Multi-scale characterization of transport in heterogeneous porous and fractured aquifer media using innovative heat and solute tracer tests

Characterization of subsurface mass and energy transport in heterogeneous porous and fractured aquifer media is particularly challenging due to existence of preferential flowpaths and dynamic matrix processes at multiple scales. The main contribution of the present work is to combine solute and highly diffusive tracer information with the objective to improve the understanding of the actual transport processes and the predictive capacity of models. The use of innovative tracers such as heat or inert dissolved gases leads to different sensitivities to preferential flowpaths and matrix processes due to the broad variation of their diffusion coefficient values. This in turn allows a more realistic consideration of the heterogeneity patterns in the geological media and a better quantification of the associated uncertainty in model predictions. Field campaigns were performed in different geological formations with injections of: (1) heat and a solute in alluvial sediments; (2) dissolved gases, heat and uranine in a porous/fractured chalk system, (3) salt and water with higher and lower temperature (than the natural groundwater background temperature) in a weathered and fissured granite system in India. These field campaigns provided information on the evolution of concentration and temperature as function of time (breakthrough curves). The observations were interpreted with regards to peak time, peak value and slope of the tailing, taking into account the variation of the diffusion coefficient value of the different tracers. First, a joint heat and solute tracer experiment, previously carried out in alluvial sediments, was used to study the heterogeneity and for prior (i.e., parameter) uncertainty investigation. Monte Carlo simulations were used to revisit the current state of knowledge. For example, an increase in hydraulic conductivity with depth (due to the local spatial heterogeneity of the sediments) could now be realistically implemented in the aquifer simulation using the numerical calculator HydroGeoSphere (HGS). The Monte Carlo simulations were also used for a Euclidean distance-based sensitivity analysis relating model input and output uncertainty. It was shown that heat transport is more sensitive to heterogeneous patterns compared to solute tracer transport. Second, the joint injection of uranine and dissolved gases and their recovery (forced gradient experiment) in a porous/fractured chalk emphasized the influence of the specific (i.e., fracturing) aquifer heterogeneity on higher diffusive tracers. A channel model (analytical solution with explicit fracture numbers and apertures) was used to simulate the transport of helium, the gas with the highest diffusion coefficient tested. Then, using HGS, the simulation of flow, transport of solutes and heat transfer allowed to test the influence of matrix diffusion at different time and spatial scales by varying the diffusion coefficient of the tracer species. Third, the injection of hot and cold water in a weathered and fissured granite system in India, allowed the study of the actual and differential release of heat from the matrix to the fracture fluid and vice versa. A deterministic model solution using HGS showed that the width of the discrete fracture that is supposed to represent the fracture tested in the field had a significant effect on the temperature simulations. The results of the present PhD thesis clearly emphasize the importance and usefulness of these new tracers as well as their added value for an advanced aquifer characterization, to obtain a realistic assessment of the inherent heterogeneity of the different investigated groundwater systems. The latter is required for reliable simulations of mass and energy transport in the dynamic subsurface.