Florent Lallier is currently working on stochastic sedimentary stratigraphy. The correlations of outcrops, well cores and logs are a fundamental step in reservoir studies, yet subject to many uncertainties. While several authors have discussed about the “right” approach to determine well correlations, an important question does not seem to have been considered: is it possible to build an algorithm that would generate several possible sets of well correlations accounting for observations along well paths (typically well cores and logs), some reference correlation lines (e.g. coming from clear biostratigraphic boundaries or from seismic stratigraphy) and sedimentological scenarii (e.g., paleogeographic parameters, paleoslope, etc). This algorithm will be based on the computation of the correlation probability between each sedimentary sequence. From these probability several possible realization of well correlation will be computed. These realisations will be used to (i) evaluate the accuracy of well interpretations in terms of facies and depositionnal environment for example; (ii) generate reservoir models integrating well correlation uncertainties.

Supervisors: Jean Borgomano (University of Provence), Sophie Viseur (University of Provence) and Guillaume Caumon (CRPG-ENSG)