ABSTRACT

Modelling the spatial distribution of the hydraulic properties of a flow domain is not an easy task as it reflects the distribution of 3D geological features, commonly inferred from insufficient combinations of 2D field mapping and one dimensional logs. The non-conservative nature of geological features rule out simple conservation laws, as for mass and energy, which could guide the evaluation of their spatial distribution. Whenever possible, point inferences for the eigenvalues and eigenvectors of hydraulic conductivities may allow interpolations or correlation to descriptors of lithologic and structural features outside the sampling points. Given a dense spatial distribution of the hydraulic head, it may sometimes be possible to infer some characteristics of these tensor components. However, hydraulic heads are seldom monitored at many sampling points. Despite these difficulties, there are few techniques to approximate the spatial distribution of hydraulic attributes and properties within a flow domain. If the “modeller’’ bypasses this important step and substitutes the actual flow domain with a simple homogeneous and isotropic one, the resulting numerical model has little value despite its computational merits. This is particularly true regarding groundwater simulation for fractured rock hydraulics.