Experimental and modeling assessment of slug tests in the presence of coupled hydraulic, thermal, and solute transport effects

Assessment of hydrogeological parameters such as aquifer permeability, thermal conductivity, and dispersion coefficients is critical for comprehensive groundwater resource assessment. In this context we provide experimental analyses of slug tests performed upon considering a confined groundwater system where hydraulic, thermal, and solute transport processes take place. Design and execution of experiments are supported by a detailed numerical modeling analysis taking into account the intimately coupled nature of these mechanisms.

The study investigates coupling mechanisms among (Darcy scale) flow, thermal, and chemical fields across the aquifer. In this sense, groundwater flow directly influences rates of heat and solute transport. Temperature impacts groundwater dynamics upon altering, e.g., water density and viscosity. Our numerical simulations are grounded on the well known and broadly tested COMSOL suite. We also explore the potential of a Physics-Informed Neural Network (PINN) approach to provide characterize complex coupling conditions of the type we analyze, thereby complementing model evaluation and parameter estimation accuracy. Doing so enables us to estimate the set of model parameters through numerical simulations performed according to two diverse strategies and anchored on experimental data.

A dedicated indoor experimental platform is then developed. Slug tests associated with coupled flow and (chemical/thermal) transport conditions are designed on the basis of preliminary numerical simulations performed using both the COMSOL-based fully coupled model and the PINN approach. The platform is equipped with excitation devices and a high-frequency, high-precision automatic data collection system tailored to meet the requirements of hydraulic-thermal-chemical coupling associated with slug tests. In this context, NaCl is employed as a tracer, its concentration being monitored through electrical conductivity signals. A cylindrical container filled with homogeneous fine sand is designed to represent the porous domain. The top is sealed with an insulating film and cement, simulating ideal confined aquifer conditions. One-dimensional column tests are also performed to enable cross-validation of interpretive modeling and parameter estimation. By integrating data such as water level, temperature, and electrical conductivity under various experimental conditions, the study qualitatively examines the temporal and spatial variations in groundwater flow, heat transport, and solute transport. These experimental results are then quantitatively employed in the context of model-based parameter estimation. The latter is performed through the full system model (as implemented in the COMSOL suite) as well as through the PINN approach.