| Popis: |
Dynamic grid refinement and amalgamation have the potential to reduce computational times for the simulation of advanced processes such as VAPEX (vapour extraction of oils), and other processes, when using a compositional simulator. An accurate representation of fluid flow near internal boundaries is required when attempting to evaluate the benefits of special processes such as VAPEX in the field. Such EOR processes are controlled by the transport of fluids (and energy) due to diffusive and/or dispersive flows at internal boundaries which are moving in the reservoir. These boundaries, or interface regions, lie between oil mobilized by the process and the more viscous oil which has yet to be contacted by the solvent (or heat). The interface regions are relatively thin when compared to the typical cell sizes used to model processes at a field or even pilot scale. Thus, there are obvious problems in properly representing fluid physics at interfaces in such models when properties are being averaged (up-scaled) over cells that are much larger than the interface region. There is a dynamic gridding approach that can be used to circumvent the aforementioned difficulties. Fine scale simulation grids allow adequate resolution for modelling interface calculations, and can be generated using relatively straightforward up-scaling techniques from an even finer gridded geological model. And there are regions, away from fronts, where high resolution is probably not required, and where again relatively straightforward, even more up-scaled, amalgamations would suffice. Thus, if fine scale grid cells could be used only where they were deemed to be needed, as determined by certain rules, then very efficient calculations would result. Such rules would need to isolate where there were large gradients in properties, such as near interfaces, or be able to select other areas of interest. Also, they should be able to include relatively fixed regions, such as regions near active wells. Possibilities for extending this determination to methods based on estimates for front advancement speeds could also be considered. Thus, dynamic grid refinement offers a possible solution to many accuracy and up-scaling issues in EOR simulation. A compositional simulator, GEM, a product of Computer Modelling Group Ltd. (CMG), was modified to include the dynamic gridding functionality suggested above. This simulator has already been used to simulate several special processes, including VAPEX and variations using hot solvents. The model can perform dynamic refinement in a multi-component setting, including the modelling of asphaltenes. The new dynamic gridding features are fully integrated with the adaptive implicit formulation already used in the simulator, as well as with its dispersion and diffusion modelling capabilities, and its facilities for higher order methods. The model is found to be capable of obtaining good run-time speed-ups, often upwards of two or three fold, depending upon the problem, with little or no change in the model's computed results. Thus, a compositional simulator has been developed that is capable of conducting accurate and efficient simulations through the effective use of dynamic gridding concepts. |
