This paper presents a new approach to constrain reservoirrnfacies models to dynamic data. This approach is mainly basedrnon the combination of an optimization method called thernsimplex method and parameterization techniques such as therngradual deformation method.rnThis parameterization technique is well known for itsrnefficiency in constraining geostatistical models to dynamicrndata. Originaly developped for continuous Gaussian models,rnthe gradual deformation method was extended to faciesrnmodels [12] and more recently to any kind of geostatisticalrnmodel. However, the numerical behavior of the inversion canrnbe strongly affected by discontinuities in the model changes inrncase of facies-based models. Our approach proposes a robustrninversion method coupled with the gradual deformationrnmethod for the calibration of facies models.rnThe methodology we present is directly inspired by thernconstruction of a reservoir facies model. We start byrngenerating a related Gaussian continuous geostatistical model.rnIn a second phase this continuous model is transformedrnaccording to the number of facies and their proportions tornobtain a consistent facies model. Since a directrnparameterization of the facies model does not allow thernpreservation of the geological properties of the model, wernperform the parameterization on the related continuous modelrnand then we proceed to the model truncation. In this manner,rnwe can optimize the deformation parameters involved in thernparameterization using the simplex method to obtain anrnhistory match.rnThe efficiency of this approach is illustrated by matchingrnan interference test. The parameterization phase wasrnperformed using the gradual deformation method. Constrainedrnfacies models were thus obtained.rnA second phase of this study was devoted to quantifyingrnthe impact of uncertainty of spatial facies distribution onrnproduction forecasts. In an integrated process, we combine thernsimplex approach with experimental design theory and thernjoint modeling technique to achieve a probabilistic productionrnforecast which takes into account the uncertainty on spatialrnfacies distribution and classical reservoir parameters whilernpreserving the history match. This application demonstratesrnthe importance of the interference test matching in reducingrnthe uncertainties on production forecasts.
展开▼