Horizontal well placement determination within a reservoir is a significant and difficult step in the reservoir development process. Determining the optimal well location is a complex problem involving many factors including geological considerations, reservoir and fluid properties, economic costs, lateral direction, and technical ability. The most thorough approach to this problem is that of an exhaustive search, in which a simulation is run for every conceivable well position in the reservoir. Although thorough and accurate, this approach is typically not used in real world applications due to the time constraints from the excessive number of simulations. This project suggests the use of a genetic algorithm applied to the horizontal well placement problem in a gas reservoir to reduce the required number of simulations. This research aims to first determine if well placement optimization is even necessary in a gas reservoir, and if so, to determine the benefit of optimization. Performance of the genetic algorithm was analyzed through five different case scenarios, one involving a vertical well and four involving horizontal wells. The genetic algorithm approach is used to evaluate the effect of well placement in heterogeneous and anisotropic reservoirs on reservoir recovery. The wells are constrained by surface gas rate and bottom-hole pressure for each case. This project's main new contribution is its application of using genetic algorithms to study the effect of well placement optimization in gas reservoirs. Two fundamental questions have been answered in this research. First, does well placement in a gas reservoir affect the reservoir performance? If so, what is an efficient method to find the optimal well location based on reservoir performance? The research provides evidence that well placement optimization is an important criterion during the reservoir development phase of a horizontal-well project in gas reservoirs, but it is less significant to vertical wells in a homogeneous reservoir. It is also shown that genetic algorithms are an extremely efficient and robust tool to find the optimal location.

Hydrocarbon use has been increasing and will continue to increase for the foreseeable future in even the most pessimistic energy scenarios. Over the past few decades, natural gas has become the major player and revenue source for many countries and multinationals. Its presence and power share will continue to grow in the world energy mix. Much of the current gas reserves are found in gas condensate reservoirs. When these reservoirs are allowed to deplete, the pressure drops below the dew point pressure and a liquid condensate will begin to form in the wellbore or near wellbore formation, possibly affecting production. A field optimization includes determining the number of wells, type (vertical, horizontal, multilateral, etc.), trajectory and location of wells. Optimum well placement has been studied extensively for oil reservoirs. However, well placement in gas condensate reservoirs has received little attention when compared to oil. In most cases involving a homogeneous gas reservoir, the optimum well location could be determined as the center of the reservoir, but when considering the complexity of a heterogeneous reservoir with initial compositional variation, the well placement dilemma does not produce such a simple result. In this research, a horizontal well placement problem is optimized by using a modified Genetic Algorithm. The algorithm presented has been modified specifically for gas condensate reservoirs. Unlike oil reservoirs, the cumulative production in gas reservoirs does not vary significantly (although the variation is not economically negligible) and there are possibly more local optimums. Therefore the possibility of finding better production scenarios in subsequent optimization steps is not much higher than the worse case scenarios, which delays finding the best production plan. The second modification is developed in order to find optimum well location in a reservoir with geological uncertainties. In this modification, for the first time, the probability of success of optimum production is defined by the user. These modifications magnify the small variations and produce a faster convergence while also giving the user the option to input the probability of success when compared to a Standard Genetic Algorithm.

Tight gas reservoirs provide almost one quarter of the current U.S. domestic gas production, with significant projected increases in the next several decades in both the U.S. and abroad. These reservoirs constitute an important play type, with opportunities for improved reservoir simulation & management, such as simulation model design, well placement. Our work develops robust and efficient strategies for improved tight gas reservoir simulation and management. Reservoir simulation models are usually acquired by upscaling the detailed 3D geologic models. Earlier studies of flow simulation have developed layer-based coarse reservoir simulation models, from the more detailed 3D geologic models. However, the layer-based approach cannot capture the essential sand and flow. We introduce and utilize the diffusive time of flight to understand the pressure continuity within the fluvial sands, and develop novel adaptive reservoir simulation grids to preserve the continuity of the reservoir sands. Combined with the high resolution transmissibility based upscaling of flow properties, and well index based upscaling of the well connections, we can build accurate simulation models with at least one order magnitude simulation speed up, but the predicted recoveries are almost indistinguishable from those of the geologic models. General practice of well placement usually requires reservoir simulation to predict the dynamic reservoir response. Numerous well placement scenarios require many reservoir simulation runs, which may have significant CPU demands. We propose a novel simulation-free screening approach to generate a quality map, based on a combination of static and dynamic reservoir properties. The geologic uncertainty is taken into consideration through an uncertainty map form the spatial connectivity analysis and variograms. Combining the quality map and uncertainty map, good infill well locations and drilling sequence can be determined for improved reservoir management. We apply this workflow to design the infill well drilling sequence and explore the impact of subsurface also, for a large-scale tight gas reservoir. Also, we evaluated an improved pressure approximation method, through the comparison with the leading order high frequency term of the asymptotic solution. The proposed pressure solution can better predict the heterogeneous reservoir depletion behavior, thus provide good opportunities for tight gas reservoir management.

