Continuum model evaluation of the effect of saturation on coalescence filtration

George Chase, The University of Akron

Abstract

Coalescing filters are widely used throughout industry for removal of liquid aerosols from gases or the separation of liquid droplets from emulsions. Typical filters are constructed of non-woven fibers. Fibrous filters are capable of efficient removal of micron and submicron sized droplets and particles. The filtration process is highly complex due to variability in fiber sizes, particle sizes, mixtures of particles and droplets, mixture of types of droplets (oil, water, etc.), and effects of viscosity, surface tension, and chemical reactions between components or with the filter fibers. Prediction of filter performance under such complex conditions is difficult.Performance of a filter depends on many factors like particle and fiber sizes, flow rate, surface properties of the fibers etc. One of those parameters is the saturation of the filter medium. Saturation is a measure of the amount of liquid present in the void space. Prior models assume that the saturation is uniform along the depth of the medium. In real media, the liquid holdup at steady state need not be uniform with position. Local velocity increases when the saturation is high.In this paper, a steady state model for a coalescing filter is used to evaluate the effects of saturation on void fraction and its subsequent effect on filter performance. Single fiber mechanisms of direct interception and diffusion deposition are used to model droplet capture efficiencies and drag forces. These mechanisms are applied to volume averaged continuum equations in which the saturation is varied linearly with position in the filter. The results show the minimum pressure drop and largest quality factor occurs with a uniform saturation profile and that variation in average saturation has a greater effect on filter performance than does the slope of the linear saturation profile. The model predicts that uniform saturation profile performs better than the other profiles.