Pasumarti, Lakshmi (2014-08). Exploring an Unstructured Lattice Representation for Carbonate Reservoir Characterization. Master's Thesis.
Carbonates for flow simulation purposes are typically characterized as grid-blocks of varying permeability, with a finer grid employed where heterogeneity is greatest. However, this manner of representation is more suited to sandstone reservoirs, as transport in carbonate reservoirs is usually far more un-geometric due to the complex types of carbonate rock pore-spaces. Far from simply flow between inter-granular pore-spaces, diagenetic processes produce carbonate reservoirs with permeability heterogeneity mainly within three distinct but yet interacting geologic features - Matrix, Vugs and Fractures - very often with each feature occurring at various length scales. This project will explore the merits of an unstructured means of representing carbonates via a lattice-network of pore-volumes connected in space in directions and connectivity properties driven by the rock fabric, as opposed to being limited by the rigid geometry of grid-blocks. With this goal in mind, some aspects related to a lattice-based characterization will be studied.
Firstly, the geologic context that motivates a non-grid based approach to carbonate reservoir modeling will be discussed in the literature review. Secondly, convective and diffusive calculations on a grid will be compared to their equivalents on a lattice in order to establish the applicability of the lattice-system. Convective time-of-flight on a grid is calculated using Pollock's method, while an approximation using the average pore-volumes between nodes will be employed on the lattice. Diffusive time-of-flight on a grid is populated using the Fast Marching Method (FMM), whereas Dijkstra's Algorithm is more appropriate for a lattice. Thirdly, ?-CT-scan data of a rock sample from an outcrop will be used to build an equivalent unstructured lattice-representation of the media at that length scale, and explored for convective and diffusive flow properties. This will be performed by using the AVIZO Suite to first binarize the ?-CT data into pore space and non-pore space, and then skeletonizing it to convert the pore-space into an unstructured set of nodes - carrying volumes - and bonds - each carrying a mean length and radius. These properties will then be used to calculate the transmissibility and diffusive time-of-flight across each bond. Once these are known, convective and diffusive floods can be initiated and the appropriate responses studied to learn about the rock properties. This project is envisaged also as laying the groundwork for a long term goal of unstructured lattice-based carbonate reservoir characterization.