Despite their significance, structural parameters are sometimes neglected in assessments of uncertainty on connected volumes and forecast production for compartmentalized reservoirs. A workflow is proposed for modelling multiple realizations of fault geometry and properties using 3D geomodelling software. Geometrical parameters that may be simulated include fault and horizon shape and location, fault displacement and fault pattern, while property variables include fault permeability, thickness and clay smears. Realizations are ranked by estimated connected volume, with selected models being exported for numerical flow simulation. Experimental design is used to assess sensitivity of forecast production and pressure to different parameters. The workflow is illustrated using a North Sea reservoir, in which structural heterogeneities cause considerable uncertainty on connected volumes, with implications for history matching and infill well planning. Fault geometry and permeability were the most important properties for all studied responses, however their relative significance could vary between early and late field life. A number of improvements are proposed, chiefly in the areas of connected volume estimation, handling of uncertain grid geometries and calculation of stress- or saturation-dependent fault permeabilities. Finally, the method can be integrated with conventional sedimentary and petrophysical uncertainties to investigate interactions and relative sensitivities with regard to structural parameters.

No abstract