Geological Consistency Over Geological Time: A New Constraint to Honor for Process-Based Stochastic Reservoir Models

Massonat, Gerard J., Total, Pau, France

When geological models are populated with facies and Petrophysics through the use of stochastic algorithms, the correlation between cell values is provided by both the prior sta­tistics and the variograms. With such a procedure, the geological consistency of the sto­chastic model cannot be tested, and this could lead to the building of nice but inconsistent realisations of geological models. The new challenge for geological modelling is then to con­strain stochastic models to geological coherency during the modelling process. This can be done by the development of methodologies and algorithms based on the respect of the geo­logical process(es) which led to the reservoir construction.

Two methodologies are presented.

In the first one, the stratigraphic signal extracted from each well is compared with the others, in order to evaluate the consistency between data and extract a common stratigraph­ic behaviour for the reservoir. This common part of the signal is propagated in the whole model in order to drive the stochastic simulation of facies in shelf marine sedimentary envi­ronments. This methodology improves the stratigraphic aspect of the grid, as well as it forces the modelling to honour the geological consistency over all the depositional time.

The second methodology consists in the development of a 4D stochastic algorithm for modelling karstic reservoirs, with respect to geological pre-existing heterogeneity and hydraulic process. Karstic conduits are generated over time consistently with facies, bed­ding and fractures discontinuities, and location in the hydraulic system. Conduits develop­ment probability field changes at each step, delivering a fully 4D stochastic model. Finally, the simulation is stopped when it honours equivalent permeability from well tests.

These two new methodologies can be considered as representative of the Future of sto­chastic modelling of geology.