Events Calendar

Petroleum Engineer Seminar

Speaker: Dr. Behnam Jafarpour, Texas A & M University

Talk Title: Feature Based Reservoir Descriptions for Improved Dynamic Data Integration

Abstract: Subsurface systems pose some of the most challenging characterization and modeling problems in science and engineering with significant hydrological, environmental, and energy security implications. The main uncertainties in characterizing these systems arise from the lack of convenient access to deep geologic formations, the multiscale heterogeneity in rock physical properties, and the complex interactions between fluids and porous rocks over a wide range of temporal and spatial scales. Consequently, significant uncertainty is introduced into modeling and prediction of the related flow and transport processes, complicating the development of subsurface energy and natural resources. Calibration of prior reservoir models through integration of dynamic flow data is an important mechanism for reducing flow modeling and prediction uncertainties. In this talk, I will discuss the advantages of posing the dynamic flow data integration as a geologic feature estimation problem. The fundamental premise of the proposed methodology is that subsurface property distributions often form connected patterns (features) that exhibit strong spatial correlations. The most salient features in the description of these correlated flow properties are amenable to sparse (or compact) representations in properly designed geologic domains (i.e., geologic dictionaries), which motivates the need for a feature estimation problem formulation. In addition, flow data often have low-resolution content and do not allow for reliable reconstruction of high resolution models. A geologic feature estimation framework is also useful for reconciling model and data resolutions during data integration. By combining advanced computational and mathematical tools with physical insight from the intrinsic properties of geologic formations and fluid flow data, integration of flow data into reservoir models can be more consistently posed as a feature estimation problem. Using several numerical experiments, I will demonstrate how the proposed geologically-inspired feature estimation framework leads to a more robust (against prior uncertainty) and geologically consistent method for solving large-scale subsurface characterization inverse problems.

Host: Mork Family Dept. , Petroleum Eng. Program

Location: Ronald Tutor Hall of Engineering (RTH) - 324

Audiences: Everyone Is Invited

Contact: Takimoto Idania