Abstract
SAGD process is highly affected by geological conditions such as reservoir heterogeneities caused by the existence of shale layers and/or Clast Breccia, impairment zones and so on. To evaluate SAGD performance with satisfactory precision, detailed geological models which reflect complex sedimentary heterogeneities should be constructed based on data such as well control points, 3D seismic data and Training Image. On the other hand, the relationships between complexity and simulation time conflict with each other. Although detailed geological models are needed, simulation time takes longer.
At Hangingstone Demo area, (Demo area), there are 24 well pairs which have been producing bitumen by SAGD process. The objective of this study is to obtain a unique set of parameters, such as permeabilities and extent of heterogeneities, from simultaneous history matching using different geological models (well pairs) from the Demo area. These parameters will be used for forecasting the performance of the nearby analogue Hangingstone Expansion area (HE area) development. At least a couple of well pairs coming from different well pads were chosen as representative models which indicate each typical type of reservoir. These are called Clean Sand rich type, Clast rich, and Sandy-IHS (Inclined Heterolithic Stratification) rich.
If different single models are applied to history matching, at least the same number of history matching studies are needed separately. Obtaining an optimum parameter combination will require a significant number of simulation runs. Furthermore, applying complex geological models for simulation requires significant time to finish running. Figure 1 shows the comparison between individual optimization and simultaneous optimization, which is our new workflow. By use of this, not only multiple different models but also different realizations for the same well pair are matched simultaneously by a single optimization without any time loss.
Our new workflow is composed of mainly 3 sections. The first one is vertical upscaling and coupling between the modeling package and reservoir simulator. The second one is the simultaneous history matching using multiple different models. The third one is the interpretation of the vertical observation well temperature data and history matching utilizing such data. This new technique along with field observation data will be discussed.