The importance of reservoir uncertainties for the prediction of petroleum recovered volumes and fluid flow performance is well known. Large uncertainties in reservoir gross volume estimation derive not only from the measurement errors, but also from the way that the static reservoir model is constructed. The aim of this article is to explore the uncertainty induced by the combination of the different steps in construction of the reservoir model. These steps can be split into three main areas: geometry, internal heterogeneities, and petrophysical properties. The first task consists of building the architecture through the delineation of units. It must deal with the sparseness of control data (layer intercepts measured at few wells, sometimes deviated) and the complexity of the stratigraphic description (several heterogeneous layers, presence of faults) and account for the seismic horizons observed. Each layer is then populated with lithofacies reflecting the geological environment. This environment is characterized by the depositional sequence as well as the number of lithotypes (facies with similar properties), their proportions, spatial trends, and arrangement (transition between facies). Interpreted seismic attributes can bring additional constraints on the lithotype proportions. Each lithotype is then assigned petrophysical properties. Finally, the hydrocarbon volumes in place are calculated above an oil-water contact. Uncertainties are evaluated within a realistic layering model, using relevant stochastic techniques (categorical and multigaussian simulations) and choosing the adequate parameters. The quantification is performed on global volumes as well as on more local criteria.

The results show that the choice of modeling workflow has an impact on the hydrocarbon volumetric assessment, affecting both the estimate of mean volume and the range of uncertainty about the mean. This needs to be taken into account when undertaking reservoir volumetric assessments.