Constraining the uncertainty in fracture geometry using tracer tests

In fractured-rock aquifers, the geometric and hydraulic properties of the fractures commonly have a dominant influence on transport. Tracer tests are often used to estimate directly the gross transport properties of a fractured rock mass. The prospects for understanding characteristics of the heterogeneities in a fractured porous medium were explored from evidence provided by tracer experiments. The approach was to simulate flow and transport on a large set of prescribed fracture networks in a two-dimensional homogeneous permeable medium, thus generating synthetic tracer test data. The fracture orientation, aperture, spacing and network geometry were systematically altered from one case to the next. A classification scheme was devised for the tracer breakthrough curves using principal component analysis and this classification was linked to the fracture pattern properties. Even under highly simplified and controlled conditions, quite different fracture patterns can produce very similar breakthrough curves. The classification scheme thus demonstrates that a single breakthrough curve cannot reveal the fracture geometry with any precision. However, the scheme provided a methodology for rejecting geometric properties that do not belong to the fracture pattern under investigation, thus reducing the uncertainty in fracture geometry.