Reservoir management, sometimes referred to as asset management in the context of petroleum reservoirs, has become recognized as an important facet of petroleum reservoir development and production operations. In the first stage of planning field development, the simulation model is calibrated to dynamic data (history matching). One of the aims of the research is to extend the streamline based generalized travel time inversion method for full field models with multimillion cells through the use of grid coarsening. This makes the streamline based inversion suitable for high resolution simulation models with decades long production history and numerous wells by significantly reducing the computational effort. In addition, a novel workflow is proposed to integrate well bottom-hole pressure data during model calibration and the approach is illustrated via application to the CO2 sequestration. In the second stage, field development strategies are optimized. The strategies are primarily focused on rate optimization followed by infill well drilling. A method is proposed to modify the streamline-based rate optimization approach which previously focused on maximizing sweep efficiency by equalizing arrival time of the waterfront to producers, to account for accelerated production for improving the net present value (NPV). Optimum compromise between maximizing sweep efficiency and maximizing NPV can be selected based on a 'trade-off curve.' The proposed method is demonstrated on field scale application considering geological uncertainty. Finally, a novel method for well placement optimization is proposed that relies on streamlines and time of flight to first locate the potential regions of poorly swept and drained oil. Specifically, the proposed approach utilizes a dynamic measure based on the total streamline time of flight combined with static and dynamic parameters to identify "Sweet-Spots" for infill drilling. The "Sweet-Spots" can be either used directly as potential well-placement locations or as starting points during application of a formal optimization technique. The main advantage of the proposed method is its computational efficiency in calculating dynamic measure map. The complete workflow was also demonstrated on a multimillion cell reservoir model of a mature carbonate field with notable success. The infill locations based on dynamic measure map have been verified by subsequent drilling.

This thesis proposes a framework for determining the optimum location of an injection well by using an inference method, Artificial Neural Networks and a search algorithm to create a search space and locate the global maxima. Theoretical foundation of the proposed framework is followed by description of the field for case study. A complex carbonate reservoir, having a recorded geothermal production history is used to evaluate the proposed framework ( Kizildere Geothermal field, Turkey). In the proposed framework, neural networks are used as a tool to replicate the behavior of commercial simulators, by capturing the response of the field given a limited number of parameters (Temperature, pressure, injection location and injection flow rate) as variables. A study on different network designs is followed by introduction of a search algorithm to generate decision surfaces. Results indicate that a combination of neural networks and an optimization algorithm (explicit search with variable stepping) to capture local maxima can be used to locate a region or a location for optimum well placement. Results also indicate shortcomings and possible pitfalls associated with the approach. With the provided flexibility of the proposed workflow, it is possible to incorporate various parameters including injection flow rate, temperature and location. For the field of study (Kizildere), optimum injection well location is found to be in the south&amp / #8209 / eastern part of the field. Specific locations resulting from the workflow indicated a consistent search space, having higher values in that particular region. When studied with fixed flow rates (2500 and 4911 m 3 /day), search run through the whole field located two locations which are in the very same region / thus resulting with consistent predictions. Further study carried on by incorporating effect of different flow rates indicates that the algorithm can be run in a particular region of interest (south&amp / #8209 / east in the case of study) and different flow rates may yield different locations. This analysis resulted with a new location in the same region and an optimum injection rate of 4000 m 3 /day). It is observed that use of neural network as a proxy to numerical simulator is viable for narrowing down or locating the area of interest for optimum well placement.

