The origin of preferential flow and non-equilibrium transport in unsaturated heterogeneous porous systems

Stabilization of waste bodies of landfill is achieved by treating the waste body using irrigation, recycling of leachate combined with landfill gas extraction and/or aeration. In order for the regulators and landfill operators to agree on a required level of after care, a quantitative estimation of remaining long-term emission potential is required. Our hypothesis is that, material heterogeneity in unsaturated systems is the origin of preferential flow and that infiltration patterns and rates are the controlling factors affecting non-equilibrium solute transport. We developed a coupled flow and transport model using a finite difference method implemented as a MATLAB toolbox, which we named Variably Saturated Flow and Transport (VarSatFT).

Using VarSatFT, we analysed water flow and solute transport in different unsaturated heterogeneous small scale systems. The origin of preferential flow and non-equilibrium solute transport lies in the funnelling of flow and advective transport through the high permeable zones. This leads to concentration gradients between the solutes in the mobile and immobile pore space due to the flushing of the mobile pore space with fresh (rain) water. The variation in infiltration rates and patterns are found to be the controlling factors for the magnitude non-equilibrium solute transport. The findings from the numerical analyses were verified in lab scale experiments. Infiltration and leachate recirculation can stimulate biodegradation if sufficient water is added to significantly increase water content Our findings indicate the severe limitations associated with single continuum modelling methods of water flow and solute transport for full scale landfills, especially when leachate concentrations need to predicted.

For the laboratory experiments we required data on the water retention parameters from well sorted sands near saturation. We developed an approach developed using the vertical distribution of water content along a TDR probe. We performed these measurements in a multi-step drainage experiment at moments when flow had ceased so that hydrostatic conditions can be assumed. This gives a direct measurement of the water retention curve. Combining the water retention curve with the model for TDR waveforms and the pressure head distribution from the hydrostatic conditions allowed for the parameters in the unsaturated water retention curve to be optimized using the Bayesian inference scheme. The approach we developed reduces the number of parameters compared with other TDR approaches which optimize water content at every node along TDR probe. This approach is suitable to quantify water retention parameters for samples with long heights and samples with uniform particle size distributions.