Abstract
For numerical reservoir simulation, the well model has always been a critical component that can have significant impact on the results and performance of the simulation. A new well model has been developed in a commercially available simulator to provide additional capabilities and improved robustness for advanced thermal simulation.
A Natural Variable (NV) formulation, similar to that used in the reservoir solution, has been adopted for the new well model. The NV formulation enables the well model to reuse many of the reservoir solution computations hence allows for rapidly adding support of new features as they get implemented in the reservoir solution. In addition, to model the steam injection process more accurately, we adopted full-upstream weighted mobility for injection connections, for which the amount of steam injected depends on the wellbore instead of reservoir cell condition. The NV well model also supports advanced features such as thermal multi-segment wells with loops for modeling annular flow, thermal drift flux model for counter-current flow, and dynamic coefficients for conductive heat transfer.
We present numerical results using real field data to demonstrate the new capabilities. Comparisons between the NV well model and the original Mass Variable (MV) well model, as well as between the full-upstream weighted and traditional cell voidage injection mobility are included. For cases using cell-voidage mobility and no advanced features, both well models produce similar results. On the other hand, for the cases tested using recommended full-upstream weighted mobility and advanced features such as thermal multi-segment wells with drift flux, and dynamic heat transfer, NV well model produces more stable results with superior convergence behavior. We also observed that, compared with cell-voidage mobility, full-upstream weighted mobility yields more realistic (higher) injectivity, which is critical in modeling steam injection processes.
With NV well model and its advanced features, we can obtain more efficient, accurate, and robust performance predictions for thermal recovery processes for better reservoir management of heavy oil fields. In addition, algorithmic reuse of reservoir calculations within the reservoir simulator enable easier extension of the well-model to support additional complex physics.
