Research Initiation Award: Uncertainty Quantification of Multi-Phase Porous Media Flows on GPUs

Central State University, Wilberforce OH

Investigators

Abstract

Research Initiation Awards provide support for junior and mid-career faculty at Historically Black Colleges and Universities who are building new research programs or redirecting and rebuilding existing research programs. It is expected that the award helps to further the faculty member's research capability and effectiveness, improves research and teaching at his home institution, and involves undergraduate students in research experiences. The award to Central State University has potential broader impact in a number of areas. The project seeks to employ a Bayesian framework on Graphics Processing Units (GPUs) for forecasting flows of the spatial distribution of subsurface properties, such as permeability and porosity. Undergraduate students will be involved in the project and a new course in high performance computing will be developed. In oil recovery, carbon dioxide sequestration, or monitoring and remediation of aquifer contamination, it is often required to forecast quantities such as the fraction of oil in the produced fluid, carbon dioxide concentration, or concentration of contaminants, using subsurface fluid flow models with limited data. In this work, a Bayesian framework on Graphics Processing Units for forecasting flows will be employed. The main objectives of the project are to: expand the subsurface characterization in an existing Bayesian framework that employs a flow simulator on GPUs; improve experimental data fitting using the Bayesian statistical framework; and extend the improved framework to include a three-dimensional flow simulator on GPUs. To improve the characterization, a huff-puff technique will be employed, which consists of injecting a tracer for a short period of time into the subsurface during the huff phase, and monitoring the tracer during the puff phase. Additionally, an ensemble Kalman filter for integrating permeability and porosity data at different spatial scales for reconstructing fine-scale spatial distributions of permeability and porosity will be used. Through collaboration with industry, the proposed Bayesian framework can be used to detect contaminants in water aquifers equipped with sensors.