This work presents the application of a fit-for-purpose history match workflow to a giant and geologically complex carbonate reservoir with over 60 years of production/injection history and 600+ wells. The target was to deliver, within schedule and spec, a high-quality sizeable model (15+ million grid blocks) that honored the underlying geologic characteristics and reproduced the distinctive production mechanisms present across the different regions of the reservoir, while keeping parametrization of uncertainties at a manageable level. Practical implementation routes were applied to efficiently translate key reservoir plumbing elements and other identified subsurface uncertainties into dynamic modeling components that could be investigated over large uncertainty ranges via Assisted History Matching (AHM) tools. To manage the history match process of this vast and mature reservoir, a sophisticated and custom-tailored sector-centered modeling scheme was adopted based on a "Divide & Conquer" approach. This tactic divides the big history match problem into smaller more manageable pieces, allowing for simultaneous history match of different sectors by different engineers while having frequent reassembling of sectors into a full-field model to ensure alignment, preserve consistent reservoir behaviors, and update (flux) boundary conditions. The iterative sector-based history match scheme applied to the giant field dynamic model made it possible to achieve a good history match within the given time and IT resources available to carry out the history match. The new dynamic model respects the conceptual understanding of the reservoir behavior and honors the available subsurface and production data of approximately 80% of the individual wells within the desired history match criteria. The use of the sector modeling workflow approach in a large full field model, allowed for faster turnaround of results for history matching purposes. The applied workflow also demonstrated that achieving a good history match in the individual sectors also resulted in a good history match for the full field model, achieved in a faster way. The final model respects the conceptual understanding of reservoir behavior as well as honors the available performance data at a scale which allows not only more reliable production forecasts but also model-based pattern-level waterflood optimization and its use for well location optimization (WLO) studies. The model supports development planning and reservoir management decisions (20+ new wells drilled annually), with waterflooding aiming to increase ultimate recovery by more than 20%. The methodology allowed significant time-savings to deliver the dynamic model within a relatively short schedule (~9 months) and required quality specifications. The successful application of the custom-made history match workflow is currently being replicated in other reservoirs of similar scale and complexity in North Kuwait and could also be applied to other massive reservoirs around the world. This work also illustrates a good example of achieving excellent HM results while keeping the parametrization of uncertainties as practical as possible.

Well testing is an integral part of successful exploration and production programs. Conventional techniques to acquire reservoir surveillance data and develop oil fields are not rigorously applicable for complex gas condensate reservoirs. Once gas condensate reservoirs are put on production, reservoir models need to be history matched as early as possible. Moreover, reservoir model should periodically be updated with gradually unfolding production data. Reservoir models will not be representative for future production forecasting unless accurately measured dynamic flow rate and pressure data are integrated into geologically consistent static reservoir models. In this paper, production testing via mobile flowmeters of gas condensate fields in central onshore Oman, is presented. In these fields, an extensive number of well tests (more than 1,000) have been performed through mobile well testing units for more than a decade. Condensate to gas ratios (CGR) do vary significantly because most of the wells are producing commingled from two to three clastic reservoirs that are continuously being depleted. Acquiring accurate liquid and gas flow rate measurements was a challenge because average producing gas volume fractions (GVF) were greater than 98%. Under these conditions, felicitous fluid characterization is paramount for unerring gas and liquid flow rate measurements. Therefore, Petroleum Development Oman (PDO) has decided to characterize the fluid properties by capturing surface fluid samples while testing the wells at various operating pressures and temperatures. Most of the wells also had production logging tool (PLT) measurements performed to allocate surface productions to various reservoir units. In summary, multiphase well testing coupled with fluid sampling and analysis significantly helped to build an accurate production history that is used for updating reservoir models. Well testing and fluid analysis greatly assisted well and reservoir management efforts. Oil and gas compositions are analyzed for numerous flowing conditions. Heptanes plus (C7+) mole percentages of the reservoir fluids are used to establish CGR correlations in addition to various pressure/volume/temperature (PVT) and flow rate relations. Frequent well testing and surface fluid sampling indicated that the producing reservoir layers are depleting at different rates after more than a decade of production. CGR trends illustrated that certain wells produce more gas, leaving behind dropped-off condensate in the reservoir. This result supported the informed decision-making process to maximize gas and condensate production from specific areas in the field. In addition, well testing results helped PDO to rank the wells and analyze the production decline. In this paper, advancements in gas well testing methodologies successfully adapted for gas condensate reservoirs, will be presented in detail in conjunction with lessons learned and good practices.

Infill drilling is a proved strategy to improve hydrocarbon recovery from reservoirs to increase production and maximize field value. Infill drilling projects address the following questions: 1) Where should the wells be drilled? 2) What should be their optimum trajectories? 3) What are the realistic ranges of incremental production of the infill wells? Answering these questions is important yet challenging as it requires the evaluation of multiple scenarios which is laborious and time intensive. This study presents an integrated workflow that allows the optimization of drilling locations using an automated approach that comprises cutting-edge optimization algorithms coupled to reservoir simulation. This workflow concurrently evaluates multiple scenarios until they are narrowed down to an optimum range according to pre-set objectives and honoring pre-established well design constraints. The simultaneous nature of the workflow makes it possible to differentiate between acceleration and real incremental recovery linked to proposed locations. In addition, the technology enables the optimization of all the elements that are relevant to the selection of drilling candidates, such as location, trajectory, inclination, and perforation interval. The well location optimization workflow was applied to a real carbonate large field; heavily faulted; with a well count of +400 active wells and subject to waterflooding. Hence the need for an automated way of finding new optimal drilling locations enabling testing of many locations. Also due to the significant full field model size; sector modelling capability was used such that the optimization, i.e. running many scenarios; could be carried out across smaller scale models within a reasonable time frame. Using powerful hardware and a fully parallelized simulation engine were also important elements in allowing the efficient evaluation of ranges of possible solutions while getting deeper insights into the field and wells responses. As a result of the study, 8 out of the original 9 well locations were moved to more optimal locations. The proposed optimized locations generate an incremental oil recovery increase of more than 70% compared to the original location (pre-optimization). In addition, the project was completed within 2 weeks of equivalent computational time which is a significant acceleration compared to a manual approach of running optimization on a full field model and it is significantly more straight forward than the conventional location selection process. The novelty of the project is introduced by customized python scripts. These scripts allow to achieve practical ways for placing the well locations to explore the solution space and at the same time, honor well design constraints, such as maximum well length, maximum step-out from the surface well-pad, and well perforation interval. Such in-built flexibility combined with automation and highly advanced optimization algorithms helped to achieve the project goals much easier and faster.