Skip to Main content Skip to Navigation

Modélisation de l'incertitude géologique par simulation stochastique de cubes de proportions de faciès - Application aux réservoirs pétroliers de type carbonaté ou silico-clastique

Abstract : After finding out a potential oil field, development decisions are based on uncertain representations of the reservoir. Indeed, its characterisation uses numerical, spatial models of the reservoir. However, if they are representative of subsoil heterogeneities, the uncertainty linked to subsoil complexity remain. Usually, uncertainty is supposed to be assessed using many equiprobable models, which represent the heterogeneities expected into the reservoir. Nevertheless, those alternative images of the underground correspond to multiple realizations of a given and a single stochastic model. Those methods ignore the uncertainty related to the choice of the underlying probabilistic model. This work aims at improving that kind of uncertainty assessment when modelling petroleum reservoir. It conveys the doubt linked with our subsoil properties understanding on probabilistic models, and proposes to integrate it on them. This thesis first defines uncertainty in the context of oil industry modelling, particularly on 3D geological models comprising several litho-types or facies. To build them, we need, before any simulations, to estimate for every point in the space the probability of occurring for each facies : this is the proportions cube. Even thought those probabilities are often poorly known, they are frozen while using current methods of uncertainty assessment. So, the impact of an uncertain geological scenario on the definition of a proportion cube is forgotten. Two methods based on stochastic simulations of alternative, equiprobable proportion cubes have been developed to sample the complete geological uncertainty space. The first one is closely linked to geology. It integrates directly uncertainty related to the parameters composing the geological scenario. Based on a multi-realisation approach, it describes its implementation on every parameters of geological scenario from information at wells to maps or global hypothesis at reservoir scale resolution. A Monte Carlo approach samples the components of the sedimentary scheme. Each drawing enables to build a proportion cube using modelling tools which integrates more or less explicitly parameters of geological scenario. That methodology is illustrated and applied to an modelling process which is used to model marine carbonate deposits. The second method appears to be more geostatistics focussing on proportion cubes. It rather aims at reconcile distinct eventual sedimentary models. In the meshed model symbolising the reservoir, it assesses the probabilistic law of facies proportion in each cells – they are supposed to follow Dirichlet's probabilistic law. That assessment is done from some models inferred from different geological scenarios. Facies proportions are sequentially simulated, cell after cell, introducing a spatial correlation model (variogram), which could be deterministic as probabilistic. Various practical cases, comprising synthetic reservoirs or real field, illustrates and specifies the different steps of the proposed method.
Document type :
Complete list of metadatas
Contributor : Abbas Zerkoune <>
Submitted on : Tuesday, August 18, 2009 - 12:50:09 AM
Last modification on : Thursday, November 19, 2020 - 3:54:26 PM
Long-term archiving on: : Tuesday, June 15, 2010 - 7:58:33 PM


  • HAL Id : tel-00410136, version 1



Abbas Zerkoune. Modélisation de l'incertitude géologique par simulation stochastique de cubes de proportions de faciès - Application aux réservoirs pétroliers de type carbonaté ou silico-clastique. Modélisation et simulation. Université Joseph-Fourier - Grenoble I, 2009. Français. ⟨tel-00410136⟩



Record views


Files downloads