Rule-Based Reservoir Modeling by Integration of Multiple Information Sources: Learning Time-Varying Geologic Processes

Placeholder Show Content


Rule-based modeling methodology has been developed to improve the integration of geologic information into geostatistical reservoir models. Quantifying modeling rules significantly aid in building geologically accurate reservoir models and reproduce the intrinsic complexity of subsurface conditions. Especially when we face up to a complexity where field data and geological knowledge are both limited the way we utilized rules is fairly important. To expand the application of rule-based reservoir modeling in various field cases, we propose a systematic methodology of creating rules from other information sources. Physical geomorphic experiments and process-based models contain time series information we need for a reservoir model. Incorporating these two information sources facilitate the rule induction of rule-based modeling and therefore help capture the underlying uncertainty. Two examples are demonstrated in our study. A reference class from experiments is created for turbidite lobe system, while a realization of process-based models is used to mimic and simulate channel network patterns and their behaviors on a delta plain. In our study, we assume if an experiment is comparable to field data at a certain interpretation scale, the sedimentary processes and associated structures are informative and provide at least some references resulting in sedimentary features at the comparable scale. Ripley’s K-function is utilized to analyze and extract spatial clustering information of lobe elements at a given scale from experimental strata. We converted K function to modeling rules allowing us to integrate clustering patterns of turbidite lobes into surface-based models. Surface-based models successfully produce a clustered point behavior and a stratigraphic framework comparable to the chosen physical tank experiment. These models can be used to better assess subsurface spatial uncertainty under such a stochastic process framework constrained by experimental information. To facilitate the utilization of process-based models, an automated channel feature extraction tool is developed and able to adjust parameters for the optimal result. Multi-Scale Line Tracking Algorithm is embedded and shows robust and accurate extraction of channel networks from Delft3D models. Space colonization algorithm is proposed to capture the developmental processes of channel network and reproduce a network pattern. It is able to integrate theoretical knowledge and simulate a network coupling with feature extraction tool and. The overall methodology is able to efficiently simulate channel networks and their progradation through time given information from one or more realizations of process-based models.


Type of resource text
Date created March 2015


Author Wang, Yinan
Primary advisor Mukerji, Tapan
Degree granting institution Stanford University, Department of Energy Resources Engineering


Subject School of Earth Energy & Environmental Sciences
Genre Thesis

Bibliographic information

Access conditions

Use and reproduction
User agrees that, where applicable, content will not be used to identify or to otherwise infringe the privacy or confidentiality rights of individuals. Content distributed via the Stanford Digital Repository may be subject to additional license and use restrictions applied by the depositor.

Preferred citation

Preferred Citation
Wang, Yinan. (2015). Rule-Based Reservoir Modeling by Integration of Multiple Information Sources: Learning Time-Varying Geologic Processes. Stanford Digital Repository. Available at:


Master's Theses, Doerr School of Sustainability

View other items in this collection in SearchWorks

Contact information

Also listed in

Loading usage metrics...