One of the main objectives in the Oil & Gas Industry is to constantly improve the reservoir management capabilities by using production optimization strategies that can positively impact the so-called net-present value (NPV) of a given project. In order to achieve this goal the industry is faced with the difficult task of maximizing hydrocarbon production and minimizing unwanted fluids, such as water, while sustaining or even enhancing the reservoir recovery factor by handling properly the fluids at surface facilities. A key element in this process is the understanding of the interactions between subsurface and subsurface dynamics in order to provide insightful production strategies which honor reservoir management surface facility constraints. The implementation of the ideal situation of fully coupling surface/subsurface has been hindered by the required computational efforts involved in the process. Consequently, various types of partially coupling that require less computational efforts are practically implemented. Due to importance of coupling surface and subsurface model on production optimization and taking the advantage of advancing computational performance, this research explores the concept of surface and subsurface model couplings and production optimization. The research aims at demonstrating the role of coupling of surface and subsurface model on production optimization under simple production constraint (i.e. production and injection pressure limit). The normal production prediction runs with various reservoir description (homogeneous-low permeability, homogeneous-high permeability, and heterogeneous permeability) and different fluid properties (dead-oil PVT and lived-oil PVT) were performed in order to understand the effect of coupling level, and coupling scheme with different reservoir descriptions and fluid properties on production and injection rate prediction. The result shows that for dead-oil PVT, the production rate from different coupling schemes in homogeneous and heterogeneous reservoir is less sensitive than lived-oil PVT cases. For lived-oil PVT, the production rate from different coupling schemes in homogeneous high permeability and heterogeneous permeability are more sensitive than homogeneous low permeability. The production optimization on water flooding under production and injection constraint cases is considered here also.

<P> This study investigates ways to improve the accuracy of 3D geologic models by assessing the impact of data quality, grid complexity, data quantity and distribution, interpolation algorithm and program selection on model accuracy. The first component of this research examines the impact of variable quality data on 3D model outputs and presents a new methodology to optimize the impact of high quality data, while minimizing the impact of low quality data on the model results. This 'Quality Weighted' modelling approach greatly improves model accuracy when compared with un-weighted models. </p> <p> The second component of the research assesses the variability and influence of data quantity, data distribution, algorithm selection, and program selection on the accuracy of 3D geologic models. A series of synthetic grids representing environments of varying complexity were created from which data subsets were extracted using specially developed MA TLAB scripts. The modelled data were compared back to the actual synthetic values and statistical tests were conducted to quantify the impact of each variable on the accuracy of the model predictions. The results indicate that grid complexity is the predominant control on model accuracy, more data do not necessarily produce more accurate models, and data distribution is particularly important when relatively simple environments are modelled. A major finding of this study is that in some situations, the software program selected for modelling can have a greater influence on model accuracy than the algorithm used for interpolation. When modelling spatial data there is always a high level of uncertainty, especially in subsurface environments where the unit(s) of interest are defined by data only available in select locations. The research presented in this thesis can be used to guide the selection of modelling parameters used in 3D subsurface investigations and will allow the more effective and efficient creation of accurate 3D models. </p> / Thesis / Doctor of Philosophy (PhD)

HydroGeoSphere (HGS) is a 3D control-volume finite element hydrologic model describing fully-integrated surface-subsurface water flow and solute and thermal energy transport. Because the model solves tightly-coupled highly-nonlinear partial differential equations, often applied at regional and continental scales (for example, to analyze the impact of climate change on water resources), high performance computing (HPC) is essential. The target parallelization includes the composition of the Jacobian matrix for the iterative linearization method and the sparse-matrix solver, preconditioned BiCGSTAB. The Jacobian matrix assembly is parallelized by using a static scheduling scheme with taking account into data racing conditions, which may occur during the matrix construction. The parallelization of the solver is achieved by partitioning the domain into equal-size sub-domains, with an efficient reordering scheme. The computational flow of the BiCGSTAB solver is also modified to reduce the parallelization overhead and to be suitable for parallel architectures. The parallelized model is tested on several benchmark cases that include linear and nonlinear problems involving various domain sizes and degrees of hydrologic complexity. The performance is evaluated in terms of computational robustness and efficiency, using standard scaling performance measures. Simulation profiling results indicate that the efficiency becomes higher for three situations: 1) with an increasing number of nodes/elements in the mesh because the work load per CPU decreases with increasing the number of nodes, which reduces the relative portion of parallel overhead in total computing time., 2) for increasingly nonlinear transient simulations because this makes the coefficient matrix diagonal dominance, and 3) with domains of irregular geometry that increases condition number. These characteristics are promising for the large-scale analysis of water resource problems that involve integrated surface-subsurface flow regimes. Large-scale real-world simulations illustrate the importance of node reordering, which is associated with the process of the domain partitioning. With node reordering, super-scalarable parallel speedup was obtained when compared to a serial simulation performed with natural node ordering. The results indicate that the number of iterations increases as the number of threads increases due to the increased number of elements in the off-diagonal blocks in the coefficient matrix. In terms of the privatization scheme, the parallel efficiency with privatization was higher than that with the shared scheme for most of simulations performed.

HydroGeoSphere (HGS) is a 3D control-volume finite element hydrologic model describing fully-integrated surface-subsurface water flow and solute and thermal energy transport. Because the model solves tightly-coupled highly-nonlinear partial differential equations, often applied at regional and continental scales (for example, to analyze the impact of climate change on water resources), high performance computing (HPC) is essential. The target parallelization includes the composition of the Jacobian matrix for the iterative linearization method and the sparse-matrix solver, preconditioned BiCGSTAB. The Jacobian matrix assembly is parallelized by using a static scheduling scheme with taking account into data racing conditions, which may occur during the matrix construction. The parallelization of the solver is achieved by partitioning the domain into equal-size sub-domains, with an efficient reordering scheme. The computational flow of the BiCGSTAB solver is also modified to reduce the parallelization overhead and to be suitable for parallel architectures. The parallelized model is tested on several benchmark cases that include linear and nonlinear problems involving various domain sizes and degrees of hydrologic complexity. The performance is evaluated in terms of computational robustness and efficiency, using standard scaling performance measures. Simulation profiling results indicate that the efficiency becomes higher for three situations: 1) with an increasing number of nodes/elements in the mesh because the work load per CPU decreases with increasing the number of nodes, which reduces the relative portion of parallel overhead in total computing time., 2) for increasingly nonlinear transient simulations because this makes the coefficient matrix diagonal dominance, and 3) with domains of irregular geometry that increases condition number. These characteristics are promising for the large-scale analysis of water resource problems that involve integrated surface-subsurface flow regimes. Large-scale real-world simulations illustrate the importance of node reordering, which is associated with the process of the domain partitioning. With node reordering, super-scalarable parallel speedup was obtained when compared to a serial simulation performed with natural node ordering. The results indicate that the number of iterations increases as the number of threads increases due to the increased number of elements in the off-diagonal blocks in the coefficient matrix. In terms of the privatization scheme, the parallel efficiency with privatization was higher than that with the shared scheme for most of simulations performed.