ABSTRACT
Although predicting the fate and transport of pollutants in the subsurface is basic in many environmental problems, mass transport modeling remains as a challenging subject of research. This is due to the limitations of the classical Advection-Dispersion Equation (ADE) approach applied to model mass transport. Anomalous transport is usually the expression applied when the ADE fails to reproduce real field as well as small-scale lab experiments. To solve this problem some authors model hydraulic conductivity (K) heterogeneity at very high resolutions with the support of stochastic simulation techniques. Besides, the non Fickian behavior of transport is another issue addressed. However, the effects of the spatial variability of dispersivity, and the influence of the model support scale on this property, have been rarely study. Moreover, dispersivity is always modeled as an averaged calibrated parameter highly correlated with the model discretization size. This is mainly due to the lack of experimental knowledge on this basic parameter.
In order to study the local behavior of the dispersivity a laboratory prototype, a porous medium tank, was designed and built at the Technical University of Valencia (Spain). This paper presents some results and conclusions obtained from a detailed analysis of a solute transport intermediate scale experiment conducted in this lab tank. The flow within the porous medium tank lab is quasi-2D with steady state conditions. The porous medium hydraulic conductivity (K) field imitates the patterns of spatial variability found in a real and highly heterogeneous formation (MADE2 site experiment). The solute transport tests are run using a visible and conservative dye tracer. The tank is monitored through a grid of pressure transducers and by the sequence of digital images that are processed to map the evolution of solute concentrations in the tank. The whole system is controlled from a computer that records pressure and images data. The set of exhaustive head and concentration data is then used to obtain detail local information of the effective dispersivity field at different time steps, and at different support scales.
We have found that the dispersivity field displays patterns of spatial variability that are related with the physical nature of the local material and also with the local evolution of concentrations at every grid block. In fact, the anomalous transport behavior observed in the lab tank can be accurately modeled using the classical ADE if the dispersivity field identified from the lab measurements is used. Besides, when this exhaustive information is processed to identify the dispersivity field at different support scales, it is found a strongly dependence of dispersivity with respect to the discretization size. When the blocks size increases there is a smoothing effect that prevents a good reproduction of concentration observations by means of the ADE equation. This highlights the crucial influence of a detailed modeling of the dispersivity field to deal with the limitations of the ADE.
