The quantification of fracture spacing and length are crucial for determination of effective block sizes to be used in reservoir simulation. Outcrop and modeling studies have shown that average values of these parameters do not adequately describe a fracture pattern, however. Observational work suggests that the cumulative number of fractures larger than a given size (in length or aperture) can be described with power-law relationships, and these empirical models are used to extrapolate from the observational scale to other scales of interest. Fracture spacing distributions are commonly described relative to their "saturation level," a measure of the regularity of the spacing between fractures of a given set that is thought to be related to strain level. As a spacing population approaches a normal distribution it is said to become more saturated or "well-developed," and this progression is assumed to be related to increasing strain imparted to a rock layer. A complication observed in sandstone outcrops from Oil Mountain, a Laramide anticlinal structure in Wyoming, is that fracture sets can reach saturation at different levels of strain even in the same bed. Assuming that average spacing indicates the level of strain accommodated by a given fracture set, the cross-fold fracture set from Oil Mountain requires very high strain to become saturated (spacing on the order of ift for a 30 ft thick bed). In contrast, the fold-parallel set displays a very high saturation at extremely low strain level (30 ft average spacing for a 30 ft thick bed). With a geomechanical model, we investigate the flindamental processes that control fracture propagation to explain spacing variability and suggest how to develop predictive models of fracture spatial arrangement.
In addition to discussing the mechanics of fracture network development, we address the sampling problems inherent in reservoir characterization. For subsurface fractures, measuring the saturation, length distribution, or connectivity is highly problematic because boreholes rarely encounter the fractures of interest. This is a serious impediment to acquiring sufficient data to constrain predictive characterization models that feed into reservoir simulations. Using Oil Mountain as an outcrop analog of a sandstone reservoir, we explore the use of microfractures as indicators of the attributes of large fractures. The advantage of microfractures is they are so much more abundant than macrofractures that they can be readily sampled from weilbores (a single thin section can produce thousands of measuments). If microfractures can be used as proxies for large fractures, then the results of predictive characterization modeling can be applied to subsurface cases much more effectively and accurately than has hitherto been possible.