Predicting Solute Transport Parameters in Saturated Porous Media Using Hybrid Algorithm

Abstract

This study aims to estimate the solute transport parameters in saturated porous media using a hybrid algorithm. In this study, the Physical Non-Equilibrium (PNE) model was used to describe the transport of solutes in porous media. A numerical solution for the PNE model is obtained using the Finite Volume Method (FVM) based on the Ttri-Diagonal Matrix Algorithm (TDMA). The developed program, written in Matlab, is capable to solve the advection-dispersion (ADE) and the PNE equations for the mobile -immobile (MIM)model with linear sorption isotherm. The Solute transport parameters, (immobile water content, mass transfer coefficient, and dispersion coefficient), are estimated using different algorithms such as the Levenberg-Marquardt algorithm (LM), genetic algorithm (GA), simulated annealing algorithm (SA). To overcome the limitations of deterministic optimization models which are rather unstable and divergent around a local minimum, a hybrid algorithm (GA+LM, SA+LM) permits to estimate of the solute transport parameters. Numerical solutions are verified using the experiments conducted by Semra (2003) which are about the transport of toluene through a column composed of impregnated Chromosorb grains at ambient temperature (20 °C) for three flow rates (1, 2 and 5ml/min). The results show that the hybrid algorithm (GA+LM, SA+LM) is more accurate than others (GA, SA, and LM). Comparing to the ADE model, The PNE with linear isotherm model gives a better description to the BeakThrough Curves (BTCs) with higher values of determination coefficient (R2) and lower values of Root Mean Square Error (RMSE). Also, the solute transport parameters tended to vary with the flow rate.

Keywords

• genetic algorithm
• finite volume method
• levenberg-marquardt algorithm
• numerical solution
• physical non-equilibrium