Étude du mécanisme des transferts d'eau et de soluté en zone non saturée avec prise en compte d'une fraction liquide immobile

Transfer of solutes in unsaturated soil in presence of stagnant water

Institut de Mécanique, Laboratoire associé au CNRS. Université Scientifique et Médicale de Grenoble

Abstract

Assuming that when a soil volume is unsaturated part of the water can be trapped by air between particles and in dead end pores, the classical scheme of hydrodynamic dispersion, represented by eq. (1), is modified in order to be more representative with the criterion that during the movement of solutes in soil lateral diffusion of chemicals in or out of the stagnant water is possible. The change of salt concentration during a steady flow is now represented by the set of equations (2) and (3) where subscripts m and im refers to the mobile and immobile fraction respectively. Three parameters must be known in order to solve this set of equations: the apparent dispersion coefficient D(cm²/hr) ; the diffusion mass transfer coefficient α: (hr^-1) and the mobile fraction ƒ = θm/θ. A series of experiments where done in a vertical soil column of soil for steady state flow of water and a pulse or step change of salt concentration at the inlet. The salt concentration where measured with time at several depths inside and at the exit of the column. Values for Dap,α ,ƒ were obtained by curve fitting observed and calculated concentration, at one depth (22 cm) using an explicit discretization of eq. (2). The same values, first determined during a step change of salt concentration, corresponding to the experiment on fig. 3, were first used to calculate salt distribution at other depths and at the exit for this experiment, and then to compute salt distribution during an other experiment, done with the same water flux, but for a pulse change of salt concentration (fig. 4). Excellent agreement was obtained in bath case ; the model presented a good description of the extensive tailing observed inside and at the outside of the column, which could not be simulated with the classical "dispersion- convective" equation (fig. 3.a). Finally a sensitivity study of the influence of each parameters was done (fig. 5 to 7). It appears that the immobile fraction of water has the strongest influence on the relative concentration curves. Further experiments should now be done in order to determine the variations of each parameter with the water content and to solve the transient infiltration problems.