Reservoir compartmentalisation, whether structural or stratigraphic, is one of the most prominent parameter for accurately characterising the distribution of hydrocarbons in the subsurface and it is a key element for optimising hydrocarbon recovery. In order to accurately characterise its compartmentalisation, a new volume-based structural modelling technique have been applied for generating a geocellular model of the complex, highly faulted east flank of the studied field (Sabah, Malaysia). Benefits over existing pillar-based and surface-based techniques are discussed.
The volume-based modelling technique consists of interpolating a continuous 3D property representing the relative stratigraphic age of the formations from all available well and seismic interpretation data. A watertight structural framework composed of faults and horizon surfaces is then extracted from this property, and converted to a geocellular grid in which the faults are stair-stepped. New workflows were developed for early integration of fluid distribution and production data during the creation of the geological framework, leading to an accurate delineation of fault compartments.
The fault network of the new east flank model is composed of over 90 synthetic and antithetic faults, forming X and Y-shaped truncations and subdivided into over 35 geological units. Fault truncations are located within the volume of interest and have key bearing on the understanding of fault sealing potential. The subsequent 3D grids include all interpreted faults and integrate fluid contacts, pressure and production data, capturing essential compartmentalisation characteristics.
The main benefit of the volume-based technique over pillar-based methodologies is that it incorporates all the fault network complexity. Indeed, as their coordinate lines must be parallel to fault surfaces, grids which geometry is extruded along curvilinear pillars can generally not be used to represent fault systems with X or Y-shaped truncations. Besides, grid cells may become excessively skewed in presence of synthetic and antithetic faults, which can trigger inaccuracies when performing flow simulations.
Finally, thanks to early integration of production data during the interpretation phase the dynamic model could easily be history matched by adjusting fluid transmissibility across nearby fault blocks and between adjacent sand layers.