Nwankwor, Emeka
A unified metaheuristic and system-theoretic framework for petroleum reservoir management.
Doctor of Philosophy thesis, University of Liverpool.
PDF
NwankworEmeka-Thesis.pdf - Submitted version Access to this file is embargoed until Unspecified. After the embargo period this will be available under License Creative Commons Attribution No Derivatives. Download (4MB) |
|
PDF
NkwankworEme_March2014_16993.pdf - Author Accepted Manuscript Available under License Creative Commons Attribution No Derivatives. Download (4MB) |
Abstract
With phenomenal rise in world population as well as robust economic growth in China, India and other emerging economies; the global demand for energy continues to grow in monumental proportions. Owing to its wide end-use capabilities, petroleum is without doubt, the world’s number one energy resource. The present demand for oil and credible future forecasts – which point to the fact that the demand is expected to increase in the coming decades – make it imperative that the E&P industry must device means to improve the present low recovery factor of hydrocarbon reservoirs. Efficiently tailored model-based optimization, estimation and control techniques within the ambit of a closed-loop reservoir management framework can play a significant role in achieving this objective. In this thesis, some fundamental reservoir engineering problems such as field development planning, production scheduling and control are formulated into different optimization problems. In this regard, field development optimization identifies the well placements that best maximizes hydrocarbon recovery, while production optimization identifies reservoir well-settings that maximizes total oil recovery or asset value, and finally, the implementation of a predictive controller algorithm which computes corrected well controls that minimizes the difference between actual outputs and simulated (or optimal) reference trajectory. We employ either deterministic or metaheuristic optimization algorithms, such that the choice of algorithm is purely based on the peculiarity of the underlying optimization problem. Altogether, we present a unified metaheuristic and system-theoretic framework for petroleum reservoir management. The proposed framework is essentially a closed-loop reservoir management approach with four key elements, namely: a new metaheuristic technique for field development optimization, a gradient-based adjoint formulation for well rates control, an effective predictive control strategy for tracking the gradient-based optimal production trajectory and an efficient model-updating (or history matching) – where well production data are used to systematically recalibrate reservoir model parameters in order to minimize the mismatch between actual and simulated measurements. Central to all of these problems is the use of white-box reservoir models which are employed in the well placement optimization and production settings optimization. However, a simple data-driven black-box model which results from the linearization of an identified nonlinear model is employed in the predictive controller algorithm. The benefits and efficiency of the approach in our work is demonstrated through the maximization of the NPV of waterflooded reservoir models that are subject to production and geological uncertainty. Our procedure provides an improvement in the NPV, and importantly, the predictive control algorithm ensures that this improved NPV are attainable as nearly as possible in practice.
Item Type: | Thesis (Doctor of Philosophy) |
---|---|
Additional Information: | Date: 2014-03 (completed) |
Subjects: | ?? QA75 ?? |
Divisions: | Faculty of Science and Engineering > School of Electrical Engineering, Electronics and Computer Science |
Depositing User: | Symplectic Admin |
Date Deposited: | 08 Aug 2014 10:55 |
Last Modified: | 16 Dec 2022 04:41 |
DOI: | 10.17638/00016993 |
Supervisors: |
|
URI: | https://livrepository.liverpool.ac.uk/id/eprint/16993 |