Continuous column condensers

Column pressure climbs, the tops get hotter, and the operators spend reflux and reboiler steam to buy back the separation the fouled surface gave away. The arithmetic is unkind. A unit with a clean overall coefficient of 150 BTU/hr-ft2-F loses about a quarter of its capacity once 0.002 hr-ft2-F/BTU of scale builds on the water side, which is an ordinary summer on untreated tower water. The pressure controller first opens the coolant valve fully, then pressure creeps a few psi, and on a close-boiling separation that can move the top product off specification long before anyone considers cleaning. Trend the approach between condensing temperature and water outlet from the first week of a run; a steady rise is scale, a sudden one is usually gas.

Yes, flooded-condenser control is stable and cheap to instrument, as long as the extra surface for the flooded zone was in the rating and the drum sits where liquid can rise into the shell. Condensate under the level only subcools, at a coefficient perhaps a fifth of the condensing value, so a unit meant to run 20 percent flooded needs more then 20 percent extra surface to keep the same condensing capacity in reserve. It does not vent itself; inerts gather above the liquid and the controller reads them as more flooding, so a small bleed on a timer or a pressure differential alarm belongs in the design. Expect subcooled reflux, and tell the column designer, because the top tray will see more liquid than the flow meter reports.

Cooling water belongs in the tubes on a continuous overhead, and the pass count should be set so the velocity holds near 5 ft/s at design flow, not just above the minimum. A worked case: 300 gpm through 3/4 inch, 16 BWG tubes at 5 ft/s needs roughly 64 tubes per pass, so a 256-tube bundle would be built four-pass rather than two. That velocity scrubs the tube wall for months, keeps the fouling factor honest, and still leaves the shell side free for the vapor, which can tolerate almost no pressure drop. Put the overhead in the tubes only when its design pressure would otherwise force a heavy shell, or when the overhead itself fouls and must be rodded at turnaround.

Set the margin with a measured fouling factor and the hottest coolant day, and resist a flat percentage on top of that. Fouling factors from 0.001 to 0.002 hr-ft2-F/BTU on treated tower water already add 15 to 30 percent of surface at typical coefficients; stacking another 25 percent on top gives a condenser that, on a 60 F winter morning, has roughly twice the driving force of a summer afternoon with the same duty. The control valve then sits almost closed, tube velocity falls, and the water side scales faster than the fouling factor assumed, so the oversizing creates the fouling it was meant to cover. If a future capacity increase is the real reason for margin, state the future case and rate it as a second design point instead.

Yes, and a continuous column is the best candidate on the site for it, because the load changes little from one month to the next. An overhead condensing at 250 F can preheat column feed or raise steam at a few psig in a first exchanger, with a smaller trim condenser on cooling water finishing the job. Keep two things in the specification. The trim unit should hold column pressure on its own at a reduced rate during startup, when there is no feed to heat, and the recovery exchanger should see the dirty fluid on the side that can be cleaned, since it now sits inside the process loop rather than at the end of it. Payback on the extra shell is often inside a year when the alternative is fired reboil.