I
Retention of particulate in/on a filter is an extremely complex entity
due to simultaneous and varied 'capture mechanisms'. that 'hold' the
particles in place: seiving - where the statistical pores of the filter
structure are smaller than the particle; inertial impaction - where the
particles leave the flow stream when the flow stream takes sharp bends;
aDsorption - where the particles are held to the filter media subtrate
by weak electronic bonding (van der waals forces); the formation of a
"cake" on the upper surface and into the upper 10-15% of the depth of
filter medium; Polarization of gel-forms, etc.

All filtration is 'particle specific' .... and depends exactly which
mechanism of capture 'predominates' for which type of particle you want
to remove. For fuel oil, etc. filtration where probably there are more
'deformable' particles or particles that can change shape under
increasing differential pressure and then are subject to extrusion
either through or partly through the media (settling deeper into the
media) thus 'blinding it' ..... vacuum filtration has historically
shown the least efficacy of on-stream service life versus pressure
filtration. Apparently vacuum feed filtration allows the deformables
and smaller than 'seiving' size particles to partly extrude more deeply
into the matrix, partly closees off the statistical pores which
increases the face velocity of the fluid in the cross sections of the
filter matrix. The increasing face velocity of the fluid through the
sections creates an untowards physical event (as per the standard
D'Arcys equation) derived to be: On stream life (T1/T2) being a
function of the ratio of face velocities to the 'n-th' power where 'n'
- approximately approaches to the 2/3 power). My 'guess' (after 35
years of observation, etc.) is that in vacuum filtration the capture
involves an *accelerating* particle; while pressure feed involves a
*decelerating* particle.

Simplistically and historically, vacuum draws the deformables and
smaller than the target 'seiving sized' particles deeper into the
matrix, shuts down the open flow paths quicker than in pressure feed
--- all apparently internal velocity dependent.
Filtration hydrodynamics is probably very similar to aerodynamics where
intuition and simple logic will always produce the wrong answer. I've
been deeply involved in ultra-purity filtration and 'separations'
engineering (and the physical chemistry of) for almost 35 years and
still dont know all the answers .... although I do know that vacuum
feed filtration will *always* have comparatively shorter service life
than pressure feed (for just about ALL filtrations) ... and for that
reason alone is good enough for me and most others to stay away from
vacuum feed filtration. There's probably a doctoral discertation
waiting for someone who can correctly figure this one out - many have
tried but none have ever been successful.

Like I posted earlier, filtration has nothing to do with 'screen doors'
and is an extremely variable complex entity at below the macroscopic
level. Dont attempt to 'rationalize' it as you will ultimately always
wind up with the wrong solution. It's really an 'art-form'.