Non-stationary transport phenomena in networks of fractures: effective simulations and stochastic analysis

Among the major challenges in performing underground flow simulations in fractured media are geometrical complexities in the domain and uncertainty in the problem parameters, including the geometrical configuration. The Discrete Fracture Network (DFN) model is largely applied in order to properly account for the directionality of the flow in fractured media. Generation of DFN configurations is usually based on stochastic data and this contributes to generate very complex geometrical configurations for which a conforming mesh generation is often infeasible. Moreover, uncertainty in the geometrical and hydro-geological properties calls for a deep uncertainty quantification analysis; the corresponding huge computational cost of the simulations requires modern efficient approaches faster and cheaper than the classical Monte Carlo approach. In this paper we numerically investigate both these aspects, proposing a viable solution for dealing with geometrical complexities arising in the computation of the hydraulic head and in the solution of the unsteady transport problem of a passive scalar in the DFN, and for dealing with uncertainties in hydro-geological parameters of the fracture distribution considered.