This paper presents a first step for incorporating heterogeneous rock properties into numerical models; our final goal is to realistically model wellbores in heterogeneous bedded media, with individual beds represented as transversely isotropic, but not necessarily having identical directions of symmetry. 3D FEA modeling and orthotropic elastic symmetry was used to represent the behavior of arbitrarily oriented wellbores on layered media. The media (presently restricted to homogeneous bedded media) is assumed to be transversely isotropic with 3D failure criterion varying with orientation to bedding and with confining pressure. This is a substantial improvement over current models using anisotropic strength and isotropic deformation (i.e., localized plane of weakness models). This first step shows deformation and failure strongly dependent on orientation to bedding. Varying bedding orientation, in-situ stress orientations, and/or the wellbore orientation, results in marked differences in the deformation and failure at the wellbore. This is particularly important for interpretation of in-situ stress orientation from wellbore failures.
Rock mechanics characterization (via laboratory measurements) and numerical modeling are fundamantal tools for understanding and forecasting wellbore and reservoir stability problems during drilling, completion and production. Lacking rock properties and in-situ stress data, rock mechanics practicioners have traditionally focused on making educated guesses on geomechanical data and on using simple models for analysis and forecasting1. Recently, the risk analysis method has been introduced to geomechanics modeling, to allow quantitative evaluations of the uncertainties associated to the data used for modeling (e.g., insitu stress, failure parameters, rock properties), and to assess the degree of relevance of data required for modeling. Risk analysis is an excellent methodology for adding a quality rating to geomechanical predictions; however, it is most relevant when sufficient data is available (i.e., a statistically representative, present or historical, data set), and when the mechanical model used in risk analysis evaluations includes the important variables and relationships that define the problem. Thus the benefits of risk analysis are also undermined by lack of data and simplicity of the mechanical models. Advancements in rock mechanics, leading to better modeling, better forecasting and risk reduction, will result from better quality and availability of data for modeling, and from better models; the latter objective being strongly limited by and dependent on the former.
Improvements in the volume and quality of rock data for modeling are currently taking place (e.g., high-resolution continuous strength measurements on core, small-scale sample testing, textural analysis for predictions of rock properties, and the emergence of comprehensive rock mechanics data collections allowing searches of rock properties based on geological or compositional descriptions). Advancements in numerical analysis and computer power have made it possible to solve complex problems on standard desktop computers. We thus feel that next important advancement to rock mechanics modeling is to re-evaluate the parameters required in the models. Granted the complex and coupled constitutive behavior of reservoir and overburden rocks (poro-thermo-elastic, plastic or visco-plastic) we strongly believe that heterogeneity is the fundamental rock property that should be introduced to computer modeling.
Rock heterogeneity is observed at scales ranging from the micro-scale (via SEM and thin-section images), through the meso scale (via visual observati