannot handle multimodal error functions
Solution Approach 1:
The invention adds a temporal dimension to the sampling process by implementing phases of intensive local sampling followed by escape clauses that broaden the search. This dimensional addition allows the algorithm to transition between exploitation and exploration, handling multimodal error functions while maintaining computational efficiency through the structured escape mechanism
Data Source
AI summary
A modified direct search Neighborhood Algorithm (NA), similar to genetic or evolutionary algorithms, is used for automatic calibration of Stratigraphic Forward Models (SFM) to data constraints. The modified NA comprises explicit escape clauses from local minima and admissible solutions, thereby preventing a natural tendency of conventional NAs to converge and oversample a single minimum. The density of models in parameter space is used to define local minima. Admissible solutions are defined by considering error thresholds taking into account that SFM may be imperfect and that calibration constraints may be uncertain. The modified NA algorithm obtains a plurality of geological scenarios that agree with the data, and which can be used for further analysis and prediction.