Joint Time-Lapse Full-Waveform Inversion with Time-Lag Cost Function
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional 4D Full-Waveform Inversion (FWI) methods require perfect repeatability of seismic data acquisitions and a good initial model, often leading to convergence issues and noise in time-lapse signal analysis, especially due to non-repeatability and crosstalk between baseline and monitor models.
Innovation Solution
The proposed method jointly inverts baseline and monitor datasets using a time-lag cost function with target regularization, including total variation and model difference norms, to stabilize the inversion process and enhance 4D signals in reservoirs while suppressing noise, without requiring perfect data repeatability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional 4D FWI methods are used to invert baseline and monitor datasets separately, then the inversion process can be performed with existing algorithms, but the method requires perfect data repeatability and a good initial model, leading to convergence issues and noise in time-lapse signal analysis
Solution Approach 1:
The patent combines baseline and monitor dataset inversions into a single joint inversion framework. The joint cost function integrates both datasets simultaneously, allowing the inversion to exploit correlations between them. This merging approach eliminates the requirement for perfect repeatability between separate acquisitions and reduces crosstalk between baseline and monitor models, directly resolving the contradiction between reliability requirements and measurement precision.
Solution Approach 2:
The patent introduces a time-lag cost function as an intermediary mechanism that accounts for temporal variations between baseline and monitor datasets. This time-lag formulation acts as a mediator that properly handles non-repeatability effects, allowing the inversion to proceed without requiring perfect data repeatability while maintaining accurate time-lapse signal extraction.
2Device complexity
If separate inversion of baseline and monitor datasets is performed, then computational simplicity is maintained, but crosstalk between baseline and monitor models occurs, causing convergence to suboptimal local minima
Solution Approach 1:
The joint inversion framework merges the inversion of baseline and monitor datasets into a unified optimization problem. By simultaneously inverting both datasets with a joint cost function, the method eliminates crosstalk between models and ensures consistent convergence to the same global minimum, improving reliability without excessive complexity increase.
Solution Approach 2:
The joint inversion process incorporates feedback mechanisms where information from both baseline and monitor datasets continuously informs the optimization. The gradient computations and model updates leverage correlations between datasets, providing feedback that guides convergence toward the global minimum and prevents陷入 suboptimal local minima.
3Productivity
If traditional FWI cost functions are used without time-lag formulation, then the inversion process is computationally efficient, but the method fails to account for non-repeatability, leading to noise in 4D signals
Solution Approach 1:
The patent modifies the traditional FWI cost function by incorporating time-lag parameters that account for temporal variations between baseline and monitor acquisitions. This parameter change allows the cost function to properly handle non-repeatability effects while maintaining computational efficiency through gradient-based optimization methods.
Solution Approach 2:
The patent converts the harmful effect of non-repeatability into a beneficial feature by formulating a time-lag cost function that explicitly models temporal variations. Instead of treating non-repeatability as noise to be eliminated, the method uses it as information to improve the accuracy of time-lapse signal extraction, turning a harmful factor into an advantage.
4Measurement precision
If perfect initial models are required for 4D FWI, then accurate velocity models can be obtained, but the method becomes impractical for real-world applications where good initial models are unavailable
Solution Approach 1:
The joint inversion framework merges information from both baseline and monitor datasets, allowing the inversion to compensate for deficiencies in individual initial models. By exploiting correlations between datasets, the method can achieve accurate velocity models even when initial models are not perfect, greatly improving practical applicability.
Solution Approach 2:
The patent performs preliminary joint inversion that simultaneously updates both baseline and monitor models, rather than requiring one model to be perfect beforehand. This preliminary joint action allows both models to improve together, eliminating the need for perfect initial conditions and making the method practical for real-world applications.
Data Source
AI summary
Methods and devices according to various embodiments perform full-wave inversion jointly for datasets acquired at different times over the same underground formation using a time-lag cost function with target regularization terms. This approach improves the 4D signal within reservoirs and suppresses 4D noise outside.