Probabilistic estimation of fracture transmissivity from Wellbore hydraulic data accounting for depth-dependent anisotropic rock stress

Abstract

A new method is introduced that incorporates the use of hydrological and rock mechanical data in assigning transmissivities for fracture-network models. The hydrological data comes from fixed-interval packer tests carried out in a borehole and the rock-mechanical data are the prevailing in situ depth-dependent stress-field and the stress-closure relationship of fractures. In the model, the fracture transmissivity distribution is considered to be constituted of two components, one deterministic stress-induced component and the other a stochastic component that describes the intrinsic variability of fractures in a network. The outcome is a tensorial description of fracture transmissivities in an anisotropic stress-regime, where the transmissivity for an arbitrarily oriented fracture in the network is determined by its orientation in relation to the ambient stress-field. These transmissivities are conditioned such that the overall results satisfy the hydraulic packer test data. The suggested procedure is applied to an example data set from a site at Sellafield, England. The results show that the probabilistic approach , relying on hydraulic data alone, may underestimate the true variability in fracture transmissivities, since the typically vertical boreholes entail a sampling bias towards horizontal fractures that are predominantly subject to vertical stress. The suggested method helps to account for the true underlying three-dimensional variability that is incompletely resolved by using the hydraulic borehole data alone. This method is likely to have the largest impact at low stress-levels, in strongly anisotropic stress-fields, for borehole directions parallel to one principal stress, and for fracture network geometries characterized by sets orthogonal to the three principal stresses.