Abstract
Electrokinetic (EK) technologies are promising solutions for the remediation of contaminated sites, particularly in low-permeability porous media. However, their widespread adoption is hindered by the challenge of predicting the complex, coupled physico-chemical processes triggered by the application of external electric fields in the subsurface. Numerical models therefore represent essential tools to interpret system behavior. Uncertainties in experimental data, as well as in the formulation of conceptual models, still pose a challenge to develop robust predictive tools. In this context, our work addresses the impact of various sources of uncertainty on model-based predictions of EK transport in porous media. We employ Monte Carlo-based global sensitivity analyses (GSA) within both single-model (SM-GSA) and multi-model (MM-GSA). The multi-model approach relies on a theoretical framework encompassing different models capable of interpreting a set of EK transport scenarios. This allows us to address the impact of model formulation besides parametric uncertainty on mass transfer and reaction dynamics of EK transport. All candidate models in our set are based on a 2D dipole electrode configuration and each model incorporates a different combination of physico-chemical processes to explore different EK remediation scenarios dominated by electromigration or electroosmosis, for both conservative and reactive transport settings. We also investigate the influence of background electrolytes, charge interactions, reactant mobility and degradation reaction kinetics on system dynamics. To overcome the computational burden of process-based modeling and GSA implementations, we develop machine learning-based surrogate models. The latter are employed within both SM- and MM-GSA frameworks, using Sobol’ and AMAE sensitivity indices, respectively. This work provides a comprehensive quantification of how multiple sources of uncertainty impact electrokinetic transport behavior in porous media.
Original language | English |
---|---|
Article number | 104887 |
Journal | Advances in Water Resources |
Volume | 196 |
Number of pages | 16 |
ISSN | 0309-1708 |
DOIs | |
Publication status | Published - 2025 |
Keywords
- Electrokinetics
- Electromigration and electroosmosis
- Global sensitivity analysis
- Uncertainty quantification
- Machine learning
- Contamination