Robust re-development and management of brown field depends on having an integrated reservoir model that can reliably replicate performance history, and predict future performance in form of acceptable and consistent production forecasts that can be validated with other methods. This often involves integration of all geosciences and petroleum engineering data in a relatively fine simulation model which, although captures the heterogeneities in the reservoir to some more detailed level, usually takes up so much CPU space and long model's runtime. Typically, a shorter runtime can be achieved by building a coarser static model, and/or upscaling finer static model that will lead to the loss of some of the reservoir's pecularities. Such coarser models are normally impacted by many shortcomings, most important of which is numerical dispersion that result in spurious gas cusping or water encroachment/coning issues. These problems are more critical in reservoirs with high dipping structure or in model with some random skewed or dipping cells. This paper demonstrates a robust methodology of managing these issues of long run time and not compromising on quality of result.
This case study was on a 67 ft oil-rim reservoir (20km × 8km) with huge overlying gas cap and long production history in the Niger Delta. The wells in this reservoir are characterized by some water coning and gas cusping problems.
In order to achieve shorter models runtimes while still preserving key reservoir details, a relatively coarse static model was constructed based on consistent geological concepts incorporating all relevant geosciences and Petroleum Engineering data. Preliminary attempts made towards using the coarse model to achieve acceptable history match failed, largely due to numerical dispersion issues arising from high dipping model cells around areas with slight increase in structural dip. The model was subsequently updated by "refinement" of the high dipping cells at the free water level interface only. The updated model, whilst still fast, achieved acceptable history match within the defined uncertainty envelop, which could only have been achieved by a very fine model that will take order of magnitude time to run.
Based on the model derived using this approach, more than 12 hours of model runtime was saved, the existing well behaviours were better replicated and few new drilling opportunities were identified. The use of this robust approach results in a better understanding of the subsurface and delivers new opportunities which will ultimately enable more optimal (faster decision making) management/development of the field.