Oscillating microfiltration of mineral and ferric floc suspensions

By applying shear to the surface of a filter during micro and ultra-filtration it is well-known that the permeate flux across the membrane increases. Additional shear, by for instance rotating the membrane can increase the flux up to 17.1x (Jaffrin et al., 2004). Oscillation has been demonstrated to have superior shear creating conditions over rotational systems (Zamani et al., 2015) due to the inertia of the fluid. However, most of the previously reported work was with Newtonian systems, and highly concentrated minerals and ferric floc are highly non-Newtonian, and investigated in this thesis.

Calcium carbonate (calcite) with a d₃₂ of 2.7 μm was the main test material in the first stage of this work. The filters were oscillated both axially and azimuthally between 20Hz and 140Hz. Several different types of filter were used: Nickel slotted, PTFE coated nickel slotted (both surface filters) and ceramic (a matrix type, i.e. depth filter to achieve its pore rating). Providing a filter cake formed, the type of filter and mode of oscillation was not significant, the equilibrium, or sustainable, flux was solely dependent on the shear generated by the oscillation. A predictive model for the sustainable flux was investigated, and found to correlate the data well provided the characteristic diameter for the particles used in the model was the 5% particle diameter on the cumulative volume distribution curve, and not the d50 or the Sauter mean value. This is likely due to the migration of fines within the cake, causing flow pore restriction.

Shear stress values ranged up to 240 Pa and the particle paste believed to be at, or near, the deposited solids showed non-Newtonian flow behaviour described by the Herschel–Bulkley equation. The shear was computed using Comsol^® at, and near, the oscillating surface. The peak shear (maximum value) was used in the correlation for flux, which appeared to fit the data well and provide a realistic prediction for sustainable flux in a force balance model (flux as a function of shear stress).

The ferric floc was a much more complex material than the minerals. The filtration properties changed with time and no distinct interface was observed between the filter cake and suspension. The material is also significantly compressible, so increasing the pressure drop over the cake significantly increased the resistance to permeate flow. However, oscillation was again shown to provide good filtration performance, under carefully controlled conditions, and it was possible to construct a test rig which completed the filtration was a single-pass fashion, i.e. without needing to recirculate the feed suspension as in the case in conventional crossflow filtration. In a single pass it was possible to recover 98% of the water to the permeate and take the feed concentration from 1 g/L of Fe to ^~50 g/L of Fe in the retentate, a remarkable degree of concentration.