Can We Sample the Complete Geological Uncertainty Space in Reservoir-Modeling Uncertainty Estimates?

Abstract

Summary The method by which the space of uncertainty should be sampled in reservoir models is an essential point of discussion that can have a major impact on the appraisal, development and economics of hydrocarbon fields. Usually, uncertainty is assessed by way of many realizations of a given stochastic model with fixed parameters but this does not adequately sample the complete geological uncertainty space and it underestimates the uncertainty. To improve the uncertainty assessment, in this article we propose a classification and hierarchy of the sources of internal reservoir architecture uncertainty. The impact of the different levels and sources of uncertainty on elements such as original hydrocarbons in place or final recovery is quantified. In this article we demonstrate that decisions as significant as stationarity, organization of reservoir heterogeneity, choice of the global statistics and variogramme range are of prime importance in relation to the other parameters. Furthermore, we also prove that the relative importance of any parameter in assessing the uncertainty depends on the type of result—volumetric or dynamic—on which the uncertainty calculation is computed. Uncertainty of the global mean provided by hard data is assessed using sampling techniques. The coefficients of variation obtained from many field cases allow a comparison between the different distributions. It lends support to the hierarchy of the sources of geological uncertainty into six levels. The evolution of this uncertainty with the quantity of hard data, wells, and the quality of knowledge, zoning, has been quantified using synthetic models and a field case. Both a general function for the decrease of uncertainty with the quantity of the data, and a possible increase—for some parameters—of uncertainty estimates with knowledge development are demonstrated herein.