Datapages, Inc.Print this page

3-D Fluvial Reservoir Characterization Using Multiple-Point Statistics


Dovera, Laura, Ernesto Della Rossa, Marco Pontiggia, Livio Ruvo, Giuseppe Serafini, Eni, San Donato Milanese (Milano), Italy


Multiple-Point statistics, an innovative technology being developed by Stanford University, enables us to build a reservoir model integrating all available information and considering different geological scenarios in order to correctly express the geological uncer­tainty. This statistical approach performs the simulation starting from a Training Image, a conceptual visual representation of how heterogeneities could be distributed in the actual reservoir.

This paper presents a 3D detailed reservoir characterization of a fluvial reservoir using Multiple-Point statistics. The reservoir consists of prograding fluvial bars separated by shales. The coarser facies are supposed to overly and sometimes to erode the finer ones. Because of the overall progradation pattern, a coarsening-upward vertical facies sequence may be inferred.

First, by integrating all the information coming from cluster analysis, seismic and sedi­mentological model, a stationary Training Image consisting of fluvial bars separated by shales and with a coarsening-upward facies trend has been built.

Next, using the Training Image built and the conditioning hard and soft data, a Multiple-Point statistical simulation has been performed.

The results show that the geological conceptual model has been reconstructed correct­ly and that the Training Image patterns geometries and facies contacts are well reproduced.

To quantify the geological uncertainty, a different Training Image has been built and a new simulation has been performed. Also in this case interesting results has been obtained.

A comparison with a traditional object based simulation is also presented.