The Mapping, 3-D Modelling and Upscaling of a Modern, Mixed-Influence Delta System and the Effect on Reservoir Heterogeneity and Connectivity: The Danube Delta, Black Sea
Engineering Honours Degree, 2013
University of Adelaide
Ancient fluviodeltaic reservoirs are thought to contain a significant proportion of the world's remaining hydrocarbon reserves; however, their complex morphologies are poorly understood. Reservoir modelling is a technique commonly utilised in industry to characterise the heterogeneities contained within a reservoir. Heterogeneity information initially input into stochastic models is often susceptible to error and misrepresentation upon upscaling for dynamic simulation. The Danube Delta, located on the western coast of the Black Sea in Eastern Europe, is a modern, mixed-influence delta that has been selected as an analogue for shallow-marine reservoirs by the WAVE consortium as the subject location for this study. A number of 3-D reservoir models of the Danube Delta are presented, in order to test the effects of upscaling on reservoir heterogeneity. Models built were based on initial mapping of the analogue delta at a number of hierarchical scales, with thicknesses inferred from previously published work. The study area is fluvial- dominated, and wave influence varies laterally across the delta. Tidal influence is assumed to be negligible.
Upscaling of these models was undertaken in both lateral and vertical directions, while volumetrics and connectivity were monitored to assess the effects. Lateral upscaling resulted in minimal changes in reservoir volume, the largest of which were observed in the upscaling of finer-scaled features. Vertical upscaling resulted in much larger errors in volume, controlled by assumed thicknesses and locations of units within the modelled vertical sequence. With increasing upscaled grid cell sizes, connectivity approximation initially decreased, until a certain point, beyond which connectivity was seen to increase. It is recommended that models created on a finer hierarchical scale are used to obtain an accurate assessment of reservoir connectivity, and that these models should be laterally upscaled no further than 50m x 50m in order to retain a reasonable estimate of heterogeneity. Other factors aside from size of the grid cells should also be assessed when undertaking upscaling, such as the orientation of grid cells relative to the features being modelled. Due to the large error observed, vertical upscaling of models with thickness assumptions as coarse as those assumed in this study is not recommended.