Abstract
There are many examples of dry oil production from portions of reservoirs where the local water saturation is high. When relative permeability data are available, predictions of the expected water cut are not zero, but typically 30-60%. This lack of agreement means that reservoir management using simulation models is questionable because such models can only mimic the observed reservoir production using data that bear little resemblance to measurements.
The first focus of this paper is to examine the uncertainties in the data that are used for the predictions. This then provides a numerically structured approach to adjustments that need to be made to data so that history matching of simulation models can be achieved. The relative permeabilities, rather than saturations and fluid properties, are shown to be the least certain of the relevant data.
The second focus is to explore the reasons why the relative permeability data are so uncertain. The evidence points to the fact that oil emplacement and subsequent geological history of the reservoirs have not been considered sufficiently in preparing core samples and making measurements. Greater reliance on drill-stem and early production tests is therefore crucial for deriving reservoir relative permeabilities until laboratories are able to mimic, within rock samples, oil emplacement as experienced in the reservoir.
The main source of data is from the abandoned UK North Sea reservoir (Maureen, block 16/29a).
