Dynamic Delumping of Reservoir Simulation
- Arif Kuntadi (Norwegian University of Science & Tech) | Curtis Hays Whitson (Norwegian University of Science & Tech) | Mohammad Faizul Hoda (Petrostreamz A/S)
- Document ID
- Society of Petroleum Engineers
- SPE Annual Technical Conference and Exhibition, 8-10 October, San Antonio, Texas, USA
- Publication Date
- Document Type
- Conference Paper
- 2012. Society of Petroleum Engineers
- 5.8.8 Gas-condensate reservoirs, 5.4.2 Gas Injection Methods, 4.1.5 Processing Equipment, 3.3.6 Integrated Modeling, 5.2.2 Fluid Modeling, Equations of State, 5.2.1 Phase Behavior and PVT Measurements, 5.6.5 Tracers, 4.1.2 Separation and Treating, 5.2 Reservoir Fluid Dynamics, 5.4 Enhanced Recovery, 2.2.2 Perforating, 4.6 Natural Gas, 5.5 Reservoir Simulation, 4.1.1 Process Simulation
- 1 in the last 30 days
- 582 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 5.00|
|SPE Non-Member Price:||USD 28.00|
Integrated modeling is becoming a necessary tool in the petroleum industry to manage the value chain of different models. Reservoir models commonly utilize a simple fluid model to reduce computational time. However, the downstream models often require a more detailed EOS fluid model to perform surface-process facility modeling. This paper presents a dynamic delumping method to generate detailed compositional streams from either black-oil or compositional (lumped-EOS) reservoir simulations, performed as a simple post-processing step.
A set of phase-specific, pressure-dependent split factors are used to perform dynamic delumping. The split factors are generated from simulated depletion PVT experiments using a detailed-EOS model. Delumping is performed phase-wise at the well-connection level, for each time step of the reservoir simulator. For gas injection processes, the amount of injection gas is estimated from stream information and, accordingly, removed from the stream before applying the phase-specific pressure-dependent split factors. Different split factor sets are used when the reservoir model has multiple PVT regions.
We have run many reservoir simulation cases using different production mechanisms and reservoir fluids. Compared with detailed-EOS simulations, the proposed method gives near-exact results for depletion, and excellent agreement in gas injection cases. Dynamic delumping also works with complex fluid systems exhibiting large in-situ compositional (GOR) variations. For injection gas cases, improved accuracy is obtained using a tracer option in the reservoir simulator, to better estimate injection-gas quantity. This approach requires negligible cpu compared with detailed-EOS reservoir simulation.
Dynamic delumping is applied as an automated post-processing for any reservoir simulator. The results of our work provide a key technology for integrating subsurface and surface petroleum models, ensuring greater consistency in the complete value chain and enabling engineers to optimize assets, both locally and globally.
In the last few years, integrated modeling has become a preferred tool in the petroleum industry to manage the value chain of different assets. It is slowly replacing the traditional modeling approach that treats each petroleum discipline model separately. Having different discipline models and applications in a single platform will ensure more consistency of the value chain from one discipline to another. Integrated modeling also enables engineers to optimize assets, both locally and globally, using an automatic approach. Coupling of different petroleum disciplines entails transferring and combining petroleum streams from one model to the others. Stream conversion is a key requirement in integrated modeling because petroleum disciplines usually have their own fluid model, and it is rare to have a single common fluid model in both the subsurface and surface simulation models.
Integrated modeling in reservoir and production management typically couples the reservoir, production network and process simulations. Reservoir simulation usually utilizes a simple fluid model to reduce the computational time due to the large numbers of grid cells being used in the reservoir model. The fluid model becomes more detailed as it moves downstream. Translating from a more detailed fluid description to a simpler description is usually a trivial process, but the converse does not hold true.
|File Size||3 MB||Number of Pages||22|