MONTE-CARLO NETWORK SIMULATION OF HORIZONTAL, UPFLOW AND DOWNFLOW DEPTH FILTRATION

A test particle trajectory approach is developed for the simulation of deep bed filtration. A 3-D network of constricted pores represents the pore space of granular filters, network-scale trajectories of a large number of non-Brownian test particles are computed, and filter coefficient predictions are obtained for horizontal, down-and upflow filtration operation. This simulator yields numerical results that agree excellently with our earlier predictions by the pore-scale trajectory-based population balance method. The new approach, however, circumvents the cumbersome step of calculating the impacted fraction in each unit cell, which the earlier method required, by providing direct statistical estimates of the local and overall deposition rates for continuous and discrete pore-size distributions. For large superficial velocities (upsilon(s)> similar to 1 mm/s) and distributed pore size, downflow filters are more efficient than horizontal flow filters, whereas for small velocities (upsilon(s)< similar to 0.5 mm/s) the opposite is observed. Horizontal flow operation is also favored by uniform packing for almost any value of the external pressure gradient. Upflow operation is the least efficient for the packings considered here over a broad range of superficial velocity and particle-size values. Observed differences among the three filtration types are maximal for uniform packings and decrease considerably with increasing packing heterogeneity.