Seismic Data Inversion With Perturbation-Based Uncertainty Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current seismic inversion techniques struggle with accurately assessing uncertainties in 4D seismic data, particularly in reservoir property changes, and fail to provide reliable quantitative estimates of elastic reservoir properties due to noise and limited consideration of uncertainties in input data and modeling.
Innovation Solution
A method involving multiple perturbations of a parameterized model and data to generate multiple sets of differences in model parameters, followed by statistical analysis to derive statistical characteristics such as mean and standard deviation, which are used to assess changes in physical properties like pore pressure and saturation, incorporating uncertainties in rock-physics models and seismic data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional seismic inversion techniques are used to process 4D seismic data, then the processing speed and simplicity are maintained, but the measurement precision and reliability of reservoir property estimates deteriorate due to noise and unaccounted uncertainties
Solution Approach 1:
The patent applies preliminary action by performing multiple perturbations of the model and data before the inversion process. Specifically, the model parameters and input data are perturbed multiple times to generate different realizations, and the inversion is performed on each realization. This preliminary perturbation step ensures that uncertainties are accounted for before the main inversion, improving the precision of reservoir property estimates without adding complexity during the core inversion process.
Solution Approach 2:
The patent applies parameter changes by systematically varying model parameters and data parameters through perturbation. The model parameters (such as elastic properties) and data parameters (seismic traces) are modified according to their respective uncertainty distributions. This allows the inversion to account for parameter uncertainties, thereby improving measurement precision while maintaining a structured approach to complexity.
2Reliability
If multiple perturbations and statistical analysis are applied to account for uncertainties, then the reliability of reservoir property estimates improves, but the processing time and computational complexity increase
Solution Approach 1:
The patent applies partial action by performing a limited number of perturbations (e.g., 10-50 realizations) rather than an exhaustive number. This provides sufficient statistical reliability for reservoir property estimates while avoiding excessive computational time. The number of perturbations is optimized to achieve the desired reliability level without unnecessary processing time expenditure.
Solution Approach 2:
The patent applies copying by generating multiple realizations (copies) of the model and data through perturbation. Each realization is a slightly modified version that accounts for uncertainties. By processing these copies and performing statistical analysis on the results, the reliability of interpretations is improved while the computational burden is distributed across manageable iterations rather than requiring a single complex exhaustive analysis.
3Measurement precision
If uncertainties in rock-physics models and seismic data are incorporated through multiple perturbations, then the accuracy of reservoir property changes is improved, but the device complexity and computational resources required increase
Solution Approach 1:
The patent applies segmentation by dividing the uncertainty analysis into separate, manageable components: model perturbations and data perturbations are handled independently. The model parameters are perturbed according to their uncertainty distributions, and the data are perturbed separately. This segmentation allows each component to be processed independently, reducing the overall complexity of the system while maintaining accuracy in reservoir property change estimates.
Solution Approach 2:
The patent applies parameter changes by systematically modifying model parameters and data parameters through perturbation based on their uncertainty characteristics. This allows the processing system to account for uncertainties in a structured manner, improving measurement precision while maintaining a manageable level of complexity through standardized perturbation procedures rather than requiring complex adaptive systems.
Data Source
AI summary
A method is provided of processing data representing a physical system, the method comprising: providing (P2) input data representing differences in the physical system between a first state and a second state of the physical system; and inverting (P5) the input data, or data derived therefrom, in accordance with a parameterized model (PI) of the physical system to obtain differences in the parameters of the model between the first state and the second state, with parameters of the model representing properties of the physical system; wherein the inverting step is performed (P3 to P6) for a plurality of different perturbations (P4) of the parameterized model and/or of the data to obtain a plurality of sets of differences in the parameters of the model; and wherein a statistical analysis (P7) of the plurality of sets of differences is performed to obtain statistical characteristics of the differences in the parameters of the model.