Stochastic modeling of channel meanders and resultant point bars
MetadataShow full item record
The evolution of river meanders over time greatly determines the distribution of point bars along alluvial streams. These point bars are curvilinear deposits and form potential reservoirs for hydrocarbons. As part of reservoir modeling, it is necessary to reconstruct alluvial stream meanders based on which point bars may have been deposited. This will allow better modeling of connectivity patterns that have significant impact on fluid flow and transport properties within hydrocarbon reservoirs. Multiple point (mp) statistics can be applied to model connectivity patterns in a spatial domain. However, controlling the migration of channels so as to model the location of point bars consistent with the conditioning information observed in the field is difficult within the mp statistics based schemes. In contrast, we generate connected centerlines for meanders, such that they pass through given sets of channel locations deduced from well logs and simultaneously, the meanders are curved so as to yield point bar at the correct locations. These centerlines are stochastically generated, geologically realistic and made to capture applicable geomorphological observations. Based on the generated meander centerlines, several realizations of 3D geologic models are developed and further used to model fluid flow. The stochastic modeling methodology is applied to a real field data set for the Cranfield field in Mississippi. In this reservoir, there are several channel meanders and consequently, the stratigraphic layers are modeled independently. The results demonstrate that the implemented algorithm is a better approach to modeling connectivity patterns in alluvial systems where flow pattern identification is often challenging.