Institute of Mechanical Process Engineering and Mechanics

Mist filters: How steady is their "steady state"?

  • Author:

    H.E, Kolb, G. Kasper

  • Source:

    Chemical Engineering Science, 2019, 225, 54-59

  • This paper addresses the phenomenon of a slow ∆p increase (“creep”) sometimes observed in mist filters during their so-called steady-state operation, especially at relatively high flow rates and aerosol loads (Kolb et al., 2017). This creep phenomenon draws into question the concept of steady state and awaits a physical explanation on the basis of more systematic data. The present study was performed with conventional glass microfiber filter media at filter face velocities in the range of 1–70 cm/s, using submicron oil aerosol at loading rates up to 1.1 g/(m2 s). Experiments started either with a dry filter (designated as S0= 0, where S stands for the liquid saturation level), or with a completely pre-saturated filter (designated as S0 = 1). The S0 = 0 runs reached “steady state” (defined here as the onset of drainage) after approximately 0.3 h and were then loaded continually for another 100 h. Runs starting at S0 = 1reached a steady state very quickly.

    With regard to ∆p, all filters featured ∆p creep when starting from S0 = 0. For coarse filter grades, ∆p creep effectively ended in less than 40 h at moderate flow velocities; finer grades crept further and did not stop creeping within any reasonable loading time. When starting from S0 = 1, a steady ∆p endpoint was reached rapidly, but was considerably higher than the level attained by creep. It was further shown that creep is associated with a slow increase in the saturation level and that S approached the (higher) level of pre-saturated filters. The increase in S during creep was associated with the formation of additional fine oil channels. Conversely, oil channels were completely absent in pre-saturated filters and the liquid was uniformly dispersed. Using simple theoretical considerations for flow in porous media, the observed differences in ∆p are explained in part by the difference in S between S0= 0 and S0 = 1 filters, and in additionally also by taking into account the different liquid distribution patterns.

    We conclude that the onset of drainage under steady external conditions – the stage conventionally viewed as the beginning of steady-state operation – does not automatically represent an internal steady state with regard to saturation levels or pressure drop; and presumably also not for the oil distribution pattern. The internal saturation reaches a stable endpoint sooner than the pressure drop, which may continue to creep for the lifetime of the filter. Of and when a transition from channel patterns to homogeneous oil distribution occurs during steady operation is not known. However, the transition can be forced rather quickly by operating the filter intermittently at a very low flow velocity.