Enhanced Reservoir Management with Multiple Realizations

  • Author / Creator
    de Barros, George
  • A numerical reservoir model is the result of studies whose main objective is to describe the dynamic behaviour of a hydrocarbon reservoir for predicting its future performance under different development and production strategies. Reservoir models are built with uncertain parameters. The available data are widely spaced, at large scale and contain error. Thus, predictions based on these models are also uncertain. Understanding the uncertainties related to the reservoir is crucial for making development and management decisions throughout the lifetime of reservoir production.
    Conventionally, decisions in hydrocarbon field development are based on a single reference case. The reference case represents the best model with the appropriate set of input parameters for predicting future reservoir performance. However, this model is only one outcome of a large ensemble of possible models describing the reservoir. Making decisions based on one reference case disregards the geological uncertainty. The complexity of many response variables requires managing multiple realizations to improve the consistency and accuracy of reservoir models.
    A methodology to facilitate the transfer of subsurface uncertainty through reservoir management is developed and demonstrated in this thesis, reaching up to 5% of improvement in the economic performance of projects. The ensemble of realizations must be used in the calculation of measures of performance and in optimization, helping the reservoir team make decisions for maximizing the value of the reservoir project.
    The number of realizations could be large and the analysis of results is largely automated; however, human inspection is still necessary for quality control. Tools are required for processing and analyzing the ensemble of realizations qualitatively and productively. Since all realizations should be considered in reservoir management, a visualization methodology is developed to facilitate the understanding of the space of uncertainty. Methods combining the distance between realizations and animations are considered.
    The computational requirements for history matching and flow simulation is an often mentioned reason to avoid dealing with all realizations all the time. Computational performance has grown exponentially over the past 30 years through faster processors, multiple cores, parallelism, and GPUs, supporting the premise of managing multiple realizations. Moreover, developments in ensemble-based history matching techniques encourage the use of a large number of reservoir models. An alternative approach to integrating production data into the geological modeling workflow is also developed. The geological consistency is preserved through an ensemble of reservoir models conditioned to the available static and dynamic data.
    The impact of considering all realizations in the decision-making process is demonstrated with a realistic case study. The improvement of the production revenue considering all realizations is significant, reaching more than $100 million, and supports the statement of this research.

  • Subjects / Keywords
  • Graduation date
    Fall 2019
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • License
    Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.