Water-soluble gases as partitioning tracers to investigate the pore volume-transmissivity correlation in a fracture

Hydraulically equivalent fractures may show striking differences when a gas-migration experiment is performed because of the different correlations between transmissivity, pore volume and entry pressure. We numerically simulate gas migration between injection and extraction boreholes in a parallel plate fracture with a heterogeneous fault gouge, in a rough-walled fracture filled with homogeneous material, and in a rough-walled empty fracture. The parallel plate model and the empty model clearly show the existence of preferential paths; for high variance of the transmissivity field, gas flow takes place only in few discrete channels separated by water-saturated regions. In contrast, in the fracture filled with homogeneous fault gouge, the gas saturation is continuous and more uniformly distributed. It appears a fundamental issue to be able to discriminate in situ among conceptual models that can yield such a different gas-saturation distribution. As in practice, the saturation distribution cannot be directly observed, tracer experiments are performed to characterize a fracture. For these reasons, we simulate the transport of tracers, which are added to the gas phase as soon as quasi-steady saturation distribution and extraction rate are achieved, and we compare the breakthrough curves obtained assuming different models. Our numerical simulations suggest that discrimination among the models on the basis of single-tracer tests is unlikely. A better tool to investigate fracture properties is provided by a gas-tracer test, in which a cocktail of gases with different water solubility is employed. These gases behave as partitioning tracers and allow us to estimate the gas saturation in the fracture. Indeed, by comparison of the residence-time distributions of different gases, we are able to compute a streamline effective saturation, which is an excellent estimate of fracture saturation. In addition, the streamline effective saturation curve contains information that is useful to identify the conceptual model that more likely applies to the fracture.