Demonstration of grid-enabled ensemble Kalman Filter data assimilation methodology for reservoir characterization

  • Authors:

  • Ravi Vadapalli;Ping Luo;Taesung Kim;Ajitabh Kumar;Shameem Siddiqui

  • Affiliations:

  • Texas Tech University, Lubbock, TX;Texas A&M University, College Station, TX;Texas A&M University, College Station, TX;Texas A&M University, College Station, TX;Texas Tech University, Lubbock, TX

  • Venue:
  • Proceedings of the 15th ACM Mardi Gras conference: From lightweight mash-ups to lambda grids: Understanding the spectrum of distributed computing requirements, applications, tools, infrastructures, interoperability, and the incremental adoption of key capabilities
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Ensemble Kalman filter data assimilation methodology is a popular approach for hydrocarbon reservoir simulations in energy exploration. In this approach, an ensemble of geological models and production data of oil fields is used to forecast the dynamic response of oil wells. The Schlumberger ECLIPSE software is used for these simulations. Since models in the ensemble do not communicate, message-passing implementation is a good choice. Each model checks out an ECLIPSE license and therefore, parallelizability of reservoir simulations depends on the number licenses available. We have Grid-enabled the ensemble Kalman filter data assimilation methodology for the TIGRE Grid computing environment. By pooling the licenses and computing resources across the collaborating institutions using GridWay metascheduler and TIGRE environment, the computational accuracy can be increased while reducing the simulation runtime. In this paper, we provide an account of our efforts in Grid-enabling the ensemble Kalman Filter data assimilation methodology. Potential benefits of this approach, observations and lessons learned will be discussed.