ABSTRACT Sedimentary facies is a key control on petrophysical properties within subsurface reservoirs and facies modelling is a critical aspect of reservoir modelling. Several modelling methods exist and selecting the best approach for a specific case is challenging and time-consuming. Outcrop analogues provide detailed information on architecture, geometry and facies connectivity that is not directly available in the subsurface. By modelling outcrop data it is possible to test and compare different modelling strategies systematically in a case where the geology is well constrained. The Eocene aged, Sant Llorenç del Munt fan-delta complex (NE Spain) is a well exposed, transgressive–regressive fan-delta. Outcrop data were used to test a variety of modelling strategies in which the density of conditioning wells, stratigraphic subdivision, modelling algorithm and trends were all varied. The results of these modelling exercises were compared against themselves and against a close-to-deterministically built Base Case reconstruction using a series of static measures including the distribution of fan-delta front reservoir facies, directional connectivity and reservoir-to-well connectivity.

The results highlight the following: (a) the impact of the conditioning well density on improving the stratigraphic architecture reproduction in the different modelling approaches; (b) that surface-based modelling subdivisions including the maximum flooding surface to separate independent grids for modelling the transgressive and regressive sequence sets can be detrimental when compared to using only the top and base of the composite sequence unless additional constraints are included; (c) that an algorithm combining a linear trend and a Gaussian field is the most suitable algorithm for reproducing this type of architecture, but requires defining a 3D trend; and (d) the need for using trends to reproduce the architecture when well data are sparse. These results provide guidelines for modelling analogue fan-delta reservoirs in the subsurface.