Representation and Modeling of Structural Uncertainty in Subsurface Systems by Means of Image Operations in Depositional Space

Abstract:

Structural uncertainty is an aspect of uncertainty quantification that is receiving increasing attention due to its zero-th order impact in forecasting flow in subsurface formations. Imperfect data and lack of geological understanding poses both challenges of conceptual uncertainty at the fault hierarchy level and parameter uncertainty at single fault level. Therefore structural uncertainty is often underrepresented in subsurface modeling. Unlike property uncertainty, such as for hydraulic conductivity & porosity few methods to address structural uncertainty in an integrated manner exist. A first bottleneck with modeling structural uncertainty is due to constraints posed by fault hierarchy, fault-horizon consistency, and structural-depositional environment consistency. A second bottleneck associated with structural uncertainty lies in the difficulty to construct grids compatible for flow & transport simulations, often requiring manual intervention. In this presentation, we propose to use approximate geological representations of faults in a simple Cartesian domain and to select appropriate structural models using flow proxies in the same Cartesian domain . Faults are first mapped into a Cartesian (depositional) space using an unfolding-unfaulting operation. Fault uncertainty is then modeled implicitly by increasing or decreasing the size of fault blocks in depositional space or by simple removing fault without affecting hierarchical/age relationships. Utilizing simple flow diagnostics in the Cartesian domain we select a small amount (10s) of structural models to represent flow & transport uncertainty. A case study illustrates how this approximate method compares well with the full method of modeling 100s of structural models.