Joint integration of time-lapse seismic, electromagnetic and production data for reservoir monitoring and management
Abstract/Contents
- Abstract
- Joint integration of time-lapse seismic, time-lapse electromagnetic, and production data can provide a powerful means of characterizing and monitoring reservoirs. Those data contain complementary information about the changes in the reservoir during operation, and, thus, their proper integration can lead to more reliable forecasts and optimal decisions in reservoir management. This dissertation focuses on developing workflows to jointly integrate time-lapse seismic, electromagnetic, and production data because this task is significantly time-consuming and very challenging. The first part of the thesis addresses a quick and efficient method, which can provide a tool for locating the changes in the reservoir and assessing the uncertainty associated with the estimation quantitatively. The developed workflow, termed statistical integration workflow, utilizes well logs to link reservoir properties with seismic and electromagnetic data by building the joint probability distribution. A new upscaling method from well logs to the scales of seismic and electromagnetic measurements is established using multiple-point geostatistical simulation. The statistical integration workflow is applied to facies classification and the detection of depleted regions. Stochastic optimization is also investigated in this dissertation. As the joint optimization of time-lapse seismic, electromagnetic, and production data requires a huge amount of computational time, we formulate a new algorithm, the probabilistic particle swarm optimization (Pro-PSO). This algorithm is designed to alleviate the time-consuming job by parallel computations of multiple candidate models and the improvement of models based on information sharing. More importantly, any probabilistic priors, such as geological information, can be incorporated into the algorithm. Applications are investigated for a synthetic example of seismic inversion and flow history matching of a Gaussian porosity field, parameterized by its spatial principal components. The result validates the effectiveness of Pro-PSO as compared with conventional PSO. Another version of Pro-PSO for discrete parameters, called Pro-DPSO, is also developed where particles (candidate models) move in the probability mass function space instead of the parameter space. Then, Pro-DPSO is hybridized with a multiple-point geostatistical algorithm, the single normal equation algorithm (SNESIM) to preserve non-Gaussian geological features. This hybridized algorithm (Pro-DPSO-SNESIM) is evaluated on a synthetic example of seismic inversion and compared with a Markov chain Monte Carlo (MCMC) optimization method. The algorithms Pro-DPSO and Pro-DPSO-SNESIM provide not only optimized models but also optimized probability mass functions (pmf) of parameters. Therefore, it also presents the variations of realizations sampled from the optimized pmfs. Lastly, we introduce the specialization workflow of Pro-PSO algorithms for the joint integration of time-lapse seismic, time-lapse electromagnetic, and production data. In this workflow, the particles of Pro-PSO are divided into several groups, and each group is specialized in the evaluation of a particular type of data misfit. Dividing up the objective function components among different groups of particles allows the algorithm to take advantage of situations where different forward simulators for each type of data require very different computational times per iteration. The optimization is implemented by sharing multiple best models from each type of data misfit, not by sharing a single best model based on the sum (or other combinations) of all the misfits. The "divide-and-conquer" workflow is evaluated on two synthetic cases of joint integration showing that it is much more efficient than an equivalent conventional workflow minimizing an integrated objective function.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2015 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Lee, Jaehoon |
|---|---|
| Associated with | Stanford University, Department of Energy Resources Engineering. |
| Primary advisor | Mukerji, Tapan, 1965- |
| Thesis advisor | Mukerji, Tapan, 1965- |
| Thesis advisor | Mavko, Gary, 1949- |
| Thesis advisor | Tompkins, Michael (Michael John Batiste) |
| Advisor | Mavko, Gary, 1949- |
| Advisor | Tompkins, Michael (Michael John Batiste) |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Jaehoon Lee. |
|---|---|
| Note | Submitted to the Department of Energy Resources Engineering. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2015. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2015 by Jaehoon Lee
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...