La quantité d’hydrocarbures récupérés peut être considérablement augmentée si un placement optimal des puits non conventionnels à forer, peut être trouvé. Pour cela, l’utilisation d’algorithmes d’optimisation, où la fonction objectif est évaluée en utilisant un simulateur de réservoir, est nécessaire. Par ailleurs, pour des réservoirs avec une géologie complexe avec des hétérogénéités élevées, le problème d’optimisation nécessite des algorithmes capables de faire face à la non-régularité de la fonction objectif. L’objectif de cette thèse est de développer une méthodologie efficace pour déterminer l’emplacement optimal des puits et leurs trajectoires, qui offre la valeur liquidative maximale en utilisant un nombre techniquement abordable de simulations de réservoir.Dans cette thèse, nous montrons une application réussie de l’algorithme “Covariance Matrix Adaptation - Evolution Strategy” (CMA-ES) qui est reconnu comme l’un des plus puissants optimiseurs sans-dérivés pour l’optimisation continue. Par ailleurs, afin de réduire le nombre de simulations de réservoir (évaluations de la fonction objectif), nous concevons deux nouveaux algorithmes. Premièrement, nous proposons une nouvelle variante de la méthode CMA-ES avec des méta-modèles, appelé le nouveau-local-méta-modèle CMA-ES (nlmm-CMA), améliorant la variante déjà existante de la méthode local-méta-modèle CMA-ES (lmm-CMA) sur la plupart des fonctions de benchmark, en particulier pour des tailles de population plus grande que celle par défaut. Ensuite, nous proposons d’exploiter la séparabilité partielle de la fonction objectif durant le processus d’optimisation afin de définir un nouvel algorithme appelé la partiellement séparable local-méta-modèle CMAES (p-sep lmm-CMA), conduisant à une réduction importante en nombre d’évaluations par rapport à la méthode CMA-ES standard.Dans cette thèse, nous appliquons également les algorithmes développés (nlmm-CMA et p-sep lmm-CMA) sur le problème de placement des puits pour montrer, à travers plusieurs exemples, une réduction significative du nombre de simulations de réservoir nécessaire pour trouver la configuration optimale des puits. Les approches proposées sont révélées prometteuses en considérant un budget restreint de simulations de réservoir, qui est le contexte imposé dans la pratique.Enfin, nous proposons une nouvelle approche pour gérer l’incertitude géologique pour le problème d’optimisation de placement des puits. L’approche proposée utilise seulement une réalisation, ainsi que le voisinage de chaque configuration, afin d’estimer sa fonction objectif au lieu d’utiliser multiples réalisations. L’approche est illustrée sur un cas de réservoir de benchmark, et se révèle être en mesure de capturer l’incertitude géologique en utilisant un nombre réduit de simulations de réservoir. / The amount of hydrocarbon recovered can be considerably increased by finding optimal placement of non-conventional wells. For that purpose, the use of optimization algorithms, where the objective function is evaluated using a reservoir simulator, is needed. Furthermore, for complex reservoir geologies with high heterogeneities, the optimization problem requires algorithms able to cope with the non-regularity of the objective function. The goal of this thesis was to develop an efficient methodology for determining optimal well locations and trajectories, that offers the maximum asset value using a technically feasible number of reservoir simulations.In this thesis, we show a successful application of the Covariance Matrix Adaptation - Evolution Strategy (CMA-ES) which is recognized as one of the most powerful derivative-free optimizers for continuous optimization. Furthermore, in order to reduce the number of reservoir simulations (objective function evaluations), we design two new algorithms. First, we propose a new variant of CMA-ES with meta-models, called the newlocal-meta-model CMA-ES (nlmm-CMA), improving over the already existing variant of the local-meta-model CMA-ES (lmm-CMA) on most benchmark functions, in particular for population sizes larger than the default one. Then, we propose to exploit the partial separability of the objective function in the optimization process to define a new algorithm called the partially separable local-meta-model CMA-ES (p-sep lmm-CMA), leading to an important speedup compared to the standard CMA-ES.In this thesis, we apply also the developed algorithms (nlmm-CMA and p-sep lmm-CMA) on the well placement problem to show, through several examples, a significant reduction of the number of reservoir simulations needed to find optimal well configurations. The proposed approaches are shown to be promising when considering a restricted budget of reservoir simulations, which is the imposed context in practice.Finally, we propose a new approach to handle geological uncertainty for the well placement optimization problem. The proposed approach uses only one realization together with the neighborhood of each well configuration in order to estimate its objective function instead of using multiple realizations. The approach is illustrated on a synthetic benchmark reservoir case, and is shown to be able to capture the geological uncertainty using a reduced number of reservoir simulations.