Next Generation Geological Modeling for Hydrocarbon Reservoir Characterization - Advances in Data, Methods, Models and Their Applications in Geoscience - page 235



2

  ( RF  RF )

Eq. 9 as . The geological realizations, corresponding to identified

kl k l

tT 

cluster-centroids are selected as the “representative samples” of the entire uncertainty space
and simulated with the full-physics simulator as the reference case. Simulation outcome is
post-processed to compute the distribution of ultimate recovery factor (URF) after a lengthy
period of production, usually selected as the target forecast uncertainty for quantification.
The cumulative density functions (CDFs) are finally constructed for the selected model
realizations with weights assigned to URFs based on the number of models in each
particular cluster and finally quantitatively compared to the reference CDF derived from
full-physics simulations.

3. Discussion and conclusions

Reservoir characterization encompasses techniques and methods that improve
understanding of geological petrophysical controls on a reservoir fluid flow. Presence of a
large number of geological uncertainties and limited well data often render recovery
forecasting a difficult task in typical appraisal and early development settings. Moreover, in
geologically-complex, heavily faulted reservoirs, quantification of the effect of stratigraphic
and structural uncertainties on the dynamic performance, fluid mobility and in situ
hydrocarbons is of principal importance. Although the generation of a sound structural
framework is one of the major contributors to uncertainty in hydrocarbons volumes, and
therefore risk, in reservoir characterization it often represents a compromise between the
actual structure and what the modern modeling technology allows (Hoffman et al. , 2007).
In this paper we focus on some of the outstanding features that have the potential to
significantly differentiate the DecisionSpace  Desktop Earth Modeling, as the next-
generation geological modeling technology, from standard industrial approaches and
workflows mainly in the areas of geologically-driven facies modeling, reservoir property
modeling in grid-less modality and the state-of-the-art workflows for dynamic quantitative
uncertainty and risk management throughout the asset lifecycle. The facies simulation
workflow utilizes a powerful combination of describing the spatial geological trends with
lithotype proportion curves and matrices, integrated with Plurigaussian simulation (PGS), a
robust and widely-tested algorithm with long industrial history. While the implementation
of PGS is not unique to DecisionSpace  Desktop Earth Modeling its geologically highly
intuitive approach to modeling, based on understanding of realistic depositional
environments puts the geologist back into the driver's seat. However, even the most recent
advances in geomodeling practice that represents 3D reservoir volumes with high-
resolution geocellular grids may only mitigate but not eliminate the fact that estimating
gridding parameters commonly results in artifacts due to topological constraints and
misrepresentation of important aspects of the structural framework, which may introduce
substantial difficulties for dynamic reservoir simulator later in the workflow. Hence we look
into the future of building geological models and present and validate the evolving
technology, with the truly game-changing industrial potential to utilize Maximum
Continuity Fields for 3D reservoir property interpolation, performed in the absence of a
geocellular grid. The selection of optimal geocellular parameters with attributes like cell
size, number of cells and layering, is postponed throughout the process and rendered at