Paul Gelderblom

Geological Constraints in Model-based Seismic Inversion

Geologically Constrained Inversion is an extension to the Shell proprietary Promise stochastic inversion engine that takes lateral continuity and well constraints into account. The new algorithm produces an ensemble of reservoir-size (as opposed to single-position) models that match the seismic data, the geological constraints and the well data. The results are geologically more realistic than those of the original single-position inversion. The algorithm is parallelized so that it can be applied to real-world size datasets.

Facies Inversion with Plurigaussian Lithotype Rules

Accurate incorporation of geological concepts such as lithological facies distributions is an important aspect of building reservoir models. Consequently, accounting for facies in seismic inversion generates models conditioned to geological concepts and plays an important role in decision-making. Shell’s proprietary probabilistic model-based seismic inversion engine Promise is generally applied to invert for continuous variables, such as NTG, saturation, and layer thickness from seismic data. In some depositional environments, the spatial variability of reservoir properties is characterized at fine geological scale by facies and the corresponding petrophysical properties; hence, an implementation of facies in seismic inversion is desirable. In this study, we propose a novel methodology for lithological facies inversion utilizing Plurigaussian rock-type rules. Direct inversion of facies may result in unrealistic facies contacts; therefore, the proposed technique instead inverts for a pair of “guide” variables using Promise. The guide variables are then classified into facies using a methodology inspired by Plurigaussian simulations, where a defined lithofacies rule map is used to constrain facies proportions and contacts. The required inputs for the workflow are the lithofacies rules and variogram estimates of the guide variables. Both of these can be derived from a prior estimate of the facies distribution and can also take into account geological constraints from a human expert. We demonstrate the workflow for a three facies case with a synthetic wedge model and seismic data of a marine survey.