The key problem addressed in this paper arises in the situation where multi-scale shale drapes are present along channel, channel-belt and/or valley-bounding surfaces, but the channel location is uncertain or unknown. In order to reduce this uncertainty, the channel and drape locations should be modified in order to calibrate the flow response to the historic production data as well as any static data from wells. We propose a coupled geological modelling and history-matching method in which the reservoir architecture composed of channels is simulated with a pre-defined stacking pattern, then the shale drapes are simulated along the bounding surfaces using a multiple-point statistics technique and, finally, channel and scour-hole locations are gradually perturbed until the corresponding flow responses match the field production data. The perturbation during the history matching must be geologically consistent, i.e. the result must honour the observed individual channel geometries and the interpreted channel stacking patterns.
A digital 3D model representing a real field in offshore West Africa was studied to demonstrate this modelling and history-matching approach in which the multi-scale shale drapes are simulated and perturbed while the reservoir geological concepts are preserved and the well data are honoured. The final history-matched geological models have better prediction power than geological models built with randomly selected input parameters.