A column can drown, and it can leak

What a tray actually is

A sieve tray is a flat deck perforated with thousands of small holes, with a low dam across the far edge and a chute beside it down to the deck below. Liquid arrives at one side, runs across, pools against the dam, spills over and falls down the chute. Vapour arrives from underneath and has nowhere to go but up through the holes, straight into the liquid running over them.

What that makes is not a pool with bubbles in it. It is a froth, and it can stand two or three times the height of the clear liquid underneath it. All of the separation happens in that froth, because that is the only place the two phases touch. Everything that follows is about keeping it there.

Jet flood: it drowns from the top

Push more vapour through the holes and the froth climbs. Past a point the spray reaches the deck above and the liquid that should have gone down the chute is carried up instead. Now the tray above is being fed liquid it already processed, separation collapses in both directions at once, and the pressure drop runs away. That is jet flood, and it is the ceiling on the chart.

Where the ceiling sits comes from Fair’s flooding correlation: a capacity factor C set by the tray spacing and by how much liquid is on the deck, then uflood = C √((ρL − ρV) / ρV). For this tray that is 87.5 kg/s of vapour. Notice the ceiling falls as you move right: the more liquid on the deck, the less vapour it takes to lift it.

Downcomer flood: it drowns from the bottom

The chute is not a free ride. Liquid coming down it has to push against the pressure drop of the tray it is feeding, and it stands up in the downcomer until it has the head to do that. Add the weir height, the crest over the weir, the tray pressure drop and the loss squeezing under the apron and you get the backup: 212 mm of liquid here.

The limit is the tray above. Once the froth in the downcomer reaches it, liquid has nowhere to drain and the column fills from the bottom up. The usual limit is half the tray spacing plus the weir, which is 263 mm for this tray, and that is the wall on the right of the chart. It arrives at 104 kg/s of liquid — 1.53× the design rate.

Weeping: it leaks

Now take vapour away. The gas coming up through the holes is the only thing holding the liquid on the deck, and below a certain hole velocity the liquid stops waiting for the weir and drains straight down through the perforations. That is weeping. Nothing bangs and nothing floods; the column just quietly stops separating, one tray at a time, because liquid that fell through a hole never met the vapour on that tray at all.

Every sieve tray weeps a little in normal operation, which is why the weep point is defined by leakage becoming excessive rather than by leakage starting: in practice, enough to cost 10 to 20% of the tray efficiency. That is why the floor on this chart is drawn at the turndown a sieve tray is credited with, about 2:1, rather than computed. Exactly where it falls comes off Fair’s weep-point chart and moves with hole size and weir height.

The box, and the one number every tray is designed to

Put the three walls on one pair of axes — liquid load across, vapour load up — and what is left is the operating box. The design job is to put the operating point inside it, and then to keep it there through startup, turndown, a fouled tray and a summer day.

The number that does that is 80% of jet flood. It is normal practice to limit tray design to a maximum of 80%, and the missing 20% is not timidity: it pays for the tower control, for errors in the vapour-liquid equilibrium data the whole column was sized on, and for the flooding correlation itself, which is a fit to other people’s columns. This tray sits at 80%.

Why that half metre turns into a skyscraper

The froth needs room. Give it too little and the spray from one deck reaches the next one at a vapour rate you wanted to run at, so the tray spacing you choose is really a choice about capacity. About half a metre is the standard answer, and the moment you accept it the height of the column stops being a design decision and becomes arithmetic.

Propane and propylene boil a few degrees apart, so a splitter needs 187 trays. 187 × 0.5 m is 93.5 m of shelves before the reboiler, the sump, the disengaging space at the top or the skirt it stands on. The crude column that shipped to the Dangote refinery is 112.5 m long and 14.0 m across, and it travelled by sea lying on its side, because standing up it does not fit under anything. Ordinary columns run 6 to 60 m. Every one of them, tall or short, is sized on the chart at the top of this page.

Sources: jet flood, downcomer backup flood, weeping and dumping, the 80% of jet flood design practice and the note that most sieve and valve trays weep in normal operation are from the KLM Technology Group distillation tray hydraulics design guideline. Turndown by tray type: sieve trays, AIChE CEP, May 2016 and bubble-cap trays. Column sizes: fractionating columns run 6 to 60 m; the 187-tray propane / propylene splitter is PRefChem; the 112.5 m Dangote crude column and its horizontal voyage are BusinessDay. The flooding line uses Fair’s correlation in the Lygeros and Magoulas closed form; the tray hydraulics follow the standard sieve-tray relations.