MULTI-SCALE STUDIES OF FRACTURED RESERVOIR ANALOGS

In this research two different but related topics are discussed that encompass characterization and modeling of data from petroleum reservoirs and their analogs by employing multiple-point statistical (MPS) based techniques. The first one is about the applications of Lacunarity, a technique that was originally developed for multiscale analysis of spatial data. In simple terms, lacunarity characterizes the distribution of spaces or gaps in a pattern as a function of scale. Although, initially the concept was associated with that of fractal objects, this technique can be applied to a variety of data including multifractals and other binary as well as non-binary data that are not necessarily scale-invariant. The talk shall demonstrate how to best capture the essence of lacunarity as a concept, one that has been “customized” in different manners to suit the needs of the type of data and the technical questions that are asked. A range of data types from reservoir analogs including 1-dimensional fracture spacing and 2-dimensional fracture networks to sedimentary data and multifractal permeability fields will be considered. For example, where the data type is vein spacing (1-dimensional binary) from geologic outcrops, it will be shown how lacunarity can be employed for identifying different patterns (e.g., fractal/uniform/random) at different scales of observation of the same data set. Where fracture aperture values are available along with spacing (1-dimensional non-binary) it will be shown how lacunarity can delineate scale-dependent changes between persistent and anti-persistent trends in the data and what kind of useful information can be extracted from such analyses. In another study involving fracture networks from outcrop maps (2-dimensional binary) it will be demonstrated how lacunarity can be used as a proxy for flow properties of such maps. The second topic shall demonstrate the use of Image-Quilting, a pattern-based MPS technique, for modeling fracture network maps starting with outcrop analogs as training images (TI). This research suggests that MPS has the potential of becoming an alternative fracture modeling technique and can directly take outcrop maps as inputs as opposed to traditional ones where individual statistics of fracture length, orientation and distribution need to be inferred.