Object-Based Modelling of the Spatial Distribution of Fluvial Sandstone Deposits

Numerical flow simulations in hydrocarbon reservoirs are based on a detailed knowledge of the geological variability in the subsurface. When well data are sparse compared to geological variability, it is difficult to predict reservoir characteristics in the interwell area. Therefore, it is argued that data of reservoir analogues should be used to support the characterization of sparsely sampled reservoirs. An integrated approach consisting of geological data acquisition and stochastic geological modelling is presented for the problem of predicting the spatial distribution of sandstone deposits (object-based modelling). Outcrop data of the distal part of the TOrtola alluvial fan are used to estimate the parameters of a Boolean model, that is used in a simulation algorithm to generate realizations of a 2D cross-section of the studied outcrop area. The simulations are conditioned on synthetic well data and on the stratigraphic sequence in which sand bodies are deposited. The results have shown that the outcrop area is appropriately reproduced with a mean well spacing of 230 m. The proposed method should also be validated for producing reservoirs that are analogues of the Tdrtola fan. The stochastic method has several shortcomings that arise from statistical considerations.