Closed‐flow column experiments: A numerical sensitivity analysis of reactive transport and parameter uncertainty

Abstract

The identification of transport parameters by inverse modeling often suffers from equifinality or parameter correlation when models are fitted to measurements of the solute breakthrough in column outflow experiments. This parameter uncertainty can be approached by performing multiple experiments with different sets of boundary conditions, each provoking observations that are uniquely attributable to the respective transport processes. A promising approach to further increase the information potential of the experimental outcome is the closed-flow column design. It is characterized by the recirculation of the column effluent into the solution supply vessel that feeds the inflow, which results in a damped sinusoidal oscillation in the breakthrough curve. In order to reveal the potential application of closed-flow experiments, we present a comprehensive sensitivity analysis using common models for adsorption and degradation. We show that the sensitivity of inverse parameter determination with respect to the apparent dispersion can be controlled by the experimenter. For optimal settings, a decrease in parameter uncertainty as compared to classical experiments by an order of magnitude is achieved. In addition, we show a reduced equifinality between rate-limited interactions and apparent dispersion. Furthermore, we illustrate the expected breakthrough curve for equilibrium and nonequilibrium adsorption, the latter showing strong similarities to the behavior found for completely mixed batch reactor experiments. Finally, breakthrough data from a reactive tracer experiment is evaluated using the proposed framework with excellent agreement of model and experimental results.

Colloidal Stability and Mobility of Extracellular Polymeric Substance Amended Hematite Nanoparticles

“…Because hematite is the only source of Fe in the column system, we used the total Fe concentration as a proxy for the hematite content. The adsorption strength was calculated from breakthrough data according to the method described by Ritschel and Totsche (2016b), assuming a linear distribution between the aqueous and solid phases (Supplemental Material, Section S.11). The total adsorption capacity of the glass surfaces was calculated assuming spherical glass beads of uniform size and homogeneously distributed HNP.…”

“…We decided on this type of experiment because it allows full establishment of a flow equilibrium, while at the same time conserving the total mass of the system and minimizing experimental artifacts (Ritschel and Totsche, 2016a). Thus, the true mobile fraction of a solute or colloid can be exhaustively determined, especially in the case where surface interactions are strongly rate limited (Ritschel and Totsche, 2016b).…”

An Improved Tandem Neural Network Architecture for Inverse Modeling of Multicomponent Reactive Transport in Porous Media

“…A significant challenge for developing effective RTMs is parameter estimation, since the existence of large uncertainties associated with reaction processes and the associated parameters in geochemical systems (Ampomah et al., 2017; Dai & Samper, 2006; Fakhreddine et al., 2016; McGrail, 2001; Oyanagi et al., 2020; Patil & McPherson, 2020; Ritschel & Totsche, 2016; Samper et al., 2008; Soltanian et al., 2015; Suescún‐Díaz et al., 2021; Vasco et al., 2018). Measurements of RTM parameters (e.g., selectivity coefficients and initial concentrations of chemical components) are often complicated, time‐consuming, and expensive (Dai & Samper, 2006; T. Xu et al., 1999).…”