Mechanisms That Affect Capacity and Retention
Using an oil-in-water emulsion as our model stream, the study team investigated the mechanisms that affect capacity and retention.
Throughput of a membrane in a typical filtration application is affected by many factors, including membrane structure, the
viscosity of the suspension, particle size, particle concentration, and filter train resistance. The filter's overall throughput
is determined by flux and capacity. Flux is determined by driving forces (e.g., inlet pressure), stream properties (viscosity),
and the membrane structure (i.e., pore size, asymmetry). Capacity is also driven by the membrane structure and by properties
of the process stream, such as particle load.
Figure 1. Relative capacity of a symmetric and an asymmetric membrane for a model oil-in-water emulsion. The asymmetric membrane
has a much higher capacity in this stream, which can be attributed to higher intial flux.
The impact of membrane structure on capacity was demonstrated by comparing two membrane structures that varied in asymmetry.
As Figure 1 shows, the more asymmetric structure had a higher capacity, which we believe was associated with higher initial
flux. Regardless of membrane structure, the flow rate was very low, even early in the filtration process. The viscosity of
the model emulsion was one of the factors that contributed to the observed throughput. As Figure 2 shows, the viscosity of
the emulsion is high compared to that of water. As a result, the flux of the emulsion is lower than the flux of a typical
aqueous stream. The reduced flux significantly affects processing time and filter sizing.
Figure 2. Viscosity of the model stream as a function of temperature. As expected, the suspension is less viscous at higher
temperatures, but at all temperatures, its viscosity is higher than that of water, which is 0.01 poise at room temperature.
Testing was done to observe the plugging behavior of membranes when filtering the emulsion. In addition to the low initial
flux, the flow declined rapidly after the filtration started, suggesting that the membranes were plugging quickly. To investigate
the impact of the model stream on membrane plugging, we manipulated the size and concentration of particles in the model stream.
Smaller particles increased capacity, as shown in Figure 3. This is an indication that particle plugging is also an important
factor in filter capacity.
Figure 3. Relative capacity increases as average particle size decreases.
To confirm that pore blockage was the mechanism of membrane plugging, we used common fouling models to fit our data. Fouling
models use approximations of how a filter plugs to fit the fouling curves (volume versus time). Typical fouling models include
caking (the build-up of particles on the filter surface), complete pore blockage (blocking of a pore opening by a single particle),
and gradual pore plugging (build up of particles within a pore to gradually restrict flow).3 By plotting our data using the standard fouling models, we found that the classic Vmax model,4,5 which is based on gradual pore plugging, best fit the data, and accurately predicted the plugging performance and overall
capacity. This analysis provided further evidence that pore blockage was the primary mechanism of flow decay.
Although our testing showed that the classic Vmax model predicted the flow decay behavior, we also found that the plugging
behavior was pressure-dependent. Capacity increased with increasing pressure drop across the membrane, in some cases disproportionately
to the increase in pressure. Figure 4 shows the impact of pressure on filter capacity, which was non-linear. A number of factors
may contribute to this behavior, one of which is suspected to be the high viscosity of the emulsion compared to a typical
aqueous stream. Several follow-up experiments confirmed this theory.
Figure 4. Dependence of capacity on pressure. Capacity increases with increasing pressure. The same membrane was challenged,
in duplicate, at constant pressures of 5, 7.5, 10, 15, 20, and 30 psi. Capacity was determined as the volume at which the
flow had declined to 10% of its initial value in the emulsion.