An assessment of particle disorder effects on smoothed particle hydrodynmaics simulations of transport in heterogeneous porous media

There is large evidence that accurate numerical simulations of reactive transport in natural porous media must capture local-scale heterogeneity. For example, in many cases reaction rates are controlled by mixing of chemical compounds, that is the result of the combined action of local-dispersion and Darcy's velocity variability at the local-scale (macro-dispersion). There are two main difficulties to incorporate local-scale heterogeneity in numerical models. First, in most real situations there is not enough data to define local-scale parameters. Second, the resolution of traditional numerical techniques based on grid or meshes is restricted by the cell size. A common solution to the first problem is to treat local-scale variables as space random functions. A solution for the second one is to use closure models to compute grid-scale parameters that capture subgrid-scale variability, e.g. macro-dispersion coefficients. Although grid-scale closure models work well for some problems, several studies have shown that its use in reactive transport simulations can lead to large errors in the estimation of reaction and transformation rates. In effect, the use of cell-averaged parameters and variables enhances the numerical homogenization of physically segregated contaminants. We present a new particle method to simulate reactive-transport in heterogeneous porous media that overcomes those problems. The method is based on the utilization of three well-known mesh-less particle methods: particle-tracking, smoothed particle hydrodynamics (SPH), and random-walk. Local-scale mass transfer between particles is modeled through the use of a local-dispersion coefficient and it is computed by using SPH. The temporal evolution of particle position is modeled as the result of the action of large-scale velocity that is computed by particle tracking, and small-scale stochastic variations simulated by using a random-walk approach. We show preliminary results of the application of this method to several scenarios. These results show that this is a promising methodology to accurate simulate reactive transport in real heterogeneous porous media.Presenters: name: Herrera, Paulo affiliation: University of British Columbi