The efficient d needed to produ carried out ba ndependent bo other considera and lay down tr Recent advance asset model wh he flow of oil simulator exten hroughout the results of simu ndividual field phasing and pla This case study of 18 stacked, i facilities. The i and constraints also discussed w ntroduction The developme bottom-up Fiel ntegrate variou 1. A larg 2. Surfac 3. Signif 4. Difficu

This paper presents the development of a tightly-coupled Integrated Asset Model (IAM) to capture the surface-subsurface interactions of 5 gas condensate reservoirs producing through a common surface facilities network. The objective of the exercise is to develop a tool that is built to combine existing compositional simulation models and the surface network model in a single platform/environment that could be used in production optimization, de-bottlenecking, field development and flow assurance. The success criteria included providing a solution that can ensure: Adequate representation of the different fluids of the 5 reservoisFull adherence to existing network constraints and field development guidelinesProviding optimal network configuration based on the use of automatic production optimization procedures.Flexibility to add complex network elements and decision logic The most powerful and unique feature that enabled all of the above was the effective use of procedures and built-in functions in Nexus that offer the capability to incorporate operational considerations into the solution. The first step involved developing a common fluid components basis to be used for the surface and subsurface models. This was achieved through adopting a suitable lumping scheme that involved minimal adjustment to the existing equations of state (EOS) and no compromise of the fluid description or the quality of the history matches Next step involved converting all the five reservoir models and their surface network model into the same simulation environment. The final step involved developing the procedures that carried out production optimization and proposed the optimal settings to fully utilize the available compression capacity. Finally, the surface elements of the IAM were calibrated to the results of production capacity tests. The calibration involved matching the observed gas production rates, tubing head pressures and flowline pressures. This step is required to validate the constrained model performance and prediction. The results showed the magnitude of interaction between the reservoirs and clearly identified the system bottlenecks. The model can be used to propose the best tie-in location of future wells in addition to providing first-pass flow assurance indications by highlighting elements of the network at risk of erosion throughout the field's life and under different network configurations.

An elusive goal of reservoir simulation has been the ability to accurately model multiple reservoirs producing through a common surface facility. In the past, loosely coupled techniques have often been used which did not fully converge the overall solution of the simulation. This led to instabilities at worst or to inaccuracies in the solution at best because it did not properly account for the complete interaction of the reservoir. In the following paper, we discuss the application of a fully implicit, tightly coupled surfacesubsurface simulation with a next-generation reservoir simulation of the north Kuwait Raudhatain multireservoir complex with common surface facilities using actual field data. In addition to surfacesubsurface simulation, the reservoir simulator provides a parallel unstructured grid capability and significant computational efficiency from a improved model formulation. For the simulations of this study, four stacked reservoir horizons form the subsurface portion of the model. The surface facilities consist of the production gathering centers, and gas lift and water injection capabilities. A unique feature of the model included the capability to automatically switch the flow lines for the more than 500 wells among six different separator trains at a gathering center, depending on the wellhead pressures and producing water cuts. The resultant 600,000 cell multireservoir model was first validated by comparing fifty years of historical performance for the individual reservoirs from the original simulation models with the nextgeneration model. In general, the matches between the original simulations and those generated with the next generation simulator were extremely close for each of the four reservoirs. With the validation completed, a simulation surface network was constructed that attempted to capture all of the salient features of the current and future surface facilities for the field by including, for example, actual flowline lengths in the model. In particular, future facilities expansions were included with gas lift, water injection, and the multiple header switching of wells at the gathering center. The complete model, including all four reservoirs and the surface facilities, was run for a prediction period of over sixty years. The resulting predictions provided, for the first time, solutions that show the interaction of the reservoirs with the surface facilities, including reallocation of production and injection based on facilities constraints. The multireservoir model forms the basis of a field planning and optimization tool whose forecasts can be used with greater confidence because of the inclusion of the comprehensive physics of the field production. A comparison of the simulation output with a spreadsheet field planning model shows interesting results which should form the basis of future work.

No abstract

A multifaceted strategy is presented which achieved significant reservoir simulation work flow efficiency gains at an operating company. The components of the strategy, including hardware and software, are detailed, and the relative efficiency gain is quantified with respect to the previous approach. The relative business value achieved by implementing the strategy is discussed. The new strategy improved efficiency by leveraging modern technology to impact those tasks of routine reservoir simulation work flows not directly related to reservoir engineering. This involved changes in three primary areas of the classic simulation routine of pre-processing, processing, and post-processing. A turn-key high-performance computing (HPC) solution for hosting data storage and executing simulation job processing was designed and implemented. A modern state-of-the-art simulator was configured for simulation job processing. An innovative simulation project management technology solution was deployed to organise and drive the reservoir simulation work flows and associated data management. The in-place HPC solution was constrained by the configuration of the compute nodes, lack of a master node, and the storage solution. These were configured inefficiently to the point that the expected gains from fast processing capacity were not realised due to other choke points in the compute processing and network. To realise the full potential of fast processing in a modern parallel environment, a custom HPC solution was designed. The custom solution allowed for better communication between the compute nodes and the storage, and incorporated an integrated master node. Next, simulation jobs were configured to run on a modern state-of-the-art simulator optimised for advanced HPC hardware. The new simulator significantly reduced model run times. Lastly, an innovative simulation project management technology was deployed on the reservoir engineer workstations. This technology allowed the engineers to reduce the time spent managing data, and to increase the time spent engineering. The preand post-processing engineering tasks were also made more efficient with the technology. The net result of the new strategy was to increase simulation throughput eight times. The higher volume of throughput led to more optimised engineering solutions and eventually allowed for the development of more complex simulation models. The multifaceted strategy to improve reservoir simulation work flow efficiency combines optimisation of the hardware and selecting software technologies conducive to modern HPC environments to allow engineers to focus the majority of their efforts on engineering rather than on systems and data management.