Outcrop information and interpretations can be used to create three-dimensional conceptual reservoir characterizations that try to predict possible ranges in reservoir behavior. In this study, outcrop models were used to address uncertainty in prediction of volumes and recovery in deep-water channelized slope reservoirs. Three-dimensional heterogeneities were added to the outcrop models that might be of importance for flow-simulation or seismic studies.

Two outcrop examples of deep-water depositional systems are discussed: the 100-Foot Channel Complex in the Brushy Canyon Formation of southwest Texas, and the Ainsa II Channel Complex in the Hecho Group of Spain. Three-dimensional geologic models were built, and waterflood simulations and geophysical studies were performed. The two outcrop examples are significant in that the 100-Foot Channel is a multistory channel complex, whereas the Ainsa II Channel Complex is both multistory and multilateral. Four models are discussed for the 100-Foot Channel Complex example, and eleven models are discussed for the Ainsa II Channel Complex example.

Based on static geologic models, examples of the possible origins of volumetric uncertainty are given, in which uncertainties of as much as several hundred percent are observed. Two key uncertainties are areal variation in net-to-gross and subseismic thin-bed effects. Based on waterflood simulations of the outcrop models, key features that influence reservoir recovery efficiency are variance in permeability distributions (characterized by the Dykstra-Parsons coefficient) and reservoir connectivity. Reservoir compartments were created when continuity of intrachannel mudstone layers or mudstone drapes on channel margins was high. Variation in permeability heterogeneity can affect recovery efficiency by more than 10% (or in these cases, about 25% relative to the lower recovery). Several examples of models with poor reservoir connectivity are presented, in which recovery efficiency is reduced proportionally to connected volume (a spread of 75–175% relative to the lower recovery efficiency). A comparison of reservoir connectivity, number of geobodies, and degree of heterogeneity, measured in terms of differential path length, showed that static characterization of the reservoir could be useful for dynamic predictions.