Selection of Representative Models: An Example of a Fluvial Sandstone Reservoir

The typical workflow for the creation of final static models is composed of many steps based on the inputs from surface seismic interpretation, geological well markers, 1D analysis of logs and a depositional model. Similarly, geological modeling processes such as facies and petrophysical modeling can simulate different properties based on the input data and parameters. Uncertainty is inherent in all these inputs and processes. For example, depth-converted seismic horizons use velocity models, which can impact the reservoir thickness and fault throws; facies modeling influences geometry, proportions, and spatial distribution of bodies; and petrophysical modeling influences porosity and permeability and fault analysis on the transmissibility between the fault blocks. Some of these uncertainties influence hydrocarbon volumes, some uncertainties influence the dynamic behavior, and some influence both. Using the latest tool for geological screening, we can select representative models from the uncertainty runs. Such screening supports the use of numerical experiments to calculate measures of both volume and dynamic connectivity, which, combined with input parameter analysis, serves the model selection. The selected representative models can be used for detailed reservoir simulation. ( Larue,2006 ). We tested the above method using synthetic dataset of a dome-shaped structure with fluvial sandstone reservoir having number of wells.