Reservoir Modeling and Uncertainty Estimation: A Comparison Between Stochastic and Deterministic Inversion.

Abstract

This study compares the uncertainty estimation based on both deterministic and stochastic inversion with the Gauss Simulation technique applied to modelling the distribution of limestone facies in a sand-shale reservoir. The final goal is to understand and quantify the benefit of using stochastic inversion in facies modeling and uncertainty estimation. The study area is the Oseberg South Field which is a producing oil field in the northern North Sea. This study focusses on the reservoir level in the Middle Jurassic Brent Group. Because of the missing high frequencies deterministic inversion provides a smooth average of the impedance which cannot reliably model the thin layers of limestone in the reservoir. Facies modeling based on deterministic inversion is superior to well-based stochastic modeling only in the case of thick layers that lie within seismic resolution. Stochastic inversion adds value by capturing the property distribution uncertainty which is show-cased by the facies modelling. Stochastic inversion is superior to deterministic inversion because possible limestone layers of thickness below seismic resolution are addressed. Furthermore, stochastic inversion provides multiple equiprobable results thus allowing for a more reliable uncertainty estimation of the reservoir facies. Because there are no published studies in the Oseberg South field, incorporating seismic inversion within the geomodelling workflow the results of this thesis could lead to a better decision making for future well placement.