Abstract

Modeling the Vaca Muerta shales in Argentina involves the analysis of large amounts of data from different knowledge domains, ranging from seismic- or reservoir-wide observations to small-scale spatial heterogeneities, including lateral fabric variations, preexisting natural fractures, thin layers (beef or ash beds), etc. This paper presents a workflow successfully used to identify such heterogeneities, assess their impact on hydrofrac growth patterns and include their effects on the fracturing models.

Fracture simulation models expect to replicate available field observations related to completion and stimulation. To provide a robust base model for these simulations detailed reservoir properties modeling was conducted, initially along the drilled-well trajectories, using data from logs, surface measurements, rock lab analysis, drilling data records and all available information provided. Current computational limitations imposed by the extrapolation of 1D-derived model(s) onto the 3D volume can cause a loss of resolution, which can render the whole modeling exercise questionable. Our workflow starts with the detailed characterization of the reservoir, including the identification of heterogeneity indicated by the data, the assessment of its expected impact in hydraulic fracture propagation, and the development of strategies to include its effects, by mapping spatial heterogeneities while retaining the ability to run successfully the eventual simulations within the resolution limits set by current technology.

The methodology, which is easily applicable to similar projects, enables the incorporation of anisotropic stresses, along with the precise placement of natural fractures, high strength "beef zones" and/or weak laminations, all ubiquitous in the studied shale play. This, in turn, allowed us to refine the model and better match the physical evidence revealed by the data collected during the stimulation and production phases of the control wells. Among other examples, our results showed that considering only isotropic materials would not correctly predict the fracture growth patterns that were eventually observed. By ensuring that our modeling efforts consider the different indicators provided by the various data types involved, an evidence-supported mechanical properties model is obtained. The same can be said regarding the stress model ultimately applied in the subsequent simulations. The adjustments or calibrations made to the model are then limited by the need to conform to those observations and retain the support of the experimental evidence.

This study stresses the need to aim, from the project outset, for a reservoir model honoring most if not all the collected experimental evidence. This approach enforces that the modelers thoroughly understand this evidence, including why it would be contradicted by a more simplistic model and devising ways in which crucial aspects, as supported by the data, can be ultimately incorporated to the working simulation model.

This content is only available via PDF.
You can access this article if you purchase or spend a download.