Wednesday, February 8, 2012

Process Wednesday: cake washing

That crack (lower left) could become a ravine (or a highway).
Photo credit: peach
Apropos of nothing, I've been reading chemical engineering texts on the filtration of solids from liquids. Did you know that "deliquoring" is a term of art in filtration? (And here I thought it was a description of what my liver was up to on a tough Thursday night in grad school.)

Here's a fairly detailed description of what happens when you wash a cake of product with some fresh solvent on large scale [1]:
In almost all cases, the requirement for maximum wash efficiency is to develop a mechanism, which removes the required amount of mother liquor with a minimum of washing liquor. Basically there are only two mechanisms, and in practice usually both are present:
• Displacement of the mother liquor by washing liquor
• Dilution of the mother liquor through mixing with washing liquors 
Displacement is by far the most efficient mechanism and under ideal conditions would cause the washing liquor to act as a piston driving the mother liquor out of the cake without any dilution. This is clearly utopia although a high degree of “piston or plug-flow effect” can be achieved with a careful preparation of the cake and a careful application of the wash liquor. The trick lies above all in the careful application of the washing liquor to the saturated face of the filter cake.
What happens if you got the chance to force your rinse through the cake with pressure?
It must be clear, but is often ignored, that washing at too high a pressure differential only leads to poor efficiency. Too high a pressure almost always leads to bypassing. Although the ideal cake has an even structure, no pinholes, and no cracks and lies firmly bedded against the retaining walls, we are not living in an ideal world and to some degree all cakes will have these faults. 
Forcing liquid through the cake at unnecessarily high pressures only makes matters worse. A pinhole will become a hole, a crack a ravine and if the cake does not bed down perfectly, the washing liquid will make a highway of the gap. [emphasis CJ's] In addition, high pressure differentials reduce the contact time. In all cases therefore the optimum pressure differential is the lowest possible.
You didn't know that washing the filter cake in a Buchner funnel had so much behind it, did you? Leave it to the chemical engineers to understand this stuff.

[1] Barry A. Perlmutter, "A Treatise of Filter Cake Washing Mechanisms In Pressure and Vacuum Filtration Systems." 


  1. It's a universal mechanism. You see it in oil extraction - that gusher that coated James Dean in "Giant" greatly cut down on the amount of oil that could be recovered from that well.

  2. Pressure is an enemy of filtration because of cake compression as well. Friable solids, fine fractions, etc. can pack under pressure and stop any deliquoring, washing or not. It is well worth checking in the lab whether a cake can compact. A filtration in a glass chromatography column, 1" or larger, is a fair model. Recording the filtration with a laptop camera while collecting the filtrate in a flask on a scale gives all the kinetic data one needs to a kilo scale-up.
    Jacketing the column is a luxory...

  3. Easy to underestimate what appears to be a very simple unit operation. You can have the highest yielding, room temperature, 6h reaction, but your process operators won't be impressed if you tie up their filter for 14 days because you didn't run a simple pre-scale-up filtration test in the lab...

    1. LabMonkey is dead on as something many new process chemist don't consider until learn "slow in lab doesn't translate to plant" especially, as the post indicates, things one might do to speed up (overhead pressure) are likely to do more harm than good.

      I would go further than the last line in stating ChemEs are often more in tune with practice as typically will advocate use of centrifuges if suitable/available. That does own issues and such but can be a nice upgrade form normal filtration modes.

    2. If you were to recommend a laboratory filtration setup to best simulate the plant, what approach would you take? Do you like Anon0755a's suggestion above?

    3. It sound good. You can build a scale-up model with a few parameters (filter area, mass, vacuum, temp, etc) using something as simple as a Buchner flask/funnel, but the column is a nice touch. Google BHS Pocket Filtration for one commercial example (no affiliation - just first I found to illustrate!)

    4. My personal filtration benchmark: If a filtration on a lab scale (<2L, Buchner funnel) takes more than 45 sec there is trouble ahead in the plant.